Issue 2109 - Allow for translation of app independent repositories

localeTools.py

Index: localeTools.py
diff --git a/localeTools.py b/localeTools.py
index 00346ad885f69dc32e58813a187ec9078e66405c..c925f800ceae9561dd9c789d2e2970c4c81d5b84 100644
--- a/localeTools.py
+++ b/localeTools.py
@@ -108,8 +108,10 @@ def unescapeEntity(value):
   return value.replace('&amp;', '&').replace('&lt;', '<').replace('&gt;', '>').replace('&quot;', '"')
 
 def mapLocale(type, locale):
-  mapping = langMappingChrome if type == 'chrome' else langMappingGecko
-  return mapping.get(locale, locale)
+  if type == 'chrome' or type == 'ISO-15897':

[Sebastian Noack, 2016/02/12 16:33:58]

Nit: type in {'chrome', 'ISO-15897'}

But why do we need to support both in the first place?

[Wladimir Palant, 2016/02/12 17:13:04]

This logic is wrong - you cannot decide on the mapping based on both name_format
and target_platforms, you have to choose one.

In fact, what we care about here is name_format - not our target platform but
rather which format we chose for locale naming. Consequently, langMappingChrome
should be renamed into langMappingISO and langMappingGecko into langMappingBCP.

[kzar, 2016/02/12 17:28:15]

Done.

+    return langMappingChrome.get(locale, locale)
+  else:
+    return langMappingGecko.get(locale, locale)
 
 def parseDTDString(data, path):
   result = []
@@ -272,27 +274,27 @@ def postprocessChromeLocale(path, data):
   json.dump(parsed, file, ensure_ascii=False, sort_keys=True, indent=2, separators=(',', ': '))
   file.close()
 
-def setupTranslations(type, locales, projectName, key):
-  # Copy locales list, we don't want to change the parameter
-  locales = set(locales)
+def setupTranslations(projectName, localeConfig, key):
+  # Make a new set from the locales list, mapping to Crowdin friendly format
+  locales = set(map(lambda locale: mapLocale(localeConfig['name_format'],

[Sebastian Noack, 2016/02/12 16:33:58]

Why not using a set expression?

  {mapLocale(localeConfig['name_format'], locale) for locale in
localeConfig['locales']}

[kzar, 2016/02/12 17:28:15]

Good point, Done.

+                                             locale),
+                    localeConfig['locales'].keys()))
 
   # Fill up with locales that we don't have but the browser supports
-  if type == 'chrome':
+  if 'chrome' in localeConfig['target_platforms']:
     for locale in chromeLocales:
-      locales.add(locale)
-  else:
+      locales.add(mapLocale('chrome', locale))

[Wladimir Palant, 2016/02/12 17:13:04]

mapLocale('ISO-15897', ...) please, that's the format we have here.

[kzar, 2016/02/12 18:25:26]

Yep, already changed those in second patch set. Same below.

+
+  if 'gecko' in localeConfig['target_platforms']:
     firefoxLocales = urllib2.urlopen('http://www.mozilla.org/en-US/firefox/all.html').read()
     for match in re.finditer(r'&lang=([\w\-]+)"', firefoxLocales):
-      locales.add(mapLocale(type, match.group(1)))
+      locales.add(mapLocale('gecko', match.group(1)))

[Wladimir Palant, 2016/02/12 17:13:04]

mapLocale('BCP-47', ...) please, that's the format we get here.

     langPacks = urllib2.urlopen('https://addons.mozilla.org/en-US/firefox/language-tools/').read()
     for match in re.finditer(r'<tr>.*?</tr>', langPacks, re.S):
       if match.group(0).find('Install Language Pack') >= 0:
         match2 = re.search(r'lang="([\w\-]+)"', match.group(0))
         if match2:
-          locales.add(mapLocale(type, match2.group(1)))
-
-  # Convert locale codes to the ones that Crowdin will understand
-  locales = set(map(lambda locale: mapLocale(type, locale), locales))
+          locales.add(mapLocale('gecko', match2.group(1)))

[Wladimir Palant, 2016/02/12 17:13:04]

mapLocale('BCP-47', ...) please, that's the format we get here.

 
   allowed = set()
   allowedLocales = urllib2.urlopen('http://crowdin.net/page/language-codes').read()
@@ -327,7 +329,7 @@ def postFiles(files, url):
   if result.find('<success') < 0:
     raise Exception('Server indicated that the operation was not successful\n' + result)
 
-def updateTranslationMaster(type, metadata, dir, projectName, key):
+def updateTranslationMaster(metadata, dir, projectName, localeConfig, key):
   result = json.load(urllib2.urlopen('http://api.crowdin.net/api/project/%s/info?key=%s&json=1' % (projectName, key)))
 
   existing = set(map(lambda f: f['name'], result['files']))
@@ -336,14 +338,15 @@ def updateTranslationMaster(type, metadata, dir, projectName, key):
   for file in os.listdir(dir):
     path = os.path.join(dir, file)
     if os.path.isfile(path):
-      if type == 'chrome' and file.endswith('.json'):
-        data = preprocessChromeLocale(path, metadata, True)
-        newName = file
-      elif type == 'chrome':
-        fileHandle = codecs.open(path, 'rb', encoding='utf-8')
-        data = json.dumps({file: {'message': fileHandle.read()}})
-        fileHandle.close()
-        newName = file + '.json'
+      if localeConfig['file_format'] == 'chrome-json':
+        if file.endswith('.json'):
+          data = preprocessChromeLocale(path, metadata, True)
+          newName = file
+        else:
+          fileHandle = codecs.open(path, 'rb', encoding='utf-8')
+          data = json.dumps({file: {'message': fileHandle.read()}})
+          fileHandle.close()
+          newName = file + '.json'
       else:
         data = toJSON(path)
         newName = file + '.json'
@@ -365,19 +368,20 @@ def updateTranslationMaster(type, metadata, dir, projectName, key):
     if result.find('<success') < 0:
       raise Exception('Server indicated that the operation was not successful\n' + result)
 
-def uploadTranslations(type, metadata, dir, locale, projectName, key):
+def uploadTranslations(metadata, dir, locale, projectName, localeConfig, key):
   files = []
   for file in os.listdir(dir):
     path = os.path.join(dir, file)
     if os.path.isfile(path):
-      if type == 'chrome' and file.endswith('.json'):
-        data = preprocessChromeLocale(path, metadata, False)
-        newName = file
-      elif type == 'chrome':
-        fileHandle = codecs.open(path, 'rb', encoding='utf-8')
-        data = json.dumps({file: {'message': fileHandle.read()}})
-        fileHandle.close()
-        newName = file + '.json'
+      if localeConfig['file_format'] == 'chrome-json':
+        if file.endswith('.json'):
+          data = preprocessChromeLocale(path, metadata, False)
+          newName = file
+        else:
+          fileHandle = codecs.open(path, 'rb', encoding='utf-8')

[Sebastian Noack, 2016/02/12 16:33:58]

Using the "with" statement?

[kzar, 2016/02/12 17:28:15]

I really want to avoid unrelated changes this time. Infact I realise to that end
refactoring the conditional logic here (and below) was a mistake. I've put it
back.

[Sebastian Noack, 2016/02/12 17:38:17]

Well, apparently you change this code anyway. Not too important though.

+          data = json.dumps({file: {'message': fileHandle.read()}})
+          fileHandle.close()
+          newName = file + '.json'
       else:
         data = toJSON(path)
         newName = file + '.json'
@@ -385,9 +389,11 @@ def uploadTranslations(type, metadata, dir, locale, projectName, key):
       if data:
         files.append((newName, data))
   if len(files):
-    postFiles(files, 'http://api.crowdin.net/api/project/%s/upload-translation?key=%s&language=%s' % (projectName, key, mapLocale(type, locale)))
+    postFiles(files, 'http://api.crowdin.net/api/project/%s/upload-translation?key=%s&language=%s' % (
+      projectName, key, mapLocale(localeConfig['name_format'], locale))
+    )
 
-def getTranslations(type, localesDir, defaultLocale, projectName, key):
+def getTranslations(localeConfig, projectName, key):
   result = urllib2.urlopen('http://api.crowdin.net/api/project/%s/export?key=%s' % (projectName, key)).read()
   if result.find('<success') < 0:
     raise Exception('Server indicated that the operation was not successful\n' + result)
@@ -400,20 +406,26 @@ def getTranslations(type, localesDir, defaultLocale, projectName, key):
       continue
 
     dir, file = os.path.split(info.filename)
-    if not re.match(r'^[\w\-]+$', dir) or dir == defaultLocale:
+    if not re.match(r'^[\w\-]+$', dir) or dir == localeConfig['default_locale']:
       continue
-    if type == 'chrome' and file.count('.') == 1:
+    if localeConfig['file_format'] == 'chrome-json' and file.count('.') == 1:
       origFile = file
     else:
       origFile = re.sub(r'\.json$', '', file)
-    if type == 'gecko' and not origFile.endswith('.dtd') and not origFile.endswith('.properties'):
+    if (localeConfig['file_format'] == 'gecko-dtd' and
+        not origFile.endswith('.dtd') and
+        not origFile.endswith('.properties')):
       continue
 
-    mapping = langMappingChrome if type == 'chrome' else langMappingGecko
+    if localeConfig['name_format'] == 'ISO-15897':
+      mapping = langMappingChrome
+    else:
+      mapping = langMappingGecko
+
     for key, value in mapping.iteritems():
       if value == dir:
         dir = key
-    if type == 'chrome':
+    if localeConfig['name_format'] == 'ISO-15897':
       dir = dir.replace('-', '_')
 
     data = zip.open(info.filename).read()
@@ -424,23 +436,24 @@ def getTranslations(type, localesDir, defaultLocale, projectName, key):
       dirs[dir] = set()
     dirs[dir].add(origFile)
 
-    path = os.path.join(localesDir, dir, origFile)
+    path = os.path.join(localeConfig['base_path'], dir, origFile)
     if not os.path.exists(os.path.dirname(path)):
       os.makedirs(os.path.dirname(path))
-    if type == 'chrome' and origFile.endswith('.json'):
-      postprocessChromeLocale(path, data)
-    elif type == 'chrome':
-      data = json.loads(data)
-      if origFile in data:
-        fileHandle = codecs.open(path, 'wb', encoding='utf-8')
-        fileHandle.write(data[origFile]['message'])
-        fileHandle.close()
+    if localeConfig['file_format'] == 'chrome-json':
+      if origFile.endswith('.json'):
+        postprocessChromeLocale(path, data)
+      else:
+        data = json.loads(data)
+        if origFile in data:
+          fileHandle = codecs.open(path, 'wb', encoding='utf-8')

[Sebastian Noack, 2016/02/12 16:33:58]

As discussed before, please use io.open instead codecs.open. The API is more
safe/explicit when it comes to unicode handling. And codecs.open() is currently
been discussed to be deprecated.

[kzar, 2016/02/12 17:28:15]

As above, want to avoid unrelated changes and this code shouldn't have been even
modified in the first place.

[Sebastian Noack, 2016/02/12 17:38:17]

Also here, as you change that code anyway, it's good time to tidy it up. But I
guess it's not too important.

+          fileHandle.write(data[origFile]['message'])
+          fileHandle.close()
     else:
       fromJSON(path, data)
 
   # Remove any extra files
   for dir, files in dirs.iteritems():
-    baseDir = os.path.join(localesDir, dir)
+    baseDir = os.path.join(localeConfig['base_path'], dir)
     if not os.path.exists(baseDir):
       continue
     for file in os.listdir(baseDir):