|
| 1 | +const Promise = require('bluebird') |
| 2 | +Promise.config({ longStackTraces: true, warnings: true }) |
| 3 | +const cheerio = require('cheerio') |
| 4 | +const needle = require('needle') |
| 5 | +const debug = require('debug')('mh:docker:kind:mirrors') |
| 6 | +const fs = Promise.promisifyAll(require('fs')) |
| 7 | +const readline = require('readline') |
| 8 | +const path = require('path') |
| 9 | + |
| 10 | +class Fetch { |
| 11 | + |
| 12 | + static async fetchAndCheerio(url){ |
| 13 | + debug('retriving url: %s', url) |
| 14 | + let response = await needle('get', url) |
| 15 | + debug('retrieved response:', response.body) |
| 16 | + return cheerio.load(response.body) |
| 17 | + } |
| 18 | + |
| 19 | + static /*async*/ fetchCentos(){ |
| 20 | + return new Promise((resolve,reject) => { |
| 21 | + let response = needle.get('https://www.centos.org/download/full-mirrorlist.csv') |
| 22 | + const line_reader = require('readline').createInterface({ |
| 23 | + input: response |
| 24 | + }) |
| 25 | + |
| 26 | + const mirrors = [] |
| 27 | + |
| 28 | + line_reader.on('line', line => { |
| 29 | + // poor mans csv parsing, remove leading and trailing `"`s |
| 30 | + // then splut the string on `","`. |
| 31 | + // Only works when every field is quoted with `"` |
| 32 | + line.replace(/^"/,'').replace(/"$/,'') |
| 33 | + let fields = line.split(/","/) |
| 34 | + if ( fields.length > 4 && fields[4] ) { |
| 35 | + //debug('fields', fields) |
| 36 | + mirrors.push(fields[4]) |
| 37 | + } |
| 38 | + else { |
| 39 | + if ( |
| 40 | + fields.length === 1 || |
| 41 | + ( fields[5] && fields[5].includes('ftp') ) |
| 42 | + ) return |
| 43 | + console.error('centos - bad line', line) |
| 44 | + } |
| 45 | + }) |
| 46 | + |
| 47 | + line_reader.on('close', ()=>{ |
| 48 | + debug('centos mirrors', mirrors.join("\n")) |
| 49 | + resolve(mirrors) |
| 50 | + }) |
| 51 | + |
| 52 | + line_reader.on('error', ()=> reject(error)) |
| 53 | + }) |
| 54 | + } |
| 55 | + |
| 56 | + static async fetchEpel(){ |
| 57 | + let $ = await this.fetchAndCheerio('https://admin.fedoraproject.org/mirrormanager/mirrors/EPEL') |
| 58 | + let $rows = $('.container table').first().find('tr') |
| 59 | + let mirrors = [] |
| 60 | + $rows.each((row,el) => { |
| 61 | + let mirror_info = $(el).children('td').slice(3,4).contents() |
| 62 | + let mode = null |
| 63 | + mirror_info.each((mirror_data, el) => { |
| 64 | + if ( el.type === 'text' ) { |
| 65 | + if ( /Fedora EPEL/.exec(el.data) ) mode = 'epel' |
| 66 | + if ( /Fedora Linux/.exec(el.data) ) mode = 'fedora' |
| 67 | + } |
| 68 | + if ( mode === 'epel' && el.type === 'tag' && el.name === 'a' && $(el).text() === 'http' ) { |
| 69 | + mirrors.push($(el).attr('href')) |
| 70 | + } |
| 71 | + }) |
| 72 | + }) |
| 73 | + debug('epel mirrors', mirrors.join('\n')) |
| 74 | + return mirrors |
| 75 | + } |
| 76 | + |
| 77 | + static async fetchFedora(){ |
| 78 | + let $ = await this.fetchAndCheerio('https://admin.fedoraproject.org/mirrormanager/mirrors/Fedora') |
| 79 | + let $rows = $('.container table').first().find('tr') |
| 80 | + let mirrors = [] |
| 81 | + /* |
| 82 | + $rows.map((row,el) => { |
| 83 | + return $(el).children('td').slice(3,4).find('a').each((linki, el)=> { |
| 84 | + if ( $(el).text() === 'http' ) mirrors.push($(el).attr('href')) |
| 85 | + }) |
| 86 | + }) |
| 87 | + */ |
| 88 | + $rows.each((row,el) => { |
| 89 | + let mirror_info = $(el).children('td').slice(3,4).contents() |
| 90 | + let mode = null |
| 91 | + mirror_info.each((mirror_data, el) => { |
| 92 | + if ( el.type === 'text' ) { |
| 93 | + if ( /Fedora EPEL/.exec(el.data) ) mode = 'epel' |
| 94 | + if ( /Fedora Linux/.exec(el.data) ) mode = 'fedora' |
| 95 | + } |
| 96 | + if ( mode === 'fedora' && el.type === 'tag' && el.name === 'a' && $(el).text() === 'http' ) { |
| 97 | + mirrors.push($(el).attr('href')) |
| 98 | + } |
| 99 | + }) |
| 100 | + }) |
| 101 | + debug('fedora mirrors', mirrors.join('\n')) |
| 102 | + return mirrors |
| 103 | + } |
| 104 | + |
| 105 | + static async fetchApache(){ |
| 106 | + let $ = await this.fetchAndCheerio('https://www.apache.org/mirrors/dist.html') |
| 107 | + let $rows = $('table tr') |
| 108 | + let mirrors = [] |
| 109 | + $rows.each((rowi, el) => { |
| 110 | + let $el = $(el) |
| 111 | + let cols = $el.find('td') |
| 112 | + debug('apache row %s: size %s: ', rowi, cols.length, $(cols).text()) |
| 113 | + if ( cols.length === 5 ) { |
| 114 | + let $mirror_col = $( $(cols).get(0) ) |
| 115 | + let $scheme_col = $( $(cols).get(1) ) |
| 116 | + let is_http = $scheme_col.text().includes('http') |
| 117 | + let mirror_link = $mirror_col.find('a').first().attr('href') |
| 118 | + debug('is_https: %s mirror: %s', is_http, mirror_link) |
| 119 | + if ( is_http === true ) mirrors.push(mirror_link) |
| 120 | + } |
| 121 | + }) |
| 122 | + return mirrors |
| 123 | + } |
| 124 | + |
| 125 | + static async writeMirror( file, promise ){ |
| 126 | + let file_path = path.resolve( __dirname, '..', 'files', file ) |
| 127 | + if ( typeof promise === 'function') promise = promise() |
| 128 | + let mirror_data = await promise |
| 129 | + debug('writeMirror has got the mirror data for file "%s"', file) |
| 130 | + return fs.writeFileAsync(file_path, mirror_data.join('\n')) |
| 131 | + } |
| 132 | + |
| 133 | + static async go(){ |
| 134 | + try { |
| 135 | + await this.writeMirror('centos_mirrors', ()=>this.fetchCentos()) |
| 136 | + await this.writeMirror('fedora_mirrors', ()=>this.fetchFedora()) |
| 137 | + await this.writeMirror('epel_mirrors', ()=>this.fetchEpel()) |
| 138 | + await this.writeMirror('apache_mirrors', ()=>this.fetchApache()) |
| 139 | + } |
| 140 | + catch (error) { |
| 141 | + console.log(error) |
| 142 | + } |
| 143 | + |
| 144 | + } |
| 145 | +} |
| 146 | + |
| 147 | +Fetch.go() |
0 commit comments