# -*- coding: utf-8 -*- """ Internationalization facilities Authors: David Malcolm """ __author__ = """David Malcolm , Zack Cerza """ import config import os import re import gettext from logging import debugLogger as logger def safeDecode(string): if not isinstance(string, unicode): try: string = string.decode('utf-8') except UnicodeDecodeError: #logger.log(traceback.format_exc()) #logger.log("The following string is invalid and caused the above error: '%s'" % string) string = string.decode('utf-8', 'replace') return string def safeEncode(string): pass """ Singleton list of TranslationDb instances, to be initialized by the script with whatever translation databases it wants. """ translationDbs = [] class TranslationDb: """ Abstract base class representing a database of translations """ def getTranslationsOf(self, srcName): """ Pure virtual method to look up the translation of a string. Returns a list of candidate strings (the translation), empty if not found. Note that a source string can map to multiple translated strings. For example, in the French translation of Evolution, the string "Forward" can translate to both (i) "Faire suivre" for forwarding an email, and (ii) "Suivant" for the next page in a wizard. """ raise NotImplementedError class GettextTranslationDb(TranslationDb): """ Implementation of TranslationDb which leverages gettext, using a single translation mo-file. """ def __init__(self, moFile): self.__moFile = moFile self.__gnutranslations = gettext.GNUTranslations(open(moFile)) def getTranslationsOf(self, srcName): srcName = safeDecode(srcName) # print "searching for translations of %s"%srcName # Use a dict to get uniqueness: results = {} result = self.__gnutranslations.ugettext(srcName) if result!=srcName: results[result]=None # Hack alert: # # Note that typical UI definition in GTK etc contains strings with # underscores to denote accelerators. # For example, the stock GTK "Add" item has text "_Add" which e.g. # translates to "A_jouter" in French # # Since these underscores have been stripped out before we see these strings, # we are looking for a translation of "Add" into "Ajouter" in this case, so # we need to fake it, by looking up the string multiple times, with underscores # inserted in all possible positions, stripping underscores out of the result. # Ugly, but it works. for index in range(len(srcName)): candidate = srcName[:index]+"_"+srcName[index:] result = self.__gnutranslations.ugettext(candidate) if result!=candidate: # Strip out the underscore, and add to the result: results[result.replace('_','')]=True return results.keys() def translate(srcString): """ Look up srcString in the various translation databases (if any), returning a list of all matches found (potentially the empty list) """ # Use a dict to get uniqueness: results = {} # Try to translate the string: for translationDb in translationDbs: for result in translationDb.getTranslationsOf(srcString): result = safeDecode(result) results[result]=True # No translations found: if len(results)==0: if config.config.debugTranslation: logger.log('Translation not found for "%s"'%srcString) return results.keys() class TranslatableString: """ Class representing a string that we want to match strings against, handling translation for us, by looking it up once at construction time. """ def __init__(self, untranslatedString): """ Constructor looks up the string in all of the translation databases, storing the various translations it finds. """ if isinstance(untranslatedString, unicode): untranslatedString = safeDecode(untranslatedString) else: untranslatedString = safeDecode(untranslatedString) self.untranslatedString = untranslatedString self.translatedStrings = translate(untranslatedString) def matchedBy(self, string): """ Compare the test string against either the translation of the original string (or simply the original string, if no translation was found). """ #print "comparing %s against %s"%(string, self) def stringsMatch(inS, outS): """ Compares a regular expression to a string inS: the regular expression (or normal string) outS: the normal string to be compared against """ inString = str(inS) outString = outS if inString == outString: return True inString = inString + '$' inString = safeDecode(inString) outString = safeDecode(outString) if inString[0] == '*': inString = "\\" + inString # Escape all parentheses, since grouping will never be needed here inString = re.sub('([\(\)])', r'\\\1', inString) match = re.match(inString, outString) matched = match is not None return matched matched = False # the 'ts' variable keeps track of whether we're working with # translated strings. it's only used for debugging purposes. #ts = 0 #print string, str(self) for translatedString in self.translatedStrings: #ts = ts + 1 matched = stringsMatch(translatedString, string) if not matched: matched = translatedString == string if matched: return matched #ts=0 return stringsMatch(self.untranslatedString, string) def __str__(self): """ Provide a meaningful debug version of the string (and the translation in use) """ if len(self.translatedStrings)>0: # build an output string, with commas in the correct places translations = "" for tString in self.translatedStrings: translations += u'"%s", ' % safeDecode(tString) result = u'"%s" (%s)' % (safeDecode(self.untranslatedString), translations) return safeDecode(result) else: return '"%s"' % (self.untranslatedString) def isMoFile(filename, language = ''): """ Does the given filename look like a gettext mo file? Optionally: Does the file also contain translations for a certain language, for example 'ja'? """ if re.match('(.*)\\.mo$', filename): if not language: return True elif re.match('/usr/share/locale(.*)/%s(.*)/LC_MESSAGES/(.*)\\.mo$' % \ language, filename): return True else: return False else: return False def loadAllTranslationsForLanguage(language): import distro result = [] for moFile in distro.packageDb.getMoFiles(language): translationDbs.append(GettextTranslationDb(moFile)) def getMoFilesForPackage(packageName, language = '', getDependencies=True): """ Look up the named package and find all gettext mo files within it and its dependencies. It is possible to restrict the results to those of a certain language, for example 'ja'. """ import distro result = [] for filename in distro.packageDb.getFiles(packageName): if isMoFile(filename, language): result.append(filename) if getDependencies: # Recurse: for dep in distro.packageDb.getDependencies(packageName): # We pass False to the inner call because getDependencies has already # walked the full tree result.extend(getMoFilesForPackage(dep, language, False)) return result def loadTranslationsFromPackageMoFiles(packageName, getDependencies=True): """ Helper function which appends all of the gettext translation mo-files used by the package (and its dependencies) to the translation database list. """ # Keep a list of mo-files that are already in use to avoid duplicates. moFiles = {} def load(packageName, language = '', getDependencies = True): for moFile in getMoFilesForPackage(packageName, language, getDependencies): # Searching the popt mo-files for translations makes gettext bail out, # so we ignore them here. This is # https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=172155 . if not('popt.mo' in moFile or moFile in moFiles): try: translationDbs.append(GettextTranslationDb(moFile)) moFiles[moFile] = None except (AttributeError, IndexError), inst: if config.config.debugTranslation: #import traceback #logger.log(traceback.format_exc()) logger.log("Warning: Failed to load mo-file for translation: " + moFile) # Hack alert: # # The following special-case is necessary for Ubuntu, since their # translations are shipped in a single huge package. The downside to # this special case, aside from the simple fact that there is one, # is that it makes automatic translations much slower. import distro language = os.environ.get('LANGUAGE', os.environ['LANG'])[0:2] if isinstance(distro.distro, distro.Ubuntu): load('language-pack-gnome-%s' % language, language) load(packageName, language, getDependencies)