kaauh_ats/find_empty_translations.py

#!/usr/bin/env python3
"""
Script to find empty msgstr entries in django.po file and organize them into batches
for systematic Arabic translation work.
"""

import re
import os
from typing import List, Tuple

def find_empty_translations(po_file_path: str) -> List[Tuple[int, str, str]]:
    """
    Find all entries with empty msgstr in the django.po file.

    Returns:
        List of tuples: (line_number, msgid, context_before)
    """
    empty_translations = []

    with open(po_file_path, 'r', encoding='utf-8') as file:
        lines = file.readlines()

    i = 0
    while i < len(lines):
        line = lines[i].strip()

        # Look for msgid
        if line.startswith('msgid '):
            msgid = line[7:-1]  # Remove 'msgid "' and ending '"'

            # Check next few lines for msgstr
            j = i + 1
            msgstr_found = False
            msgstr_empty = False

            while j < len(lines) and j < i + 5:  # Look ahead max 5 lines
                next_line = lines[j].strip()

                if next_line.startswith('msgstr '):
                    msgstr_found = True
                    if next_line == 'msgstr ""':
                        msgstr_empty = True
                    break
                elif next_line.startswith('msgid ') or next_line.startswith('#'):
                    # Found next entry or comment, no msgstr for current msgid
                    break
                j += 1

            if msgstr_found and msgstr_empty:
                # Get context (previous 2-3 lines)
                context_start = max(0, i - 3)
                context = ''.join(lines[context_start:i])
                empty_translations.append((i + 1, msgid, context))

            i = j  # Skip to after msgstr
        else:
            i += 1

    return empty_translations

def create_batch_files(empty_translations: List[Tuple[int, str, str]], batch_size: int = 25):
    """
    Create batch files with empty translations for systematic work.
    """
    total_batches = (len(empty_translations) + batch_size - 1) // batch_size

    print(f"Found {len(empty_translations)} empty translations")
    print(f"Creating {total_batches} batches of ~{batch_size} translations each")

    for batch_num in range(total_batches):
        start_idx = batch_num * batch_size
        end_idx = min(start_idx + batch_size, len(empty_translations))
        batch_translations = empty_translations[start_idx:end_idx]

        batch_filename = f"translation_batch_{batch_num + 1:02d}.txt"

        with open(batch_filename, 'w', encoding='utf-8') as batch_file:
            batch_file.write(f"=== TRANSLATION BATCH {batch_num + 1:02d} ===\n")
            batch_file.write(f"Translations {start_idx + 1}-{end_idx} of {len(empty_translations)}\n")
            batch_file.write("=" * 60 + "\n\n")

            for line_num, msgid, context in batch_translations:
                batch_file.write(f"Line {line_num}:\n")
                batch_file.write(f"msgid: \"{msgid}\"\n")
                batch_file.write(f"msgstr: \"\"\n")
                batch_file.write(f"\nArabic Translation: \n")
                batch_file.write(f"msgstr: \"\"\n")
                batch_file.write("-" * 40 + "\n\n")

        print(f"Created {batch_filename} with {len(batch_translations)} translations")

def create_summary_report(empty_translations: List[Tuple[int, str, str]]):
    """
    Create a summary report of all empty translations.
    """
    with open("empty_translations_summary.txt", 'w', encoding='utf-8') as report:
        report.write("EMPTY TRANSLATIONS SUMMARY REPORT\n")
        report.write("=" * 50 + "\n\n")
        report.write(f"Total empty translations: {len(empty_translations)}\n\n")

        # Group by type/pattern for better organization
        ui_elements = []
        form_fields = []
        messages = []
        navigation = []
        other = []

        for line_num, msgid, context in empty_translations:
            msgid_lower = msgid.lower()

            if any(word in msgid_lower for word in ['button', 'btn', 'click', 'select']):
                ui_elements.append((line_num, msgid))
            elif any(word in msgid_lower for word in ['field', 'form', 'input', 'enter']):
                form_fields.append((line_num, msgid))
            elif any(word in msgid_lower for word in ['error', 'success', 'warning', 'message']):
                messages.append((line_num, msgid))
            elif any(word in msgid_lower for word in ['menu', 'nav', 'page', 'dashboard']):
                navigation.append((line_num, msgid))
            else:
                other.append((line_num, msgid))

        report.write(f"UI Elements (Buttons, Links): {len(ui_elements)}\n")
        report.write(f"Form Fields & Inputs: {len(form_fields)}\n")
        report.write(f"Messages (Error/Success/Warning): {len(messages)}\n")
        report.write(f"Navigation & Pages: {len(navigation)}\n")
        report.write(f"Other: {len(other)}\n\n")

        report.write("SAMPLE ENTRIES:\n")
        report.write("-" * 30 + "\n")

        for category, name, sample_count in [
            (ui_elements, "UI Elements", 5),
            (form_fields, "Form Fields", 5),
            (messages, "Messages", 5),
            (navigation, "Navigation", 5),
            (other, "Other", 5)
        ]:
            if category:
                report.write(f"\n{name} (showing first {min(len(category), sample_count)}):\n")
                for line_num, msgid in category[:sample_count]:
                    report.write(f"  Line {line_num}: \"{msgid}\"\n")

def main():
    """Main function to process the django.po file."""
    po_file_path = "locale/ar/LC_MESSAGES/django.po"

    if not os.path.exists(po_file_path):
        print(f"Error: {po_file_path} not found!")
        return

    print("Scanning for empty translations...")
    empty_translations = find_empty_translations(po_file_path)

    if not empty_translations:
        print("No empty translations found! All msgstr entries have translations.")
        return

    # Create batch files
    create_batch_files(empty_translations, batch_size=25)

    # Create summary report
    create_summary_report(empty_translations)

    print(f"\nProcess completed!")
    print(f"Check the generated batch files: translation_batch_*.txt")
    print(f"Summary report: empty_translations_summary.txt")
    print(f"\nNext steps:")
    print(f"1. Work through each batch file systematically")
    print(f"2. Add Arabic translations to each empty msgstr")
    print(f"3. Update the main django.po file with completed translations")

if __name__ == "__main__":
    main()