Skip to content

dr_py
dr_py

TimBL / dr_py
与 Fork 源项目一致

Fork自 晚风拂柳颜 / dr_py

通知 1
Star 1
Fork 0
dr_py
dr_py

体验新版 GitCode，发现更多精彩内容 >>

提交 87a5059a 编写于 作者: H hjdhnx
浏览文件
操作

增加线上测试

上级 c0d02cbb
隐藏空白更改
内联 并排
Showing with 236 addition and 107 deletion
js/version.txt
浏览文件 @ 87a5059a
3.7.19beta3 3.7.19beta4
\ No newline at end of file \ No newline at end of file
libs/drpy.js
浏览文件 @ 87a5059a
...@@ -2,12 +2,15 @@ import 'https://gitcode.net/qq_32394351/dr_py/-/raw/master/libs/es6py.js'; ...@@ -2,12 +2,15 @@ import 'https://gitcode.net/qq_32394351/dr_py/-/raw/master/libs/es6py.js';
// import 'http://192.168.10.103:5705/libs/es6py.js'; // import 'http://192.168.10.103:5705/libs/es6py.js';
import cheerio from 'https://gitcode.net/qq_32394351/dr_py/-/raw/master/libs/cheerio.min.js'; import cheerio from 'https://gitcode.net/qq_32394351/dr_py/-/raw/master/libs/cheerio.min.js';
// import cheerio from 'http://192.168.10.103:5705/libs/cheerio.min.js'; // import cheerio from 'http://192.168.10.103:5705/libs/cheerio.min.js';
import {parseTags,urljoin,stringify} from 'https://gitcode.net/qq_32394351/dr_py/-/raw/master/libs/parseTags.js';
// import {parseTags,urljoin,stringify} from 'http://192.168.10.103:5705/libs/parseTags.js';
import 'https://gitcode.net/qq_32394351/dr_py/-/raw/master/libs/drT.js'; import 'https://gitcode.net/qq_32394351/dr_py/-/raw/master/libs/drT.js';
// import 'http://192.168.10.103:5705/libs/drT.js'; // import 'http://192.168.10.103:5705/libs/drT.js';
import muban from 'https://gitcode.net/qq_32394351/dr_py/-/raw/master/js/模板.js'; import muban from 'https://gitcode.net/qq_32394351/dr_py/-/raw/master/js/模板.js';
import pa from "./cheerio.min";
// import muban from 'http://192.168.10.103:5705/admin/view/模板.js'; // import muban from 'http://192.168.10.103:5705/admin/view/模板.js';
// const key = 'drpy_zbk'; // const key = 'drpy_zbk';
function init_test(){ function init_test(){
...@@ -28,7 +31,7 @@ let rule = {}; ...@@ -28,7 +31,7 @@ let rule = {};
/** 已知问题记录 /** 已知问题记录
* 1.影魔的jinjia2引擎不支持 {{fl}}对象直接渲染 * 1.影魔的jinjia2引擎不支持 {{fl}}对象直接渲染
* 2.import es6py.js但是里面的函数没有被装载进来.比如drpy规则报错setResult2 is undefiend * 2.import es6py.js但是里面的函数没有被装载进来.比如drpy规则报错setResult2 is undefiend
* * 3.无法重复导入cheerio(怎么解决drpy和parseTag里都需要导入cheerio的问题) 无法在副文件导入cheerio
* *
* todo: jsp:{pdfa,pdfh,pd},json:{pdfa,pdfh,pd},jq:{pdfa,pdfh,pd} * todo: jsp:{pdfa,pdfh,pd},json:{pdfa,pdfh,pd},jq:{pdfa,pdfh,pd}
* **/ * **/
...@@ -48,7 +51,9 @@ const OCR_RETRY = 3;//ocr验证重试次数 ...@@ -48,7 +51,9 @@ const OCR_RETRY = 3;//ocr验证重试次数
// const OCR_API = 'http://dm.mudery.com:10000';//ocr在线识别接口 // const OCR_API = 'http://dm.mudery.com:10000';//ocr在线识别接口
// const OCR_API = 'http://192.168.3.239:5705/parse/ocr';//ocr在线识别接口 // const OCR_API = 'http://192.168.3.239:5705/parse/ocr';//ocr在线识别接口
const OCR_API = 'http://cms.nokia.press/parse/ocr';//ocr在线识别接口 const OCR_API = 'http://cms.nokia.press/parse/ocr';//ocr在线识别接口
var MY_URL; // 全局注入变量,pd函数需要 if(typeof(MY_URL)==='undefined'){
var MY_URL; // 全局注入变量,pd函数需要
}
var VODS = [];// 一级或者搜索需要的数据列表 var VODS = [];// 一级或者搜索需要的数据列表
var vod = {};//二级用单个影片详情 var vod = {};//二级用单个影片详情
var RKEY; // 源的唯一标识 var RKEY; // 源的唯一标识
...@@ -57,6 +62,10 @@ var print; ...@@ -57,6 +62,10 @@ var print;
var log; var log;
var fetch_params; var fetch_params;
var oheaders; var oheaders;
var _pdfh;
var _pdfa;
var _pd;
const jsp = parseTags.jsp;
/*** 后台需要实现的java方法并注入到js中 ***/ /*** 后台需要实现的java方法并注入到js中 ***/
...@@ -167,63 +176,6 @@ function clearItem(k){ ...@@ -167,63 +176,6 @@ function clearItem(k){
local.delete(RKEY,k); local.delete(RKEY,k);
} }
/**
* url拼接(暂未实现)
* @param fromPath 初始当前页面url
* @param nowPath 相对当前页面url
* @returns {*}
*/
function urljoin(fromPath, nowPath) {
fromPath = fromPath||'';
nowPath = nowPath||'';
return joinUrl(fromPath, nowPath);
// try {
// // import Uri from './uri.min.js';
// // var Uri = require('./uri.min.js');
// // eval(request('https://cdn.bootcdn.net/ajax/libs/URI.js/1.19.11/URI.min.js'));
// // let new_uri = URI(nowPath, fromPath);
// let new_uri = Uri(nowPath, fromPath);
// new_uri = new_uri.toString();
// // console.log(new_uri);
// // return fromPath + nowPath
// return new_uri
// }
// catch (e) {
// console.log('urljoin发生错误:'+e.message);
// if(nowPath.startsWith('http')){
// return nowPath
// }if(nowPath.startsWith('/')){
// return getHome(fromPath)+nowPath
// }
// return fromPath+nowPath
// }
}
/**
* 重写pd方法-增加自动urljoin(没法重写,改个名继续骗)
* @param html
* @param parse
* @param uri
* @returns {*}
*/
function pD(html,parse,uri){
let ret = pdfh(html,parse);
if(typeof(uri)==='undefined'||!uri){
uri = '';
}
if(/(url|src|href|data-original|data-src)$/.test(parse)){
if(/http/.test(ret)){
ret = ret.substr(ret.indexOf('http'));
}else{
ret = urljoin(MY_URL,ret)
}
}
// MY_URL = getItem('MY_URL',MY_URL);
// console.log(`规则${RKEY}打印MY_URL:${MY_URL},uri:${uri}`);
return ret
}
/*** js自封装的方法 ***/ /*** js自封装的方法 ***/
/** /**
...@@ -461,14 +413,15 @@ function homeVodParse(homeVodObj){ ...@@ -461,14 +413,15 @@ function homeVodParse(homeVodObj){
// setItem('MY_URL',MY_URL); // setItem('MY_URL',MY_URL);
console.log(MY_URL); console.log(MY_URL);
let p = homeVodObj.推荐; let p = homeVodObj.推荐;
if(!p){ if(!p||typeof(p)!=='string'){
return '{}' return '{}'
} }
if(typeof(p)==='string'&&p.trim().startsWith('js:')){ p = p.trim();
if(p.startsWith('js:')){
const TYPE = 'home'; const TYPE = 'home';
var input = MY_URL; var input = MY_URL;
const HOST = rule.host; const HOST = rule.host;
eval(p.trim().replace('js:','')); eval(p.replace('js:',''));
d = VODS; d = VODS;
}else { }else {
p = p.split(';'); p = p.split(';');
...@@ -477,30 +430,34 @@ function homeVodParse(homeVodObj){ ...@@ -477,30 +430,34 @@ function homeVodParse(homeVodObj){
} else if (homeVodObj.double && p.length < 6) { } else if (homeVodObj.double && p.length < 6) {
return '{}' return '{}'
} }
let _ps = parseTags.getParse(p[0]);
_pdfa = _ps.pdfa;
_pdfh = _ps.pdfh;
_pd = _ps.pd;
p[0] = p[0].replace(/^(jsp:|json:|jq:)/,'');
// print(p[0]);
let html = getHtml(MY_URL); let html = getHtml(MY_URL);
try { try {
console.log('double:' + homeVodObj.double); console.log('double:' + homeVodObj.double);
if (homeVodObj.double) { if (homeVodObj.double) {
p[0] = p[0].trim().startsWith('json:') ? p[0].replace('json:', '') : p[0]; let items = _pdfa(html, p[0]);
// console.log(p[0]);
let items = pdfa(html, p[0]);
// console.log(items.length); // console.log(items.length);
for (let item of items) { for (let item of items) {
// console.log(p[1]); // console.log(p[1]);
let items2 = pdfa(item, p[1]); let items2 = _pdfa(item, p[1]);
// console.log(items2.length); // console.log(items2.length);
for (let item2 of items2) { for (let item2 of items2) {
try { try {
let title = pdfh(item2, p[2]); let title = _pdfh(item2, p[2]);
let img = ''; let img = '';
try { try {
img = pD(item2, p[3]) img = _pd(item2, p[3])
} catch (e) { } catch (e) {
} }
let desc = pdfh(item2, p[4]); let desc = _pdfh(item2, p[4]);
let links = []; let links = [];
for (let p5 of p[5].split('+')) { for (let p5 of p[5].split('+')) {
let link = !homeVodObj.detailUrl ? pD(item2, p5, MY_URL) : pdfh(item2, p5); let link = !homeVodObj.detailUrl ? _pd(item2, p5, MY_URL) : _pdfh(item2, p5);
links.push(link); links.push(link);
} }
let vod = { let vod = {
...@@ -509,9 +466,10 @@ function homeVodParse(homeVodObj){ ...@@ -509,9 +466,10 @@ function homeVodParse(homeVodObj){
vod_remarks: desc, vod_remarks: desc,
vod_id: links.join('$') vod_id: links.join('$')
}; };
// print(vod);
d.push(vod); d.push(vod);
} catch (e) { } catch (e) {
console.log('首页列表处理发生错误:'+e.message);
} }
} }
...@@ -521,21 +479,20 @@ function homeVodParse(homeVodObj){ ...@@ -521,21 +479,20 @@ function homeVodParse(homeVodObj){
} else { } else {
p[0] = p[0].trim().startsWith('json:') ? p[0].replace('json:', '') : p[0]; let items = _pdfa(html, p[0]);
let items = pdfa(html, p[0]);
for (let item of items) { for (let item of items) {
try { try {
let title = pdfh(item, p[1]); let title = _pdfh(item, p[1]);
let img = ''; let img = '';
try { try {
img = pD(item, p[2], MY_URL); img = _pd(item, p[2], MY_URL);
} catch (e) { } catch (e) {
} }
let desc = pdfh(item, p[3]); let desc = _pdfh(item, p[3]);
let links = []; let links = [];
for (let p5 of p[4].split('+')) { for (let p5 of p[4].split('+')) {
let link = !homeVodObj.detailUrl ? pD(item, p5, MY_URL) : pdfh(item, p5); let link = !homeVodObj.detailUrl ? _pd(item, p5, MY_URL) : _pdfh(item, p5);
links.push(link); links.push(link);
} }
let vod = { let vod = {
...@@ -571,6 +528,9 @@ function homeVodParse(homeVodObj){ ...@@ -571,6 +528,9 @@ function homeVodParse(homeVodObj){
*/ */
function categoryParse(cateObj) { function categoryParse(cateObj) {
let p = cateObj.一级; let p = cateObj.一级;
if(!p||typeof(p)!=='string'){
return '{}'
}
let d = []; let d = [];
// let url = cateObj.url.replaceAll('fyclass', cateObj.tid).replaceAll('fypage', cateObj.pg); // let url = cateObj.url.replaceAll('fyclass', cateObj.tid).replaceAll('fypage', cateObj.pg);
let url = cateObj.url.replaceAll('fyclass', cateObj.tid); let url = cateObj.url.replaceAll('fyclass', cateObj.tid);
...@@ -611,8 +571,8 @@ function categoryParse(cateObj) { ...@@ -611,8 +571,8 @@ function categoryParse(cateObj) {
MY_URL = url; MY_URL = url;
// setItem('MY_URL',MY_URL); // setItem('MY_URL',MY_URL);
console.log(MY_URL); console.log(MY_URL);
p = p.trim();
if(typeof(p)==='string'&&p.trim().startsWith('js:')){ if(p.startsWith('js:')){
const MY_CATE = cateObj.tid; const MY_CATE = cateObj.tid;
const MY_FL = cateObj.extend; const MY_FL = cateObj.extend;
const TYPE = 'cate'; const TYPE = 'cate';
...@@ -625,16 +585,21 @@ function categoryParse(cateObj) { ...@@ -625,16 +585,21 @@ function categoryParse(cateObj) {
if (p.length < 5) { if (p.length < 5) {
return '{}' return '{}'
} }
let _ps = parseTags.getParse(p[0]);
_pdfa = _ps.pdfa;
_pdfh = _ps.pdfh;
_pd = _ps.pd;
p[0] = p[0].replace(/^(jsp:|json:|jq:)/,'');
try { try {
let html = getHtml(MY_URL); let html = getHtml(MY_URL);
if (html) { if (html) {
let list = pdfa(html, p[0]); let list = _pdfa(html, p[0]);
list.forEach(it => { list.forEach(it => {
d.push({ d.push({
'vod_id': pD(it, p[4],MY_URL), 'vod_id': _pd(it, p[4],MY_URL),
'vod_name': pdfh(it, p[1]), 'vod_name': _pdfh(it, p[1]),
'vod_pic': pD(it, p[2],MY_URL), 'vod_pic': _pd(it, p[2],MY_URL),
'vod_remarks': pdfh(it, p[3]), 'vod_remarks': _pdfh(it, p[3]),
}); });
}); });
} }
...@@ -663,11 +628,15 @@ function searchParse(searchObj) { ...@@ -663,11 +628,15 @@ function searchParse(searchObj) {
return '{}' return '{}'
} }
let p = searchObj.搜索==='*'&&rule.一级 ? rule.一级 : searchObj.搜索; let p = searchObj.搜索==='*'&&rule.一级 ? rule.一级 : searchObj.搜索;
if(!p||typeof(p)!=='string'){
return '{}'
}
p = p.trim();
let url = searchObj.searchUrl.replaceAll('**', searchObj.wd).replaceAll('fypage', searchObj.pg); let url = searchObj.searchUrl.replaceAll('**', searchObj.wd).replaceAll('fypage', searchObj.pg);
MY_URL = url; MY_URL = url;
console.log(MY_URL); console.log(MY_URL);
// setItem('MY_URL',MY_URL); // setItem('MY_URL',MY_URL);
if(typeof(p)==='string'&&p.trim().startsWith('js:')){ if(p.startsWith('js:')){
const TYPE = 'search'; const TYPE = 'search';
const MY_PAGE = searchObj.pg; const MY_PAGE = searchObj.pg;
const KEY = searchObj.wd; const KEY = searchObj.wd;
...@@ -680,6 +649,11 @@ function searchParse(searchObj) { ...@@ -680,6 +649,11 @@ function searchParse(searchObj) {
if (p.length < 5) { if (p.length < 5) {
return '{}' return '{}'
} }
let _ps = parseTags.getParse(p[0]);
_pdfa = _ps.pdfa;
_pdfh = _ps.pdfh;
_pd = _ps.pd;
p[0] = p[0].replace(/^(jsp:|json:|jq:)/,'');
try { try {
let html = getHtml(MY_URL); let html = getHtml(MY_URL);
if (html) { if (html) {
...@@ -698,16 +672,16 @@ function searchParse(searchObj) { ...@@ -698,16 +672,16 @@ function searchParse(searchObj) {
console.log('搜索结果源码未包含关键字,疑似搜索失败,正为您打印结果源码'); console.log('搜索结果源码未包含关键字,疑似搜索失败,正为您打印结果源码');
console.log(html); console.log(html);
} }
let list = pdfa(html, p[0]); let list = _pdfa(html, p[0]);
list.forEach(it => { list.forEach(it => {
let ob = { let ob = {
'vod_id': pD(it, p[4],MY_URL), 'vod_id': _pd(it, p[4],MY_URL),
'vod_name': pdfh(it, p[1]), 'vod_name': _pdfh(it, p[1]),
'vod_pic': pD(it, p[2],MY_URL), 'vod_pic': _pd(it, p[2],MY_URL),
'vod_remarks': pdfh(it, p[3]), 'vod_remarks': _pdfh(it, p[3]),
}; };
if (p.length > 5 && p[5]) { if (p.length > 5 && p[5]) {
ob.vod_content = pdfh(it, p[5]); ob.vod_content = _pdfh(it, p[5]);
} }
d.push(ob); d.push(ob);
}); });
...@@ -767,20 +741,33 @@ function detailParse(detailObj){ ...@@ -767,20 +741,33 @@ function detailParse(detailObj){
if(!html){ if(!html){
html = getHtml(MY_URL); html = getHtml(MY_URL);
} }
let _ps;
if(p.is_json){
_ps = parseTags.json;
}else if(p.is_jsp){
_ps = parseTags.jsp;
}else if(p.is_jq){
_ps = parseTags.jq;
}else{
_ps = parseTags.jq;
}
_pdfa = _ps.pdfa;
_pdfh = _ps.pdfh;
_pd = _ps.pd;
if(p.title){ if(p.title){
let p1 = p.title.split(';'); let p1 = p.title.split(';');
vod.vod_name = pdfh(html, p1[0]).replaceAll('\n', ' ').trim(); vod.vod_name = _pdfh(html, p1[0]).replaceAll('\n', ' ').trim();
let type_name = p1.length > 1 ? pdfh(html, p1[1]).replaceAll('\n', ' ').trim():''; let type_name = p1.length > 1 ? _pdfh(html, p1[1]).replaceAll('\n', ' ').trim():'';
vod.type_name = type_name||vod.type_name; vod.type_name = type_name||vod.type_name;
} }
if(p.desc){ if(p.desc){
try{ try{
let p1 = p.desc.split(';'); let p1 = p.desc.split(';');
vod.vod_remarks = pdfh(html, p1[0]).replaceAll('\n', ' ').trim(); vod.vod_remarks = _pdfh(html, p1[0]).replaceAll('\n', ' ').trim();
vod.vod_year = p1.length > 1 ? pdfh(html, p1[1]).replaceAll('\n', ' ').trim():''; vod.vod_year = p1.length > 1 ? _pdfh(html, p1[1]).replaceAll('\n', ' ').trim():'';
vod.vod_area = p1.length > 2 ? pdfh(html, p1[2]).replaceAll('\n', ' ').trim():''; vod.vod_area = p1.length > 2 ? _pdfh(html, p1[2]).replaceAll('\n', ' ').trim():'';
vod.vod_actor = p1.length > 3 ? pdfh(html, p1[3]).replaceAll('\n', ' ').trim():''; vod.vod_actor = p1.length > 3 ? _pdfh(html, p1[3]).replaceAll('\n', ' ').trim():'';
vod.vod_director = p1.length > 4 ? pdfh(html, p1[4]).replaceAll('\n', ' ').trim():''; vod.vod_director = p1.length > 4 ? _pdfh(html, p1[4]).replaceAll('\n', ' ').trim():'';
} }
catch (e) { catch (e) {
...@@ -789,14 +776,14 @@ function detailParse(detailObj){ ...@@ -789,14 +776,14 @@ function detailParse(detailObj){
if(p.content){ if(p.content){
try{ try{
let p1 = p.content.split(';'); let p1 = p.content.split(';');
vod.vod_content = pdfh(html, p1[0]).replaceAll('\n', ' ').trim(); vod.vod_content = _pdfh(html, p1[0]).replaceAll('\n', ' ').trim();
} }
catch (e) {} catch (e) {}
} }
if(p.img){ if(p.img){
try{ try{
let p1 = p.img.split(';'); let p1 = p.img.split(';');
vod.vod_pic = pD(html, p1[0],MY_URL); vod.vod_pic = _pd(html, p1[0],MY_URL);
} }
catch (e) {} catch (e) {}
} }
...@@ -811,11 +798,11 @@ function detailParse(detailObj){ ...@@ -811,11 +798,11 @@ function detailParse(detailObj){
if(p.tabs){ if(p.tabs){
let p_tab = p.tabs.split(';')[0]; let p_tab = p.tabs.split(';')[0];
console.log(p_tab); console.log(p_tab);
let vHeader = pdfa(html, p_tab); let vHeader = _pdfa(html, p_tab);
console.log(vHeader.length); console.log(vHeader.length);
for(let v of vHeader){ for(let v of vHeader){
let v_title = pdfh(v,'body&&Text'); let v_title = _pdfh(v,'body&&Text');
console.log(v_title); console.log(v_title);
if(tab_exclude&& (new RegExp(tab_exclude)).test(v_title)){ if(tab_exclude&& (new RegExp(tab_exclude)).test(v_title)){
continue; continue;
...@@ -841,16 +828,17 @@ function detailParse(detailObj){ ...@@ -841,16 +828,17 @@ function detailParse(detailObj){
// console.log(html); // console.log(html);
let vodList = []; let vodList = [];
try { try {
vodList = pdfa(html, p1); vodList = _pdfa(html, p1);
console.log('len(vodList):'+vodList.length); console.log('len(vodList):'+vodList.length);
}catch (e) { }catch (e) {
// console.log(e.message); // console.log(e.message);
} }
let new_vod_list = []; let new_vod_list = [];
let tabName = tab_ext?pdfh(html, tab_ext):tab_name; let tabName = tab_ext?_pdfh(html, tab_ext):tab_name;
console.log(tabName); console.log(tabName);
vodList.forEach(it=>{ vodList.forEach(it=>{
new_vod_list.push(pdfh(it,'body&&Text')+'$'+pD(it,'a&&href',MY_URL)); // new_vod_list.push(_pdfh(it,'body&&Text')+'$'+_pd(it,'a&&href',MY_URL));
new_vod_list.push(_pdfh(it,'Text')+'$'+_pd(it,'a&&href',MY_URL));
}); });
let vlist = new_vod_list.join('#'); let vlist = new_vod_list.join('#');
vod_tab_list.push(vlist); vod_tab_list.push(vlist);
......
libs/parseTags.js 0 → 100644
浏览文件 @ 87a5059a
// import cheerio from 'https://gitcode.net/qq_32394351/dr_py/-/raw/master/libs/cheerio.min.js';
if(typeof(MY_URL)==='undefined'){
var MY_URL; // 全局注入变量,pd函数需要
}
/**
* url拼接
* @param fromPath 初始当前页面url
* @param nowPath 相对当前页面url
* @returns {*}
*/
export function urljoin(fromPath, nowPath) {
fromPath = fromPath||'';
nowPath = nowPath||'';
return joinUrl(fromPath, nowPath);
// try {
// // import Uri from './uri.min.js';
// // var Uri = require('./uri.min.js');
// // eval(request('https://cdn.bootcdn.net/ajax/libs/URI.js/1.19.11/URI.min.js'));
// // let new_uri = URI(nowPath, fromPath);
// let new_uri = Uri(nowPath, fromPath);
// new_uri = new_uri.toString();
// // console.log(new_uri);
// // return fromPath + nowPath
// return new_uri
// }
// catch (e) {
// console.log('urljoin发生错误:'+e.message);
// if(nowPath.startsWith('http')){
// return nowPath
// }if(nowPath.startsWith('/')){
// return getHome(fromPath)+nowPath
// }
// return fromPath+nowPath
// }
}
/**
* 重写pd方法-增加自动urljoin(没法重写,改个名继续骗)
* @param html
* @param parse
* @param uri
* @returns {*}
*/
export function pD(html,parse,uri){
let ret = pdfh(html,parse);
if(typeof(uri)==='undefined'||!uri){
uri = '';
}
if(/(url|src|href|data-original|data-src)$/.test(parse)){
if(/http/.test(ret)){
ret = ret.substr(ret.indexOf('http'));
}else{
ret = urljoin(MY_URL,ret)
}
}
// MY_URL = getItem('MY_URL',MY_URL);
// console.log(`规则${RKEY}打印MY_URL:${MY_URL},uri:${uri}`);
return ret
}
export var parseTags = {
jsp:{
pdfh:pdfh,
pdfa:pdfa,
pd:pD,
},
json:{
pdfh(html, parse) {
if (!parse || !parse.trim()){
return '';
}
if (typeof (html) === 'string'){
html = JSON.parse(html);
}
parse = parse.trim();
if (!parse.startsWith('$.')){
parse = '$.' + parse;
}
parse = parse.split('||');
for (let ps of parse) {
let ret = cheerio.jp(ps, html);
if (Array.isArray(ret)){
ret = ret[0] || '';
} else{
ret = ret || ''
}
if (ret && typeof (ret) !== 'string'){
ret = ret.toString();
}
if(ret){
return ret
}
}
return '';
},
pdfa(html, parse) {
if (!parse || !parse.trim()){
return '';
}
if (typeof (html) === 'string'){
html = JSON.parse(html);
}
parse = parse.trim()
if (!parse.startsWith('$.')){
parse = '$.' + parse;
}
let ret = cheerio.jp(parse, html);
if (Array.isArray(ret) && Array.isArray(ret[0]) && ret.length === 1){
return ret[0] || []
}
return ret || []
},
pd:function (html,parse){
let ret = this.pdfh(html,parse);
if(ret){
return urljoin(MY_URL,ret);
}
return ret
},
},
jq:{
pdfh:pdfh,
pdfa:pdfa,
pd:pD,
},
getParse(p0){//非js开头的情况自动获取解析标签
if(p0.startsWith('jsp:')){
return this.jsp
}else if(p0.startsWith('json:')){
return this.json
}else if(p0.startsWith('jq:')){
return this.jq
}else {
return this.jq
}
}
};
export var stringify = JSON.stringify;
\ No newline at end of file
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或