#!/usr/bin/python3

# Note: This script has to be in the same directory with sync_po.py (.PO synchronizer).

"""
Made for combining Django's JavaScript Catalog (translation) implementation with AngularJS templates.

Code (AngularJS):
---------------
.filter('translate', function() { return function(input, params) { return params != null ? interpolate(gettext(input), params, params.constructor == Object) : gettext(input) } })
.directive('translate', function($compile) {
    return {
        compile: function(element) {
            var html = element.html().split('<br plural="">');
            return function (scope, element, attrs) {
                function c(h){
                    element.html(h);
                    $compile(element.contents())(scope);
                }
                if (attrs.translate != '') {
                    scope.$watch(function () { return scope.$eval(attrs.translate) }, function (val) {
                        var p = html[2];
                        html[2] = ngettext(html[0], html[1], attrs.add !== undefined ? val + attrs.add : attrs.subtract !== undefined ? val - attrs.subtract : val);
                        if (p != html[2]) c(html[2]);
                    });
                } else c(gettext(html[0]));
            }
        }
    }
})
---------------

Usage in templates and equivalent commands in JS:
{{ "Save" | translate }}
-> gettext("Save")

{{ "May" | translate : null : 'month name' }}
-> pgettext('month name', "May")

{{ "Comments (%d)" | translate : [obj.comments.length] }}
-> interpolate(gettext("Comments (%d)"), [obj.comments.length])

{{ "User(s) (%(user_count)d) are found %(distance)s away" | translate : {user_count: users.length, distance: obj.distance.value + obj.distance.unit} }}
-> interpolate(gettext("User(s) (%(user_count)d) are found %(distance)s away"), {user_count: users.length, distance: obj.distance.value + obj.distance.unit}, true)

<a href="/friends/" translate="friends.length">You have {{ friends.length }} friend.<br plural>You have {{ friends.length }} friends.</a>
-> ngettext("You have {{ friends.length }} friend.", "You have {{ friends.length }} friends.", friends.length)
"""

EXT = 'html'
TAG = 'translate'
MSGSTR_NUM = 2

from re import compile
from html.parser import HTMLParser

prog = [compile(r'("|\')(.*?)\1'), compile(r'%(\(.*?\))?(s|d)')]
class TemplateParser(HTMLParser):
    recording = 0
    f = ['', '']
    tmp = [None, '']

    def __init__(self, *args, **kwargs):
        self.tag = kwargs.pop('tag')
        self.msgstr_num = kwargs.pop('msgstr_num')
        self.prog = compile(r'\{{2} *([^\s|}]+(?: +[^\s|}]+)*) *\| *'+self.tag+r'(?: *:(?:[^:]*): *(["\'])([^"\'}]*(?:(?!\2)["\'][^"\'}]*)*)\2)?[^}]*\}{2}')
        super().__init__()

    def genc(self):
        return self.f[0][0]+':'+str(self.getpos()[0])

    def addnew(self, s=''):
        return {'#': {':': [self.genc()]}, 'msgid': [s]}

    def appendpos(self, p, s):
        for i in s:
            if p in i:
                return
        for i, _ in enumerate(s):
            if ' '+self.f[0][0] in s[i]:
                s[i] += ' '+p
                return
        s.append(p)

    def repl(self, s):
        if self.f[0][1]:
            s = s.replace('{% verbatim %}', '').replace('{% endverbatim %}', '')
        return s.replace('"', '\\"').replace('\\', '\\\\')

    def extr_filter(self, t):
        for e in self.prog.findall(t):
            for s in prog[0].findall(e[0]):
                r = self.repl(s[1])
                for i in self.data:
                    if not i.get('msgid_plural', None) and ''.join(i['msgid']) == r and ''.join(i.get('msgctxt', [])) == e[2]:
                        self.appendpos(self.genc(), i['#'][':'])
                        return
                self.data.append(self.addnew(r))
                if prog[1].search(s[1]):
                    self.data[-1]['#'][','] = ['javascript-format']
                if e[2] != '':
                    self.data[-1]['msgctxt'] = [e[2]]
                self.data[-1]['msgstr'] = ['']

    def handle_starttag(self, tag, attrs):
        s = self.get_starttag_text()
        self.extr_filter(s)
        if self.recording > 0:
            if s == '<br plural>':
                self.f[1] = 'msgid_plural'
                self.tmp[0][self.f[1]] = ['']
                return
            elif tag not in ('br', 'img', 'input', 'hr'):
                self.recording += 1
            self.tmp[0][self.f[1]][0] += self.repl(s)
            return
        for name, _ in attrs:
            if name == self.tag:
                self.tmp[0] = self.addnew()
                self.f[1] = 'msgid'
                self.recording = 1
                break

    def handle_data(self, data):
        self.tmp[1] += data
        if self.recording > 0:
            self.tmp[0][self.f[1]][0] += self.repl(data)

    def handle_endtag(self, tag):
        if self.recording > 0:
            self.recording -= 1
            if self.recording == 0:
                for i in self.data:
                    if ''.join(i['msgid']) == ''.join(self.tmp[0]['msgid']) and ''.join(i.get('msgid_plural', [])) == ''.join(self.tmp[0].get('msgid_plural', [])):
                        self.appendpos(self.tmp[0]['#'][':'][0], i['#'][':'])
                        self.tmp[0] = None
                        return
                if 'msgid_plural' in self.tmp[0]:
                    self.tmp[0]['msgstr'] = {}
                    for i in range(0, self.msgstr_num):
                        self.tmp[0]['msgstr'][str(i)] = ['']
                else:
                    self.tmp[0]['msgstr'] = ['']
                self.data.append(self.tmp[0].copy())
                self.tmp[0] = None
            else:
                self.tmp[0][self.f[1]][0] += '</'+tag+'>'
        self.extr_filter(self.tmp[1])
        self.tmp[1] = ''

    def handle_entityref(self, name):
        self.tmp[1] += '&'+name+';'
        if self.recording > 0:
            self.tmp[0][self.f[1]][0] += '&'+name+';'

def main(target_dir, output_file, source_file=None, overwrite=False, tag=TAG, ext=EXT, msgstr_num=MSGSTR_NUM):
    import os
    import sync_po

    parser = TemplateParser(tag=tag, msgstr_num=msgstr_num)

    if not overwrite and os.path.exists(output_file):
        with open(output_file, 'r') as f:
            parser.data = sync_po.extract_strings(f.readlines())
    else:
        parser.data = []

    ext = '.'+ext
    for subdir, _, files in os.walk(target_dir):
        for file in files:
            filepath = subdir + os.sep + file
            if filepath.endswith(ext):
                with open(filepath, 'r') as f:
                    parser.f[0] = (filepath if filepath[:2] != './' else filepath[2:], os.sep+'static'+os.sep not in filepath)
                    parser.feed(f.read())
                    parser.reset()

    if source_file and not os.path.samefile(output_file, source_file.name):
        with source_file as f:
            sync_po.main(sync_po.extract_strings(f.readlines()), parser.data)

    with open(output_file, 'w') as f:
        sync_po.output_strings(parser.data, f)

if __name__ == '__main__':
    import argparse
    parser = argparse.ArgumentParser(description="Generates .PO file from AngularJS templates (optionally with translations from other .PO file), or updates existing. By MikiSoft")
    parser.add_argument('target_dir', type=str)
    parser.add_argument('output_file', type=str)
    parser.add_argument('-s', dest='source_file', metavar="FILE", type=argparse.FileType('r'), help="source file")
    parser.add_argument('-o', dest='overwrite', action='store_true', help="don't update, overwrite output file if exists")
    parser.add_argument('-tag', type=str, default=TAG, help="default: "+TAG)
    parser.add_argument('-ext', type=str, default=EXT, help="default: "+EXT)
    parser.add_argument('-msgstr_num', metavar="N", type=int, default=MSGSTR_NUM, help="default: "+str(MSGSTR_NUM))
    args = parser.parse_args()

    main(args.target_dir, args.output_file, args.source_file, args.overwrite, args.tag, args.ext, args.msgstr_num)