Update: 2026-06-13 00:57:17
This commit is contained in:
153
scratch/align_and_generate_driver_translations.py
Normal file
153
scratch/align_and_generate_driver_translations.py
Normal file
@@ -0,0 +1,153 @@
|
||||
import json
|
||||
import re
|
||||
import os
|
||||
|
||||
def parse_dart_map(filepath):
|
||||
translations = {}
|
||||
if not os.path.exists(filepath):
|
||||
return translations
|
||||
with open(filepath, 'r', encoding='utf-8') as f:
|
||||
content = f.read()
|
||||
|
||||
# Matches maps of form: "key": "value" or 'key': 'value'
|
||||
# Handles escapes and multi-line content matched on single lines
|
||||
pattern = re.compile(r'^\s*([\'"])(.*?)\1\s*:\s*([\'"])(.*?)\3\s*,?\s*$', re.MULTILINE)
|
||||
matches = pattern.findall(content)
|
||||
for match in matches:
|
||||
key = match[1]
|
||||
val = match[3]
|
||||
# Unescape simple sequences (single and double quotes)
|
||||
key = key.replace('\\"', '"').replace("\\'", "'")
|
||||
val = val.replace('\\"', '"').replace("\\'", "'")
|
||||
translations[key] = val
|
||||
|
||||
return translations
|
||||
|
||||
def to_dart_string(s):
|
||||
# Escape characters to be safe inside double quotes in Dart source code
|
||||
result = []
|
||||
for char in s:
|
||||
if char == '\\':
|
||||
result.append('\\\\')
|
||||
elif char == '"':
|
||||
result.append('\\"')
|
||||
elif char == '$':
|
||||
result.append('\\$')
|
||||
elif char == '\n':
|
||||
result.append('\\n')
|
||||
elif char == '\r':
|
||||
result.append('\\r')
|
||||
elif char == '\t':
|
||||
result.append('\\t')
|
||||
else:
|
||||
result.append(char)
|
||||
return "".join(result)
|
||||
|
||||
# Load current Arabic files
|
||||
ar_eg = parse_dart_map("siro_driver/lib/controller/local/ar_eg.dart")
|
||||
ar_jo = parse_dart_map("siro_driver/lib/controller/local/ar_jo.dart")
|
||||
ar_sy = parse_dart_map("siro_driver/lib/controller/local/ar_sy.dart")
|
||||
|
||||
# Load JSON data
|
||||
with open("siro_driver_translations_data.json", "r", encoding="utf-8") as f:
|
||||
json_data = json.load(f)
|
||||
|
||||
existing_syrian = json_data.get('existing_syrian', {})
|
||||
missing_keys = json_data.get('missing_keys', [])
|
||||
|
||||
# Load legacy non-Arabic languages
|
||||
with open("scratch/legacy_extracted_languages.json", "r", encoding="utf-8") as f:
|
||||
legacy_extracted = json.load(f)
|
||||
|
||||
# Compute master keys
|
||||
master_keys = set(ar_sy.keys())
|
||||
master_keys.update(ar_jo.keys())
|
||||
master_keys.update(ar_eg.keys())
|
||||
master_keys.update(existing_syrian.keys())
|
||||
master_keys.update(missing_keys)
|
||||
|
||||
master_keys = sorted(list(master_keys))
|
||||
print(f"Master keys count: {len(master_keys)}")
|
||||
|
||||
# Let's build translation maps
|
||||
aligned_maps = {
|
||||
'ar-EG': {},
|
||||
'ar-JO': {},
|
||||
'ar-SY': {},
|
||||
'en': {},
|
||||
'de': {},
|
||||
'el': {},
|
||||
'es': {},
|
||||
'fa': {},
|
||||
'fr': {},
|
||||
'hi': {},
|
||||
'it': {},
|
||||
'ru': {},
|
||||
'tr': {},
|
||||
'ur': {},
|
||||
'zh': {}
|
||||
}
|
||||
|
||||
for key in master_keys:
|
||||
# 1. English is always the key itself
|
||||
aligned_maps['en'][key] = key
|
||||
|
||||
# 2. Syrian Arabic (ar-SY)
|
||||
val_sy = ar_sy.get(key)
|
||||
if not val_sy:
|
||||
val_sy = existing_syrian.get(key)
|
||||
if not val_sy:
|
||||
val_sy = ar_jo.get(key)
|
||||
if not val_sy:
|
||||
val_sy = ar_eg.get(key)
|
||||
if not val_sy:
|
||||
val_sy = key # fallback to key
|
||||
aligned_maps['ar-SY'][key] = val_sy
|
||||
|
||||
# 3. Jordanian Arabic (ar-JO)
|
||||
val_jo = ar_jo.get(key)
|
||||
if not val_jo:
|
||||
val_jo = val_sy # fallback to Syrian
|
||||
aligned_maps['ar-JO'][key] = val_jo
|
||||
|
||||
# 4. Egyptian Arabic (ar-EG)
|
||||
val_eg = ar_eg.get(key)
|
||||
if not val_eg:
|
||||
val_eg = val_jo # fallback to Jordanian/Syrian
|
||||
aligned_maps['ar-EG'][key] = val_eg
|
||||
|
||||
# 5. Non-Arabic languages
|
||||
for lang in ['de', 'el', 'es', 'fa', 'fr', 'hi', 'it', 'ru', 'tr', 'ur', 'zh']:
|
||||
val_lang = legacy_extracted.get(lang, {}).get(key)
|
||||
if not val_lang:
|
||||
val_lang = key # fallback to English
|
||||
aligned_maps[lang][key] = val_lang
|
||||
|
||||
# Directories and file writes
|
||||
output_dir = "siro_driver/lib/controller/local"
|
||||
|
||||
def write_dart_file(filename, map_name, data_map):
|
||||
filepath = os.path.join(output_dir, filename)
|
||||
with open(filepath, 'w', encoding='utf-8') as f:
|
||||
# Write final Map<String, String> map_name = { ... };
|
||||
f.write(f"final Map<String, String> {map_name} = {{\n")
|
||||
for k, v in data_map.items():
|
||||
k_escaped = to_dart_string(k)
|
||||
v_escaped = to_dart_string(v)
|
||||
f.write(f' "{k_escaped}": "{v_escaped}",\n')
|
||||
f.write("};\n")
|
||||
print(f"Wrote {filepath} with {len(data_map)} keys.")
|
||||
|
||||
# Write Arabic dialects
|
||||
write_dart_file("ar_eg.dart", "ar_eg", aligned_maps['ar-EG'])
|
||||
write_dart_file("ar_jo.dart", "ar_jo", aligned_maps['ar-JO'])
|
||||
write_dart_file("ar_sy.dart", "ar_sy", aligned_maps['ar-SY'])
|
||||
|
||||
# Write English
|
||||
write_dart_file("en.dart", "en", aligned_maps['en'])
|
||||
|
||||
# Write non-Arabic languages
|
||||
for lang in ['de', 'el', 'es', 'fa', 'fr', 'hi', 'it', 'ru', 'tr', 'ur', 'zh']:
|
||||
write_dart_file(f"{lang}.dart", lang, aligned_maps[lang])
|
||||
|
||||
print("Alignment and file writing complete.")
|
||||
Reference in New Issue
Block a user