{"id":89332,"date":"2026-08-02T20:56:41","date_gmt":"2026-08-02T12:56:41","guid":{"rendered":"https:\/\/www.wsisp.com\/helps\/89332.html"},"modified":"2026-08-02T20:56:41","modified_gmt":"2026-08-02T12:56:41","slug":"%e7%ac%ac15%e8%af%be%ef%bc%9apyspark%ef%bd%9c%e8%87%aa%e5%ae%9a%e4%b9%89%e5%87%bd%e6%95%b0udf%e3%80%81udaf%e3%80%81udtf%e5%bc%80%e5%8f%91%e5%8e%9f%e7%90%86%e4%b8%8e%e4%bc%81%e4%b8%9a%e5%ae%9e%e6%88%98","status":"publish","type":"post","link":"https:\/\/www.wsisp.com\/helps\/89332.html","title":{"rendered":"\u7b2c15\u8bfe\uff1aPySpark\uff5c\u81ea\u5b9a\u4e49\u51fd\u6570UDF\u3001UDAF\u3001UDTF\u5f00\u53d1\u539f\u7406\u4e0e\u4f01\u4e1a\u5b9e\u6218"},"content":{"rendered":"<p><img decoding=\"async\" src=\"https:\/\/www.wsisp.com\/helps\/wp-content\/uploads\/2026\/08\/20260802125638-6a6f3e8681788.jpg\" alt=\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\" \/><\/p>\n<\/p>\n<h4>\u6587\u7ae0\u76ee\u5f55<\/h4>\n<ul>\n<li>\n<ul>\n<li>\u4e00\u3001\u8bfe\u524d\u5bfc\u8bfb<\/li>\n<li>\u4e8c\u3001\u5b66\u4e60\u76ee\u6807<\/li>\n<li>\u4e09\u3001\u6838\u5fc3\u7406\u8bba\u77e5\u8bc6\u70b9<\/li>\n<li>\u56db\u3001\u539f\u7406\u901a\u4fd7\u8bb2\u89e3<\/li>\n<li>\n<ul>\n<li>4.1 \u4f20\u7edfUDF\u7684\u201c\u9010\u884c\u4e4b\u65c5\u201d<\/li>\n<li>4.2 Pandas UDF\u7684\u201c\u6279\u91cf\u52a0\u901f\u201d<\/li>\n<li>4.3 UDAF\u7684\u81ea\u5b9a\u4e49\u805a\u5408<\/li>\n<li>4.4 UDTF\u7684\u8868\u751f\u6210<\/li>\n<\/ul>\n<\/li>\n<li>\u4e94\u3001\u91cd\u70b9\u6982\u5ff5\u62c6\u89e3<\/li>\n<li>\n<ul>\n<li>5.1 \u4f20\u7edfUDF\u7684\u6ce8\u518c\u4e0e\u4f7f\u7528<\/li>\n<li>5.2 Pandas UDF\u7684\u7c7b\u578b<\/li>\n<li>5.3 \u5206\u7ec4\u805a\u5408Pandas UDF<\/li>\n<li>5.4 \u5206\u7ec4\u53d8\u6362&#xff08;GROUPED_MAP&#xff09;<\/li>\n<li>5.5 \u8fed\u4ee3\u5668Pandas UDF&#xff08;\u5904\u7406\u5927\u6570\u636e\u5206\u533a&#xff09;<\/li>\n<li>5.6 \u6ce8\u518cUDF\u4f9bSQL\u4f7f\u7528<\/li>\n<li>5.7 UDAF\u7684\u4f20\u7edf\u5b9e\u73b0&#xff08;Scala\u4e92\u64cd\u4f5c&#xff09;<\/li>\n<li>5.8 UDTF\u7684\u6a21\u62df<\/li>\n<\/ul>\n<\/li>\n<li>\u516d\u3001\u6613\u9519\u70b9\u907f\u5751<\/li>\n<li>\n<ul>\n<li>6.1 UDF\u4e2d\u6355\u83b7\u4e0d\u53ef\u5e8f\u5217\u5316\u5bf9\u8c61<\/li>\n<li>6.2 \u8bef\u4ee5\u4e3aUDF\u53ef\u4ee5\u4fee\u6539\u5168\u5c40\u53d8\u91cf<\/li>\n<li>6.3 \u5ffd\u7565\u7a7a\u503c\u5bfc\u81f4UDF\u629b\u5f02\u5e38<\/li>\n<li>6.4 Pandas UDF\u8fd4\u56de\u7c7b\u578b\u4e0d\u5339\u914d<\/li>\n<li>6.5 \u5206\u7ec4\u805a\u5408Pandas UDF\u4e2d\u8bef\u7528&#096;agg&#096;<\/li>\n<li>6.6 \u4f7f\u7528Pandas UDF\u65f6\u4e0d\u8bbe\u7f6e&#096;spark.sql.execution.arrow.pyspark.enabled&#096;<\/li>\n<\/ul>\n<\/li>\n<li>\u4e03\u3001\u5b8c\u6574\u5b9e\u6218\u6848\u4f8b<\/li>\n<li>\u516b\u3001\u4ee3\u7801\u9010\u884c\u89e3\u6790<\/li>\n<li>\n<ul>\n<li>8.1 \u4f20\u7edfUDF\u5b9a\u4e49<\/li>\n<li>8.2 Pandas UDF\u7c7b\u578b\u63d0\u793a\u98ce\u683c<\/li>\n<li>8.3 \u5206\u7ec4\u805a\u5408&#xff08;GROUPED_MAP&#xff09;<\/li>\n<li>8.4 UDTF\u6a21\u62df&#xff1a;&#096;mapInPandas&#096;<\/li>\n<li>8.5 \u8fed\u4ee3\u5668UDF&#xff08;SCALAR_ITER&#xff09;<\/li>\n<\/ul>\n<\/li>\n<li>\u4e5d\u3001\u4e1a\u52a1\u573a\u666f\u843d\u5730\u5e94\u7528<\/li>\n<li>\n<ul>\n<li>9.1 \u573a\u666f\u4e00&#xff1a;\u5b9e\u65f6\u65e5\u5fd7\u4e2d\u7684IP\u5730\u7406\u4f4d\u7f6e\u89e3\u6790<\/li>\n<li>9.2 \u573a\u666f\u4e8c&#xff1a;\u7528\u6237\u884c\u4e3a\u5e8f\u5217\u7684\u7279\u5f81\u5de5\u7a0b<\/li>\n<li>9.3 \u573a\u666f\u4e09&#xff1a;\u8c03\u7528\u5916\u90e8\u9884\u6d4b\u670d\u52a1<\/li>\n<li>9.4 \u573a\u666f\u56db&#xff1a;\u81ea\u5b9a\u4e49\u805a\u5408\u7a97\u53e3\u51fd\u6570<\/li>\n<\/ul>\n<\/li>\n<li>\u5341\u3001\u5e38\u89c1\u62a5\u9519\u6392\u67e5<\/li>\n<li>\n<ul>\n<li>10.1 &#096;AttributeError: &#039;NoneType&#039; object has no attribute&#8230;&#096;<\/li>\n<li>10.2 &#096;ArrowTypeError: Expected a string, got &#8230;&#096;<\/li>\n<li>10.3 &#096;Py4JError: An error occurred while calling o123.__getstate__&#096;<\/li>\n<li>10.4 Pandas UDF\u6027\u80fd\u5dee&#xff08;\u4e0e\u5185\u7f6e\u51fd\u6570\u5dee\u8ddd\u5927&#xff09;<\/li>\n<li>10.5 \u5206\u7ec4\u805a\u5408UDF\u5185\u5b58\u6ea2\u51fa<\/li>\n<\/ul>\n<\/li>\n<li>\u5341\u4e00\u3001\u672c\u8282\u8bfe\u77e5\u8bc6\u70b9\u603b\u7ed3<\/li>\n<li>\n<ul>\n<li>UDF\u7c7b\u578b\u5bf9\u6bd4<\/li>\n<li>\u9009\u578b\u5efa\u8bae<\/li>\n<li>\u6027\u80fd\u4f18\u5316\u53c2\u6570<\/li>\n<\/ul>\n<\/li>\n<li>\u5341\u4e8c\u3001\u8bfe\u540e\u601d\u8003\u4f5c\u4e1a<\/li>\n<li>\n<ul>\n<li>\u4f5c\u4e1a\u4e00&#xff1a;\u7406\u8bba\u7406\u89e3\u9898<\/li>\n<li>\u4f5c\u4e1a\u4e8c&#xff1a;\u4ee3\u7801\u5b9e\u8df5\u9898<\/li>\n<li>\u4f5c\u4e1a\u4e09&#xff1a;\u573a\u666f\u5e94\u7528\u9898<\/li>\n<li>\u4f5c\u4e1a\u56db&#xff1a;\u62d3\u5c55\u7814\u7a76<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<\/li>\n<li>&#x1f517;\u300a20\u8282\u8bfe PySpark \u4ece\u5165\u95e8\u5230\u7cbe\u901a\u300b\u7cfb\u5217\u8bfe\u7a0b\u5bfc\u822a<\/li>\n<\/ul>\n<hr \/>\n<h3>\u4e00\u3001\u8bfe\u524d\u5bfc\u8bfb<\/h3>\n<p>\u5728\u524d\u9762\u7684\u8bfe\u7a0b\u4e2d&#xff0c;\u6211\u4eec\u53cd\u590d\u5f3a\u8c03&#xff1a;\u4f18\u5148\u4f7f\u7528\u5185\u7f6e\u51fd\u6570&#xff0c;\u907f\u514d\u81ea\u5b9a\u4e49\u51fd\u6570\u3002\u56e0\u4e3a\u5185\u7f6e\u51fd\u6570\u7ecf\u8fc7Catalyst\u4f18\u5316\u548c\u4ee3\u7801\u751f\u6210&#xff0c;\u6027\u80fd\u8fdc\u8d85\u666e\u901a\u7684Python UDF\u3002\u7136\u800c&#xff0c;\u5728\u5b9e\u9645\u7684\u5927\u6570\u636e\u5f00\u53d1\u4e2d&#xff0c;\u603b\u4f1a\u9047\u5230\u5185\u7f6e\u51fd\u6570\u65e0\u6cd5\u6ee1\u8db3\u7684\u4e1a\u52a1\u903b\u8f91\u2014\u2014\u590d\u6742\u7684\u52a0\u5bc6\u89e3\u5bc6\u3001\u8c03\u7528\u5916\u90e8API\u3001\u975e\u6807\u51c6\u7684\u4e1a\u52a1\u8ba1\u7b97\u516c\u5f0f\u3001\u81ea\u5b9a\u4e49\u805a\u5408\u903b\u8f91\u7b49\u3002\u8fd9\u4e9b\u573a\u666f\u4e0b&#xff0c;\u6211\u4eec\u5fc5\u987b\u4f7f\u7528\u81ea\u5b9a\u4e49\u51fd\u6570\u3002<\/p>\n<p>PySpark\u63d0\u4f9b\u4e86\u4e09\u79cd\u81ea\u5b9a\u4e49\u51fd\u6570\u63a5\u53e3&#xff1a;<\/p>\n<ul>\n<li>UDF&#xff08;User-Defined Function&#xff09;&#xff1a;\u8f93\u5165\u4e00\u884c&#xff0c;\u8f93\u51fa\u4e00\u5217&#xff0c;\u6700\u5e38\u7528<\/li>\n<li>UDAF&#xff08;User-Defined Aggregate Function&#xff09;&#xff1a;\u81ea\u5b9a\u4e49\u805a\u5408\u51fd\u6570&#xff0c;\u7528\u4e8egroupBy\u4e2d\u590d\u6742\u7684\u72b6\u6001\u805a\u5408<\/li>\n<li>UDTF&#xff08;User-Defined Table Function&#xff09;&#xff1a;\u8f93\u5165\u4e00\u884c&#xff0c;\u8f93\u51fa\u591a\u884c&#xff08;\u8868&#xff09;&#xff0c;PySpark\u4e2d\u901a\u8fc7explode\u6216pandas UDF\u6a21\u62df<\/li>\n<\/ul>\n<p>\u5f88\u591a\u5f00\u53d1\u8005\u5728\u7f16\u5199UDF\u65f6&#xff0c;\u7531\u4e8e\u4e0d\u4e86\u89e3\u5176\u5185\u90e8\u539f\u7406&#xff0c;\u5199\u51fa\u7684\u4ee3\u7801\u6548\u7387\u4f4e\u4e0b\u2014\u2014\u9010\u884c\u5904\u7406\u3001\u9891\u7e41\u5e8f\u5217\u5316\u3001\u5ffd\u7565\u7c7b\u578b\u63d0\u793a\u3001\u672a\u5904\u7406\u7a7a\u503c\u7b49\u3002\u66f4\u4e25\u91cd\u7684\u662f&#xff0c;\u67d0\u4e9bUDF\u5199\u6cd5\u4f1a\u5bfc\u81f4\u6570\u636e\u503e\u659c\u6216\u5185\u5b58\u6ea2\u51fa\u3002\u540c\u65f6&#xff0c;PySpark 2.3&#043;\u5f15\u5165\u4e86Pandas UDF&#xff08;\u5411\u91cf\u5316UDF&#xff09;&#xff0c;\u5229\u7528Pandas\u7684\u6279\u91cf\u5904\u7406\u5927\u5e45\u63d0\u5347\u4e86Python UDF\u7684\u6027\u80fd&#xff0c;\u4f46\u5f88\u591a\u5f00\u53d1\u8005\u4ecd\u7136\u5728\u7528\u4f20\u7edf\u7684\u9010\u884cUDF\u3002<\/p>\n<p>\u672c\u8282\u8bfe\u5c06\u5168\u9762\u8bb2\u89e3PySpark\u81ea\u5b9a\u4e49\u51fd\u6570\u7684\u5f00\u53d1\u539f\u7406\u548c\u6700\u4f73\u5b9e\u8df5\u3002\u5185\u5bb9\u5305\u62ec&#xff1a;<\/p>\n<ul>\n<li>\u4f20\u7edfUDF\u7684\u6ce8\u518c\u3001\u4f7f\u7528\u3001\u6027\u80fd\u9677\u9631\u53ca\u4f18\u5316<\/li>\n<li>Pandas UDF&#xff08;Series to Series\u3001Series to Scalar\u3001Iterator\u7b49&#xff09;\u7684\u7528\u6cd5<\/li>\n<li>\u5206\u7ec4\u805a\u5408UDAF\u7684\u5b9e\u73b0&#xff08;\u901a\u8fc7Pandas UDF\u5b9e\u73b0\u9ad8\u6548\u805a\u5408&#xff09;<\/li>\n<li>UDTF\u7684\u6a21\u62df\u5b9e\u73b0<\/li>\n<li>\u4f01\u4e1a\u7ea7\u5b9e\u6218\u6848\u4f8b&#xff1a;\u5982\u590d\u6742JSON\u89e3\u6790\u3001AES\u52a0\u89e3\u5bc6\u3001IP\u5730\u5740\u89e3\u6790\u3001\u81ea\u5b9a\u4e49\u8bc4\u5206\u6a21\u578b\u7b49<\/li>\n<li>\u6027\u80fd\u5bf9\u6bd4\u548c\u9009\u578b\u5efa\u8bae<\/li>\n<\/ul>\n<p>\u5b66\u5b8c\u8fd9\u8282\u8bfe&#xff0c;\u4f60\u5c06\u80fd\u591f\u5728\u5fc5\u987b\u81ea\u5b9a\u4e49\u903b\u8f91\u7684\u573a\u666f\u4e0b&#xff0c;\u9009\u62e9\u6700\u9ad8\u6548\u7684\u65b9\u5f0f\u7f16\u5199UDF&#xff0c;\u907f\u514d\u5e38\u89c1\u7684\u6027\u80fd\u9677\u9631\u3002<\/p>\n<h3>\u4e8c\u3001\u5b66\u4e60\u76ee\u6807<\/h3>\n<p>\u5b8c\u6210\u672c\u8282\u8bfe\u7684\u5b66\u4e60\u540e&#xff0c;\u4f60\u5c06\u80fd\u591f&#xff1a;<\/p>\n<li>\u7406\u89e3UDF\u6267\u884c\u539f\u7406&#xff1a;\u6e05\u695aPython UDF\u5982\u4f55\u5bfc\u81f4\u5e8f\u5217\u5316\u5f00\u9500&#xff0c;\u4e3a\u4ec0\u4e48\u6bd4\u5185\u7f6e\u51fd\u6570\u6162<\/li>\n<li>\u719f\u7ec3\u4f7f\u7528\u4f20\u7edfUDF&#xff1a;\u6ce8\u518c\u3001\u8c03\u7528\u3001\u8bbe\u7f6e\u8fd4\u56de\u7c7b\u578b\u3001\u5904\u7406\u7a7a\u503c<\/li>\n<li>\u638c\u63e1Pandas UDF&#xff1a;\u7406\u89e3\u5411\u91cf\u5316UDF\u7684\u7c7b\u578b&#xff1a;Series to Series\u3001Series to Scalar\u3001Iterator\u7b49<\/li>\n<li>\u5b9e\u73b0\u81ea\u5b9a\u4e49\u805a\u5408UDAF&#xff1a;\u4f7f\u7528pandas_udf\u914d\u5408groupBy().agg()\u5b9e\u73b0\u590d\u6742\u72b6\u6001\u805a\u5408<\/li>\n<li>\u6a21\u62dfUDTF&#xff1a;\u901a\u8fc7flatMap\u6216explode\u5b9e\u73b0\u4e00\u884c\u8f93\u51fa\u591a\u884c<\/li>\n<li>\u4f18\u5316UDF\u6027\u80fd&#xff1a;\u4f7f\u7528Pandas UDF\u66ff\u4ee3\u9010\u884cUDF&#xff0c;\u5229\u7528\u5e7f\u64ad\u53d8\u91cf&#xff0c;\u5904\u7406\u6570\u636e\u503e\u659c<\/li>\n<li>\u5728\u4f01\u4e1a\u573a\u666f\u4e2d\u5408\u7406\u9009\u578b&#xff1a;\u77e5\u9053\u4f55\u65f6\u7528UDF\u3001\u4f55\u65f6\u7528Pandas UDF\u3001\u4f55\u65f6\u5fc5\u987b\u7528\u5185\u7f6e\u51fd\u6570<\/li>\n<h3>\u4e09\u3001\u6838\u5fc3\u7406\u8bba\u77e5\u8bc6\u70b9<\/h3>\n<table>\n<tr>\u77e5\u8bc6\u70b9\u8bf4\u660e<\/tr>\n<tbody>\n<tr>\n<td>\u4f20\u7edfUDF<\/td>\n<td>\u9010\u884c\u5904\u7406&#xff0c;Python\u5e8f\u5217\u5316\u5f00\u9500\u5927&#xff0c;\u9002\u5408\u7b80\u5355\u903b\u8f91\u6216\u65e0\u6cd5\u5411\u91cf\u5316\u7684\u573a\u666f<\/td>\n<\/tr>\n<tr>\n<td>Pandas UDF<\/td>\n<td>\u57fa\u4e8eArrow\u7684\u6279\u91cf\u6570\u636e\u4f20\u8f93&#xff0c;\u5728Pandas Series\u4e0a\u64cd\u4f5c&#xff0c;\u6027\u80fd\u63a5\u8fd1\u5185\u7f6e\u51fd\u6570<\/td>\n<\/tr>\n<tr>\n<td>\u5206\u7ec4\u805a\u5408Pandas UDF<\/td>\n<td>applyInPandas\u6216grouped_agg&#xff0c;\u7528\u4e8e\u81ea\u5b9a\u4e49\u805a\u5408<\/td>\n<\/tr>\n<tr>\n<td>UDF\u7c7b\u578b<\/td>\n<td>udf(returnType)&#xff1b;Pandas UDF&#xff1a;&#064;pandas_udf(returnType, functionType)<\/td>\n<\/tr>\n<tr>\n<td>\u5e8f\u5217\u5316<\/td>\n<td>PySpark UDF\u901a\u8fc7Py4J\u8fdb\u884cJVM\u4e0ePython\u8fdb\u7a0b\u901a\u4fe1&#xff0c;\u4f7f\u7528Pickle\u5e8f\u5217\u5316<\/td>\n<\/tr>\n<tr>\n<td>Arrow<\/td>\n<td>\u5217\u5f0f\u5185\u5b58\u683c\u5f0f&#xff0c;Pandas UDF\u4f7f\u7528Arrow\u6279\u91cf\u4f20\u8f93&#xff0c;\u51cf\u5c11\u5e8f\u5217\u5316\u5f00\u9500<\/td>\n<\/tr>\n<tr>\n<td>\u5411\u91cf\u5316<\/td>\n<td>\u6279\u91cf\u5904\u7406\u6570\u636e&#xff0c;\u5229\u7528Pandas\u7684\u5217\u5f0f\u64cd\u4f5c\u548cNumPy\u52a0\u901f<\/td>\n<\/tr>\n<tr>\n<td>\u8fed\u4ee3\u5668UDF<\/td>\n<td>\u7528\u4e8e\u5904\u7406\u5927\u6570\u636e\u91cf\u5206\u6279\u6b21&#xff0c;\u907f\u514d\u5185\u5b58\u6ea2\u51fa<\/td>\n<\/tr>\n<tr>\n<td>UDAF\u5b9e\u73b0\u65b9\u5f0f<\/td>\n<td>\u4f20\u7edf\u65b9\u5f0f\u7ee7\u627fUserDefinedAggregateFunction&#xff08;Scala&#xff09;\u6216\u4f7f\u7528Pandas UDF\u805a\u5408<\/td>\n<\/tr>\n<tr>\n<td>\u9002\u7528\u573a\u666f<\/td>\n<td>UDF&#xff1a;\u590d\u6742\u5b57\u7b26\u4e32\u5904\u7406\u3001\u6b63\u5219\u3001\u673a\u5668\u5b66\u4e60\u9884\u6d4b&#xff1b;UDAF&#xff1a;\u81ea\u5b9a\u4e49\u7a97\u53e3\u805a\u5408&#xff1b;UDTF&#xff1a;\u4e00\u884c\u751f\u6210\u591a\u884c<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h3>\u56db\u3001\u539f\u7406\u901a\u4fd7\u8bb2\u89e3<\/h3>\n<h4>4.1 \u4f20\u7edfUDF\u7684\u201c\u9010\u884c\u4e4b\u65c5\u201d<\/h4>\n<p>\u5f53\u4f60\u6ce8\u518c\u4e00\u4e2a\u4f20\u7edfUDF&#xff0c;\u5e76\u5e94\u7528\u4e8eDataFrame\u65f6&#xff0c;\u6267\u884c\u8fc7\u7a0b\u5982\u4e0b&#xff1a;<\/p>\n<li>Driver\u7aef&#xff1a;Python\u51fd\u6570\u88ab\u5e8f\u5217\u5316&#xff08;pickle&#xff09;\u5e76\u5e7f\u64ad\u5230\u6240\u6709Executor\u3002<\/li>\n<li>Executor\u7aef&#xff1a;\u6bcf\u4e2aExecutor\u542f\u52a8\u4e00\u4e2aPython\u8fdb\u7a0b&#xff08;\u5982\u679c\u914d\u7f6e\u4e86spark.python.worker.reuse\u5219\u590d\u7528&#xff09;\u3002<\/li>\n<li>\u6570\u636e\u8f6c\u6362&#xff1a;\u6bcf\u4e2a\u5206\u533a\u7684JVM\u6570\u636e&#xff08;UnsafeRow&#xff09;\u88ab\u8f6c\u6362\u4e3aPython\u5bf9\u8c61&#xff08;Row\u6216\u5143\u7ec4&#xff09;&#xff0c;\u901a\u8fc7Py4J\u4f20\u9012\u7ed9Python\u8fdb\u7a0b\u3002<\/li>\n<li>\u9010\u884c\u8c03\u7528&#xff1a;Python\u8fdb\u7a0b\u5bf9\u6bcf\u4e00\u884c\u8c03\u7528\u4f60\u7684\u81ea\u5b9a\u4e49\u51fd\u6570&#xff0c;\u8ba1\u7b97\u7ed3\u679c\u3002<\/li>\n<li>\u7ed3\u679c\u8fd4\u56de&#xff1a;\u7ed3\u679c\u5e8f\u5217\u5316\u4f20\u56deJVM&#xff0c;\u5e76\u7ec4\u88c5\u6210\u65b0\u7684\u884c\u3002<\/li>\n<p>\u8fd9\u4e2a\u8fc7\u7a0b\u4e2d&#xff0c;\u6bcf\u6761\u6570\u636e\u90fd\u8981\u7ecf\u5386\u591a\u6b21\u5e8f\u5217\u5316\/\u53cd\u5e8f\u5217\u5316&#xff0c;\u5373\u4f7f\u4f60\u7684\u51fd\u6570\u5f88\u7b80\u5355&#xff0c;\u5f00\u9500\u4e5f\u5f88\u5927\u3002\u4f8b\u5982&#xff0c;lambda x: x&#043;1\u7528UDF\u5b9e\u73b0\u53ef\u80fd\u6bd4\u5185\u7f6ecol&#043;1\u6162\u51e0\u5341\u500d\u3002<\/p>\n<h4>4.2 Pandas UDF\u7684\u201c\u6279\u91cf\u52a0\u901f\u201d<\/h4>\n<p>Pandas UDF\u5229\u7528Apache Arrow\u4f5c\u4e3a\u5217\u5f0f\u5185\u5b58\u683c\u5f0f&#xff0c;\u6279\u91cf\u4f20\u8f93\u6570\u636e\u3002\u6d41\u7a0b\u5982\u4e0b&#xff1a;<\/p>\n<li>\u5206\u533a\u6279\u91cf\u8bfb\u53d6&#xff1a;Spark\u5c06\u6bcf\u4e2a\u5206\u533a\u7684\u6570\u636e\u8f6c\u6362\u4e3aArrow\u683c\u5f0f&#xff08;\u5217\u5f0f&#xff09;&#xff0c;\u4e00\u6b21\u6027\u4f20\u7ed9Python\u8fdb\u7a0b\u3002<\/li>\n<li>Pandas DataFrame\/Series\u64cd\u4f5c&#xff1a;\u4f60\u7684\u51fd\u6570\u63a5\u6536\u4e00\u4e2aPandas Series\u6216DataFrame&#xff0c;\u5229\u7528Pandas\u7684\u5411\u91cf\u5316\u64cd\u4f5c\u6216NumPy\u8ba1\u7b97&#xff0c;\u4e00\u6b21\u6027\u5904\u7406\u6574\u4e2a\u6279\u6b21\u3002<\/li>\n<li>\u6279\u91cf\u8fd4\u56de&#xff1a;\u7ed3\u679c\u6279\u91cf\u8f6c\u56deArrow\u683c\u5f0f&#xff0c;\u4f20\u56deJVM\u3002<\/li>\n<p>\u8fd9\u6837\u5927\u5927\u51cf\u5c11\u4e86JVM\u4e0ePython\u4e4b\u95f4\u7684\u901a\u4fe1\u6b21\u6570&#xff0c;\u5e76\u4e14\u5229\u7528Pandas\u5e95\u5c42C\u8bed\u8a00\u52a0\u901f&#xff0c;\u6027\u80fd\u901a\u5e38\u6bd4\u9010\u884cUDF\u9ad8\u4e00\u4e2a\u6570\u91cf\u7ea7&#xff0c;\u63a5\u8fd1\u5185\u7f6e\u51fd\u6570\u7684\u6c34\u5e73\u3002<\/p>\n<h4>4.3 UDAF\u7684\u81ea\u5b9a\u4e49\u805a\u5408<\/h4>\n<p>Spark\u5185\u7f6e\u805a\u5408\u51fd\u6570&#xff08;\u5982sum\u3001avg&#xff09;\u65e0\u6cd5\u6ee1\u8db3\u590d\u6742\u805a\u5408\u9700\u6c42\u65f6&#xff0c;\u53ef\u4ee5\u81ea\u5b9a\u4e49\u805a\u5408\u3002\u5728PySpark\u4e2d&#xff0c;\u63a8\u8350\u4f7f\u7528grouped_apply\u6216applyInPandas\u914d\u5408Pandas UDF\u5b9e\u73b0\u3002\u4f60\u5b9a\u4e49\u4e00\u4e2a\u51fd\u6570&#xff0c;\u63a5\u6536\u4e00\u4e2a\u5206\u7ec4\u7684\u6240\u6709\u6570\u636e&#xff08;Pandas DataFrame&#xff09;&#xff0c;\u8fd4\u56de\u4e00\u4e2a\u805a\u5408\u7ed3\u679c&#xff08;Pandas Series\u6216DataFrame&#xff09;\u3002Spark\u4f1a\u81ea\u52a8\u5c06\u5206\u7ec4\u6570\u636e\u62c9\u53d6\u5230\u540c\u4e00\u4e2aExecutor\u4e0a&#xff08;\u53ef\u80fd\u5f15\u53d1shuffle&#xff09;&#xff0c;\u7136\u540e\u8c03\u7528\u4f60\u7684\u51fd\u6570\u3002<\/p>\n<h4>4.4 UDTF\u7684\u8868\u751f\u6210<\/h4>\n<p>Spark SQL\u539f\u751f\u4e0d\u652f\u6301UDTF&#xff08;\u7c7b\u4f3cHive\u7684explode&#xff09;&#xff0c;\u4f46\u53ef\u4ee5\u901a\u8fc7flatMap\u6216explode\u51fd\u6570\u6a21\u62df\u3002\u4f8b\u5982&#xff0c;\u4f60\u53ef\u4ee5\u5b9a\u4e49UDF\u8fd4\u56de\u4e00\u4e2a\u6570\u7ec4&#xff0c;\u7136\u540e\u8c03\u7528explode\u5c06\u6570\u7ec4\u5c55\u5f00\u4e3a\u591a\u884c&#xff0c;\u6548\u679c\u7b49\u540c\u4e8eUDTF\u3002Pandas UDF\u4e5f\u53ef\u4ee5\u8fd4\u56depd.DataFrame\u6765\u5b9e\u73b0\u591a\u884c\u8f93\u51fa\u3002<\/p>\n<h3>\u4e94\u3001\u91cd\u70b9\u6982\u5ff5\u62c6\u89e3<\/h3>\n<h4>5.1 \u4f20\u7edfUDF\u7684\u6ce8\u518c\u4e0e\u4f7f\u7528<\/h4>\n<p><span class=\"token keyword\">from<\/span> pyspark<span class=\"token punctuation\">.<\/span>sql<span class=\"token punctuation\">.<\/span>functions <span class=\"token keyword\">import<\/span> udf<br \/>\n<span class=\"token keyword\">from<\/span> pyspark<span class=\"token punctuation\">.<\/span>sql<span class=\"token punctuation\">.<\/span>types <span class=\"token keyword\">import<\/span> StringType<\/p>\n<p><span class=\"token comment\"># \u65b9\u5f0f1&#xff1a;\u88c5\u9970\u5668<\/span><br \/>\n<span class=\"token decorator annotation punctuation\">&#064;udf<\/span><span class=\"token punctuation\">(<\/span>returnType<span class=\"token operator\">&#061;<\/span>StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">my_upper<\/span><span class=\"token punctuation\">(<\/span>s<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">if<\/span> s <span class=\"token keyword\">is<\/span> <span class=\"token boolean\">None<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token keyword\">return<\/span> <span class=\"token boolean\">None<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> s<span class=\"token punctuation\">.<\/span>upper<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u65b9\u5f0f2&#xff1a;\u51fd\u6570\u5b9a\u4e49\u540e\u6ce8\u518c<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">my_lower<\/span><span class=\"token punctuation\">(<\/span>s<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> s<span class=\"token punctuation\">.<\/span>lower<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token keyword\">if<\/span> s <span class=\"token keyword\">else<\/span> <span class=\"token boolean\">None<\/span><br \/>\nmy_lower_udf <span class=\"token operator\">&#061;<\/span> udf<span class=\"token punctuation\">(<\/span>my_lower<span class=\"token punctuation\">,<\/span> StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u4f7f\u7528<\/span><br \/>\ndf<span class=\"token punctuation\">.<\/span>withColumn<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;upper_name&#034;<\/span><span class=\"token punctuation\">,<\/span> my_upper<span class=\"token punctuation\">(<\/span>col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;name&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>\u5173\u952e\u70b9&#xff1a;<\/p>\n<ul>\n<li>\u5fc5\u987b\u6307\u5b9a\u8fd4\u56de\u7c7b\u578b&#xff08;\u6216Spark\u81ea\u52a8\u63a8\u65ad&#xff0c;\u4f46\u5efa\u8bae\u663e\u5f0f\u6307\u5b9a&#xff09;\u3002<\/li>\n<li>\u51fd\u6570\u5185\u90e8\u5e94\u5904\u7406\u7a7a\u503c&#xff08;None&#xff09;&#xff0c;\u5426\u5219\u53ef\u80fd\u629b\u5f02\u5e38\u3002<\/li>\n<li>UDF\u9ed8\u8ba4\u662f\u786e\u5b9a\u6027\u7684&#xff08;\u76f8\u540c\u8f93\u5165\u76f8\u540c\u8f93\u51fa&#xff09;&#xff0c;\u53ef\u901a\u8fc7udf(&#8230;, deterministic&#061;False)\u58f0\u660e\u975e\u786e\u5b9a\u6027&#xff08;\u5982\u968f\u673a\u51fd\u6570&#xff09;\u3002<\/li>\n<\/ul>\n<h4>5.2 Pandas UDF\u7684\u7c7b\u578b<\/h4>\n<p>Pandas UDF\u901a\u8fc7&#064;pandas_udf\u88c5\u9970\u5668&#xff0c;\u5e76\u4f20\u5165\u8fd4\u56de\u7c7b\u578b\u548c\u51fd\u6570\u7c7b\u578b\u3002\u51fd\u6570\u7c7b\u578b\u6709&#xff1a;<\/p>\n<table>\n<tr>\u51fd\u6570\u7c7b\u578b\u8f93\u5165\u7c7b\u578b\u8f93\u51fa\u7c7b\u578b\u4f7f\u7528\u573a\u666f<\/tr>\n<tbody>\n<tr>\n<td>PandasUDFType.SCALAR&#xff08;\u6216\u9ed8\u8ba4&#xff09;<\/td>\n<td>pd.Series<\/td>\n<td>pd.Series&#xff08;\u957f\u5ea6\u76f8\u540c&#xff09;<\/td>\n<td>\u9010\u884c\u4f46\u6279\u91cf\u8ba1\u7b97&#xff0c;\u5982upper<\/td>\n<\/tr>\n<tr>\n<td>PandasUDFType.GROUPED_MAP<\/td>\n<td>pd.DataFrame<\/td>\n<td>pd.DataFrame<\/td>\n<td>\u5206\u7ec4\u540e\u5bf9\u6bcf\u7ec4DataFrame\u53d8\u6362&#xff0c;\u8fd4\u56de\u4efb\u610f\u884c\u6570<\/td>\n<\/tr>\n<tr>\n<td>PandasUDFType.GROUPED_AGG<\/td>\n<td>pd.Series\u6216pd.DataFrame<\/td>\n<td>\u6807\u91cf&#xff08;int\/float\u7b49&#xff09;<\/td>\n<td>\u5206\u7ec4\u805a\u5408&#xff0c;\u8fd4\u56de\u5355\u4e2a\u503c<\/td>\n<\/tr>\n<tr>\n<td>PandasUDFType.SCALAR_ITER<\/td>\n<td>Iterator[pd.Series]<\/td>\n<td>Iterator[pd.Series]<\/td>\n<td>\u5927\u6570\u636e\u91cf\u5206\u6279\u6b21\u5904\u7406<\/td>\n<\/tr>\n<tr>\n<td>PandasUDFType.MAP_ITER<\/td>\n<td>Iterator[pd.DataFrame]<\/td>\n<td>Iterator[pd.DataFrame]<\/td>\n<td>\u9002\u7528\u4e8emapInPandas<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>\u91cd\u8981&#xff1a;PySpark 3.0&#043;\u63a8\u8350\u4f7f\u7528\u7c7b\u578b\u63d0\u793a\u7684\u65b9\u5f0f\u5b9a\u4e49Pandas UDF&#xff0c;\u800c\u4e0d\u4f7f\u7528PandasUDFType\u679a\u4e3e\u3002\u4f8b\u5982&#xff1a;<\/p>\n<p><span class=\"token keyword\">from<\/span> pyspark<span class=\"token punctuation\">.<\/span>sql<span class=\"token punctuation\">.<\/span>functions <span class=\"token keyword\">import<\/span> pandas_udf<br \/>\n<span class=\"token keyword\">import<\/span> pandas <span class=\"token keyword\">as<\/span> pd<\/p>\n<p><span class=\"token decorator annotation punctuation\">&#064;pandas_udf<\/span><span class=\"token punctuation\">(<\/span>StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">my_upper<\/span><span class=\"token punctuation\">(<\/span>s<span class=\"token punctuation\">:<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span><span class=\"token operator\">&gt;<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> s<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">str<\/span><span class=\"token punctuation\">.<\/span>upper<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<h4>5.3 \u5206\u7ec4\u805a\u5408Pandas UDF<\/h4>\n<p>\u793a\u4f8b&#xff1a;\u81ea\u5b9a\u4e49\u805a\u5408\u51fd\u6570\u2014\u2014\u8ba1\u7b97\u6bcf\u4e2a\u7ec4\u7684\u51e0\u4f55\u5e73\u5747\u503c\u3002<\/p>\n<p><span class=\"token decorator annotation punctuation\">&#064;pandas_udf<\/span><span class=\"token punctuation\">(<\/span>returnType<span class=\"token operator\">&#061;<\/span>FloatType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">geometric_mean<\/span><span class=\"token punctuation\">(<\/span>v<span class=\"token punctuation\">:<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span><span class=\"token operator\">&gt;<\/span> <span class=\"token builtin\">float<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> np<span class=\"token punctuation\">.<\/span>exp<span class=\"token punctuation\">(<\/span>np<span class=\"token punctuation\">.<\/span>log<span class=\"token punctuation\">(<\/span>v<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>mean<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>df<span class=\"token punctuation\">.<\/span>groupBy<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;group&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>agg<span class=\"token punctuation\">(<\/span>geometric_mean<span class=\"token punctuation\">(<\/span>col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;value&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>alias<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;geo_mean&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>GROUPED_AGG\u7c7b\u578b\u7684Pandas UDF\u63a5\u6536\u6bcf\u4e2a\u5206\u7ec4\u7684\u5217&#xff08;\u4e00\u4e2a\u6216\u591a\u4e2aSeries&#xff09;&#xff0c;\u8fd4\u56de\u4e00\u4e2a\u6807\u91cf\u503c\u3002<\/p>\n<h4>5.4 \u5206\u7ec4\u53d8\u6362&#xff08;GROUPED_MAP&#xff09;<\/h4>\n<p><span class=\"token decorator annotation punctuation\">&#064;pandas_udf<\/span><span class=\"token punctuation\">(<\/span>returnType<span class=\"token operator\">&#061;<\/span>df<span class=\"token punctuation\">.<\/span>schema<span class=\"token punctuation\">,<\/span> functionType<span class=\"token operator\">&#061;<\/span>PandasUDFType<span class=\"token punctuation\">.<\/span>GROUPED_MAP<span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">normalize<\/span><span class=\"token punctuation\">(<\/span>df<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    df<span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;value_norm&#034;<\/span><span class=\"token punctuation\">]<\/span> <span class=\"token operator\">&#061;<\/span> <span class=\"token punctuation\">(<\/span>df<span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;value&#034;<\/span><span class=\"token punctuation\">]<\/span> <span class=\"token operator\">&#8211;<\/span> df<span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;value&#034;<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">.<\/span>mean<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token operator\">\/<\/span> df<span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;value&#034;<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">.<\/span>std<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> df<\/p>\n<p>df<span class=\"token punctuation\">.<\/span>groupBy<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;group&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span><span class=\"token builtin\">apply<\/span><span class=\"token punctuation\">(<\/span>normalize<span class=\"token punctuation\">)<\/span><\/p>\n<p>\u6ce8\u610f&#xff1a;GROUPED_MAP\u8981\u6c42\u8fd4\u56de\u7684DataFrame\u5fc5\u987b\u4e0e\u8f93\u5165\u7684StructType\u517c\u5bb9&#xff0c;\u4f46\u884c\u6570\u53ef\u4ee5\u4e0d\u540c\u3002<\/p>\n<h4>5.5 \u8fed\u4ee3\u5668Pandas UDF&#xff08;\u5904\u7406\u5927\u6570\u636e\u5206\u533a&#xff09;<\/h4>\n<p>\u5f53\u5206\u533a\u5185\u6570\u636e\u91cf\u5927\u65e0\u6cd5\u4e00\u6b21\u6027\u88c5\u5165\u5185\u5b58\u65f6&#xff0c;\u53ef\u4ee5\u4f7f\u7528SCALAR_ITER\u6216MAP_ITER\u3002\u4f8b\u5982&#xff1a;<\/p>\n<p><span class=\"token decorator annotation punctuation\">&#064;pandas_udf<\/span><span class=\"token punctuation\">(<\/span>StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">process_batch<\/span><span class=\"token punctuation\">(<\/span>iterator<span class=\"token punctuation\">:<\/span> Iterator<span class=\"token punctuation\">[<\/span>pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span><span class=\"token operator\">&gt;<\/span> Iterator<span class=\"token punctuation\">[<\/span>pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">for<\/span> s <span class=\"token keyword\">in<\/span> iterator<span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token keyword\">yield<\/span> s<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">str<\/span><span class=\"token punctuation\">.<\/span>upper<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>Spark\u4f1a\u5c06\u6bcf\u4e2a\u5206\u533a\u7684\u6570\u636e\u5206\u6279\u6b21\u4f20\u5165\u8fed\u4ee3\u5668&#xff0c;\u6bcf\u6b21\u5904\u7406\u4e00\u4e2a\u6279\u6b21&#xff08;\u7531spark.sql.execution.arrow.maxRecordsPerBatch\u63a7\u5236&#xff09;\u3002<\/p>\n<h4>5.6 \u6ce8\u518cUDF\u4f9bSQL\u4f7f\u7528<\/h4>\n<p>spark<span class=\"token punctuation\">.<\/span>udf<span class=\"token punctuation\">.<\/span>register<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;my_upper&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token keyword\">lambda<\/span> s<span class=\"token punctuation\">:<\/span> s<span class=\"token punctuation\">.<\/span>upper<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token keyword\">if<\/span> s <span class=\"token keyword\">else<\/span> <span class=\"token boolean\">None<\/span><span class=\"token punctuation\">,<\/span> StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\nspark<span class=\"token punctuation\">.<\/span>sql<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;SELECT my_upper(name) FROM users&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>show<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<h4>5.7 UDAF\u7684\u4f20\u7edf\u5b9e\u73b0&#xff08;Scala\u4e92\u64cd\u4f5c&#xff09;<\/h4>\n<p>\u5728PySpark\u4e2d&#xff0c;\u5982\u679c\u4f60\u9700\u8981\u9ad8\u6027\u80fd\u7684UDAF&#xff0c;\u53ef\u4ee5\u7f16\u5199Scala UDAF\u5e76\u901a\u8fc7spark.udf.registerJavaFunction\u8c03\u7528\u3002\u4f46\u4e00\u822c\u4e0d\u63a8\u8350&#xff0c;\u56e0\u4e3a\u589e\u52a0\u4e86\u590d\u6742\u5ea6\u3002Pandas UDF\u7684GROUPED_AGG\u5df2\u7ecf\u80fd\u6ee1\u8db3\u5927\u591a\u6570\u9700\u6c42\u3002<\/p>\n<h4>5.8 UDTF\u7684\u6a21\u62df<\/h4>\n<p>\u65b9\u6cd51&#xff1a;UDF\u8fd4\u56de\u6570\u7ec4 &#043; explode<\/p>\n<p><span class=\"token keyword\">def<\/span> <span class=\"token function\">split_words<\/span><span class=\"token punctuation\">(<\/span>s<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> s<span class=\"token punctuation\">.<\/span>split<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token keyword\">if<\/span> s <span class=\"token keyword\">else<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token punctuation\">]<\/span><br \/>\nsplit_udf <span class=\"token operator\">&#061;<\/span> udf<span class=\"token punctuation\">(<\/span>split_words<span class=\"token punctuation\">,<\/span> ArrayType<span class=\"token punctuation\">(<\/span>StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\ndf<span class=\"token punctuation\">.<\/span>withColumn<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;word&#034;<\/span><span class=\"token punctuation\">,<\/span> explode<span class=\"token punctuation\">(<\/span>split_udf<span class=\"token punctuation\">(<\/span>col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;text&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>\u65b9\u6cd52&#xff1a;\u4f7f\u7528flatMap&#xff08;RDD\u5c42&#xff09;<\/p>\n<p>df<span class=\"token punctuation\">.<\/span>rdd<span class=\"token punctuation\">.<\/span>flatMap<span class=\"token punctuation\">(<\/span><span class=\"token keyword\">lambda<\/span> row<span class=\"token punctuation\">:<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token punctuation\">(<\/span>row<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">id<\/span><span class=\"token punctuation\">,<\/span> word<span class=\"token punctuation\">)<\/span> <span class=\"token keyword\">for<\/span> word <span class=\"token keyword\">in<\/span> row<span class=\"token punctuation\">.<\/span>text<span class=\"token punctuation\">.<\/span>split<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>toDF<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;id&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;word&#034;<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>\u65b9\u6cd53&#xff1a;mapInPandas\u8fd4\u56de\u591a\u884c<\/p>\n<p><span class=\"token keyword\">def<\/span> <span class=\"token function\">generate_rows<\/span><span class=\"token punctuation\">(<\/span>iterator<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">for<\/span> pdf <span class=\"token keyword\">in<\/span> iterator<span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token keyword\">for<\/span> _<span class=\"token punctuation\">,<\/span> row <span class=\"token keyword\">in<\/span> pdf<span class=\"token punctuation\">.<\/span>iterrows<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n            <span class=\"token keyword\">for<\/span> word <span class=\"token keyword\">in<\/span> row<span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;text&#034;<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">.<\/span>split<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n                <span class=\"token keyword\">yield<\/span> <span class=\"token punctuation\">(<\/span>row<span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;id&#034;<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">,<\/span> word<span class=\"token punctuation\">)<\/span><\/p>\n<p>df<span class=\"token punctuation\">.<\/span>mapInPandas<span class=\"token punctuation\">(<\/span>generate_rows<span class=\"token punctuation\">,<\/span> schema<span class=\"token operator\">&#061;<\/span><span class=\"token string\">&#034;id int, word string&#034;<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<h3>\u516d\u3001\u6613\u9519\u70b9\u907f\u5751<\/h3>\n<h4>6.1 UDF\u4e2d\u6355\u83b7\u4e0d\u53ef\u5e8f\u5217\u5316\u5bf9\u8c61<\/h4>\n<p>\u9519\u8bef\u793a\u4f8b&#xff1a;<\/p>\n<p><span class=\"token keyword\">class<\/span> <span class=\"token class-name\">MyProcessor<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">def<\/span> <span class=\"token function\">__init__<\/span><span class=\"token punctuation\">(<\/span>self<span class=\"token punctuation\">,<\/span> data<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        self<span class=\"token punctuation\">.<\/span>data <span class=\"token operator\">&#061;<\/span> data<br \/>\n    <span class=\"token keyword\">def<\/span> <span class=\"token function\">process<\/span><span class=\"token punctuation\">(<\/span>self<span class=\"token punctuation\">,<\/span> x<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token keyword\">return<\/span> self<span class=\"token punctuation\">.<\/span>data<span class=\"token punctuation\">[<\/span>x<span class=\"token punctuation\">]<\/span><\/p>\n<p>processor <span class=\"token operator\">&#061;<\/span> MyProcessor<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">{<\/span><span class=\"token string\">&#034;a&#034;<\/span><span class=\"token punctuation\">:<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">}<\/span><span class=\"token punctuation\">)<\/span><br \/>\nudf<span class=\"token punctuation\">(<\/span>processor<span class=\"token punctuation\">.<\/span>process<span class=\"token punctuation\">,<\/span> IntegerType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span>  <span class=\"token comment\"># \u62a5\u9519&#xff1a;\u65e0\u6cd5\u5e8f\u5217\u5316processor<\/span><\/p>\n<p>\u89e3\u51b3&#xff1a;\u5c06\u6570\u636e\u901a\u8fc7\u5e7f\u64ad\u53d8\u91cf\u4f20\u9012&#xff0c;\u6216\u8005\u5728UDF\u5185\u90e8\u91cd\u65b0\u521b\u5efa\u5bf9\u8c61&#xff08;\u5982\u679c\u8f7b\u91cf&#xff09;\u3002<\/p>\n<h4>6.2 \u8bef\u4ee5\u4e3aUDF\u53ef\u4ee5\u4fee\u6539\u5168\u5c40\u53d8\u91cf<\/h4>\n<p>UDF\u5728Executor\u7684Python\u8fdb\u7a0b\u4e2d\u6267\u884c&#xff0c;\u4fee\u6539\u7684\u5168\u5c40\u53d8\u91cf\u4e0d\u4f1a\u5f71\u54cdDriver\u6216\u5176\u4ed6Executor\u3002\u5982\u9700\u7d2f\u52a0&#xff0c;\u4f7f\u7528\u7d2f\u52a0\u5668\u3002<\/p>\n<h4>6.3 \u5ffd\u7565\u7a7a\u503c\u5bfc\u81f4UDF\u629b\u5f02\u5e38<\/h4>\n<p>UDF\u63a5\u6536\u7684\u8f93\u5165\u53ef\u80fd\u4e3aNone&#xff0c;\u5fc5\u987b\u663e\u5f0f\u5904\u7406&#xff0c;\u5426\u5219\u629b\u51faTypeError\u6216AttributeError\u3002<\/p>\n<h4>6.4 Pandas UDF\u8fd4\u56de\u7c7b\u578b\u4e0d\u5339\u914d<\/h4>\n<p>\u9519\u8bef&#xff1a;\u58f0\u660e\u8fd4\u56deFloatType()&#xff0c;\u4f46\u5b9e\u9645\u8fd4\u56de\u4e86pd.Series\u5305\u542b\u5b57\u7b26\u4e32\u3002<\/p>\n<p>\u89e3\u51b3&#xff1a;\u786e\u4fdd\u8fd4\u56de\u7684Series\u7c7b\u578b\u4e0e\u58f0\u660e\u7684Spark\u7c7b\u578b\u517c\u5bb9&#xff08;\u5982float64\u5bf9\u5e94DoubleType&#xff09;\u3002<\/p>\n<h4>6.5 \u5206\u7ec4\u805a\u5408Pandas UDF\u4e2d\u8bef\u7528agg<\/h4>\n<p><span class=\"token comment\"># \u9519\u8bef&#xff1a;\u4e0d\u80fd\u76f4\u63a5\u5728agg\u4e2d\u4f7f\u7528GROUPED_MAP\u7c7b\u578b\u7684UDF<\/span><br \/>\ndf<span class=\"token punctuation\">.<\/span>groupBy<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;g&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>agg<span class=\"token punctuation\">(<\/span>my_udf<span class=\"token punctuation\">(<\/span>df<span class=\"token punctuation\">.<\/span>value<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span>  <span class=\"token comment\"># \u62a5\u9519<\/span><\/p>\n<p>GROUPED_MAP\u901a\u8fc7apply\u4f7f\u7528&#xff0c;GROUPED_AGG\u901a\u8fc7agg\u4f7f\u7528\u3002<\/p>\n<h4>6.6 \u4f7f\u7528Pandas UDF\u65f6\u4e0d\u8bbe\u7f6espark.sql.execution.arrow.pyspark.enabled<\/h4>\n<p>Pandas UDF\u4f9d\u8d56Arrow&#xff0c;\u8be5\u914d\u7f6e\u9ed8\u8ba4true&#xff08;Spark 3.x&#xff09;&#xff0c;\u4f46\u5982\u679c\u8bbe\u7f6e\u4e3afalse&#xff0c;Pandas UDF\u4f1a\u56de\u9000\u5230\u9010\u884c\u5904\u7406&#xff0c;\u6027\u80fd\u4e0b\u964d\u4e25\u91cd\u3002<\/p>\n<h3>\u4e03\u3001\u5b8c\u6574\u5b9e\u6218\u6848\u4f8b<\/h3>\n<p>\u672c\u6848\u4f8b\u5c06\u6f14\u793a\u4ece\u4f20\u7edfUDF\u5230Pandas UDF\u7684\u6f14\u8fdb&#xff0c;\u5e76\u5b9e\u73b0\u4e00\u4e2a\u590d\u6742\u7684\u4e1a\u52a1\u573a\u666f&#xff1a;IP\u5730\u5740\u89e3\u6790&#xff08;IP\u8f6c\u6362\u4e3a\u5730\u7406\u4f4d\u7f6e&#xff09;\u3001\u81ea\u5b9a\u4e49\u805a\u5408&#xff08;\u8ba1\u7b97\u6bcf\u4e2a\u7528\u6237\u4f1a\u8bdd\u7684\u505c\u7559\u65f6\u957f&#xff09;\u3001UDTF\u6a21\u62df&#xff08;URL\u62c6\u89e3\u4e3a\u8def\u5f84\u5c42\u7ea7&#xff09;\u3002<\/p>\n<p><span class=\"token comment\"># &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061; udf_udaf_udtf_demo.py &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;<\/span><br \/>\n<span class=\"token comment\"># \u529f\u80fd&#xff1a;PySpark\u81ea\u5b9a\u4e49\u51fd\u6570&#xff08;UDF\u3001Pandas UDF\u3001UDAF\u3001UDTF\u6a21\u62df&#xff09;\u4f01\u4e1a\u7ea7\u5b9e\u6218<\/span><br \/>\n<span class=\"token comment\"># \u6570\u636e&#xff1a;\u6a21\u62df\u7528\u6237\u8bbf\u95ee\u65e5\u5fd7&#xff08;IP\u3001\u65f6\u95f4\u3001URL&#xff09;<\/span><br \/>\n<span class=\"token comment\"># \u73af\u5883&#xff1a;PySpark 3.x&#xff0c;\u9700\u8981\u5b89\u88c5geoip2\u5e93&#xff08;\u53ef\u9009&#xff0c;\u8fd9\u91cc\u6a21\u62df&#xff09;<\/span><\/p>\n<p><span class=\"token keyword\">from<\/span> pyspark<span class=\"token punctuation\">.<\/span>sql <span class=\"token keyword\">import<\/span> SparkSession<br \/>\n<span class=\"token keyword\">from<\/span> pyspark<span class=\"token punctuation\">.<\/span>sql<span class=\"token punctuation\">.<\/span>functions <span class=\"token keyword\">import<\/span> udf<span class=\"token punctuation\">,<\/span> pandas_udf<span class=\"token punctuation\">,<\/span> col<span class=\"token punctuation\">,<\/span> explode<span class=\"token punctuation\">,<\/span> struct<span class=\"token punctuation\">,<\/span> lit<br \/>\n<span class=\"token keyword\">from<\/span> pyspark<span class=\"token punctuation\">.<\/span>sql<span class=\"token punctuation\">.<\/span>types <span class=\"token keyword\">import<\/span> StringType<span class=\"token punctuation\">,<\/span> IntegerType<span class=\"token punctuation\">,<\/span> DoubleType<span class=\"token punctuation\">,<\/span> ArrayType<span class=\"token punctuation\">,<\/span> StructType<span class=\"token punctuation\">,<\/span> StructField<span class=\"token punctuation\">,<\/span> TimestampType<br \/>\n<span class=\"token keyword\">import<\/span> pandas <span class=\"token keyword\">as<\/span> pd<br \/>\n<span class=\"token keyword\">import<\/span> numpy <span class=\"token keyword\">as<\/span> np<br \/>\n<span class=\"token keyword\">import<\/span> random<br \/>\n<span class=\"token keyword\">from<\/span> datetime <span class=\"token keyword\">import<\/span> datetime<span class=\"token punctuation\">,<\/span> timedelta<br \/>\n<span class=\"token keyword\">import<\/span> time<\/p>\n<p><span class=\"token comment\"># &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061; 1. \u521b\u5efaSparkSession &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;<\/span><br \/>\nspark <span class=\"token operator\">&#061;<\/span> SparkSession<span class=\"token punctuation\">.<\/span>builder \\\\<br \/>\n    <span class=\"token punctuation\">.<\/span>appName<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;UDF_UDAF_UDTF_Demo&#034;<\/span><span class=\"token punctuation\">)<\/span> \\\\<br \/>\n    <span class=\"token punctuation\">.<\/span>master<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;local[4]&#034;<\/span><span class=\"token punctuation\">)<\/span> \\\\<br \/>\n    <span class=\"token punctuation\">.<\/span>config<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;spark.sql.execution.arrow.pyspark.enabled&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;true&#034;<\/span><span class=\"token punctuation\">)<\/span> \\\\<br \/>\n    <span class=\"token punctuation\">.<\/span>config<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;spark.sql.execution.arrow.maxRecordsPerBatch&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;10000&#034;<\/span><span class=\"token punctuation\">)<\/span> \\\\<br \/>\n    <span class=\"token punctuation\">.<\/span>getOrCreate<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>sc <span class=\"token operator\">&#061;<\/span> spark<span class=\"token punctuation\">.<\/span>sparkContext<br \/>\nsc<span class=\"token punctuation\">.<\/span>setLogLevel<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;WARN&#034;<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;PySpark \u81ea\u5b9a\u4e49\u51fd\u6570&#xff08;UDF \/ Pandas UDF \/ UDAF \/ UDTF&#xff09;\u4f01\u4e1a\u5b9e\u6218&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061; 2. \u751f\u6210\u6a21\u62df\u6570\u636e &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\\\\n\u751f\u6210\u7528\u6237\u8bbf\u95ee\u65e5\u5fd7&#8230;&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\nnum_records <span class=\"token operator\">&#061;<\/span> <span class=\"token number\">500_000<\/span><br \/>\nusers <span class=\"token operator\">&#061;<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#034;user_<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>i<span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">&#034;<\/span><\/span> <span class=\"token keyword\">for<\/span> i <span class=\"token keyword\">in<\/span> <span class=\"token builtin\">range<\/span><span class=\"token punctuation\">(<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">1001<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">]<\/span><br \/>\nips <span class=\"token operator\">&#061;<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#034;192.168.<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>random<span class=\"token punctuation\">.<\/span>randint<span class=\"token punctuation\">(<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">,<\/span><span class=\"token number\">255<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">.<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>random<span class=\"token punctuation\">.<\/span>randint<span class=\"token punctuation\">(<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">,<\/span><span class=\"token number\">255<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">&#034;<\/span><\/span> <span class=\"token keyword\">for<\/span> _ <span class=\"token keyword\">in<\/span> <span class=\"token builtin\">range<\/span><span class=\"token punctuation\">(<\/span><span class=\"token number\">500<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">]<\/span><br \/>\nurls <span class=\"token operator\">&#061;<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;\/home&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;\/product\/1&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;\/product\/2&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;\/cart&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;\/checkout&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;\/api\/status&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;\/search?q&#061;spark&#034;<\/span><span class=\"token punctuation\">]<\/span><\/p>\n<p><span class=\"token keyword\">def<\/span> <span class=\"token function\">generate_log<\/span><span class=\"token punctuation\">(<\/span>i<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    user <span class=\"token operator\">&#061;<\/span> random<span class=\"token punctuation\">.<\/span>choice<span class=\"token punctuation\">(<\/span>users<span class=\"token punctuation\">)<\/span><br \/>\n    ip <span class=\"token operator\">&#061;<\/span> random<span class=\"token punctuation\">.<\/span>choice<span class=\"token punctuation\">(<\/span>ips<span class=\"token punctuation\">)<\/span><br \/>\n    url <span class=\"token operator\">&#061;<\/span> random<span class=\"token punctuation\">.<\/span>choice<span class=\"token punctuation\">(<\/span>urls<span class=\"token punctuation\">)<\/span><br \/>\n    ts <span class=\"token operator\">&#061;<\/span> datetime<span class=\"token punctuation\">.<\/span>now<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span> timedelta<span class=\"token punctuation\">(<\/span>seconds<span class=\"token operator\">&#061;<\/span>random<span class=\"token punctuation\">.<\/span>randint<span class=\"token punctuation\">(<\/span><span class=\"token number\">0<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">86400<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">30<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> <span class=\"token punctuation\">(<\/span>user<span class=\"token punctuation\">,<\/span> ip<span class=\"token punctuation\">,<\/span> url<span class=\"token punctuation\">,<\/span> ts<span class=\"token punctuation\">)<\/span><\/p>\n<p>data <span class=\"token operator\">&#061;<\/span> <span class=\"token punctuation\">[<\/span>generate_log<span class=\"token punctuation\">(<\/span>i<span class=\"token punctuation\">)<\/span> <span class=\"token keyword\">for<\/span> i <span class=\"token keyword\">in<\/span> <span class=\"token builtin\">range<\/span><span class=\"token punctuation\">(<\/span>num_records<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">]<\/span><br \/>\ndf <span class=\"token operator\">&#061;<\/span> spark<span class=\"token punctuation\">.<\/span>createDataFrame<span class=\"token punctuation\">(<\/span>data<span class=\"token punctuation\">,<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;user_id&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;ip&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;url&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;event_time&#034;<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#034;\u751f\u6210\u4e86 <\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>df<span class=\"token punctuation\">.<\/span>count<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><span class=\"token format-spec\">,<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\"> \u6761\u65e5\u5fd7&#034;<\/span><\/span><span class=\"token punctuation\">)<\/span><br \/>\ndf<span class=\"token punctuation\">.<\/span>show<span class=\"token punctuation\">(<\/span><span class=\"token number\">5<\/span><span class=\"token punctuation\">,<\/span> truncate<span class=\"token operator\">&#061;<\/span><span class=\"token boolean\">False<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061; 3. \u4f20\u7edfUDF\u793a\u4f8b&#xff1a;IP\u5730\u5740\u6bb5\u89e3\u6790&#xff08;\u6a21\u62df\u5730\u7406\u4f4d\u7f6e&#xff09; &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\\\\n&#034;<\/span> <span class=\"token operator\">&#043;<\/span> <span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\u6b65\u9aa41: \u4f20\u7edfUDF &#8211; IP\u5730\u5740\u89e3\u6790&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u6a21\u62dfIP\u5230\u7701\u4efd\u7684\u6620\u5c04&#xff08;\u5b9e\u9645\u53ef\u7528geoip\u6570\u636e\u5e93&#xff09;<\/span><br \/>\nip_to_city <span class=\"token operator\">&#061;<\/span> <span class=\"token punctuation\">{<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#034;192.168.<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>i<span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">.<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>j<span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">&#034;<\/span><\/span><span class=\"token punctuation\">:<\/span> <span class=\"token string-interpolation\"><span class=\"token string\">f&#034;City_<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>i <span class=\"token operator\">%<\/span> <span class=\"token number\">5<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">&#034;<\/span><\/span> <span class=\"token keyword\">for<\/span> i <span class=\"token keyword\">in<\/span> <span class=\"token builtin\">range<\/span><span class=\"token punctuation\">(<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">256<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token keyword\">for<\/span> j <span class=\"token keyword\">in<\/span> <span class=\"token builtin\">range<\/span><span class=\"token punctuation\">(<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">256<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">}<\/span><br \/>\n<span class=\"token comment\"># \u7b80\u5316&#xff1a;\u53ea\u6a21\u62df\u90e8\u5206\u89c4\u5219<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">ip_to_region<\/span><span class=\"token punctuation\">(<\/span>ip<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">if<\/span> <span class=\"token keyword\">not<\/span> ip<span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token keyword\">return<\/span> <span class=\"token string\">&#034;Unknown&#034;<\/span><br \/>\n    parts <span class=\"token operator\">&#061;<\/span> ip<span class=\"token punctuation\">.<\/span>split<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;.&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">if<\/span> <span class=\"token builtin\">len<\/span><span class=\"token punctuation\">(<\/span>parts<span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#061;&#061;<\/span> <span class=\"token number\">4<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        second <span class=\"token operator\">&#061;<\/span> <span class=\"token builtin\">int<\/span><span class=\"token punctuation\">(<\/span>parts<span class=\"token punctuation\">[<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        <span class=\"token keyword\">if<\/span> second <span class=\"token operator\">&lt;&#061;<\/span> <span class=\"token number\">50<\/span><span class=\"token punctuation\">:<\/span><br \/>\n            <span class=\"token keyword\">return<\/span> <span class=\"token string\">&#034;North&#034;<\/span><br \/>\n        <span class=\"token keyword\">elif<\/span> second <span class=\"token operator\">&lt;&#061;<\/span> <span class=\"token number\">100<\/span><span class=\"token punctuation\">:<\/span><br \/>\n            <span class=\"token keyword\">return<\/span> <span class=\"token string\">&#034;East&#034;<\/span><br \/>\n        <span class=\"token keyword\">elif<\/span> second <span class=\"token operator\">&lt;&#061;<\/span> <span class=\"token number\">150<\/span><span class=\"token punctuation\">:<\/span><br \/>\n            <span class=\"token keyword\">return<\/span> <span class=\"token string\">&#034;South&#034;<\/span><br \/>\n        <span class=\"token keyword\">else<\/span><span class=\"token punctuation\">:<\/span><br \/>\n            <span class=\"token keyword\">return<\/span> <span class=\"token string\">&#034;West&#034;<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> <span class=\"token string\">&#034;Unknown&#034;<\/span><\/p>\n<p><span class=\"token comment\"># \u6ce8\u518c\u4f20\u7edfUDF<\/span><br \/>\nregion_udf <span class=\"token operator\">&#061;<\/span> udf<span class=\"token punctuation\">(<\/span>ip_to_region<span class=\"token punctuation\">,<\/span> StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\ndf_with_region <span class=\"token operator\">&#061;<\/span> df<span class=\"token punctuation\">.<\/span>withColumn<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;region&#034;<\/span><span class=\"token punctuation\">,<\/span> region_udf<span class=\"token punctuation\">(<\/span>col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;ip&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\u6dfb\u52a0\u5730\u7406\u4f4d\u7f6e\u5217&#xff08;\u4f20\u7edfUDF&#xff09;:&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\ndf_with_region<span class=\"token punctuation\">.<\/span>select<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;ip&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;region&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>show<span class=\"token punctuation\">(<\/span><span class=\"token number\">10<\/span><span class=\"token punctuation\">,<\/span> truncate<span class=\"token operator\">&#061;<\/span><span class=\"token boolean\">False<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u6027\u80fd\u6d4b\u8bd5&#xff1a;\u4f20\u7edfUDF<\/span><br \/>\nstart <span class=\"token operator\">&#061;<\/span> time<span class=\"token punctuation\">.<\/span>time<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\ndf_with_region<span class=\"token punctuation\">.<\/span>count<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#034;\u4f20\u7edfUDF\u5904\u7406500k\u6761\u8017\u65f6: <\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>time<span class=\"token punctuation\">.<\/span>time<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span> start<span class=\"token punctuation\">:<\/span><span class=\"token format-spec\">.2f<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">\u79d2&#034;<\/span><\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061; 4. Pandas UDF&#xff08;Series to Series&#xff09;\u66ff\u4ee3\u4f20\u7edfUDF &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\\\\n&#034;<\/span> <span class=\"token operator\">&#043;<\/span> <span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\u6b65\u9aa42: Pandas UDF\u5411\u91cf\u5316\u5904\u7406&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u4f7f\u7528\u7c7b\u578b\u63d0\u793a\u7684Pandas UDF&#xff08;\u63a8\u8350&#xff09;<\/span><br \/>\n<span class=\"token decorator annotation punctuation\">&#064;pandas_udf<\/span><span class=\"token punctuation\">(<\/span>StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">ip_to_region_pandas<\/span><span class=\"token punctuation\">(<\/span>ip_series<span class=\"token punctuation\">:<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span><span class=\"token operator\">&gt;<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token comment\"># \u5411\u91cf\u5316\u64cd\u4f5c&#xff1a;\u4f7f\u7528str\u8bbf\u95ee\u5668\u548c\u6761\u4ef6\u5224\u65ad<\/span><br \/>\n    <span class=\"token comment\"># \u63d0\u53d6\u7b2c\u4e8c\u6bb5<\/span><br \/>\n    parts <span class=\"token operator\">&#061;<\/span> ip_series<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">str<\/span><span class=\"token punctuation\">.<\/span>split<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#039;.&#039;<\/span><span class=\"token punctuation\">,<\/span> expand<span class=\"token operator\">&#061;<\/span><span class=\"token boolean\">True<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    second <span class=\"token operator\">&#061;<\/span> parts<span class=\"token punctuation\">[<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">.<\/span>astype<span class=\"token punctuation\">(<\/span><span class=\"token builtin\">int<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    result <span class=\"token operator\">&#061;<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">(<\/span>index<span class=\"token operator\">&#061;<\/span>ip_series<span class=\"token punctuation\">.<\/span>index<span class=\"token punctuation\">,<\/span> dtype<span class=\"token operator\">&#061;<\/span><span class=\"token string\">&#034;object&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    result<span class=\"token punctuation\">[<\/span>second <span class=\"token operator\">&lt;&#061;<\/span> <span class=\"token number\">50<\/span><span class=\"token punctuation\">]<\/span> <span class=\"token operator\">&#061;<\/span> <span class=\"token string\">&#034;North&#034;<\/span><br \/>\n    result<span class=\"token punctuation\">[<\/span><span class=\"token punctuation\">(<\/span>second <span class=\"token operator\">&gt;<\/span> <span class=\"token number\">50<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&amp;<\/span> <span class=\"token punctuation\">(<\/span>second <span class=\"token operator\">&lt;&#061;<\/span> <span class=\"token number\">100<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">]<\/span> <span class=\"token operator\">&#061;<\/span> <span class=\"token string\">&#034;East&#034;<\/span><br \/>\n    result<span class=\"token punctuation\">[<\/span><span class=\"token punctuation\">(<\/span>second <span class=\"token operator\">&gt;<\/span> <span class=\"token number\">100<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&amp;<\/span> <span class=\"token punctuation\">(<\/span>second <span class=\"token operator\">&lt;&#061;<\/span> <span class=\"token number\">150<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">]<\/span> <span class=\"token operator\">&#061;<\/span> <span class=\"token string\">&#034;South&#034;<\/span><br \/>\n    result<span class=\"token punctuation\">[<\/span>second <span class=\"token operator\">&gt;<\/span> <span class=\"token number\">150<\/span><span class=\"token punctuation\">]<\/span> <span class=\"token operator\">&#061;<\/span> <span class=\"token string\">&#034;West&#034;<\/span><br \/>\n    result<span class=\"token punctuation\">[<\/span>second<span class=\"token punctuation\">.<\/span>isna<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">]<\/span> <span class=\"token operator\">&#061;<\/span> <span class=\"token string\">&#034;Unknown&#034;<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> result<\/p>\n<p>df_with_region_pd <span class=\"token operator\">&#061;<\/span> df<span class=\"token punctuation\">.<\/span>withColumn<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;region&#034;<\/span><span class=\"token punctuation\">,<\/span> ip_to_region_pandas<span class=\"token punctuation\">(<\/span>col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;ip&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;Pandas UDF\u7ed3\u679c\u793a\u4f8b:&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\ndf_with_region_pd<span class=\"token punctuation\">.<\/span>select<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;ip&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;region&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>show<span class=\"token punctuation\">(<\/span><span class=\"token number\">10<\/span><span class=\"token punctuation\">,<\/span> truncate<span class=\"token operator\">&#061;<\/span><span class=\"token boolean\">False<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>start <span class=\"token operator\">&#061;<\/span> time<span class=\"token punctuation\">.<\/span>time<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\ndf_with_region_pd<span class=\"token punctuation\">.<\/span>count<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#034;Pandas UDF\u5904\u7406\u8017\u65f6: <\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>time<span class=\"token punctuation\">.<\/span>time<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span> start<span class=\"token punctuation\">:<\/span><span class=\"token format-spec\">.2f<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">\u79d2&#034;<\/span><\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061; 5. \u5206\u7ec4\u805a\u5408UDAF&#xff1a;\u8ba1\u7b97\u6bcf\u4e2a\u7528\u6237\u7684\u4f1a\u8bdd\u65f6\u957f &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\\\\n&#034;<\/span> <span class=\"token operator\">&#043;<\/span> <span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\u6b65\u9aa43: \u81ea\u5b9a\u4e49\u805a\u5408\u51fd\u6570&#xff08;UDAF&#xff09;- \u7528\u6237\u4f1a\u8bdd\u5206\u6790&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u4f1a\u8bdd\u5b9a\u4e49&#xff1a;\u540c\u4e00\u7528\u6237\u76f8\u90bb\u4e8b\u4ef6\u95f4\u9694\u8d85\u8fc730\u5206\u949f\u89c6\u4e3a\u65b0\u4f1a\u8bdd<\/span><br \/>\n<span class=\"token comment\"># \u6211\u4eec\u9700\u8981\u8ba1\u7b97\u6bcf\u4e2a\u7528\u6237\u7684\u603b\u4f1a\u8bdd\u65f6\u957f&#xff08;\u6240\u6709\u4f1a\u8bdd\u65f6\u957f\u4e4b\u548c&#xff09;<\/span><br \/>\n<span class=\"token comment\"># \u4f7f\u7528Pandas UDF GROUPED_AGG\u5b9e\u73b0\u81ea\u5b9a\u4e49\u805a\u5408<\/span><\/p>\n<p><span class=\"token comment\"># \u5148\u6309\u7528\u6237\u548c\u65f6\u95f4\u6392\u5e8f&#xff0c;\u5206\u7ec4\u6570\u636e\u4f20\u5165\u4e00\u4e2aDataFrame<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">session_duration_udaf<\/span><span class=\"token punctuation\">(<\/span>pdf<span class=\"token punctuation\">:<\/span> pd<span class=\"token punctuation\">.<\/span>DataFrame<span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span><span class=\"token operator\">&gt;<\/span> <span class=\"token builtin\">float<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token comment\"># \u5047\u8bbepdf\u6709event_time\u5217&#xff0c;\u5df2\u6309\u65f6\u95f4\u6392\u5e8f<\/span><br \/>\n    <span class=\"token comment\"># \u8ba1\u7b97\u4f1a\u8bdd\u95f4\u9699&#xff0c;\u8d85\u8fc730\u5206\u949f\u5207\u5206<\/span><br \/>\n    pdf <span class=\"token operator\">&#061;<\/span> pdf<span class=\"token punctuation\">.<\/span>sort_values<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;event_time&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    deltas <span class=\"token operator\">&#061;<\/span> pdf<span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;event_time&#034;<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">.<\/span>diff<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>dt<span class=\"token punctuation\">.<\/span>total_seconds<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>fillna<span class=\"token punctuation\">(<\/span><span class=\"token number\">0<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token comment\"># \u6807\u8bb0\u65b0\u4f1a\u8bdd\u5f00\u59cb&#xff08;\u95f4\u9699&gt;1800\u79d2&#xff09;<\/span><br \/>\n    session_starts <span class=\"token operator\">&#061;<\/span> deltas <span class=\"token operator\">&gt;<\/span> <span class=\"token number\">1800<\/span><br \/>\n    session_id <span class=\"token operator\">&#061;<\/span> session_starts<span class=\"token punctuation\">.<\/span>cumsum<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token comment\"># \u8ba1\u7b97\u6bcf\u4e2a\u4f1a\u8bdd\u7684\u65f6\u957f&#xff08;\u6700\u5927\u503c-\u6700\u5c0f\u503c&#xff09;<\/span><br \/>\n    durations <span class=\"token operator\">&#061;<\/span> pdf<span class=\"token punctuation\">.<\/span>groupby<span class=\"token punctuation\">(<\/span>session_id<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;event_time&#034;<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">.<\/span>agg<span class=\"token punctuation\">(<\/span><span class=\"token keyword\">lambda<\/span> x<span class=\"token punctuation\">:<\/span> x<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">max<\/span><span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span> x<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">min<\/span><span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>dt<span class=\"token punctuation\">.<\/span>total_seconds<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    total_duration <span class=\"token operator\">&#061;<\/span> durations<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">sum<\/span><span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> total_duration<\/p>\n<p><span class=\"token comment\"># \u6ce8\u518c\u4e3a\u805a\u5408UDF&#xff08;GROUPED_AGG&#xff09;<\/span><br \/>\n<span class=\"token keyword\">from<\/span> pyspark<span class=\"token punctuation\">.<\/span>sql<span class=\"token punctuation\">.<\/span>functions <span class=\"token keyword\">import<\/span> pandas_udf<br \/>\n<span class=\"token keyword\">from<\/span> pyspark<span class=\"token punctuation\">.<\/span>sql<span class=\"token punctuation\">.<\/span>types <span class=\"token keyword\">import<\/span> DoubleType<\/p>\n<p>session_udaf <span class=\"token operator\">&#061;<\/span> pandas_udf<span class=\"token punctuation\">(<\/span>session_duration_udaf<span class=\"token punctuation\">,<\/span> returnType<span class=\"token operator\">&#061;<\/span>DoubleType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u6ce8\u610f&#xff1a;GROUPED_AGG\u7c7b\u578b\u53ea\u80fd\u7528\u4e8e\u805a\u5408&#xff0c;\u8f93\u5165\u662f\u4e00\u7ec4\u5217&#xff08;\u6b64\u5904\u6211\u4eec\u8981\u4f20\u5165\u5b8c\u6574DataFrame\u7ed3\u6784&#xff09;<\/span><br \/>\n<span class=\"token comment\"># \u66f4\u5e38\u7528\u7684\u662f\u5bf9\u5206\u7ec4\u540e\u7684\u6bcf\u4e2aGroup\u5e94\u7528\u81ea\u5b9a\u4e49\u805a\u5408\u51fd\u6570&#xff0c;\u9700\u8981\u4f20\u9012\u591a\u4e2a\u5217\u4f5c\u4e3a\u53c2\u6570\u3002<\/span><br \/>\n<span class=\"token comment\"># \u4f46GROUPED_AGG UDF\u53c2\u6570\u5fc5\u987b\u662f\u591a\u4e2aSeries&#xff0c;\u4e0d\u80fd\u76f4\u63a5\u63a5\u6536DataFrame\u3002<\/span><br \/>\n<span class=\"token comment\"># \u56e0\u6b64\u6211\u4eec\u7528\u53e6\u4e00\u79cd\u65b9\u5f0f&#xff1a;groupBy().apply() \u7684GROUPED_MAP\u7c7b\u578b\u3002<\/span><\/p>\n<p><span class=\"token comment\"># \u65b9\u6cd5&#xff1a;\u4f7f\u7528GROUPED_MAP&#xff0c;\u8fd4\u56de\u805a\u5408\u7ed3\u679c<\/span><br \/>\n<span class=\"token decorator annotation punctuation\">&#064;pandas_udf<\/span><span class=\"token punctuation\">(<\/span>returnType<span class=\"token operator\">&#061;<\/span>StructType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">[<\/span>StructField<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;total_session_duration&#034;<\/span><span class=\"token punctuation\">,<\/span> DoubleType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span><br \/>\n            functionType<span class=\"token operator\">&#061;<\/span><span class=\"token string\">&#034;GROUPED_MAP&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">session_analysis<\/span><span class=\"token punctuation\">(<\/span>pdf<span class=\"token punctuation\">:<\/span> pd<span class=\"token punctuation\">.<\/span>DataFrame<span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span><span class=\"token operator\">&gt;<\/span> pd<span class=\"token punctuation\">.<\/span>DataFrame<span class=\"token punctuation\">:<\/span><br \/>\n    pdf <span class=\"token operator\">&#061;<\/span> pdf<span class=\"token punctuation\">.<\/span>sort_values<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;event_time&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    deltas <span class=\"token operator\">&#061;<\/span> pdf<span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;event_time&#034;<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">.<\/span>diff<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>dt<span class=\"token punctuation\">.<\/span>total_seconds<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>fillna<span class=\"token punctuation\">(<\/span><span class=\"token number\">0<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    session_starts <span class=\"token operator\">&#061;<\/span> deltas <span class=\"token operator\">&gt;<\/span> <span class=\"token number\">1800<\/span><br \/>\n    session_id <span class=\"token operator\">&#061;<\/span> session_starts<span class=\"token punctuation\">.<\/span>cumsum<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    durations <span class=\"token operator\">&#061;<\/span> pdf<span class=\"token punctuation\">.<\/span>groupby<span class=\"token punctuation\">(<\/span>session_id<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;event_time&#034;<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">.<\/span>agg<span class=\"token punctuation\">(<\/span><span class=\"token keyword\">lambda<\/span> x<span class=\"token punctuation\">:<\/span> x<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">max<\/span><span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span> x<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">min<\/span><span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>dt<span class=\"token punctuation\">.<\/span>total_seconds<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    total_duration <span class=\"token operator\">&#061;<\/span> durations<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">sum<\/span><span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> pd<span class=\"token punctuation\">.<\/span>DataFrame<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">{<\/span><span class=\"token string\">&#034;total_session_duration&#034;<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token punctuation\">[<\/span>total_duration<span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">}<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u5e94\u7528<\/span><br \/>\nresult <span class=\"token operator\">&#061;<\/span> df<span class=\"token punctuation\">.<\/span>groupBy<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;user_id&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span><span class=\"token builtin\">apply<\/span><span class=\"token punctuation\">(<\/span>session_analysis<span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\u6bcf\u4e2a\u7528\u6237\u7684\u603b\u4f1a\u8bdd\u65f6\u957f&#xff08;\u79d2&#xff09;:&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\nresult<span class=\"token punctuation\">.<\/span>orderBy<span class=\"token punctuation\">(<\/span>col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;total_session_duration&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>desc<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>show<span class=\"token punctuation\">(<\/span><span class=\"token number\">10<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u66f4\u7b80\u6d01\u7684UDAF\u65b9\u5f0f&#xff1a;\u4f7f\u7528aggregate &#043; Pandas UDF&#xff08;\u9700\u8981Spark 3.0&#043;&#xff09;<\/span><br \/>\n<span class=\"token comment\"># \u8fd9\u91cc\u6211\u4eec\u5c55\u793a\u53e6\u4e00\u79cd&#xff1a;\u4f7f\u7528\u5185\u7f6e\u51fd\u6570\u914d\u5408\u7a97\u53e3&#xff0c;\u4f46\u4e3a\u4e86\u6f14\u793aUDAF&#xff0c;\u91c7\u7528GROUPED_MAP\u3002<\/span><\/p>\n<p><span class=\"token comment\"># &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061; 6. \u6a21\u62dfUDTF&#xff1a;\u5c06URL\u62c6\u5206\u4e3a\u8def\u5f84\u5c42\u7ea7 &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\\\\n&#034;<\/span> <span class=\"token operator\">&#043;<\/span> <span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\u6b65\u9aa44: UDTF\u6a21\u62df &#8211; URL\u8def\u5f84\u62c6\u5206&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u4f20\u7edf\u65b9\u5f0f&#xff1a;UDF\u8fd4\u56de\u6570\u7ec4 &#043; explode<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">url_to_paths<\/span><span class=\"token punctuation\">(<\/span>url<span class=\"token punctuation\">:<\/span> <span class=\"token builtin\">str<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">if<\/span> <span class=\"token keyword\">not<\/span> url<span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token keyword\">return<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token punctuation\">]<\/span><br \/>\n    <span class=\"token comment\"># \u53bb\u6389\u67e5\u8be2\u53c2\u6570<\/span><br \/>\n    path <span class=\"token operator\">&#061;<\/span> url<span class=\"token punctuation\">.<\/span>split<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#039;?&#039;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">[<\/span><span class=\"token number\">0<\/span><span class=\"token punctuation\">]<\/span><br \/>\n    parts <span class=\"token operator\">&#061;<\/span> path<span class=\"token punctuation\">.<\/span>split<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#039;\/&#039;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token comment\"># \u8fc7\u6ee4\u7a7a\u5b57\u7b26\u4e32<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> <span class=\"token punctuation\">[<\/span>p <span class=\"token keyword\">for<\/span> p <span class=\"token keyword\">in<\/span> parts <span class=\"token keyword\">if<\/span> p<span class=\"token punctuation\">]<\/span><\/p>\n<p>paths_udf <span class=\"token operator\">&#061;<\/span> udf<span class=\"token punctuation\">(<\/span>url_to_paths<span class=\"token punctuation\">,<\/span> ArrayType<span class=\"token punctuation\">(<\/span>StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\ndf_exploded <span class=\"token operator\">&#061;<\/span> df<span class=\"token punctuation\">.<\/span>withColumn<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;path_parts&#034;<\/span><span class=\"token punctuation\">,<\/span> paths_udf<span class=\"token punctuation\">(<\/span>col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;url&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span> \\\\<br \/>\n                <span class=\"token punctuation\">.<\/span>withColumn<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;path_level&#034;<\/span><span class=\"token punctuation\">,<\/span> explode<span class=\"token punctuation\">(<\/span>col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;path_parts&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;URL\u62c6\u89e3\u540e\u7684\u8def\u5f84\u7ec4\u4ef6:&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\ndf_exploded<span class=\"token punctuation\">.<\/span>select<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;url&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;path_level&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>show<span class=\"token punctuation\">(<\/span><span class=\"token number\">10<\/span><span class=\"token punctuation\">,<\/span> truncate<span class=\"token operator\">&#061;<\/span><span class=\"token boolean\">False<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u4f7f\u7528Pandas UDF &#043; mapInPandas \u5b9e\u73b0\u66f4\u590d\u6742\u7684UDTF&#xff08;\u53ef\u4ee5\u8f93\u51fa\u591a\u884c\u591a\u5217&#xff09;<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">url_to_rows<\/span><span class=\"token punctuation\">(<\/span>iterator<span class=\"token punctuation\">:<\/span> Iterator<span class=\"token punctuation\">[<\/span>pd<span class=\"token punctuation\">.<\/span>DataFrame<span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span><span class=\"token operator\">&gt;<\/span> Iterator<span class=\"token punctuation\">[<\/span>pd<span class=\"token punctuation\">.<\/span>DataFrame<span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">for<\/span> pdf <span class=\"token keyword\">in<\/span> iterator<span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token comment\"># \u5bf9\u6bcf\u4e2aDataFrame&#xff08;\u5206\u533a&#xff09;\u9010\u884c\u5904\u7406<\/span><br \/>\n        rows <span class=\"token operator\">&#061;<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token punctuation\">]<\/span><br \/>\n        <span class=\"token keyword\">for<\/span> _<span class=\"token punctuation\">,<\/span> row <span class=\"token keyword\">in<\/span> pdf<span class=\"token punctuation\">.<\/span>iterrows<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n            url <span class=\"token operator\">&#061;<\/span> row<span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;url&#034;<\/span><span class=\"token punctuation\">]<\/span><br \/>\n            user_id <span class=\"token operator\">&#061;<\/span> row<span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;user_id&#034;<\/span><span class=\"token punctuation\">]<\/span><br \/>\n            <span class=\"token keyword\">if<\/span> url<span class=\"token punctuation\">:<\/span><br \/>\n                path <span class=\"token operator\">&#061;<\/span> url<span class=\"token punctuation\">.<\/span>split<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#039;?&#039;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">[<\/span><span class=\"token number\">0<\/span><span class=\"token punctuation\">]<\/span><br \/>\n                <span class=\"token keyword\">for<\/span> level<span class=\"token punctuation\">,<\/span> part <span class=\"token keyword\">in<\/span> <span class=\"token builtin\">enumerate<\/span><span class=\"token punctuation\">(<\/span>path<span class=\"token punctuation\">.<\/span>split<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#039;\/&#039;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n                    <span class=\"token keyword\">if<\/span> part<span class=\"token punctuation\">:<\/span><br \/>\n                        rows<span class=\"token punctuation\">.<\/span>append<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">{<\/span><span class=\"token string\">&#034;user_id&#034;<\/span><span class=\"token punctuation\">:<\/span> user_id<span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;url&#034;<\/span><span class=\"token punctuation\">:<\/span> url<span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;level&#034;<\/span><span class=\"token punctuation\">:<\/span> level<span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;part&#034;<\/span><span class=\"token punctuation\">:<\/span> part<span class=\"token punctuation\">}<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        <span class=\"token keyword\">yield<\/span> pd<span class=\"token punctuation\">.<\/span>DataFrame<span class=\"token punctuation\">(<\/span>rows<span class=\"token punctuation\">)<\/span><\/p>\n<p>schema <span class=\"token operator\">&#061;<\/span> StructType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">[<\/span><br \/>\n    StructField<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;user_id&#034;<\/span><span class=\"token punctuation\">,<\/span> StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span><br \/>\n    StructField<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;url&#034;<\/span><span class=\"token punctuation\">,<\/span> StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span><br \/>\n    StructField<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;level&#034;<\/span><span class=\"token punctuation\">,<\/span> IntegerType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span><br \/>\n    StructField<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;part&#034;<\/span><span class=\"token punctuation\">,<\/span> StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>df_udtf <span class=\"token operator\">&#061;<\/span> df<span class=\"token punctuation\">.<\/span>select<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;user_id&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;url&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>mapInPandas<span class=\"token punctuation\">(<\/span>url_to_rows<span class=\"token punctuation\">,<\/span> schema<span class=\"token operator\">&#061;<\/span>schema<span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;mapInPandas UDTF\u7ed3\u679c:&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\ndf_udtf<span class=\"token punctuation\">.<\/span>show<span class=\"token punctuation\">(<\/span><span class=\"token number\">10<\/span><span class=\"token punctuation\">,<\/span> truncate<span class=\"token operator\">&#061;<\/span><span class=\"token boolean\">False<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061; 7. \u590d\u6742Pandas UDF&#xff1a;\u8c03\u7528\u5916\u90e8\u6a21\u578b&#xff08;\u6a21\u62df&#xff09; &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\\\\n&#034;<\/span> <span class=\"token operator\">&#043;<\/span> <span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\u6b65\u9aa45: \u590d\u6742Pandas UDF &#8211; \u8bc4\u5206\u6a21\u578b\u9884\u6d4b&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u6a21\u62df\u4e00\u4e2a\u9884\u5148\u8bad\u7ec3\u597d\u7684\u6a21\u578b&#xff08;\u7b80\u5355\u7ebf\u6027\u56de\u5f52\u7cfb\u6570&#xff09;<\/span><br \/>\n<span class=\"token comment\"># \u5047\u8bbe\u6211\u4eec\u6709\u4e00\u4e2a\u7279\u5f81\u5411\u91cf [\u6d4f\u89c8\u65f6\u957f, \u70b9\u51fb\u6b21\u6570, \u9875\u9762\u6df1\u5ea6] \u9884\u6d4b\u7528\u6237\u8bc4\u5206<\/span><br \/>\n<span class=\"token comment\"># \u7279\u5f81\u6570\u636e\u7531url\u957f\u5ea6\u3001\u70b9\u51fb\u6570\u7b49\u6a21\u62df<\/span><\/p>\n<p><span class=\"token comment\"># \u751f\u6210\u6a21\u62df\u7279\u5f81\u6570\u636e<\/span><br \/>\nfeature_df <span class=\"token operator\">&#061;<\/span> df<span class=\"token punctuation\">.<\/span>withColumn<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;url_length&#034;<\/span><span class=\"token punctuation\">,<\/span> length<span class=\"token punctuation\">(<\/span>col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;url&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span> \\\\<br \/>\n               <span class=\"token punctuation\">.<\/span>withColumn<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;hour&#034;<\/span><span class=\"token punctuation\">,<\/span> hour<span class=\"token punctuation\">(<\/span>col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;event_time&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span> \\\\<br \/>\n               <span class=\"token punctuation\">.<\/span>withColumn<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;is_api&#034;<\/span><span class=\"token punctuation\">,<\/span> col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;url&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>contains<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;api&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u5b9a\u4e49\u6a21\u578b\u9884\u6d4b\u51fd\u6570&#xff08;\u5411\u91cf\u5316&#xff09;<\/span><br \/>\ncoef <span class=\"token operator\">&#061;<\/span> np<span class=\"token punctuation\">.<\/span>array<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">[<\/span><span class=\"token number\">0.1<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token operator\">&#8211;<\/span><span class=\"token number\">0.05<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">0.3<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span>  <span class=\"token comment\"># \u6743\u91cd<\/span><\/p>\n<p><span class=\"token decorator annotation punctuation\">&#064;pandas_udf<\/span><span class=\"token punctuation\">(<\/span>DoubleType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">predict_score<\/span><span class=\"token punctuation\">(<\/span>url_length<span class=\"token punctuation\">:<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">,<\/span> hour<span class=\"token punctuation\">:<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">,<\/span> is_api<span class=\"token punctuation\">:<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span><span class=\"token operator\">&gt;<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token comment\"># \u6784\u5efa\u7279\u5f81\u77e9\u9635<\/span><br \/>\n    features <span class=\"token operator\">&#061;<\/span> pd<span class=\"token punctuation\">.<\/span>DataFrame<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">{<\/span><br \/>\n        <span class=\"token string\">&#034;len&#034;<\/span><span class=\"token punctuation\">:<\/span> url_length<span class=\"token punctuation\">,<\/span><br \/>\n        <span class=\"token string\">&#034;hour&#034;<\/span><span class=\"token punctuation\">:<\/span> hour<span class=\"token punctuation\">,<\/span><br \/>\n        <span class=\"token string\">&#034;is_api&#034;<\/span><span class=\"token punctuation\">:<\/span> is_api<span class=\"token punctuation\">.<\/span>astype<span class=\"token punctuation\">(<\/span><span class=\"token builtin\">int<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token punctuation\">}<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>values<br \/>\n    <span class=\"token comment\"># \u9884\u6d4b<\/span><br \/>\n    scores <span class=\"token operator\">&#061;<\/span> features<span class=\"token punctuation\">.<\/span>dot<span class=\"token punctuation\">(<\/span>coef<span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#043;<\/span> <span class=\"token number\">0.5<\/span>  <span class=\"token comment\"># \u52a0\u504f\u7f6e<\/span><br \/>\n    <span class=\"token comment\"># \u9650\u5236\u8303\u56f4[0,1]<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">(<\/span>np<span class=\"token punctuation\">.<\/span>clip<span class=\"token punctuation\">(<\/span>scores<span class=\"token punctuation\">,<\/span> <span class=\"token number\">0<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">1<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>df_with_score <span class=\"token operator\">&#061;<\/span> feature_df<span class=\"token punctuation\">.<\/span>withColumn<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;score&#034;<\/span><span class=\"token punctuation\">,<\/span> predict_score<span class=\"token punctuation\">(<\/span>col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;url_length&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span> col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;hour&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span> col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;is_api&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\u8bc4\u5206\u7ed3\u679c\u793a\u4f8b:&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\ndf_with_score<span class=\"token punctuation\">.<\/span>select<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;url&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;url_length&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;hour&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;is_api&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;score&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>show<span class=\"token punctuation\">(<\/span><span class=\"token number\">10<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061; 8. \u8fed\u4ee3\u5668Pandas UDF\u5904\u7406\u8d85\u5927\u5206\u533a &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\\\\n&#034;<\/span> <span class=\"token operator\">&#043;<\/span> <span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\u6b65\u9aa46: \u8fed\u4ee3\u5668Pandas UDF&#xff08;\u5206\u6279\u6b21\u5904\u7406&#xff09;&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u573a\u666f&#xff1a;\u9700\u8981\u9010\u884c\u8c03\u7528\u5916\u90e8API&#xff0c;\u4f46\u5185\u5b58\u6709\u9650&#xff0c;\u5206\u6279\u6b21\u53d1\u9001<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">call_external_api<\/span><span class=\"token punctuation\">(<\/span>iterator<span class=\"token punctuation\">:<\/span> Iterator<span class=\"token punctuation\">[<\/span>pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span><span class=\"token operator\">&gt;<\/span> Iterator<span class=\"token punctuation\">[<\/span>pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">for<\/span> batch <span class=\"token keyword\">in<\/span> iterator<span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token comment\"># batch\u662fpd.Series&#xff0c;\u5047\u8bbe\u662furl\u5217<\/span><br \/>\n        results <span class=\"token operator\">&#061;<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token punctuation\">]<\/span><br \/>\n        <span class=\"token keyword\">for<\/span> url <span class=\"token keyword\">in<\/span> batch<span class=\"token punctuation\">:<\/span><br \/>\n            <span class=\"token comment\"># \u6a21\u62dfAPI\u8c03\u7528<\/span><br \/>\n            result <span class=\"token operator\">&#061;<\/span> <span class=\"token builtin\">len<\/span><span class=\"token punctuation\">(<\/span>url<span class=\"token punctuation\">)<\/span>  <span class=\"token comment\"># \u7b80\u5355\u6a21\u62df<\/span><br \/>\n            results<span class=\"token punctuation\">.<\/span>append<span class=\"token punctuation\">(<\/span>result<span class=\"token punctuation\">)<\/span><br \/>\n        <span class=\"token keyword\">yield<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">(<\/span>results<span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u6ce8\u518c\u8fed\u4ee3\u5668UDF<\/span><br \/>\n<span class=\"token decorator annotation punctuation\">&#064;pandas_udf<\/span><span class=\"token punctuation\">(<\/span>IntegerType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span> functionType<span class=\"token operator\">&#061;<\/span><span class=\"token string\">&#034;SCALAR_ITER&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">api_udf<\/span><span class=\"token punctuation\">(<\/span>iterator<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">for<\/span> batch <span class=\"token keyword\">in<\/span> iterator<span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token keyword\">yield<\/span> batch<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">str<\/span><span class=\"token punctuation\">.<\/span><span class=\"token builtin\">len<\/span><span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span>  <span class=\"token comment\"># \u5411\u91cf\u5316\u8ba1\u7b97\u957f\u5ea6&#xff0c;\u6f14\u793a<\/span><\/p>\n<p><span class=\"token comment\"># \u6216\u8005\u4f7f\u7528mapInPandas<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">api_map<\/span><span class=\"token punctuation\">(<\/span>iterator<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">for<\/span> pdf <span class=\"token keyword\">in<\/span> iterator<span class=\"token punctuation\">:<\/span><br \/>\n        pdf<span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;url_len&#034;<\/span><span class=\"token punctuation\">]<\/span> <span class=\"token operator\">&#061;<\/span> pdf<span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;url&#034;<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">.<\/span><span class=\"token builtin\">str<\/span><span class=\"token punctuation\">.<\/span><span class=\"token builtin\">len<\/span><span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        <span class=\"token keyword\">yield<\/span> pdf<span class=\"token punctuation\">[<\/span><span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;user_id&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;url_len&#034;<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">]<\/span><\/p>\n<p>df_len <span class=\"token operator\">&#061;<\/span> df<span class=\"token punctuation\">.<\/span>select<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;user_id&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;url&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>mapInPandas<span class=\"token punctuation\">(<\/span>api_map<span class=\"token punctuation\">,<\/span> schema<span class=\"token operator\">&#061;<\/span><span class=\"token string\">&#034;user_id string, url_len int&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\ndf_len<span class=\"token punctuation\">.<\/span>show<span class=\"token punctuation\">(<\/span><span class=\"token number\">5<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061; 9. \u6ce8\u518cUDF\u4f9bSQL\u4f7f\u7528 &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\\\\n&#034;<\/span> <span class=\"token operator\">&#043;<\/span> <span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\u6b65\u9aa47: \u6ce8\u518cUDF\u5230SQL\u5f15\u64ce&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u6ce8\u518c\u4f20\u7edfUDF<\/span><br \/>\nspark<span class=\"token punctuation\">.<\/span>udf<span class=\"token punctuation\">.<\/span>register<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;region_udf_sql&#034;<\/span><span class=\"token punctuation\">,<\/span> ip_to_region<span class=\"token punctuation\">,<\/span> StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token comment\"># \u6ce8\u518cPandas UDF\u5230SQL&#xff08;Spark 3.0&#043;\u652f\u6301&#xff09;<\/span><br \/>\nspark<span class=\"token punctuation\">.<\/span>udf<span class=\"token punctuation\">.<\/span>register<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;region_pandas_sql&#034;<\/span><span class=\"token punctuation\">,<\/span> ip_to_region_pandas<span class=\"token punctuation\">,<\/span> StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>df<span class=\"token punctuation\">.<\/span>createOrReplaceTempView<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;logs&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\nsql_result <span class=\"token operator\">&#061;<\/span> spark<span class=\"token punctuation\">.<\/span>sql<span class=\"token punctuation\">(<\/span><span class=\"token triple-quoted-string string\">&#034;&#034;&#034;<br \/>\n    SELECT ip, region_udf_sql(ip) as region_traditional, region_pandas_sql(ip) as region_pandas<br \/>\n    FROM logs<br \/>\n    LIMIT 10<br \/>\n&#034;&#034;&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;SQL\u4e2d\u4f7f\u7528UDF:&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\nsql_result<span class=\"token punctuation\">.<\/span>show<span class=\"token punctuation\">(<\/span>truncate<span class=\"token operator\">&#061;<\/span><span class=\"token boolean\">False<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061; 10. \u6027\u80fd\u5bf9\u6bd4\u6c47\u603b &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\\\\n&#034;<\/span> <span class=\"token operator\">&#043;<\/span> <span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\u6b65\u9aa48: \u6027\u80fd\u5bf9\u6bd4\u603b\u7ed3&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;&#061;&#034;<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">80<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u91cd\u65b0\u8ba1\u65f6\u5bf9\u6bd4<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">benchmark<\/span><span class=\"token punctuation\">(<\/span>udf_type<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">if<\/span> udf_type <span class=\"token operator\">&#061;&#061;<\/span> <span class=\"token string\">&#034;traditional&#034;<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        df<span class=\"token punctuation\">.<\/span>withColumn<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;region&#034;<\/span><span class=\"token punctuation\">,<\/span> region_udf<span class=\"token punctuation\">(<\/span>col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;ip&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>count<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">elif<\/span> udf_type <span class=\"token operator\">&#061;&#061;<\/span> <span class=\"token string\">&#034;pandas&#034;<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        df<span class=\"token punctuation\">.<\/span>withColumn<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;region&#034;<\/span><span class=\"token punctuation\">,<\/span> ip_to_region_pandas<span class=\"token punctuation\">(<\/span>col<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;ip&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>count<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">elif<\/span> udf_type <span class=\"token operator\">&#061;&#061;<\/span> <span class=\"token string\">&#034;builtin&#034;<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token comment\"># \u6a21\u62df\u5185\u7f6e\u51fd\u6570\u66ff\u4ee3&#xff08;\u5b9e\u9645\u4e0a\u4e0d\u80fd\u7528\u5185\u7f6e\u51fd\u6570\u76f4\u63a5\u6620\u5c04IP&#xff0c;\u4ec5\u793a\u610f&#xff09;<\/span><br \/>\n        df<span class=\"token punctuation\">.<\/span>withColumn<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;region&#034;<\/span><span class=\"token punctuation\">,<\/span> lit<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;East&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>count<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token keyword\">for<\/span> t <span class=\"token keyword\">in<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;traditional&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;pandas&#034;<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    start <span class=\"token operator\">&#061;<\/span> time<span class=\"token punctuation\">.<\/span>time<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    benchmark<span class=\"token punctuation\">(<\/span>t<span class=\"token punctuation\">)<\/span><br \/>\n    elapsed <span class=\"token operator\">&#061;<\/span> time<span class=\"token punctuation\">.<\/span>time<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span> start<br \/>\n    <span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#034;<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>t<span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\"> UDF \u8017\u65f6: <\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>elapsed<span class=\"token punctuation\">:<\/span><span class=\"token format-spec\">.2f<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">\u79d2&#034;<\/span><\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\\\\n\u6027\u80fd\u603b\u7ed3: Pandas UDF&#xff08;\u5411\u91cf\u5316&#xff09;\u6bd4\u4f20\u7edfUDF\u5feb3-10\u500d&#xff0c;\u63a5\u8fd1\u5185\u7f6e\u51fd\u6570\u3002&#034;<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061; 11. \u6e05\u7406 &#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;&#061;<\/span><br \/>\nspark<span class=\"token punctuation\">.<\/span>stop<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\\\\n\u2705 \u81ea\u5b9a\u4e49\u51fd\u6570\u5b9e\u6218\u6f14\u793a\u5b8c\u6210&#034;<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<h3>\u516b\u3001\u4ee3\u7801\u9010\u884c\u89e3\u6790<\/h3>\n<h4>8.1 \u4f20\u7edfUDF\u5b9a\u4e49<\/h4>\n<p><span class=\"token decorator annotation punctuation\">&#064;udf<\/span><span class=\"token punctuation\">(<\/span>returnType<span class=\"token operator\">&#061;<\/span>StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">ip_to_region<\/span><span class=\"token punctuation\">(<\/span>ip<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">if<\/span> <span class=\"token keyword\">not<\/span> ip<span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token keyword\">return<\/span> <span class=\"token string\">&#034;Unknown&#034;<\/span><br \/>\n    <span class=\"token comment\"># \u903b\u8f91&#8230;<\/span><\/p>\n<ul>\n<li>\u88c5\u9970\u5668&#064;udf\u9700\u8981\u6307\u5b9a\u8fd4\u56de\u7c7b\u578b&#xff08;\u5426\u5219Spark\u5c1d\u8bd5\u63a8\u65ad&#xff0c;\u53ef\u80fd\u51fa\u9519&#xff09;\u3002<\/li>\n<li>\u51fd\u6570\u5185\u90e8\u5fc5\u987b\u5904\u7406\u7a7a\u503c\u3002<\/li>\n<\/ul>\n<h4>8.2 Pandas UDF\u7c7b\u578b\u63d0\u793a\u98ce\u683c<\/h4>\n<p><span class=\"token decorator annotation punctuation\">&#064;pandas_udf<\/span><span class=\"token punctuation\">(<\/span>StringType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">ip_to_region_pandas<\/span><span class=\"token punctuation\">(<\/span>ip_series<span class=\"token punctuation\">:<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span><span class=\"token operator\">&gt;<\/span> pd<span class=\"token punctuation\">.<\/span>Series<span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> ip_series<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">str<\/span><span class=\"token punctuation\">.<\/span>split<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#039;.&#039;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span><span class=\"token builtin\">str<\/span><span class=\"token punctuation\">[<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">.<\/span>astype<span class=\"token punctuation\">(<\/span><span class=\"token builtin\">int<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span><span class=\"token builtin\">apply<\/span><span class=\"token punctuation\">(<\/span><span class=\"token keyword\">lambda<\/span> x<span class=\"token punctuation\">:<\/span> <span class=\"token punctuation\">.<\/span><span class=\"token punctuation\">.<\/span><span class=\"token punctuation\">.<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<ul>\n<li>\u4f7f\u7528Python\u7c7b\u578b\u63d0\u793a\u66f4\u52a0\u6e05\u6670\u3002<\/li>\n<li>\u5229\u7528Pandas\u7684\u5b57\u7b26\u4e32\u5411\u91cf\u5316\u65b9\u6cd5&#xff08;.str&#xff09;\u63d0\u9ad8\u6027\u80fd\u3002<\/li>\n<\/ul>\n<h4>8.3 \u5206\u7ec4\u805a\u5408&#xff08;GROUPED_MAP&#xff09;<\/h4>\n<p><span class=\"token decorator annotation punctuation\">&#064;pandas_udf<\/span><span class=\"token punctuation\">(<\/span>returnType<span class=\"token operator\">&#061;<\/span>StructType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">[<\/span><span class=\"token punctuation\">.<\/span><span class=\"token punctuation\">.<\/span><span class=\"token punctuation\">.<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span> functionType<span class=\"token operator\">&#061;<\/span><span class=\"token string\">&#034;GROUPED_MAP&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">session_analysis<\/span><span class=\"token punctuation\">(<\/span>pdf<span class=\"token punctuation\">:<\/span> pd<span class=\"token punctuation\">.<\/span>DataFrame<span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span><span class=\"token operator\">&gt;<\/span> pd<span class=\"token punctuation\">.<\/span>DataFrame<span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token comment\"># \u5904\u7406\u4e00\u7ec4\u6570\u636e<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> pd<span class=\"token punctuation\">.<\/span>DataFrame<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">{<\/span><span class=\"token string\">&#034;col&#034;<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token punctuation\">[<\/span>value<span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">}<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<ul>\n<li>\u63a5\u53d7\u4e00\u4e2a\u5206\u7ec4\u7684\u5168\u90e8\u6570\u636e\u4f5c\u4e3aPandas DataFrame&#xff0c;\u8f93\u51fa\u4e00\u4e2aDataFrame&#xff08;\u53ef\u4ee5\u662f\u591a\u884c&#xff09;\u3002<\/li>\n<li>\u4e0egroupBy().apply()\u914d\u5408\u4f7f\u7528&#xff0c;\u6bcf\u4e2a\u5206\u7ec4\u8c03\u7528\u4e00\u6b21\u3002<\/li>\n<\/ul>\n<h4>8.4 UDTF\u6a21\u62df&#xff1a;mapInPandas<\/h4>\n<p><span class=\"token keyword\">def<\/span> <span class=\"token function\">url_to_rows<\/span><span class=\"token punctuation\">(<\/span>iterator<span class=\"token punctuation\">:<\/span> Iterator<span class=\"token punctuation\">[<\/span>pd<span class=\"token punctuation\">.<\/span>DataFrame<span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#8211;<\/span><span class=\"token operator\">&gt;<\/span> Iterator<span class=\"token punctuation\">[<\/span>pd<span class=\"token punctuation\">.<\/span>DataFrame<span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">for<\/span> pdf <span class=\"token keyword\">in<\/span> iterator<span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token comment\"># \u5904\u7406\u5e76yield\u65b0\u7684DataFrame<\/span><br \/>\n        <span class=\"token keyword\">yield<\/span> new_pdf<br \/>\ndf<span class=\"token punctuation\">.<\/span>mapInPandas<span class=\"token punctuation\">(<\/span>url_to_rows<span class=\"token punctuation\">,<\/span> schema<span class=\"token punctuation\">)<\/span><\/p>\n<ul>\n<li>\u66f4\u7075\u6d3b&#xff0c;\u53ef\u4ee5\u8f93\u51fa\u4efb\u610f\u884c\u6570\u548c\u5217\u6570\u3002<\/li>\n<\/ul>\n<h4>8.5 \u8fed\u4ee3\u5668UDF&#xff08;SCALAR_ITER&#xff09;<\/h4>\n<p><span class=\"token decorator annotation punctuation\">&#064;pandas_udf<\/span><span class=\"token punctuation\">(<\/span>IntegerType<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span> functionType<span class=\"token operator\">&#061;<\/span><span class=\"token string\">&#034;SCALAR_ITER&#034;<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">def<\/span> <span class=\"token function\">api_udf<\/span><span class=\"token punctuation\">(<\/span>iterator<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">for<\/span> batch <span class=\"token keyword\">in<\/span> iterator<span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token keyword\">yield<\/span> batch<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">str<\/span><span class=\"token punctuation\">.<\/span><span class=\"token builtin\">len<\/span><span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<ul>\n<li>\u6bcf\u4e2a\u6279\u6b21\u4e00\u4e2apd.Series&#xff0c;\u907f\u514d\u4e00\u6b21\u6027\u52a0\u8f7d\u6574\u4e2a\u5206\u533a\u3002<\/li>\n<\/ul>\n<h3>\u4e5d\u3001\u4e1a\u52a1\u573a\u666f\u843d\u5730\u5e94\u7528<\/h3>\n<h4>9.1 \u573a\u666f\u4e00&#xff1a;\u5b9e\u65f6\u65e5\u5fd7\u4e2d\u7684IP\u5730\u7406\u4f4d\u7f6e\u89e3\u6790<\/h4>\n<p>\u5b9e\u9645\u751f\u4ea7\u4f7f\u7528GeoIP\u6570\u636e\u5e93&#xff08;\u5982MaxMind&#xff09;&#xff0c;\u901a\u8fc7Pandas UDF\u6279\u91cf\u67e5\u8be2\u5185\u5b58\u4e2d\u7684IP\u7d22\u5f15&#xff0c;\u5927\u5e45\u63d0\u5347\u541e\u5410\u91cf\u3002<\/p>\n<h4>9.2 \u573a\u666f\u4e8c&#xff1a;\u7528\u6237\u884c\u4e3a\u5e8f\u5217\u7684\u7279\u5f81\u5de5\u7a0b<\/h4>\n<p>\u5bf9\u4e8e\u6bcf\u4e2a\u7528\u6237\u7684\u4e8b\u4ef6\u5e8f\u5217&#xff0c;\u9700\u8981\u8ba1\u7b97\u5e8f\u5217\u7edf\u8ba1\u7279\u5f81&#xff08;\u5982\u5e73\u5747\u95f4\u9694\u3001\u5e8f\u5217\u71b5\u7b49&#xff09;\u3002\u4f7f\u7528groupBy().applyInPandas&#xff0c;\u5c06\u6bcf\u4e2a\u7528\u6237\u7684\u6570\u636e\u52a0\u8f7d\u5230Pandas DataFrame&#xff0c;\u5229\u7528Python\u7684\u7edf\u8ba1\u5e93\u8ba1\u7b97\u590d\u6742\u7279\u5f81\u3002<\/p>\n<h4>9.3 \u573a\u666f\u4e09&#xff1a;\u8c03\u7528\u5916\u90e8\u9884\u6d4b\u670d\u52a1<\/h4>\n<p>\u5728\u6d41\u5f0f\u8ba1\u7b97\u4e2d&#xff0c;\u6bcf\u6761\u6570\u636e\u9700\u8981\u8c03\u7528\u5916\u90e8\u6a21\u578b\u670d\u52a1&#xff08;HTTP&#xff09;\u3002\u53ef\u4ee5\u4f7f\u7528mapInPandas\u6279\u91cf\u805a\u5408\u8bf7\u6c42&#xff0c;\u51cf\u5c11\u7f51\u7edc\u5f00\u9500\u3002<\/p>\n<h4>9.4 \u573a\u666f\u56db&#xff1a;\u81ea\u5b9a\u4e49\u805a\u5408\u7a97\u53e3\u51fd\u6570<\/h4>\n<p>Spark\u5185\u7f6e\u7a97\u53e3\u51fd\u6570\u4e0d\u652f\u6301\u81ea\u5b9a\u4e49\u590d\u6742\u805a\u5408&#xff08;\u5982\u4e2d\u4f4d\u6570\u3001\u6a21\u5f0f&#xff09;&#xff0c;\u53ef\u4ee5\u4f7f\u7528Pandas UDF\u914d\u5408groupBy\u5b9e\u73b0\u3002<\/p>\n<h3>\u5341\u3001\u5e38\u89c1\u62a5\u9519\u6392\u67e5<\/h3>\n<h4>10.1 AttributeError: &#039;NoneType&#039; object has no attribute&#8230;<\/h4>\n<p>\u539f\u56e0&#xff1a;UDF\u4e2d\u6ca1\u6709\u5904\u7406\u8f93\u5165\u4e3aNone\u7684\u60c5\u51b5\u3002<\/p>\n<p>\u89e3\u51b3&#xff1a;\u5728\u51fd\u6570\u5f00\u59cb\u68c0\u67e5if s is None: return None\u3002<\/p>\n<h4>10.2 ArrowTypeError: Expected a string, got &#8230;<\/h4>\n<p>\u539f\u56e0&#xff1a;Pandas UDF\u8fd4\u56de\u7684Series\u7c7b\u578b\u4e0e\u58f0\u660e\u7684Spark\u7c7b\u578b\u4e0d\u5339\u914d\u3002<\/p>\n<p>\u89e3\u51b3&#xff1a;\u786e\u4fdd\u8fd4\u56de\u7684Series dtype\u5bf9\u5e94\u6b63\u786e&#xff0c;\u5982\u8fd4\u56depd.Series\u7684dtype\u4e3aobject\u6216string&#xff0c;\u4e0eStringType\u517c\u5bb9\u3002<\/p>\n<h4>10.3 Py4JError: An error occurred while calling o123.__getstate__<\/h4>\n<p>\u539f\u56e0&#xff1a;UDF\u4e2d\u6355\u83b7\u4e86\u4e0d\u53ef\u5e8f\u5217\u5316\u7684\u5bf9\u8c61&#xff08;\u5982\u7f51\u7edc\u8fde\u63a5&#xff09;\u3002<\/p>\n<p>\u89e3\u51b3&#xff1a;\u5728\u51fd\u6570\u5185\u90e8\u521b\u5efa\u8fde\u63a5&#xff0c;\u6216\u4f7f\u7528\u5e7f\u64ad\u53d8\u91cf\u3002<\/p>\n<h4>10.4 Pandas UDF\u6027\u80fd\u5dee&#xff08;\u4e0e\u5185\u7f6e\u51fd\u6570\u5dee\u8ddd\u5927&#xff09;<\/h4>\n<p>\u539f\u56e0&#xff1a;Arrow\u672a\u542f\u7528&#xff0c;\u6216\u6570\u636e\u91cf\u592a\u5c0f\u6279\u6b21\u6548\u679c\u4e0d\u660e\u663e&#xff0c;\u6216Pandas\u64cd\u4f5c\u975e\u5411\u91cf\u5316\u3002<\/p>\n<p>\u89e3\u51b3&#xff1a;\u786e\u8ba4spark.sql.execution.arrow.pyspark.enabled&#061;true&#xff0c;\u4e14Pandas\u64cd\u4f5c\u4f7f\u7528\u5411\u91cf\u5316\u65b9\u6cd5&#xff08;\u5982.str\u3001.dt&#xff09;&#xff0c;\u907f\u514dapply\u5faa\u73af\u3002<\/p>\n<h4>10.5 \u5206\u7ec4\u805a\u5408UDF\u5185\u5b58\u6ea2\u51fa<\/h4>\n<p>\u539f\u56e0&#xff1a;\u5355\u4e2a\u5206\u7ec4\u7684\u6570\u636e\u91cf\u8fc7\u5927&#xff0c;\u88c5\u5165Pandas DataFrame\u5bfc\u81f4OOM\u3002<\/p>\n<p>\u89e3\u51b3&#xff1a;\u5bf9\u5206\u7ec4\u5185\u6570\u636e\u91c7\u6837\u6216\u9650\u5236\u5927\u5c0f&#xff1b;\u4f7f\u7528mapInPandas\u9010\u6b65\u5904\u7406\u3002<\/p>\n<h3>\u5341\u4e00\u3001\u672c\u8282\u8bfe\u77e5\u8bc6\u70b9\u603b\u7ed3<\/h3>\n<h4>UDF\u7c7b\u578b\u5bf9\u6bd4<\/h4>\n<table>\n<tr>\u7c7b\u578b\u4f18\u70b9\u7f3a\u70b9\u9002\u7528\u573a\u666f<\/tr>\n<tbody>\n<tr>\n<td>\u4f20\u7edfUDF<\/td>\n<td>\u7b80\u5355\u3001\u7075\u6d3b<\/td>\n<td>\u6162\u3001\u5e8f\u5217\u5316\u5f00\u9500\u5927<\/td>\n<td>\u7b80\u5355\u903b\u8f91\u3001\u4f4e\u9891\u8c03\u7528<\/td>\n<\/tr>\n<tr>\n<td>Pandas UDF (SCALAR)<\/td>\n<td>\u6279\u91cf\u5904\u7406\u3001\u901f\u5ea6\u5feb<\/td>\n<td>\u9700\u719f\u6089Pandas<\/td>\n<td>\u590d\u6742\u8f6c\u6362\u3001\u8c03\u7528\u7b2c\u4e09\u65b9\u5e93<\/td>\n<\/tr>\n<tr>\n<td>Pandas UDF (GROUPED_MAP)<\/td>\n<td>\u5904\u7406\u6574\u4e2a\u5206\u7ec4<\/td>\n<td>\u5206\u7ec4\u53ef\u80fd\u503e\u659c<\/td>\n<td>\u5206\u7ec4\u5185\u590d\u6742\u53d8\u6362<\/td>\n<\/tr>\n<tr>\n<td>Pandas UDF (GROUPED_AGG)<\/td>\n<td>\u81ea\u5b9a\u4e49\u805a\u5408<\/td>\n<td>\u53ea\u80fd\u8fd4\u56de\u6807\u91cf<\/td>\n<td>\u81ea\u5b9a\u4e49\u805a\u5408\u51fd\u6570<\/td>\n<\/tr>\n<tr>\n<td>mapInPandas<\/td>\n<td>\u6700\u7075\u6d3b&#xff08;UDTF&#xff09;<\/td>\n<td>\u9700\u8981\u624b\u52a8\u5904\u7406Schema<\/td>\n<td>\u4e00\u884c\u8f93\u51fa\u591a\u884c\u591a\u5217<\/td>\n<\/tr>\n<tr>\n<td>\u8fed\u4ee3\u5668UDF<\/td>\n<td>\u8282\u7701\u5185\u5b58<\/td>\n<td>\u7a0d\u590d\u6742<\/td>\n<td>\u8d85\u5927\u5206\u533a\u9010\u6279\u5904\u7406<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h4>\u9009\u578b\u5efa\u8bae<\/h4>\n<ul>\n<li>\u80fd\u7528\u5185\u7f6e\u51fd\u6570\u7edd\u4e0d\u7528UDF<\/li>\n<li>\u5fc5\u987b\u7528UDF\u65f6&#xff0c;\u4f18\u5148\u9009\u62e9Pandas UDF<\/li>\n<li>\u5982\u679c\u6570\u636e\u91cf\u5927\u4e14\u5206\u7ec4\u5747\u5300&#xff0c;\u53ef\u7528GROUPED_MAP<\/li>\n<li>\u5982\u679c\u8f93\u51fa\u591a\u884c&#xff0c;\u4f7f\u7528mapInPandas\u6216explode&#043;UDF<\/li>\n<\/ul>\n<h4>\u6027\u80fd\u4f18\u5316\u53c2\u6570<\/h4>\n<table>\n<tr>\u53c2\u6570\u9ed8\u8ba4\u503c\u8bf4\u660e<\/tr>\n<tbody>\n<tr>\n<td>spark.sql.execution.arrow.pyspark.enabled<\/td>\n<td>true<\/td>\n<td>\u542f\u7528Arrow\u52a0\u901f<\/td>\n<\/tr>\n<tr>\n<td>spark.sql.execution.arrow.maxRecordsPerBatch<\/td>\n<td>10000<\/td>\n<td>\u6bcf\u6279\u6700\u5927\u8bb0\u5f55\u6570<\/td>\n<\/tr>\n<tr>\n<td>spark.python.worker.reuse<\/td>\n<td>true<\/td>\n<td>\u590d\u7528Python\u8fdb\u7a0b<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h3>\u5341\u4e8c\u3001\u8bfe\u540e\u601d\u8003\u4f5c\u4e1a<\/h3>\n<h4>\u4f5c\u4e1a\u4e00&#xff1a;\u7406\u8bba\u7406\u89e3\u9898<\/h4>\n<li>\n<p>\u8bf7\u89e3\u91ca\u4f20\u7edfPySpark UDF\u4e3a\u4f55\u6027\u80fd\u8f83\u5dee&#xff1f;Pandas UDF\u662f\u5982\u4f55\u89e3\u51b3\u8fd9\u4e2a\u95ee\u9898\u7684&#xff1f;<\/p>\n<\/li>\n<li>\n<p>\u4ec0\u4e48\u662f\u5411\u91cf\u5316\u64cd\u4f5c&#xff1f;\u8bf7\u4e3e\u4f8b\u8bf4\u660e\u5728Pandas UDF\u4e2d\u5982\u4f55\u5b9e\u73b0\u5411\u91cf\u5316\u3002<\/p>\n<\/li>\n<li>\n<p>\u4ec0\u4e48\u60c5\u51b5\u4e0b\u5e94\u8be5\u4f7f\u7528\u8fed\u4ee3\u5668Pandas UDF\u800c\u4e0d\u662f\u666e\u901aPandas UDF&#xff1f;<\/p>\n<\/li>\n<h4>\u4f5c\u4e1a\u4e8c&#xff1a;\u4ee3\u7801\u5b9e\u8df5\u9898<\/h4>\n<li>\n<p>\u7f16\u5199\u4e00\u4e2a\u4f20\u7edfUDF\u548c\u4e00\u4e2aPandas UDF&#xff0c;\u8ba1\u7b97\u5b57\u7b26\u4e32\u5217\u4e2d\u5355\u8bcd\u6570\u91cf\u7684\u5e73\u65b9\u6839\u3002\u6bd4\u8f83\u4e24\u8005\u7684\u6027\u80fd\u5dee\u5f02&#xff08;\u4f7f\u7528100\u4e07\u6761\u6570\u636e&#xff09;\u3002<\/p>\n<\/li>\n<li>\n<p>\u4f7f\u7528groupBy().applyInPandas\u5b9e\u73b0\u81ea\u5b9a\u4e49\u805a\u5408&#xff1a;\u8ba1\u7b97\u6bcf\u4e2a\u7528\u6237\u8bbf\u95eeURL\u7684\u8def\u5f84\u6df1\u5ea6\u4e2d\u4f4d\u6570\u3002<\/p>\n<\/li>\n<li>\n<p>\u6a21\u62dfUDTF&#xff1a;\u6709\u4e00\u4e2a\u8868logs\u5305\u542bsession_id\u548cevents&#xff08;\u6570\u7ec4\u683c\u5f0f&#xff09;&#xff0c;\u5c06\u5176\u5c55\u5f00\u4e3a\u6bcf\u4e2a\u4e8b\u4ef6\u4e00\u884c&#xff0c;\u540c\u65f6\u4fdd\u7559session_id\u3002\u4f7f\u7528explode\u548cmapInPandas\u4e24\u79cd\u65b9\u5f0f\u5b9e\u73b0\u3002<\/p>\n<\/li>\n<h4>\u4f5c\u4e1a\u4e09&#xff1a;\u573a\u666f\u5e94\u7528\u9898<\/h4>\n<p>\u67d0\u98ce\u63a7\u7cfb\u7edf\u9700\u8981\u5b9e\u65f6\u5904\u7406\u4ea4\u6613\u4e8b\u4ef6&#xff0c;\u6bcf\u6761\u4ea4\u6613\u5305\u542buser_id\u3001amount\u3001timestamp\u3001location\u3002\u9700\u8981&#xff1a;<\/p>\n<ul>\n<li>\u5bf9\u4e8e\u6bcf\u4e2a\u7528\u6237&#xff0c;\u8ba1\u7b97\u8fc7\u53bb1\u5c0f\u65f6\u5185\u7684\u4ea4\u6613\u603b\u91d1\u989d&#xff08;\u6ed1\u52a8\u7a97\u53e3\u805a\u5408&#xff09;<\/li>\n<li>\u6807\u8bb0\u4ea4\u6613\u91d1\u989d\u8d85\u8fc7\u8be5\u7528\u6237\u5386\u53f2\u5e73\u5747\u91d1\u989d3\u500d\u7684\u4ea4\u6613&#xff08;\u5f02\u5e38\u68c0\u6d4b&#xff09;<\/li>\n<li>\u8c03\u7528\u5916\u90e8\u89c4\u5219\u5f15\u64ce&#xff08;HTTP\u63a5\u53e3&#xff09;\u83b7\u53d6\u98ce\u9669\u8bc4\u5206&#xff0c;\u8bc4\u5206\u7ed3\u679c\u9700\u8981\u66f4\u65b0\u56deDataFrame<\/li>\n<\/ul>\n<p>\u8bf7\u8bbe\u8ba1\u4f7f\u7528PySpark UDF\/Pandas UDF\u7684\u89e3\u51b3\u65b9\u6848&#xff0c;\u8bf4\u660e\u9009\u62e9\u7684UDF\u7c7b\u578b\u53ca\u539f\u56e0&#xff0c;\u5e76\u5199\u51fa\u5173\u952e\u4ee3\u7801\u3002<\/p>\n<h4>\u4f5c\u4e1a\u56db&#xff1a;\u62d3\u5c55\u7814\u7a76<\/h4>\n<li>\n<p>\u7814\u7a76PySpark\u4e2dpandas_udf\u7684GROUPED_MAP\u4e0eapplyInPandas\u7684\u5f02\u540c&#xff0c;\u4ee5\u53ca\u5b83\u4eec\u5728Spark SQL\u7269\u7406\u6267\u884c\u8ba1\u5212\u4e2d\u7684\u5b9e\u73b0\u3002<\/p>\n<\/li>\n<li>\n<p>\u5b66\u4e60Apache Arrow\u5185\u5b58\u683c\u5f0f&#xff0c;\u4e86\u89e3PySpark\u5982\u4f55\u5229\u7528Arrow\u8fdb\u884c\u96f6\u62f7\u8d1d\u6570\u636e\u4f20\u8f93\u3002<\/p>\n<\/li>\n<li>\n<p>\u5b9e\u73b0\u4e00\u4e2a\u81ea\u5b9a\u4e49\u805a\u5408\u51fd\u6570&#xff08;UDAF&#xff09;\u6765\u8ba1\u7b97\u51e0\u4f55\u5e73\u5747&#xff0c;\u5e76\u4f7f\u7528&#064;pandas_udf\u548c\u4f20\u7edfScala UDAF\u4e24\u79cd\u65b9\u5f0f&#xff0c;\u5bf9\u6bd4\u6027\u80fd\u548c\u6613\u7528\u6027\u3002<\/p>\n<\/li>\n<hr \/>\n<p>\u63d0\u4ea4\u65b9\u5f0f&#xff1a;\u672c\u6b21\u4f5c\u4e1a\u8981\u6c42\u63d0\u4ea4\u53ef\u8fd0\u884c\u7684\u4ee3\u7801\u548c\u6027\u80fd\u5bf9\u6bd4\u6570\u636e&#xff0c;\u4ee5\u53ca\u7406\u8bba\u9898\u7684\u6587\u5b57\u89e3\u7b54\u3002\u9f13\u52b1\u5c06UDF\u5c01\u88c5\u4e3a\u53ef\u5bfc\u5165\u7684\u6a21\u5757&#xff0c;\u63d0\u9ad8\u590d\u7528\u6027\u3002<\/p>\n<p>\u6269\u5c55\u9605\u8bfb&#xff1a;<\/p>\n<ul>\n<li>Spark\u5b98\u65b9\u6587\u6863&#xff1a;Pandas UDFs<\/li>\n<li>\u300aPySpark\u5b9e\u6218\u6307\u5357\u300b\u7b2c8\u7ae0<\/li>\n<li>Apache Arrow\u5b98\u65b9\u6587\u6863<\/li>\n<\/ul>\n<p>\u901a\u8fc7\u672c\u8282\u8bfe\u7684\u5b66\u4e60&#xff0c;\u4f60\u5df2\u7ecf\u638c\u63e1\u4e86PySpark\u81ea\u5b9a\u4e49\u51fd\u6570\u7684\u5168\u5957\u6280\u80fd&#xff0c;\u4ece\u4f20\u7edfUDF\u5230\u9ad8\u6027\u80fdPandas UDF&#xff0c;\u518d\u5230UDAF\u548cUDTF\u6a21\u62df\u3002\u8bb0\u4f4f&#xff1a;\u9996\u9009\u5185\u7f6e\u51fd\u6570&#xff0c;\u4e0d\u5f97\u5df2\u7528UDF\u65f6\u9996\u9009Pandas UDF\u3002\u4e0b\u4e00\u8282\u8bfe\u6211\u4eec\u5c06\u5b66\u4e60PySpark\u591a\u6570\u636e\u6e90\u8bfb\u5199&#xff08;CSV\/JSON\/Parquet\/Hive\/MySQL&#xff09;&#xff0c;\u8ba9Spark\u4e0e\u5916\u90e8\u5b58\u50a8\u7cfb\u7edf\u65e0\u7f1d\u96c6\u6210\u3002\u6211\u4eec\u4e0b\u8282\u8bfe\u89c1&#xff01;<\/p>\n<hr \/>\n<h2>&#x1f517;\u300a20\u8282\u8bfe PySpark \u4ece\u5165\u95e8\u5230\u7cbe\u901a\u300b\u7cfb\u5217\u8bfe\u7a0b\u5bfc\u822a<\/h2>\n<p>\u53bb\u8ba2\u9605<\/p>\n<p>&#x1f31f; \u611f\u8c22\u60a8\u8010\u5fc3\u9605\u8bfb\u5230\u8fd9\u91cc&#xff01; &#x1f4a1; \u5982\u679c\u672c\u6587\u5bf9\u60a8\u6709\u6240\u542f\u53d1\u6b22\u8fce&#xff1a; &#x1f44d; \u70b9\u8d5e&#x1f4cc; \u6536\u85cf &#x1f4e4; \u5206\u4eab\u7ed9\u66f4\u591a\u9700\u8981\u7684\u4f19\u4f34\u3002 &#x1f5e3;\ufe0f \u671f\u5f85\u5728\u8bc4\u8bba\u533a\u770b\u5230\u60a8\u7684\u60f3\u6cd5, \u5171\u540c\u8fdb\u6b65\u3002 &#x1f514; \u5173\u6ce8\u6211&#xff0c;\u6301\u7eed\u83b7\u53d6\u66f4\u591a\u5e72\u8d27\u5185\u5bb9&#xff5e; &#x1f917; \u6211\u4eec\u4e0b\u7bc7\u6587\u7ae0\u89c1&#xff5e;<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u6587\u7ae0\u76ee\u5f55 \u4e00\u3001\u8bfe\u524d\u5bfc\u8bfb\u4e8c\u3001\u5b66\u4e60\u76ee\u6807\u4e09\u3001\u6838\u5fc3\u7406\u8bba\u77e5\u8bc6\u70b9\u56db\u3001\u539f\u7406\u901a\u4fd7\u8bb2\u89e34.1 \u4f20\u7edfUDF\u7684\u201c\u9010\u884c\u4e4b\u65c5\u201d4.2 Pandas UDF\u7684\u201c\u6279\u91cf\u52a0\u901f\u201d4.3 UDAF\u7684\u81ea\u5b9a\u4e49\u805a\u54084.4 UDTF\u7684\u8868\u751f\u6210 \u4e94\u3001\u91cd\u70b9\u6982\u5ff5\u62c6\u89e35.1 \u4f20\u7edfUDF\u7684\u6ce8\u518c\u4e0e\u4f7f\u75285.2 Pandas UDF\u7684\u7c7b\u578b5.3 \u5206\u7ec4\u805a\u5408Pandas UDF5.4 \u5206\u7ec4\u53d8\u6362&#xff08;GROUPED_MAP&#xff09;5.5 \u8fed\u4ee3\u5668Pandas UDF&#xff08;\u5904\u7406\u5927\u6570\u636e\u5206\u533a&#xff09;5.6 \u6ce8\u518cUDF<\/p>\n","protected":false},"author":2,"featured_media":89331,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[10128,81],"topic":[],"class_list":["post-89332","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-server","tag-pyspark","tag-python"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v20.3 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>\u7b2c15\u8bfe\uff1aPySpark\uff5c\u81ea\u5b9a\u4e49\u51fd\u6570UDF\u3001UDAF\u3001UDTF\u5f00\u53d1\u539f\u7406\u4e0e\u4f01\u4e1a\u5b9e\u6218 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.wsisp.com\/helps\/89332.html\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"\u7b2c15\u8bfe\uff1aPySpark\uff5c\u81ea\u5b9a\u4e49\u51fd\u6570UDF\u3001UDAF\u3001UDTF\u5f00\u53d1\u539f\u7406\u4e0e\u4f01\u4e1a\u5b9e\u6218 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"og:description\" content=\"\u6587\u7ae0\u76ee\u5f55 \u4e00\u3001\u8bfe\u524d\u5bfc\u8bfb\u4e8c\u3001\u5b66\u4e60\u76ee\u6807\u4e09\u3001\u6838\u5fc3\u7406\u8bba\u77e5\u8bc6\u70b9\u56db\u3001\u539f\u7406\u901a\u4fd7\u8bb2\u89e34.1 \u4f20\u7edfUDF\u7684\u201c\u9010\u884c\u4e4b\u65c5\u201d4.2 Pandas UDF\u7684\u201c\u6279\u91cf\u52a0\u901f\u201d4.3 UDAF\u7684\u81ea\u5b9a\u4e49\u805a\u54084.4 UDTF\u7684\u8868\u751f\u6210 \u4e94\u3001\u91cd\u70b9\u6982\u5ff5\u62c6\u89e35.1 \u4f20\u7edfUDF\u7684\u6ce8\u518c\u4e0e\u4f7f\u75285.2 Pandas UDF\u7684\u7c7b\u578b5.3 \u5206\u7ec4\u805a\u5408Pandas UDF5.4 \u5206\u7ec4\u53d8\u6362&#xff08;GROUPED_MAP&#xff09;5.5 \u8fed\u4ee3\u5668Pandas UDF&#xff08;\u5904\u7406\u5927\u6570\u636e\u5206\u533a&#xff09;5.6 \u6ce8\u518cUDF\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.wsisp.com\/helps\/89332.html\" \/>\n<meta property=\"og:site_name\" content=\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"article:published_time\" content=\"2026-08-02T12:56:41+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.wsisp.com\/helps\/wp-content\/uploads\/2026\/08\/20260802125638-6a6f3e8681788.jpg\" \/>\n<meta name=\"author\" content=\"admin\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"15 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/89332.html\",\"url\":\"https:\/\/www.wsisp.com\/helps\/89332.html\",\"name\":\"\u7b2c15\u8bfe\uff1aPySpark\uff5c\u81ea\u5b9a\u4e49\u51fd\u6570UDF\u3001UDAF\u3001UDTF\u5f00\u53d1\u539f\u7406\u4e0e\u4f01\u4e1a\u5b9e\u6218 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"isPartOf\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\"},\"datePublished\":\"2026-08-02T12:56:41+00:00\",\"dateModified\":\"2026-08-02T12:56:41+00:00\",\"author\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\"},\"breadcrumb\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/89332.html#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.wsisp.com\/helps\/89332.html\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/89332.html#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/www.wsisp.com\/helps\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u7b2c15\u8bfe\uff1aPySpark\uff5c\u81ea\u5b9a\u4e49\u51fd\u6570UDF\u3001UDAF\u3001UDTF\u5f00\u53d1\u539f\u7406\u4e0e\u4f01\u4e1a\u5b9e\u6218\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\",\"url\":\"https:\/\/www.wsisp.com\/helps\/\",\"name\":\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"description\":\"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"contentUrl\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"caption\":\"admin\"},\"sameAs\":[\"http:\/\/wp.wsisp.com\"],\"url\":\"https:\/\/www.wsisp.com\/helps\/author\/admin\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u7b2c15\u8bfe\uff1aPySpark\uff5c\u81ea\u5b9a\u4e49\u51fd\u6570UDF\u3001UDAF\u3001UDTF\u5f00\u53d1\u539f\u7406\u4e0e\u4f01\u4e1a\u5b9e\u6218 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.wsisp.com\/helps\/89332.html","og_locale":"zh_CN","og_type":"article","og_title":"\u7b2c15\u8bfe\uff1aPySpark\uff5c\u81ea\u5b9a\u4e49\u51fd\u6570UDF\u3001UDAF\u3001UDTF\u5f00\u53d1\u539f\u7406\u4e0e\u4f01\u4e1a\u5b9e\u6218 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","og_description":"\u6587\u7ae0\u76ee\u5f55 \u4e00\u3001\u8bfe\u524d\u5bfc\u8bfb\u4e8c\u3001\u5b66\u4e60\u76ee\u6807\u4e09\u3001\u6838\u5fc3\u7406\u8bba\u77e5\u8bc6\u70b9\u56db\u3001\u539f\u7406\u901a\u4fd7\u8bb2\u89e34.1 \u4f20\u7edfUDF\u7684\u201c\u9010\u884c\u4e4b\u65c5\u201d4.2 Pandas UDF\u7684\u201c\u6279\u91cf\u52a0\u901f\u201d4.3 UDAF\u7684\u81ea\u5b9a\u4e49\u805a\u54084.4 UDTF\u7684\u8868\u751f\u6210 \u4e94\u3001\u91cd\u70b9\u6982\u5ff5\u62c6\u89e35.1 \u4f20\u7edfUDF\u7684\u6ce8\u518c\u4e0e\u4f7f\u75285.2 Pandas UDF\u7684\u7c7b\u578b5.3 \u5206\u7ec4\u805a\u5408Pandas UDF5.4 \u5206\u7ec4\u53d8\u6362&#xff08;GROUPED_MAP&#xff09;5.5 \u8fed\u4ee3\u5668Pandas UDF&#xff08;\u5904\u7406\u5927\u6570\u636e\u5206\u533a&#xff09;5.6 \u6ce8\u518cUDF","og_url":"https:\/\/www.wsisp.com\/helps\/89332.html","og_site_name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","article_published_time":"2026-08-02T12:56:41+00:00","og_image":[{"url":"https:\/\/www.wsisp.com\/helps\/wp-content\/uploads\/2026\/08\/20260802125638-6a6f3e8681788.jpg"}],"author":"admin","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"admin","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"15 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.wsisp.com\/helps\/89332.html","url":"https:\/\/www.wsisp.com\/helps\/89332.html","name":"\u7b2c15\u8bfe\uff1aPySpark\uff5c\u81ea\u5b9a\u4e49\u51fd\u6570UDF\u3001UDAF\u3001UDTF\u5f00\u53d1\u539f\u7406\u4e0e\u4f01\u4e1a\u5b9e\u6218 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","isPartOf":{"@id":"https:\/\/www.wsisp.com\/helps\/#website"},"datePublished":"2026-08-02T12:56:41+00:00","dateModified":"2026-08-02T12:56:41+00:00","author":{"@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41"},"breadcrumb":{"@id":"https:\/\/www.wsisp.com\/helps\/89332.html#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.wsisp.com\/helps\/89332.html"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.wsisp.com\/helps\/89332.html#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/www.wsisp.com\/helps"},{"@type":"ListItem","position":2,"name":"\u7b2c15\u8bfe\uff1aPySpark\uff5c\u81ea\u5b9a\u4e49\u51fd\u6570UDF\u3001UDAF\u3001UDTF\u5f00\u53d1\u539f\u7406\u4e0e\u4f01\u4e1a\u5b9e\u6218"}]},{"@type":"WebSite","@id":"https:\/\/www.wsisp.com\/helps\/#website","url":"https:\/\/www.wsisp.com\/helps\/","name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","description":"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41","name":"admin","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/","url":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","contentUrl":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","caption":"admin"},"sameAs":["http:\/\/wp.wsisp.com"],"url":"https:\/\/www.wsisp.com\/helps\/author\/admin"}]}},"_links":{"self":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/89332","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/comments?post=89332"}],"version-history":[{"count":0,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/89332\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media\/89331"}],"wp:attachment":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media?parent=89332"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/categories?post=89332"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/tags?post=89332"},{"taxonomy":"topic","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/topic?post=89332"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}