xwords/xwords4/android/scripts/copy-strings.py

#!/usr/bin/python

# Go through all the res_src strings.xml files, and copy them over
# into the world where they'll get used in a build. This is meant to
# allow them to be built-in as an alternative to the
# locutils/downloadable system.

import re, sys, os, getopt
from lxml import etree

s_prefix = 'XLATE ME: '


sComment = """
    DO NOT EDIT THIS FILE!!!!
    It was generated (from %s).
    Any changes you make to it will be lost.
"""

def sameOrSameWithPrefix( str1, str2 ):
    result = str1 == str2
    if not result:
        if str1.startswith(s_prefix):
            result = str1[len(s_prefix):] == str2
    return result

def sameAsEnglishPlural(engNames, strElem):
    strs = engNames[strElem.get('name')]['strings']
    str = strElem.text
    result = 1 == len(strs) and 'other' in strs \
             and sameOrSameWithPrefix( str, strs['other'] )
    return result

def loadPlural(plural):
    items = {}
    for child in plural.getchildren():
        items[child.get('quantity')] = child.text
    return items

def checkAndCopy( engNames, src, dest, verbose ):
    parser = etree.XMLParser(remove_blank_text=True, encoding="utf-8")
    doc = etree.parse(src, parser)
    for elem in doc.getroot().iter():
        if 'resources' == elem.tag:
            pass
        elif 'item' == elem.tag:
            pass
        elif 'string' == elem.tag:
            name = elem.get('name')
            if not name in engNames or elem.text.startswith(s_prefix):
                elem.getparent().remove( elem )
            elif not 'string' == engNames[name]['type']:
                if 'plurals' == engNames[name]['type']:
                    if sameAsEnglishPlural( engNames, elem ):
                        elem.getparent().remove( elem )
                    else:
                        elem.tag = 'plurals'
                        item = etree.Element("item")
                        item.text = elem.text
                        elem.text = None
                        item.set('quantity', 'other')
                        elem.append( item )
                        if verbose: print 'translated string', name, 'to plural'
                else:
                    elem.getparent().remove( elem )
            elif engNames[name]['string'] == elem.text:
                if verbose: print "Same as english: name: %s; text: %s" % (name, elem.text)
                elem.getparent().remove( elem )
        elif 'plurals' == elem.tag:
            name = elem.get('name')
            if not name in engNames or not 'plurals' == engNames[name]['type']:
                # print 'removing', name
                elem.getparent().remove( elem )
        elif not isinstance( elem.tag, basestring ): # comment
            elem.getparent().remove(elem)
        else:
            print 'unexpected elem:', elem.tag
            sys.exit(1)
    
    if True:
        comment = etree.Comment(sComment % (src))
        doc.getroot().insert( 0, comment )
        dir = os.path.dirname( dest )
        try: os.makedirs( dir )
        except: pass
        out = open( dest, "w" )
        out.write( etree.tostring( doc, pretty_print=True, encoding="utf-8", xml_declaration=True ) )

def main():
    # add these via params later
    excepts = ['values-ca_PS', 'values-ba_CK']
    verboses = ['values-fr']

    # summarize the english file
    wd = os.path.dirname(sys.argv[0])
    path = wd + '/../XWords4/res/values/strings.xml'
    engNames = {}

    engFormats = {}
    parser = etree.XMLParser(remove_blank_text=True, encoding="utf-8")
    doc = etree.parse(path, parser)
    pat = re.compile( '(%\d\$[sd])', re.DOTALL | re.MULTILINE )
    for typ in ['string', 'plurals']:
        for elem in doc.findall(typ):
            name = elem.get('name')
            item = { 'type' : typ }
            if typ == 'string':
                item['string'] = elem.text
            else:
                item['strings'] = loadPlural(elem)
            engNames[name] = item
    # print engNames
    
    # iterate over src files
    for subdir, dirs, files in os.walk('res_src'):
        for file in [file for file in files if file == "strings.xml"]:
            path = "%s/%s" % (subdir, file)
            for excpt in excepts:
                if excpt in path :
                    path = None
                    break
            if path: 
                verbose = 0 < len([verb for verb in verboses if verb in path])
                print "*** looking at %s ***" % (path)
                dest = path.replace( 'res_src', 'res', 1 )
                checkAndCopy( engNames, path, dest, verbose )

##############################################################################
if __name__ == '__main__':
    main()