Merge remote-tracking branch 'upstream/master'

.github/workflows/test.yml

@@ -0,0 +1,15 @@
+name: test
+on:
+  push:
+    branches:
+      - master
+  workflow_dispatch:
+  pull_request:
+    types: [opened, synchronize, reopened, edited]
+jobs:
+  main:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v3
+      - run: npm install
+      - run: npm test

.travis.yml

@@ -1,8 +0,0 @@
-language: node_js
-
-node_js:
-  - '10'
-
-script:
-  - npm run lint
-  - npm run test

README.md

@@ -1,4 +1,4 @@
-# EPG Grabber [![Build Status](https://app.travis-ci.com/freearhey/epg-grabber.svg?branch=master)](https://app.travis-ci.com/freearhey/epg-grabber)
+# EPG Grabber [![test](https://github.com/freearhey/epg-grabber/actions/workflows/test.yml/badge.svg)](https://github.com/freearhey/epg-grabber/actions/workflows/test.yml)
 
 Node.js CLI tool for grabbing EPG from different websites.
 
@@ -74,29 +74,31 @@ epg-grabber --config=example.com.config.js
 Arguments:
 
 - `-c, --config`: path to config file
-- `-o, --output`: path to output file (default: 'guide.xml')
+- `-o, --output`: path to output file or path template (example: `guides/{site}.{lang}.xml`; default: `guide.xml`)
-- `--channels`: path to list of channels (can be specified via config file)
+- `--channels`: path to list of channels; you can also use wildcard to specify the path to multiple files at once (example: `example.com_*.channels.xml`)
-- `--lang`: set default language for all programs (default: 'en')
+- `--lang`: set default language for all programs (default: `en`)
-- `--days`: number of days for which to grab the program (default: 1)
+- `--days`: number of days for which to grab the program (default: `1`)
-- `--delay`: delay between requests in milliseconds (default: 3000)
+- `--delay`: delay between requests in milliseconds (default: `3000`)
-- `--timeout`: set a timeout for each request in milliseconds (default: 5000)
+- `--timeout`: set a timeout for each request in milliseconds (default: `5000`)
-- `--cache-ttl`: maximum time for storing each request in milliseconds (default: 0)
+- `--max-connections`: set a limit on the number of concurrent requests per site (default: `1`)
-- `--gzip`: compress the output (default: false)
+- `--cache-ttl`: maximum time for storing each request in milliseconds (default: `0`)
-- `--debug`: enable debug mode (default: false)
+- `--gzip`: compress the output (default: `false`)
-- `--curl`: display current request as CURL (default: false)
+- `--debug`: enable debug mode (default: `false`)
+- `--curl`: display current request as CURL (default: `false`)
 - `--log`: path to log file (optional)
-- `--log-level`: set the log level (default: 'info')
+- `--log-level`: set the log level (default: `info`)
 
 ## Site Config
 
 ```js
 module.exports = {
   site: 'example.com', // site domain name (required)
-  output: 'example.com.guide.xml', // path to output file (default: 'guide.xml')
+  output: 'example.com.guide.xml', // path to output file or path template (example: 'guides/{site}.{lang}.xml'; default: 'guide.xml')
-  channels: 'example.com.channels.xml', // path to channels.xml file (required)
+  channels: 'example.com.channels.xml', // path to list of channels; you can also use an array to specify the path to multiple files at once (example: ['channels1.xml', 'channels2.xml']; required)
   lang: 'fr', // default language for all programs (default: 'en')
   days: 3, // number of days for which to grab the program (default: 1)
   delay: 5000, // delay between requests (default: 3000)
+  maxConnections: 200, // limit on the number of concurrent requests (default: 1)
 
   request: { // request options (details: https://github.com/axios/axios#request-config)
 
@@ -235,6 +237,34 @@ You can also specify the language and logo for each channel individually, like s
 >France 24</channel>
 ```
 
+## How to use SOCKS proxy?
+
+First, you need to install [socks-proxy-agent](https://www.npmjs.com/package/socks-proxy-agent):
+
+```sh
+npm install socks-proxy-agent
+```
+
+Then you can use it to create an agent that acts as a SOCKS proxy. Here is an example of how to do it with the Tor SOCKS proxy:
+
+```js
+const { SocksProxyAgent } = require('socks-proxy-agent')
+
+const torProxyAgent = new SocksProxyAgent('socks://localhost:9050')
+
+module.exports = {
+  site: 'example.com',
+  url: 'https://example.com/epg.json',
+  request: {
+    httpsAgent: torProxyAgent,
+    httpAgent: torProxyAgent
+  },
+  parser(context) {
+    // ...
+  }
+}
+```
+
 ## Contribution
 
 If you find a bug or want to contribute to the code or documentation, you can help by submitting an [issue](https://github.com/freearhey/epg-grabber/issues) or a [pull request](https://github.com/freearhey/epg-grabber/pulls).

bin/epg-grabber.js

@@ -12,6 +12,7 @@ const { name, version, description } = require('../package.json')
 const _ = require('lodash')
 const dayjs = require('dayjs')
 const utc = require('dayjs/plugin/utc')
+const { TaskQueue } = require('cwait')
 
 dayjs.extend(utc)
 
@@ -21,11 +22,16 @@ program
   .description(description)
   .requiredOption('-c, --config <config>', 'Path to [site].config.js file')
   .option('-o, --output <output>', 'Path to output file')
-  .option('--channels <channels>', 'Path to channels.xml file')
+  .option('--channels <channels>', 'Path to list of channels')
   .option('--lang <lang>', 'Set default language for all programs')
   .option('--days <days>', 'Number of days for which to grab the program', parseNumber)
   .option('--delay <delay>', 'Delay between requests (in milliseconds)', parseNumber)
   .option('--timeout <timeout>', 'Set a timeout for each request (in milliseconds)', parseNumber)
+  .option(
+    '--max-connections <maxConnections>',
+    'Set a limit on the number of concurrent requests per site',
+    parseNumber
+  )
   .option(
     '--cache-ttl <cacheTtl>',
     'Maximum time for storing each request (in milliseconds)',
@@ -53,6 +59,7 @@ async function main() {
     curl: options.curl,
     lang: options.lang,
     delay: options.delay,
+    maxConnections: options.maxConnections,
     request: {
         ignoreCookieErrors: true,
     }
@@ -60,34 +67,75 @@ async function main() {
 
   if (options.timeout) config.request.timeout = options.timeout
   if (options.cacheTtl) config.request.cache.ttl = options.cacheTtl
-  if (options.channels) config.channels = options.channels
-  else if (config.channels)
-    config.channels = file.join(file.dirname(options.config), config.channels)
-  else throw new Error("The required 'channels' property is missing")
 
-  if (!config.channels) return logger.error('Path to [site].channels.xml is missing')
+  if (options.channels) config.channels = options.channels
-  logger.info(`Loading '${config.channels}'...`)
+
+  let parsedChannels = []
+  if (config.channels) {
+    const dir = file.dirname(options.config)
+
+    let files = []
+    if (Array.isArray(config.channels)) {
+      files = config.channels.map(path => file.join(dir, path))
+    } else if (typeof config.channels === 'string') {
+      files = await file.list(config.channels)
+    } else {
+      throw new Error('The "channels" attribute must be of type array or string')
+    }
+
+    for (let filepath of files) {
+      logger.info(`Loading '${filepath}'...`)
+      const channelsXML = file.read(filepath)
+      const { channels } = parseChannels(channelsXML)
+      parsedChannels = parsedChannels.concat(channels)
+    }
+  } else throw new Error('Path to "channels" is missing')
+
   const grabber = new EPGGrabber(config)
 
-  const channelsXML = file.read(config.channels)
+  let template = options.output || config.output
-  const { channels } = parseChannels(channelsXML)
+  const variables = file.templateVariables(template)
 
+  const groups = _.groupBy(parsedChannels, channel => {
+    let groupId = ''
+    for (let key in channel) {
+      if (variables.includes(key)) {
+        groupId += channel[key]
+      }
+    }
+
+    return groupId
+  })
+
+  for (let groupId in groups) {
+    const channels = groups[groupId]
     let programs = []
     let i = 1
     let days = config.days || 1
+    const maxConnections = config.maxConnections || 1
     const total = channels.length * days
     const utcDate = getUTCDate()
     const dates = Array.from({ length: days }, (_, i) => utcDate.add(i, 'd'))
+    const taskQueue = new TaskQueue(Promise, maxConnections)
+
+    let queue = []
     for (let channel of channels) {
       if (!channel.logo && config.logo) {
         channel.logo = await grabber.loadLogo(channel)
       }
 
       for (let date of dates) {
+        queue.push({ channel, date })
+      }
+    }
+
+    await Promise.all(
+      queue.map(
+        taskQueue.wrap(async ({ channel, date }) => {
           await grabber
             .grab(channel, date, (data, err) => {
               logger.info(
-            `[${i}/${total}] ${config.site} - ${data.channel.id} - ${dayjs
+                `[${i}/${total}] ${config.site} - ${data.channel.xmltv_id} - ${dayjs
                   .utc(data.date)
                   .format('MMM D, YYYY')} (${data.programs.length} programs)`
               )
@@ -99,13 +147,14 @@ async function main() {
             .then(results => {
               programs = programs.concat(results)
             })
-    }
+        })
-  }
+      )
+    )
 
     programs = _.uniqBy(programs, p => p.start + p.channel)
 
     const xml = generateXMLTV({ channels, programs })
-  let outputPath = options.output || config.output
+    let outputPath = file.templateFormat(template, channels[0])
     if (options.gzip) {
       outputPath = outputPath || 'guide.xml.gz'
       const compressed = await gzip(xml)
@@ -116,6 +165,8 @@ async function main() {
     }
 
     logger.info(`File '${outputPath}' successfully saved`)
+  }
+
   logger.info('Finish')
 }
 

package.json

@@ -1,6 +1,6 @@
 {
   "name": "epg-grabber",
-  "version": "0.29.8",
+  "version": "0.32.0",
   "description": "Node.js CLI tool for grabbing EPG from different sites",
   "main": "src/index.js",
   "preferGlobal": true,
@@ -34,8 +34,10 @@
     "axios-mock-adapter": "^1.20.0",
     "commander": "^7.1.0",
     "curl-generator": "^0.2.0",
+    "cwait": "^1.1.2",
     "dayjs": "^1.10.4",
     "epg-parser": "^0.1.6",
+    "fs-extra": "^11.1.1",
     "glob": "^7.1.6",
     "lodash": "^4.17.21",
     "node-gzip": "^1.1.2",

src/Channel.js

@@ -1,7 +1,7 @@
 class Channel {
   constructor(c) {
     const data = {
-			id: c.id || c.xmltv_id,
+      xmltv_id: c.xmltv_id,
       name: c.name,
       site: c.site || '',
       site_id: c.site_id,

src/Program.js

@@ -10,7 +10,7 @@ class Program {
 
     const data = {
       site: c.site || '',
-			channel: c.id || '',
+      channel: c.xmltv_id || '',
       titles: toArray(p.titles || p.title).map(text => toTextObject(text, c.lang)),
       sub_titles: toArray(p.sub_titles || p.sub_title).map(text => toTextObject(text, c.lang)),
       descriptions: toArray(p.descriptions || p.description || p.desc).map(text =>

src/file.js

@@ -1,11 +1,23 @@
 const fs = require('fs')
 const path = require('path')
+const glob = require('glob')
 
+module.exports.list = list
 module.exports.read = read
 module.exports.write = write
 module.exports.resolve = resolve
 module.exports.join = join
 module.exports.dirname = dirname
+module.exports.templateVariables = templateVariables
+module.exports.templateFormat = templateFormat
+
+function list(pattern) {
+  return new Promise(resolve => {
+    glob(pattern, function (err, files) {
+      resolve(files)
+    })
+  })
+}
 
 function read(filepath) {
   return fs.readFileSync(path.resolve(filepath), { encoding: 'utf-8' })
@@ -31,3 +43,20 @@ function join(path1, path2) {
 function dirname(filepath) {
   return path.dirname(filepath)
 }
+
+function templateVariables(template) {
+  const match = template.match(/{[^}]+}/g)
+
+  return Array.isArray(match) ? match.map(s => s.substring(1, s.length - 1)) : []
+}
+
+function templateFormat(template, obj) {
+  let output = template
+  for (let key in obj) {
+    const regex = new RegExp(`{${key}}`, 'g')
+    const value = obj[key] || undefined
+    output = output.replace(regex, value)
+  }
+
+  return output
+}

src/xmltv.js

@@ -29,7 +29,7 @@ function createElements(channels, programs, date) {
     ...channels.map(channel => {
       return (
         '\r\n' +
-        el('channel', { id: channel.id }, [
+        el('channel', { id: channel.xmltv_id }, [
           el('display-name', {}, [escapeString(channel.name)]),
           el('icon', { src: channel.logo }),
           el('url', {}, [channel.url])

tests/Channel.test.js

@@ -12,28 +12,7 @@ it('can create new Channel', () => {
 
   expect(channel).toMatchObject({
     name: '1 TV',
-    id: '1TV.com',
+    xmltv_id: '1TV.com',
-    site_id: '1',
-    site: 'example.com',
-    url: 'https://example.com',
-    lang: 'fr',
-    logo: 'https://example.com/logos/1TV.png'
-  })
-})
-
-it('can create channel from exist object', () => {
-  const channel = new Channel({
-    name: '1 TV',
-    id: '1TV.com',
-    site_id: '1',
-    site: 'example.com',
-    lang: 'fr',
-    logo: 'https://example.com/logos/1TV.png'
-  })
-
-  expect(channel).toMatchObject({
-    name: '1 TV',
-    id: '1TV.com',
     site_id: '1',
     site: 'example.com',
     url: 'https://example.com',

tests/__data__/input/example.config.js

@@ -7,7 +7,7 @@ module.exports = {
   site: 'example.com',
   days: 2,
   channels: 'example.channels.xml',
-  output: 'tests/output/guide.xml',
+  output: 'tests/__data__/output/guide.xml',
   url: () => 'http://example.com/20210319/1tv.json',
   request: {
     method: 'POST',

tests/__data__/input/example_2.channels.xml

@@ -0,0 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<site site="example2.com">
+  <channels>
+    <channel xmltv_id="3TV.com" site_id="3">3 TV</channel>
+    <channel xmltv_id="4TV.com" site_id="4">4 TV</channel>
+  </channels>
+</site>

tests/__data__/input/example_channels.config.js

@@ -0,0 +1,32 @@
+const dayjs = require('dayjs')
+const utc = require('dayjs/plugin/utc')
+
+dayjs.extend(utc)
+
+module.exports = {
+  site: 'example.com',
+  days: 2,
+  channels: ['example.channels.xml', 'example_2.channels.xml'],
+  output: 'tests/__data__/output/guide.xml',
+  url: () => 'http://example.com/20210319/1tv.json',
+  request: {
+    method: 'POST',
+    headers: {
+      'Content-Type': 'application/json',
+      Cookie: 'abc=123'
+    },
+    data() {
+      return { accountID: '123' }
+    }
+  },
+  parser: () => {
+    return [
+      {
+        title: 'Program1',
+        start: 1640995200000,
+        stop: 1640998800000
+      }
+    ]
+  },
+  logo: () => 'http://example.com/logos/1TV.png?x=шеллы&sid=777'
+}

tests/bin.test.js

@@ -1,5 +1,5 @@
 const { execSync } = require('child_process')
-const fs = require('fs')
+const fs = require('fs-extra')
 const path = require('path')
 const epgParser = require('epg-parser')
 
@@ -11,6 +11,10 @@ function stdoutResultTester(stdout) {
   })
 }
 
+beforeEach(() => {
+  fs.emptyDirSync('tests/__data__/output')
+})
+
 it('can load config', () => {
   const stdout = execSync(
     `node ${pwd}/bin/epg-grabber.js --config=tests/__data__/input/example.config.js --delay=0`,
@@ -62,6 +66,26 @@ it('can generate gzip version', () => {
   )
 })
 
+it('can produce multiple outputs', () => {
+  const stdout = execSync(
+    `node ${pwd}/bin/epg-grabber.js \
+      --config=tests/__data__/input/mini.config.js \
+      --channels=tests/__data__/input/example.channels.xml \
+      --output=tests/__data__/output/{lang}/{xmltv_id}.xml`,
+    {
+      encoding: 'utf8'
+    }
+  )
+
+  expect(stdoutResultTester(stdout)).toBe(true)
+  expect(stdout.includes("File 'tests/__data__/output/fr/1TV.com.xml' successfully saved")).toBe(
+    true
+  )
+  expect(
+    stdout.includes("File 'tests/__data__/output/undefined/2TV.com.xml' successfully saved")
+  ).toBe(true)
+})
+
 it('removes duplicates of the program', () => {
   const stdout = execSync(
     `node ${pwd}/bin/epg-grabber.js \
@@ -81,3 +105,25 @@ it('removes duplicates of the program', () => {
 
   expect(output.programs).toEqual(expected.programs)
 })
+
+it('can load multiple "channels.xml" files at once', () => {
+  const stdout = execSync(
+    `node ${pwd}/bin/epg-grabber.js --config=tests/__data__/input/example.config.js --channels=tests/__data__/input/example*.channels.xml --timeout=1`,
+    {
+      encoding: 'utf8'
+    }
+  )
+
+  expect(stdoutResultTester(stdout)).toBe(true)
+})
+
+it('can parse list of "channels.xml" from array', () => {
+  const stdout = execSync(
+    `node ${pwd}/bin/epg-grabber.js --config=tests/__data__/input/example_channels.config.js --timeout=1`,
+    {
+      encoding: 'utf8'
+    }
+  )
+
+  expect(stdoutResultTester(stdout)).toBe(true)
+})