From 876def43928241ca33b064732ae32a316cef2abb Mon Sep 17 00:00:00 2001
From: Godzil <godzil@godzil.net>
Date: Sat, 11 Aug 2018 20:42:12 +0100
Subject: [PATCH] Add code to check what langage CR is serving the page, and
 try to adapt some regexp to that. The langage can be forced by the user

Fix #1 and Fix #76
---
 src/batch.ts               |  1 +
 src/interface/IConfig.d.ts |  1 +
 src/languages.ts           | 47 ++++++++++++++++++++++++++++++++++++++
 src/my_request.ts          | 17 ++++++++++++++
 src/series.ts              |  5 ++--
 5 files changed, 69 insertions(+), 2 deletions(-)
 create mode 100644 src/languages.ts

diff --git a/src/batch.ts b/src/batch.ts
index d88d632..fb435f8 100644
--- a/src/batch.ts
+++ b/src/batch.ts
@@ -349,6 +349,7 @@ function parse(args: string[]): IConfigLine
     // Episode filter
     .option('-e, --episodes <s>', 'Episode list. Read documentation on how to use')
     // Settings
+    .option('-l, --crlang <s>', 'CR page language (valid: en, fr, es, it, pt, de, ru).')
     .option('-f, --format <s>', 'The subtitle format.', 'ass')
     .option('-o, --output <s>', 'The output path.')
     .option('-s, --series <s>', 'The series name override.')
diff --git a/src/interface/IConfig.d.ts b/src/interface/IConfig.d.ts
index 957c840..7952070 100644
--- a/src/interface/IConfig.d.ts
+++ b/src/interface/IConfig.d.ts
@@ -7,6 +7,7 @@ interface IConfig {
   merge?: boolean;
   episodes?: string;
   // Settings
+  crlang?: string;
   format?: string;
   output?: string;
   series?: string;
diff --git a/src/languages.ts b/src/languages.ts
new file mode 100644
index 0000000..9c89b43
--- /dev/null
+++ b/src/languages.ts
@@ -0,0 +1,47 @@
+'use strict';
+
+const localeCC: { [id: string]: string; } =
+{
+  enUS: 'en', enGB: 'en',
+  esLA: 'es', esES: 'es',
+  ptPT: 'pt', ptBR: 'pt',
+  frFR: 'fr',
+  deDE: 'de',
+  itIT: 'it',
+  ruRU: 'ru',
+};
+
+export function localeToCC(locale: string): string
+{
+  let ret = localeCC.enGB;
+
+  if (locale in localeCC)
+  {
+    ret = localeCC[locale];
+  }
+
+  return ret;
+}
+
+const episodes_regexp: { [id: string]: RegExp; } =
+{
+  en: /Episode\s+((OVA)|(PV )?[S0-9][\-P0-9.]*[a-fA-F]?)\s*$/i,
+  fr: /Épisode\s+((OVA)|(PV )?[S0-9][\-P0-9.]*[a-fA-F]?)\s*$/i,
+  de: /Folge\s+((OVA)|(PV )?[S0-9][\-P0-9.]*[a-fA-F]?)\s*$/i,
+  es: /Episodio\s+((OVA)|(PV )?[S0-9][\-P0-9.]*[a-fA-F]?)\s*$/i,
+  it: /Episodio\s+((OVA)|(PV )?[S0-9][\-P0-9.]*[a-fA-F]?)\s*$/i,
+  pt: /Episódio\s+((OVA)|(PV )?[S0-9][\-P0-9.]*[a-fA-F]?)\s*$/i,
+  ru: /Серия\s+((OVA)|(PV )?[S0-9][\-P0-9.]*[a-fA-F]?)\s*$/i,
+};
+
+export function get_epregexp(config: IConfig): RegExp
+{
+  let ret = episodes_regexp.en;
+
+  if (config.crlang in episodes_regexp)
+  {
+    ret = episodes_regexp[config.crlang];
+  }
+
+  return ret;
+}
\ No newline at end of file
diff --git a/src/my_request.ts b/src/my_request.ts
index 5acb4dd..331a545 100644
--- a/src/my_request.ts
+++ b/src/my_request.ts
@@ -6,6 +6,7 @@ import Promise = require('bluebird');
 import uuid = require('uuid');
 import path = require('path');
 import fs = require('fs-extra');
+import languages = require('./languages');
 import log = require('./log');
 
 import { RequestPromise } from 'request-promise';
@@ -111,6 +112,22 @@ function checkIfUserIsAuth(config: IConfig, done: (err: Error) => void): void
 
     const $ = cheerio.load(body);
 
+    /* As we are here, try to detect which locale CR tell us */
+    const localeRE = /LOCALE = "([a-zA-Z]+)",/g;
+    const locale = localeRE.exec($('script').text())[1];
+    const countryCode = languages.localeToCC(locale);
+
+    if (config.crlang === undefined)
+    {
+      log.info('No locale set. Setting to the one reported by CR: "' + countryCode + '"');
+      config.crlang = countryCode;
+    }
+    else if (config.crlang !== countryCode)
+    {
+      log.warn('Crunchy is configured for locale "' + config.crlang + '" but CR report "' + countryCode + '" (LOCALE = ' + locale + ')');
+      log.warn('Check if it is correct or rerun (once) with "-l ' + countryCode + '" to correct.');
+    }
+
     /* Check if auth worked */
     const regexps = /ga\('set', 'dimension[5-8]', '([^']*)'\);/g;
     const dims = regexps.exec($('script').text());
diff --git a/src/series.ts b/src/series.ts
index 3d28129..693351f 100644
--- a/src/series.ts
+++ b/src/series.ts
@@ -1,12 +1,13 @@
 'use strict';
 import cheerio = require('cheerio');
 import episode from './episode';
-// import fs = require('fs');
 import fs = require('fs-extra');
 import my_request = require('./my_request');
 import path = require('path');
 import url = require('url');
 import log  = require('./log');
+import languages = require('./languages');
+
 const persistent = '.crpersistent';
 
 /**
@@ -234,7 +235,7 @@ function pageScrape(config: IConfig, task: IConfigTask, done: (err: any, result?
 
         const season_name = $(el).closest('ul').prev('a').text();
         const volume = /([0-9]+)\s*$/.exec($(el).closest('ul').prev('a').text());
-        const regexp = /Episode\s+((PV )?[S0-9][\-P0-9.]*[a-fA-F]?)\s*$/i;
+        const regexp = languages.get_epregexp(config);
         const episode = regexp.exec($(el).children('.series-title').text());
         const url = $(el).attr('href');