From 1e5ecf88745bf35c18a0c33fc2b49fe4ace4e7e8 Mon Sep 17 00:00:00 2001 From: naibo Date: Thu, 23 Nov 2023 03:14:31 +0800 Subject: [PATCH] Deal with data: --- .../EasySpider_windows_x64/execution_instances/123.json | 1 + .../EasySpider_windows_x64/execution_instances/124.json | 1 + .../EasySpider_windows_x64/execution_instances/125.json | 1 + .temp_to_pub/EasySpider_windows_x64/tasks/209.json | 2 +- ExecuteStage/easyspider_executestage.py | 6 ++++++ 5 files changed, 10 insertions(+), 1 deletion(-) create mode 100644 .temp_to_pub/EasySpider_windows_x64/execution_instances/123.json create mode 100644 .temp_to_pub/EasySpider_windows_x64/execution_instances/124.json create mode 100644 .temp_to_pub/EasySpider_windows_x64/execution_instances/125.json diff --git a/.temp_to_pub/EasySpider_windows_x64/execution_instances/123.json b/.temp_to_pub/EasySpider_windows_x64/execution_instances/123.json new file mode 100644 index 0000000..2930d04 --- /dev/null +++ b/.temp_to_pub/EasySpider_windows_x64/execution_instances/123.json @@ -0,0 +1 @@ +{"id":123,"name":"【软科排名】-中国最好学科排名|最权威的大学学科|高校学科排名","url":"https://www.shanghairanking.cn/rankings/bcsr/2023","links":"https://www.shanghairanking.cn/rankings/bcsr/2023","create_time":"2023/11/22 21:51:29","update_time":"2023/11/22 21:51:29","version":"0.5.0","saveThreshold":10,"cloudflare":0,"environment":0,"maxViewLength":15,"recordLog":1,"outputFormat":"xlsx","saveName":"current_time","inputExcel":"","startFromExit":0,"containJudge":false,"desc":"https://www.shanghairanking.cn/rankings/bcsr/2023","inputParameters":[{"id":0,"name":"urlList_0","nodeId":1,"nodeName":"打开网页","value":"https://www.shanghairanking.cn/rankings/bcsr/2023","desc":"要采集的网址列表，多行以\\n分开","type":"text","exampleValue":"https://www.shanghairanking.cn/rankings/bcsr/2023"},{"id":1,"name":"loopTimes_循环_1","nodeId":4,"nodeName":"循环","desc":"循环循环执行的次数（0代表无限循环）","type":"int","exampleValue":0,"value":0}],"outputParameters":[{"id":0,"name":"参数1_链接文本","desc":"","type":"text","recordASField":1,"exampleValue":"湖北大学\n "},{"id":1,"name":"参数3_文本","desc":"","type":"text","recordASField":1,"exampleValue":"（哲学）"}],"graph":[{"index":0,"id":0,"parentId":0,"type":-1,"option":0,"title":"root","sequence":[1,2],"parameters":{"history":1,"tabIndex":0,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0},"isInLoop":false},{"id":1,"index":1,"parentId":0,"type":0,"option":1,"title":"打开网页","sequence":[],"isInLoop":false,"position":0,"parameters":{"useLoop":false,"xpath":"","wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"url":"https://www.shanghairanking.cn/rankings/bcsr/2023","links":"https://www.shanghairanking.cn/rankings/bcsr/2023","maxWaitTime":10,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"cookies":""}},{"id":2,"index":2,"parentId":0,"type":1,"option":8,"title":"循环","sequence":[3,4],"isInLoop":false,"position":1,"parameters":{"history":4,"tabIndex":-1,"useLoop":false,"xpath":"/html/body/div[1]/div[1]/div[1]/div[2]/div[2]/div[1]/div/div[2]/div/a[1]/span[2]","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":1,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":""}},{"id":3,"index":3,"parentId":2,"type":0,"option":2,"title":"点击元素","sequence":[],"isInLoop":true,"position":0,"parameters":{"history":4,"tabIndex":-1,"useLoop":true,"xpath":"","iframe":false,"wait":2,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"clickWay":0,"maxWaitTime":10,"paras":[],"allXPaths":"","loopType":1}},{"id":4,"index":4,"parentId":2,"type":1,"option":8,"title":"循环","sequence":[6,5],"isInLoop":true,"position":1,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"//*[contains(@class, \"ant-pagination-next\")]/a[1]","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":0,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/ul[1]/li[5]/a[1]","//a[contains(., '')]","//A[@class='ant-pagination-item-link']","/html/body/div[last()-3]/div/div/div[last()-2]/div/div/div/div[last()-1]/div/ul/li[last()-1]/a"]}},{"id":6,"index":5,"parentId":4,"type":0,"option":2,"title":"点击元素","sequence":[],"isInLoop":true,"position":1,"parameters":{"history":5,"tabIndex":-1,"useLoop":true,"xpath":"","iframe":false,"wait":2,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"clickWay":0,"maxWaitTime":10,"paras":[],"allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/ul[1]/li[5]/a[1]","//a[contains(., '')]","//A[@class='ant-pagination-item-link']","/html/body/div[last()-3]/div/div/div[last()-2]/div/div/div/div[last()-1]/div/ul/li[last()-1]/a"],"loopType":0}},{"id":5,"index":6,"parentId":4,"type":1,"option":8,"title":"循环","sequence":[8,7],"isInLoop":true,"position":0,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[2]/table[1]/tbody[1]/tr/td[4]/div[1]/div[2]/div[1]/div[1]/div[1]/a[1]","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":1,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[2]/table[1]/tbody[1]/tr[1]/td[4]/div[1]/div[2]/div[1]/div[1]/div[1]/a[1]","//a[contains(., '湖北大学')]","//A[@class='name-cn']","/html/body/div[last()-3]/div/div/div[last()-1]/div/div/div/div[last()-1]/div/div[last()-1]/table/tbody/tr[last()-29]/td[last()-1]/div/div/div[last()-1]/div/div/a"]}},{"id":8,"index":7,"parentId":5,"type":0,"option":3,"title":"提取数据","sequence":[],"isInLoop":true,"position":1,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"clear":0,"paras":[{"nodeType":1,"contentType":0,"relative":true,"name":"参数1_链接文本","desc":"","extractType":0,"relativeXPath":"","allXPaths":"","exampleValues":[{"num":0,"value":"湖北大学\n "}],"unique_index":"dgby6tuc79nlp9toe7t","iframe":false,"default":"","paraType":"text","recordASField":1,"beforeJS":"","beforeJSWaitTime":0,"JS":"","JSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"downloadPic":0}],"loopType":1}},{"id":7,"index":8,"parentId":5,"type":0,"option":3,"title":"提取数据","sequence":[],"isInLoop":true,"position":0,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"clear":0,"paras":[{"nodeType":0,"contentType":0,"relative":false,"name":"参数3_文本","desc":"","extractType":0,"relativeXPath":"/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[1]/div[1]/div[3]","allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[1]/div[1]/div[3]","//div[contains(., '（哲学）')]","/html/body/div[last()-3]/div/div/div[last()-1]/div/div/div/div[last()-1]/div/div[last()-2]/div[last()-1]/div[last()-1]"],"exampleValues":[{"num":0,"value":"（哲学）"}],"unique_index":"xgoa18n26rllp9toqoa","iframe":false,"default":"","paraType":"text","recordASField":1,"beforeJS":"","beforeJSWaitTime":0,"JS":"","JSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"downloadPic":0}]}}]} \ No newline at end of file diff --git a/.temp_to_pub/EasySpider_windows_x64/execution_instances/124.json b/.temp_to_pub/EasySpider_windows_x64/execution_instances/124.json new file mode 100644 index 0000000..56f8caa --- /dev/null +++ b/.temp_to_pub/EasySpider_windows_x64/execution_instances/124.json @@ -0,0 +1 @@ +{"id":124,"name":"【软科排名】-中国最好学科排名|最权威的大学学科|高校学科排名","url":"https://www.shanghairanking.cn/rankings/bcsr/2023","links":"https://www.shanghairanking.cn/rankings/bcsr/2023","create_time":"2023/11/22 21:51:29","update_time":"2023/11/22 21:51:29","version":"0.5.0","saveThreshold":10,"cloudflare":0,"environment":0,"maxViewLength":15,"recordLog":1,"outputFormat":"xlsx","saveName":"current_time","inputExcel":"","startFromExit":0,"containJudge":false,"desc":"https://www.shanghairanking.cn/rankings/bcsr/2023","inputParameters":[{"id":0,"name":"urlList_0","nodeId":1,"nodeName":"打开网页","value":"https://www.shanghairanking.cn/rankings/bcsr/2023","desc":"要采集的网址列表，多行以\\n分开","type":"text","exampleValue":"https://www.shanghairanking.cn/rankings/bcsr/2023"},{"id":1,"name":"loopTimes_循环_1","nodeId":4,"nodeName":"循环","desc":"循环循环执行的次数（0代表无限循环）","type":"int","exampleValue":0,"value":0}],"outputParameters":[{"id":0,"name":"参数1_链接文本","desc":"","type":"text","recordASField":1,"exampleValue":"湖北大学\n "},{"id":1,"name":"参数3_文本","desc":"","type":"text","recordASField":1,"exampleValue":"（哲学）"}],"graph":[{"index":0,"id":0,"parentId":0,"type":-1,"option":0,"title":"root","sequence":[1,2],"parameters":{"history":1,"tabIndex":0,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0},"isInLoop":false},{"id":1,"index":1,"parentId":0,"type":0,"option":1,"title":"打开网页","sequence":[],"isInLoop":false,"position":0,"parameters":{"useLoop":false,"xpath":"","wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"url":"https://www.shanghairanking.cn/rankings/bcsr/2023","links":"https://www.shanghairanking.cn/rankings/bcsr/2023","maxWaitTime":10,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"cookies":""}},{"id":2,"index":2,"parentId":0,"type":1,"option":8,"title":"循环","sequence":[3,4],"isInLoop":false,"position":1,"parameters":{"history":4,"tabIndex":-1,"useLoop":false,"xpath":"/html/body/div[1]/div[1]/div[1]/div[2]/div[2]/div[1]/div/div[2]/div/a[1]/span[2]","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":1,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":""}},{"id":3,"index":3,"parentId":2,"type":0,"option":2,"title":"点击元素","sequence":[],"isInLoop":true,"position":0,"parameters":{"history":4,"tabIndex":-1,"useLoop":true,"xpath":"","iframe":false,"wait":2,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"clickWay":0,"maxWaitTime":10,"paras":[],"allXPaths":"","loopType":1}},{"id":4,"index":4,"parentId":2,"type":1,"option":8,"title":"循环","sequence":[6,5],"isInLoop":true,"position":1,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"//*[contains(@class, \"ant-pagination-next\")]/a[1]","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":0,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/ul[1]/li[5]/a[1]","//a[contains(., '')]","//A[@class='ant-pagination-item-link']","/html/body/div[last()-3]/div/div/div[last()-2]/div/div/div/div[last()-1]/div/ul/li[last()-1]/a"]}},{"id":6,"index":5,"parentId":4,"type":0,"option":2,"title":"点击元素","sequence":[],"isInLoop":true,"position":1,"parameters":{"history":5,"tabIndex":-1,"useLoop":true,"xpath":"","iframe":false,"wait":2,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"clickWay":0,"maxWaitTime":10,"paras":[],"allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/ul[1]/li[5]/a[1]","//a[contains(., '')]","//A[@class='ant-pagination-item-link']","/html/body/div[last()-3]/div/div/div[last()-2]/div/div/div/div[last()-1]/div/ul/li[last()-1]/a"],"loopType":0}},{"id":5,"index":6,"parentId":4,"type":1,"option":8,"title":"循环","sequence":[8,7],"isInLoop":true,"position":0,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[2]/table[1]/tbody[1]/tr/td[4]/div[1]/div[2]/div[1]/div[1]/div[1]/a[1]","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":1,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[2]/table[1]/tbody[1]/tr[1]/td[4]/div[1]/div[2]/div[1]/div[1]/div[1]/a[1]","//a[contains(., '湖北大学')]","//A[@class='name-cn']","/html/body/div[last()-3]/div/div/div[last()-1]/div/div/div/div[last()-1]/div/div[last()-1]/table/tbody/tr[last()-29]/td[last()-1]/div/div/div[last()-1]/div/div/a"]}},{"id":8,"index":7,"parentId":5,"type":0,"option":3,"title":"提取数据","sequence":[],"isInLoop":true,"position":1,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"clear":0,"paras":[{"nodeType":1,"contentType":0,"relative":true,"name":"参数1_链接文本","desc":"","extractType":0,"relativeXPath":"","allXPaths":"","exampleValues":[{"num":0,"value":"湖北大学\n "}],"unique_index":"dgby6tuc79nlp9toe7t","iframe":false,"default":"","paraType":"text","recordASField":1,"beforeJS":"","beforeJSWaitTime":0,"JS":"","JSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"downloadPic":0}],"loopType":1}},{"id":7,"index":8,"parentId":5,"type":0,"option":3,"title":"提取数据","sequence":[],"isInLoop":true,"position":0,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"clear":0,"paras":[{"nodeType":0,"contentType":0,"relative":false,"name":"参数3_文本","desc":"","extractType":0,"relativeXPath":"/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[1]/div[1]/div[3]","allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[1]/div[1]/div[3]","//div[contains(., '（哲学）')]","/html/body/div[last()-3]/div/div/div[last()-1]/div/div/div/div[last()-1]/div/div[last()-2]/div[last()-1]/div[last()-1]"],"exampleValues":[{"num":0,"value":"（哲学）"}],"unique_index":"xgoa18n26rllp9toqoa","iframe":false,"default":"","paraType":"text","recordASField":1,"beforeJS":"","beforeJSWaitTime":0,"JS":"","JSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"downloadPic":0}]}}]} \ No newline at end of file diff --git a/.temp_to_pub/EasySpider_windows_x64/execution_instances/125.json b/.temp_to_pub/EasySpider_windows_x64/execution_instances/125.json new file mode 100644 index 0000000..3676d76 --- /dev/null +++ b/.temp_to_pub/EasySpider_windows_x64/execution_instances/125.json @@ -0,0 +1 @@ +{"id":125,"name":"【软科排名】-中国最好学科排名|最权威的大学学科|高校学科排名","url":"https://www.shanghairanking.cn/rankings/bcsr/2023","links":"https://www.shanghairanking.cn/rankings/bcsr/2023","create_time":"2023/11/22 21:51:29","update_time":"2023/11/22 21:51:29","version":"0.5.0","saveThreshold":10,"cloudflare":0,"environment":0,"maxViewLength":15,"recordLog":1,"outputFormat":"xlsx","saveName":"current_time","inputExcel":"","startFromExit":0,"containJudge":false,"desc":"https://www.shanghairanking.cn/rankings/bcsr/2023","inputParameters":[{"id":0,"name":"urlList_0","nodeId":1,"nodeName":"打开网页","value":"https://www.shanghairanking.cn/rankings/bcsr/2023","desc":"要采集的网址列表，多行以\\n分开","type":"text","exampleValue":"https://www.shanghairanking.cn/rankings/bcsr/2023"},{"id":1,"name":"loopTimes_循环_1","nodeId":4,"nodeName":"循环","desc":"循环循环执行的次数（0代表无限循环）","type":"int","exampleValue":0,"value":0}],"outputParameters":[{"id":0,"name":"参数1_链接文本","desc":"","type":"text","recordASField":1,"exampleValue":"湖北大学\n "},{"id":1,"name":"参数3_文本","desc":"","type":"text","recordASField":1,"exampleValue":"（哲学）"}],"graph":[{"index":0,"id":0,"parentId":0,"type":-1,"option":0,"title":"root","sequence":[1,2],"parameters":{"history":1,"tabIndex":0,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0},"isInLoop":false},{"id":1,"index":1,"parentId":0,"type":0,"option":1,"title":"打开网页","sequence":[],"isInLoop":false,"position":0,"parameters":{"useLoop":false,"xpath":"","wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"url":"https://www.shanghairanking.cn/rankings/bcsr/2023","links":"https://www.shanghairanking.cn/rankings/bcsr/2023","maxWaitTime":10,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"cookies":""}},{"id":2,"index":2,"parentId":0,"type":1,"option":8,"title":"循环","sequence":[3,4],"isInLoop":false,"position":1,"parameters":{"history":4,"tabIndex":-1,"useLoop":false,"xpath":"/html/body/div[1]/div[1]/div[1]/div[2]/div[2]/div[1]/div/div[2]/div/a[1]/span[2]","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":1,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":""}},{"id":3,"index":3,"parentId":2,"type":0,"option":2,"title":"点击元素","sequence":[],"isInLoop":true,"position":0,"parameters":{"history":4,"tabIndex":-1,"useLoop":true,"xpath":"","iframe":false,"wait":2,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"clickWay":0,"maxWaitTime":10,"paras":[],"allXPaths":"","loopType":1}},{"id":4,"index":4,"parentId":2,"type":1,"option":8,"title":"循环","sequence":[6,5],"isInLoop":true,"position":1,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"//*[contains(@class, \"ant-pagination-next\")]/a[1]","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":0,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/ul[1]/li[5]/a[1]","//a[contains(., '')]","//A[@class='ant-pagination-item-link']","/html/body/div[last()-3]/div/div/div[last()-2]/div/div/div/div[last()-1]/div/ul/li[last()-1]/a"]}},{"id":6,"index":5,"parentId":4,"type":0,"option":2,"title":"点击元素","sequence":[],"isInLoop":true,"position":1,"parameters":{"history":5,"tabIndex":-1,"useLoop":true,"xpath":"","iframe":false,"wait":2,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"clickWay":0,"maxWaitTime":10,"paras":[],"allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/ul[1]/li[5]/a[1]","//a[contains(., '')]","//A[@class='ant-pagination-item-link']","/html/body/div[last()-3]/div/div/div[last()-2]/div/div/div/div[last()-1]/div/ul/li[last()-1]/a"],"loopType":0}},{"id":5,"index":6,"parentId":4,"type":1,"option":8,"title":"循环","sequence":[8,7],"isInLoop":true,"position":0,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[2]/table[1]/tbody[1]/tr/td[4]/div[1]/div[2]/div[1]/div[1]/div[1]/a[1]","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":1,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[2]/table[1]/tbody[1]/tr[1]/td[4]/div[1]/div[2]/div[1]/div[1]/div[1]/a[1]","//a[contains(., '湖北大学')]","//A[@class='name-cn']","/html/body/div[last()-3]/div/div/div[last()-1]/div/div/div/div[last()-1]/div/div[last()-1]/table/tbody/tr[last()-29]/td[last()-1]/div/div/div[last()-1]/div/div/a"]}},{"id":8,"index":7,"parentId":5,"type":0,"option":3,"title":"提取数据","sequence":[],"isInLoop":true,"position":1,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"clear":0,"paras":[{"nodeType":1,"contentType":0,"relative":true,"name":"参数1_链接文本","desc":"","extractType":0,"relativeXPath":"","allXPaths":"","exampleValues":[{"num":0,"value":"湖北大学\n "}],"unique_index":"dgby6tuc79nlp9toe7t","iframe":false,"default":"","paraType":"text","recordASField":1,"beforeJS":"","beforeJSWaitTime":0,"JS":"","JSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"downloadPic":0}],"loopType":1}},{"id":7,"index":8,"parentId":5,"type":0,"option":3,"title":"提取数据","sequence":[],"isInLoop":true,"position":0,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"clear":0,"paras":[{"nodeType":0,"contentType":0,"relative":false,"name":"参数3_文本","desc":"","extractType":0,"relativeXPath":"/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[1]/div[1]/div[3]","allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[1]/div[1]/div[3]","//div[contains(., '（哲学）')]","/html/body/div[last()-3]/div/div/div[last()-1]/div/div/div/div[last()-1]/div/div[last()-2]/div[last()-1]/div[last()-1]"],"exampleValues":[{"num":0,"value":"（哲学）"}],"unique_index":"xgoa18n26rllp9toqoa","iframe":false,"default":"","paraType":"text","recordASField":1,"beforeJS":"","beforeJSWaitTime":0,"JS":"","JSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"downloadPic":0}]}}]} \ No newline at end of file diff --git a/.temp_to_pub/EasySpider_windows_x64/tasks/209.json b/.temp_to_pub/EasySpider_windows_x64/tasks/209.json index 492f9de..ed8e100 100644 --- a/.temp_to_pub/EasySpider_windows_x64/tasks/209.json +++ b/.temp_to_pub/EasySpider_windows_x64/tasks/209.json @@ -1 +1 @@ -{"id":209,"name":"京东全球版-专业的综合网上购物商城","url":"https://www.jd.com","links":"https://www.jd.com","create_time":"8/14/2023, 5:40:55 PM","update_time":"8/14/2023, 5:49:34 PM","version":"0.5.0","saveThreshold":10,"cloudflare":0,"environment":0,"maxViewLength":15,"recordLog":1,"outputFormat":"xlsx","saveName":"current_time","inputExcel":"","startFromExit":0,"containJudge":false,"desc":"https://www.jd.com","inputParameters":[{"id":0,"name":"urlList_0","nodeId":1,"nodeName":"打开网页","value":"https://www.jd.com","desc":"要采集的网址列表，多行以\\n分开","type":"text","exampleValue":"https://www.jd.com"}],"outputParameters":[{"id":0,"name":"自定义操作","desc":"自定义操作返回的数据","type":"text","recordASField":0,"exampleValue":""},{"id":1,"name":"参数1_链接文本","desc":"","type":"text","recordASField":1,"exampleValue":"数码"},{"id":2,"name":"自定义操作2","desc":"自定义操作返回的数据","type":"text","recordASField":0,"exampleValue":""}],"graph":[{"index":0,"id":0,"parentId":0,"type":-1,"option":0,"title":"root","sequence":[1,2,4,3],"parameters":{"history":1,"tabIndex":0,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0},"isInLoop":false},{"id":1,"index":1,"parentId":0,"type":0,"option":1,"title":"打开网页","sequence":[],"isInLoop":false,"position":0,"parameters":{"useLoop":false,"xpath":"","wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"url":"https://www.jd.com","links":"https://www.jd.com","maxWaitTime":10,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"cookies":""}},{"id":2,"index":2,"parentId":0,"type":0,"option":5,"title":"自定义操作","sequence":[],"isInLoop":false,"position":1,"parameters":{"history":1,"tabIndex":0,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"clear":0,"codeMode":"5","code":"self.a = 1\nself.b = 2","waitTime":0,"recordASField":0,"paraType":"text"}},{"id":4,"index":3,"parentId":0,"type":0,"option":3,"title":"提取数据","sequence":[],"isInLoop":false,"position":3,"parameters":{"history":4,"tabIndex":-1,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"clear":0,"paras":[{"nodeType":1,"contentType":0,"relative":false,"name":"参数1_链接文本","desc":"","extractType":0,"relativeXPath":"/html/body/div[5]/div[Field[\"自定义操作2\"]]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/a[EVAL(\"self.a + self.b\")]","allXPaths":["/html/body/div[5]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/a[2]","//a[contains(., '数码')]","/html/body/div[last()-5]/div/div[last()-4]/div/div[last()-2]/div/div/div/div[last()-1]/div[last()-12]/a"],"exampleValues":[{"num":0,"value":"数码"}],"unique_index":"603bndbum96llaop8rs","iframe":false,"default":"","paraType":"text","recordASField":1,"beforeJS":"","beforeJSWaitTime":0,"JS":"","JSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"downloadPic":0}]}},{"id":3,"index":4,"parentId":0,"type":0,"option":5,"title":"自定义操作2","sequence":[],"isInLoop":false,"position":2,"parameters":{"history":1,"tabIndex":0,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"clear":0,"codeMode":"6","code":"self.a","waitTime":0,"recordASField":0,"paraType":"text"}}]} \ No newline at end of file +{"id":209,"name":"【软科排名】-中国最好学科排名|最权威的大学学科|高校学科排名","url":"https://www.shanghairanking.cn/rankings/bcsr/2023","links":"https://www.shanghairanking.cn/rankings/bcsr/2023","create_time":"2023/11/22 21:51:29","update_time":"2023/11/22 21:51:29","version":"0.5.0","saveThreshold":10,"cloudflare":0,"environment":0,"maxViewLength":15,"recordLog":1,"outputFormat":"xlsx","saveName":"current_time","inputExcel":"","startFromExit":0,"containJudge":false,"desc":"https://www.shanghairanking.cn/rankings/bcsr/2023","inputParameters":[{"id":0,"name":"urlList_0","nodeId":1,"nodeName":"打开网页","value":"https://www.shanghairanking.cn/rankings/bcsr/2023","desc":"要采集的网址列表，多行以\\n分开","type":"text","exampleValue":"https://www.shanghairanking.cn/rankings/bcsr/2023"},{"id":1,"name":"loopTimes_循环_1","nodeId":4,"nodeName":"循环","desc":"循环循环执行的次数（0代表无限循环）","type":"int","exampleValue":0,"value":0}],"outputParameters":[{"id":0,"name":"参数1_链接文本","desc":"","type":"text","recordASField":1,"exampleValue":"湖北大学\n "},{"id":1,"name":"参数3_文本","desc":"","type":"text","recordASField":1,"exampleValue":"（哲学）"}],"graph":[{"index":0,"id":0,"parentId":0,"type":-1,"option":0,"title":"root","sequence":[1,2],"parameters":{"history":1,"tabIndex":0,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0},"isInLoop":false},{"id":1,"index":1,"parentId":0,"type":0,"option":1,"title":"打开网页","sequence":[],"isInLoop":false,"position":0,"parameters":{"useLoop":false,"xpath":"","wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"url":"https://www.shanghairanking.cn/rankings/bcsr/2023","links":"https://www.shanghairanking.cn/rankings/bcsr/2023","maxWaitTime":10,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"cookies":""}},{"id":2,"index":2,"parentId":0,"type":1,"option":8,"title":"循环","sequence":[3,4],"isInLoop":false,"position":1,"parameters":{"history":4,"tabIndex":-1,"useLoop":false,"xpath":"/html/body/div[1]/div[1]/div[1]/div[2]/div[2]/div[1]/div/div[2]/div/a[1]/span[2]","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":1,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":""}},{"id":3,"index":3,"parentId":2,"type":0,"option":2,"title":"点击元素","sequence":[],"isInLoop":true,"position":0,"parameters":{"history":4,"tabIndex":-1,"useLoop":true,"xpath":"","iframe":false,"wait":2,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"clickWay":0,"maxWaitTime":10,"paras":[],"allXPaths":"","loopType":1}},{"id":4,"index":4,"parentId":2,"type":1,"option":8,"title":"循环","sequence":[6,5],"isInLoop":true,"position":1,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"//*[contains(@class, \"ant-pagination-next\")]/a[1]","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":0,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/ul[1]/li[5]/a[1]","//a[contains(., '')]","//A[@class='ant-pagination-item-link']","/html/body/div[last()-3]/div/div/div[last()-2]/div/div/div/div[last()-1]/div/ul/li[last()-1]/a"]}},{"id":6,"index":5,"parentId":4,"type":0,"option":2,"title":"点击元素","sequence":[],"isInLoop":true,"position":1,"parameters":{"history":5,"tabIndex":-1,"useLoop":true,"xpath":"","iframe":false,"wait":2,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"clickWay":0,"maxWaitTime":10,"paras":[],"allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/ul[1]/li[5]/a[1]","//a[contains(., '')]","//A[@class='ant-pagination-item-link']","/html/body/div[last()-3]/div/div/div[last()-2]/div/div/div/div[last()-1]/div/ul/li[last()-1]/a"],"loopType":0}},{"id":5,"index":6,"parentId":4,"type":1,"option":8,"title":"循环","sequence":[8,7],"isInLoop":true,"position":0,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[2]/table[1]/tbody[1]/tr/td[4]/div[1]/div[2]/div[1]/div[1]/div[1]/a[1]","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":1,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[2]/table[1]/tbody[1]/tr[1]/td[4]/div[1]/div[2]/div[1]/div[1]/div[1]/a[1]","//a[contains(., '湖北大学')]","//A[@class='name-cn']","/html/body/div[last()-3]/div/div/div[last()-1]/div/div/div/div[last()-1]/div/div[last()-1]/table/tbody/tr[last()-29]/td[last()-1]/div/div/div[last()-1]/div/div/a"]}},{"id":8,"index":7,"parentId":5,"type":0,"option":3,"title":"提取数据","sequence":[],"isInLoop":true,"position":1,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"clear":0,"paras":[{"nodeType":1,"contentType":0,"relative":true,"name":"参数1_链接文本","desc":"","extractType":0,"relativeXPath":"","allXPaths":"","exampleValues":[{"num":0,"value":"湖北大学\n "}],"unique_index":"dgby6tuc79nlp9toe7t","iframe":false,"default":"","paraType":"text","recordASField":1,"beforeJS":"","beforeJSWaitTime":0,"JS":"","JSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"downloadPic":0}],"loopType":1}},{"id":7,"index":8,"parentId":5,"type":0,"option":3,"title":"提取数据","sequence":[],"isInLoop":true,"position":0,"parameters":{"history":5,"tabIndex":-1,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"waitElement":"","waitElementTime":10,"waitElementIframeIndex":0,"clear":0,"paras":[{"nodeType":0,"contentType":0,"relative":false,"name":"参数3_文本","desc":"","extractType":0,"relativeXPath":"/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[1]/div[1]/div[3]","allXPaths":["/html/body/div[1]/div[1]/div[1]/div[2]/div[1]/div[3]/div[2]/div[1]/div[1]/div[1]/div[1]/div[3]","//div[contains(., '（哲学）')]","/html/body/div[last()-3]/div/div/div[last()-1]/div/div/div/div[last()-1]/div/div[last()-2]/div[last()-1]/div[last()-1]"],"exampleValues":[{"num":0,"value":"（哲学）"}],"unique_index":"xgoa18n26rllp9toqoa","iframe":false,"default":"","paraType":"text","recordASField":1,"beforeJS":"","beforeJSWaitTime":0,"JS":"","JSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"downloadPic":0}]}}]} \ No newline at end of file diff --git a/ExecuteStage/easyspider_executestage.py b/ExecuteStage/easyspider_executestage.py index 2fc1808..11835cd 100644 --- a/ExecuteStage/easyspider_executestage.py +++ b/ExecuteStage/easyspider_executestage.py @@ -1032,6 +1032,9 @@ class BrowserThread(Thread): self.browser.execute_script('window.stop()') except: pass + if self.browser.current_url.startswith("data:"): + self.browser.execute_script("history.go(1)") # 如果是data:开头的网址，就前进一步 + time.sleep(2) if int(node["parameters"]["breakMode"]) > 0: # 如果设置了退出循环的脚本条件 output = self.execute_code(int( node["parameters"]["breakMode"]) - 1, node["parameters"]["breakCode"], node["parameters"]["breakCodeWaitTime"], iframe=node["parameters"]["iframe"]) @@ -1094,6 +1097,9 @@ class BrowserThread(Thread): self.browser.execute_script('window.stop()') except: pass + if self.browser.current_url.startswith("data:"): + self.browser.execute_script("history.go(1)") # 如果是data:开头的网址，就前进一步 + time.sleep(2) except NoSuchElementException: self.print_and_log("Loop element not found: ", path) self.print_and_log("找不到循环元素: ", path)