Automatic translation of odoo files by using existing translation (e.g. from Ukrainian to Russian, English to Russian, Russian to Russian)

00-install-odoo-translation.sh

# install dependencies
sudo pip install polib
sudo pip install google-cloud-translate==2.0.0

# https://docs.transifex.com/client/installing-the-client
sudo pip install transifex-client

# prepare and export your google cloud credentials
# See https://cloud.google.com/translate/docs/basic/setup-basic
# and https://cloud.google.com/docs/authentication/getting-started
export GOOGLE_APPLICATION_CREDENTIALS=path/to/credentials.json

# prepare and export transifex token
# https://www.transifex.com/user/settings/api/
export TX_TOKEN=<your_Transifex_API_token>

cd 
wget https://gist.github.com/yelizariev/5cc80c634cce9271132fb141971fe23a/raw/translate.py
wget https://gist.github.com/yelizariev/5cc80c634cce9271132fb141971fe23a/raw/fix-translation.py
wget https://gist.github.com/yelizariev/5cc80c634cce9271132fb141971fe23a/raw/move-translation.py

fix-translation.py

import sys
import re

import polib

TAGS = ['menuselection', 'doc', 'ref', 'class', 'meth', 'func']

def fix_po_item(msgid, msgstr):
    res = msgstr

    for t in TAGS:
        search = re.search(r"(:%s:`[^`]*`)" % t, msgid)
        if not search:
            # no tag in source
            continue
        original_value = search.group(1)
        res = re.sub(":\s*%s\s*:\s*`[^`]*`" % t, ' %s' % original_value, res)
        print ("tag :%s: is replaced with following:\n%s" % (t, original_value))

    # fix replaced quotes
    res = re.sub(""", "`", res)
    res = re.sub("'", "'", res)

    # fix variables
    res = re.sub("% \(", "%(", res)
    # TODO: this doesn't work well in latin aplhabet
    res = re.sub("\) s", ")s ", res)
    res = re.sub("% \.", "%.", res)
    res = re.sub("% S", "%s", res)
    res = re.sub("% s", "%s", res)
    res = re.sub("% D", " %d", res)
    res = re.sub("% d", " %d", res)
    res = re.sub("% R", " %r", res)
    res = re.sub("% r", " %r", res)
    res = re.sub("%set", "% set", res)
    res = re.sub("%S", "%s ", res)
    res = re.sub("\$ {", "${", res)
    #res = re.sub("# ", "#", res)

    # fix broken links
    res = re.sub("\> `_", ">`_", res)

    if res != msgstr:
        return res


def fix_po(fname):
    dest_po = polib.pofile(fname)

    changed = False
    for dest_entry in dest_po:
        res = fix_po_item(dest_entry.msgid, dest_entry.msgstr)
        # python3 check
        if isinstance(res, str):
            changed = True
            dest_entry.msgstr = res

    if changed:
        dest_po.save()


if __name__ == '__main__':
    print (sys.argv)
    fname = sys.argv[1]
    fix_po(fname)

move-translation.py

import sys
import polib
import os



def move_po(lang, source_path, dest_fname):
    src_fname = os.path.join(source_path, dest_fname)
    if not os.path.exists(src_fname):
        print ("Source file doesn't exist: %s" % src_fname)
        return
    src_po = polib.pofile(src_fname)
    dest_po = polib.pofile(dest_fname)

    changed = False
    for dest_entry in dest_po:
        src_entry = src_po.find(dest_entry.msgid)
        if not src_entry:
            # print ("%s: source entry not found or empty: \n%s\n" % (src_fname, dest_entry.msgid))
            continue
        if dest_entry.msgstr == src_entry.msgstr:
            # same translation
            continue
        if not any(s in dest_entry.msgstr for s in {'&', '%', '$'}):
            # we update only entries with special characters See https://github.com/odoo/odoo/issues/43161#issuecomment-575015099
            continue
        #print (u'========================================\nUpdate \n%s\n--->\n%s ' % (str(dest_entry.msgstr), str(src_entry.msgstr)))
        dest_entry.msgstr = src_entry.msgstr
        changed = True
    if changed:
        dest_po.save()


if __name__ == '__main__':
    print (sys.argv)
    lang = sys.argv[1]
    source_path = sys.argv[2]
    dest_fname = sys.argv[3]

    move_po(lang, source_path, dest_fname)

move-translation.sh

# Transifex usually syncs translation by itself: https://docs.transifex.com/translation-memory/tm-introduction
# You may need this script only to fix previously translated terms

# You need to use this script twice: for CE and EE
# Each time you need two odoo folders: source and target

LANG=ru
VERSION_SOURCE=odoo-12
VERSION_TARGET=odoo-13

# download existing translations
cd path/to/odoo-source
tx pull -r "$VERSION_SOURCE.*" -l $LANG -f


cd path/to/odoo-target
tx pull -r "$VERSION_TARGET.*" -l $LANG -f
# make a commit to easily compare changes
git commit -a -m "sync translations with transifex"


# move translations
#export PYTHONIOENCODING=utf-8
find . -iname "ru.po" | xargs -L1 python3 ~/move-translation.py $LANG ../path/to/odoo-source


# upload translation
tx push -r "$VERSION_TARGET.*" -l ru -t

translate-from-en.py

import sys
import polib
from google.cloud import translate_v2 as translate
translate_client = translate.Client()


def translate_po(lang, fname):
    po = polib.pofile(fname)

    for entry in po:
        if entry.msgstr:
            # already translated
            continue

        try:
            tr_text = translate_text(entry.msgid, 'en', lang)
        except Exception as e:
            # save file and exit
            print ('ERROR on translation: ', e)
            break
        entry.msgstr = tr_text

    po.save()


def translate_text(text, src, dest):
    result = translate_client.translate(text, source_language=src, target_language=dest)
    return result['translatedText']


if __name__ == '__main__':
    print (sys.argv)
    lang = sys.argv[1]
    fname = sys.argv[2]

    translate_po(lang, fname)

translate-odoo-doc.sh

VERSION=odoo-13-doc
SRC_LANG=uk
DEST_LANG=ru

git clone https://github.com/odoo/documentation-user.git
cd documentation-user
tx pull -r "$VERSION.*" -l $SRC_LANG -f
tx pull -r "$VERSION.*" -l $DEST_LANG -f

# translate
find . -type d -iname i18n | xargs -L1 python3 ~/translate.py $SRC_LANG $DEST_LANG

# fix
find locale/$DEST_LANG/LC_MESSAGES -iname "*.po" | xargs -L1 python3 ~/fix-translation.py

# upload
tx push -r "$VERSION.*" -l $DEST_LANG -t

translate-odoo-source.sh

# You need to use this script twice: for CE and EE
cd path/to/odoo


SRC_LANG=uk
DEST_LANG=ru
VERSION=odoo-12

# download source
tx pull -r "$VERSION.*" -l $SRC_LANG -f
tx pull -r "$VERSION.*" -l $DEST_LANG -f

# translate
find . -type d -iname i18n | xargs -L1 python3 ~/translate.py uk ru

# fix
find . -iname "ru.po" | xargs -L1 python3 ~/fix-translation.py

# upload translation
tx push -r "$VERSION.*" -l ru -t

translate-odoo.com.sh

RESOURSES=(website_event_stand saas_trial release_notes openerp_website openerp_enterprise main help features)
SRC_LANG=uk
DEST_LANG=ru

# TODO: use command "tx" instead

# get it from transifex.com
COOKIE="Cookie: ......"

mkdir /tmp/odoo.com/
cd /tmp/odoo.com/

# download
for r in ${RESOURSES[@]}; do
  mkdir -p $r/i18n/
  wget https://www.transifex.com/odoo/odoo-com/$r/$SRC_LANG/download/for_use/ --header="$COOKIE" -O $r/i18n/$SRC_LANG.po
  wget https://www.transifex.com/odoo/odoo-com/$r/$DEST_LANG/download/for_translation/ --header="$COOKIE" -O $r/i18n/$DEST_LANG.po
done

# translate
find . -type d -iname i18n | xargs -L1 python3 ~/translate.py uk ru

translate.py

import sys
import os
import subprocess

import polib
from google.cloud import translate_v2 as translate
translate_client = translate.Client()


def translate_po(src_lang, dest_lang, src_fname, dest_fname, template_fname):

    if not os.path.exists(src_fname):
        print ("Source file doesn't exist: %s" % src_fname)
        return

    if not os.path.exists(dest_fname):
        print ("Target file doesn't exist: %s" % dest_fname)
        subprocess.check_output(["cp %s" % template_fname, dest_fname])

    src_po = polib.pofile(src_fname)
    dest_po = polib.pofile(dest_fname)

    for dest_entry in dest_po:
        if dest_entry.msgstr:
            # already translated
            continue
        if any(s in dest_entry.msgid for s in ["%", "<", ">", "$", "&", "#"]):
            # don't translate messages with special symbols
            continue
        src_entry = src_po.find(dest_entry.msgid)
        if not src_entry:
            print ("%s: source entry not found or empty: \n%s\n" % (src_fname, dest_entry.msgid))
            continue

        if src_entry.msgstr == src_entry.msgid:
            # Not for translattion.
            # Just copy original.
            # If we leave it empty, transifex will always offer to translate it, which is annoying
            dest_entry.msgstr = src_entry.msgid
            continue

        try:
            tr_text = translate_text(src_entry.msgstr, src_lang, dest_lang)
        except Exception as e:
            # save file and exit
            print ('ERROR on translation: ', e)
            break
        dest_entry.msgstr = tr_text

    dest_po.save()


def translate_text(text, src, dest):
    result = translate_client.translate(text, source_language=src, target_language=dest)
    return result['translatedText']


if __name__ == '__main__':
    print (sys.argv)
    src_lang = sys.argv[1]
    dest_lang = sys.argv[2]
    if len(sys.argv) == 4:
        folder = sys.argv[3]
        src_fname = os.path.join(folder, '%s.po' % src_lang)
        dest_fname = os.path.join(folder, '%s.po' % dest_lang)
        template_fname = os.path.join(folder, '*.pot')
    else:
        src_fname = sys.argv[3]
        dest_fname = sys.argv[4]
        template_fname = sys.argv[5]

    translate_po(src_lang, dest_lang, src_fname, dest_fname, template_fname)