|
- <mxGraphModel>
- <root>
- <mxCell id="0">
- <JsonProperty as="data">
- {"spiderName":"未定义名称","submit-strategy":"random","threadCount":"1"}
- </JsonProperty>
- </mxCell>
- <mxCell id="1" parent="0"/>
- <mxCell id="2" value="开始" style="start" parent="1" vertex="1">
- <mxGeometry x="10" y="100" width="32" height="32" as="geometry"/>
- <JsonProperty as="data">
- {"shape":"start"}
- </JsonProperty>
- </mxCell>
- <mxCell id="15" value="抓取列表" style="request" parent="1" vertex="1">
- <mxGeometry x="559" y="90" width="32" height="32" as="geometry"/>
- <JsonProperty as="data">
- {"value":"抓取列表","loopVariableName":"","method":"GET","sleep":"200","timeout":"3000","response-charset":"","retryCount":"3","retryInterval":"200","cookie-name":["Cookie"],"cookie-description":[""],"body-type":"none","body-content-type":"text/plain","loopCount":"","url":"${'https://so.toutiao.com/search?keyword='+keyword+'&pd=information&source=search_subtab_switch&dvpf=pc&aid=4916&page_num='+(page-1)}","proxy":"","cookie-value":["MONITOR_WEB_ID=7136060905808446989; _S_DPR=2; _S_IPAD=0; _S_WIN_WH=1372_739; ttwid=1%7CDU8QP2eAdg2wjFnB_ggNlqGibkM_DPTXGw1bvFbqUt8%7C1661493660%7C3e9e6f880b1d240b10612e18d44b39f2787f082f5438e6719efbb984559f96f4"],"request-body":"","follow-redirect":"1","tls-validate":"1","cookie-auto-set":"1","repeat-enable":"0","shape":"request"}
- </JsonProperty>
- </mxCell>
- <mxCell id="17" value="定义变量" style="variable" parent="1" vertex="1">
- <mxGeometry x="450" y="90" width="32" height="32" as="geometry"/>
- <JsonProperty as="data">
- {"value":"定义变量","loopVariableName":"","ziloopVariableName":"","variable-name":["page","keyword"],"variable-description":["",""],"loopCount":"","ziloopCount":"","variable-value":["${pageNoData}","北京"],"shape":"variable"}
- </JsonProperty>
- </mxCell>
- <mxCell id="33" value="定义变量" style="variable" parent="1" vertex="1">
- <mxGeometry x="689" y="90" width="32" height="32" as="geometry"/>
- <JsonProperty as="data">
- {"value":"定义变量","loopVariableName":"","ziloopVariableName":"","variable-name":["orignalhtml","datalist"],"variable-description":["",""],"loopCount":"","ziloopCount":"","variable-value":["${resp.html}","${orignalhtml.selectors('script[data-for=s-result-json]')}"],"shape":"variable"}
- </JsonProperty>
- </mxCell>
- <mxCell id="54" value="" style="strokeWidth=2;sharp=1;" parent="1" source="15" target="33" edge="1">
- <mxGeometry relative="1" as="geometry"/>
- <JsonProperty as="data">
- {"value":"","exception-flow":"0","lineWidth":"2","line-style":"sharp","lineColor":"black","condition":"","transmit-variable":"1"}
- </JsonProperty>
- </mxCell>
- <mxCell id="55" value="循环" style="loop" parent="1" vertex="1">
- <mxGeometry x="826" y="90" width="32" height="32" as="geometry"/>
- <JsonProperty as="data">
- {"value":"循环","loopItem":"","loopVariableName":"j","loopCount":"${datalist.size()}","loopStart":"0","loopEnd":"-1","awaitSleep":"0","shape":"loop"}
- </JsonProperty>
- </mxCell>
- <mxCell id="64" value="定义变量" style="variable" parent="1" vertex="1">
- <mxGeometry x="916" y="90" width="32" height="32" as="geometry"/>
- <JsonProperty as="data">
- {"value":"定义变量","loopVariableName":"","ziloopVariableName":"","variable-name":["html","sub_html","data_json_object","title","source_url","datetime","media_name","abstract","imgurl","videojson","imglist","imgjson","","","",""],"variable-description":["","","","","","","","","","","","","","","",""],"loopCount":"","ziloopCount":"","variable-value":["${datalist[j]+''}","${strUtils.substring(html,strUtils.indexOf(html,'\"hydrate\">')+10,strUtils.length(html)-9)}","${jsonUtils.parseObject(sub_html)}","${data_json_object.data.title}","${data_json_object.data.source_url}","${data_json_object.data.datetime}","${data_json_object.data.media_name}","${data_json_object.data.abstract}","${data_json_object.data.image_url}","${jsonUtils.createJsonObject()}","${jsonUtils.createJsonArray()}","${jsonUtils.createJsonObject()}","${imgjson.put(\"imgurl\",imgurl)}","${imglist.add(imgjson)}","${videojson.put(\"imglist\",imglist)}","${videojson.put(\"vediourl\",'')}"],"shape":"variable"}
- </JsonProperty>
- </mxCell>
- <mxCell id="65" value="" style="strokeWidth=2;sharp=1;" parent="1" source="55" target="64" edge="1">
- <mxGeometry relative="1" as="geometry"/>
- <JsonProperty as="data">
- {"value":"","exception-flow":"0","lineWidth":"2","line-style":"sharp","lineColor":"black","condition":"","transmit-variable":"1"}
- </JsonProperty>
- </mxCell>
- <mxCell id="78" value="输出" style="output" parent="1" vertex="1">
- <mxGeometry x="916" y="196" width="32" height="32" as="geometry"/>
- <JsonProperty as="data">
- {"value":"输出","loopVariableName":"","tableName":"","csvName":"","csvEncoding":"GBK","output-name":["title","source_url","datetime","datalist"],"loopCount":"","output-value":["${title}","${source_url}","${datetime}","${data_json_object}"],"output-all":"0","output-database":"0","output-csv":"0","shape":"output"}
- </JsonProperty>
- </mxCell>
- <mxCell id="93" value="定义变量" style="variable" parent="1" vertex="1">
- <mxGeometry x="916" y="372" width="32" height="32" as="geometry"/>
- <JsonProperty as="data">
- {"value":"定义变量","loopVariableName":"","ziloopVariableName":"","variable-name":["json","classify","website_id","websitelogo"],"variable-description":["","","",""],"loopCount":"","ziloopCount":"","variable-value":["${resp.html.json()}","${json.new_website_type}","${json.id}","${json.website_ico}"],"shape":"variable"}
- </JsonProperty>
- </mxCell>
- <mxCell id="95" value="输出" style="output" parent="1" vertex="1">
- <mxGeometry x="810" y="282" width="32" height="32" as="geometry"/>
- <JsonProperty as="data">
- {"value":"输出","loopVariableName":"","tableName":"","csvName":"","csvEncoding":"GBK","output-name":["",""],"loopCount":"","output-value":["${resp.html}","${url}"],"output-all":"0","output-database":"0","output-csv":"0","shape":"output"}
- </JsonProperty>
- </mxCell>
- <mxCell id="104" value="修改关键词状态" style="request" parent="1" vertex="1">
- <mxGeometry x="1281" y="250" width="32" height="32" as="geometry"/>
- <JsonProperty as="data">
- {"value":"修改关键词状态","loopVariableName":"","method":"GET","sleep":"","timeout":"","response-charset":"","retryCount":"","retryInterval":"","body-type":"none","body-content-type":"text/plain","loopCount":"","url":"${\"http://192.168.71.61:6135/spider_factory/baidu/keywordStatus?keyword=\"+keyword}","proxy":"","request-body":"","follow-redirect":"1","tls-validate":"1","cookie-auto-set":"1","repeat-enable":"0","shape":"request"}
- </JsonProperty>
- </mxCell>
- <mxCell id="106" value="" style="strokeWidth=2;sharp=1;" parent="1" source="33" target="55" edge="1">
- <mxGeometry relative="1" as="geometry"/>
- <JsonProperty as="data">
- {"value":"","exception-flow":"0","lineWidth":"2","line-style":"sharp","lineColor":"black","condition":"","transmit-variable":"1"}
- </JsonProperty>
- </mxCell>
- <mxCell id="110" value="开始抓取" style="request" parent="1" vertex="1">
- <mxGeometry x="314" y="380" width="32" height="32" as="geometry"/>
- <JsonProperty as="data">
- {"value":"开始抓取","loopVariableName":"","method":"GET","sleep":"","timeout":"","response-charset":"","retryCount":"","retryInterval":"","body-type":"none","body-content-type":"text/plain","loopCount":"","url":"${\"http://192.168.71.61:6135/spider_factory/baidu/keywordStatus?keyword=\"+rs[i].keyword}","proxy":"","request-body":"","follow-redirect":"1","tls-validate":"1","cookie-auto-set":"1","repeat-enable":"0","shape":"request"}
- </JsonProperty>
- </mxCell>
- <mxCell id="120" value="" style="strokeWidth=2;sharp=1;" parent="1" source="17" target="15" edge="1">
- <mxGeometry relative="1" as="geometry"/>
- <JsonProperty as="data">
- {"value":"","exception-flow":"0","lineWidth":"2","line-style":"sharp","lineColor":"black","condition":"","transmit-variable":"1"}
- </JsonProperty>
- </mxCell>
- <mxCell id="129" value="" style="strokeWidth=2;sharp=1;" parent="1" source="2" target="17" edge="1">
- <mxGeometry relative="1" as="geometry"/>
- <JsonProperty as="data">
- {"value":"","exception-flow":"0","lineWidth":"2","line-style":"sharp","lineColor":"black","condition":"","transmit-variable":"1"}
- </JsonProperty>
- </mxCell>
- <mxCell id="130" value="输出" style="output" parent="1" vertex="1">
- <mxGeometry x="1020" y="84" width="32" height="32" as="geometry"/>
- <JsonProperty as="data">
- {"value":"输出","loopVariableName":"","tableName":"","csvName":"","csvEncoding":"GBK","output-name":["title","url","publish_time","source","abstract","author","videojson"],"loopCount":"","output-value":["${strUtils.contains(title,keyword)?(strUtils.replaceAll(title,keyword,'<b class=\"key\" style=\"color:red\">'+keyword+'</b>')):(title)}","${source_url}","${datetime}","${'今日头条'}","${strUtils.contains(abstract,keyword)?(strUtils.replaceAll(abstract,keyword,'<b class=\"key\" style=\"color:red\">'+keyword+'</b>')):(abstract)}","${media_name}","${videojson}"],"output-all":"0","output-database":"0","output-csv":"0","shape":"output"}
- </JsonProperty>
- </mxCell>
- <mxCell id="131" value="" style="strokeWidth=2;strokeColor=blue;sharp=1;" parent="1" source="64" target="130" edge="1">
- <mxGeometry relative="1" as="geometry"/>
- <JsonProperty as="data">
- {"value":"","exception-flow":"0","lineWidth":"2","line-style":"sharp","lineColor":"blue","condition":"${abstract!=null||title!=null}","transmit-variable":"1"}
- </JsonProperty>
- </mxCell>
- </root>
- </mxGraphModel>
|