Add script to sync errors with Telegram's JSON

telethon_generator/parsers/errors.py

@@ -44,6 +44,7 @@ class Error:
         # Should these be split into different files or doesn't really matter?
         # Telegram isn't exactly consistent with returned errors anyway.
         self.int_code = codes[0]
+        self.int_codes = codes
         self.str_code = name
         self.subclass = _get_class_name(codes[0])
         self.subclass_exists = abs(codes[0]) in KNOWN_BASE_CLASSES

telethon_generator/parsers/methods.py

@@ -9,6 +9,15 @@ class Usability(enum.Enum):
     BOT = 2
     BOTH = 4
 
+    @property
+    def key(self):
+        return {
+            Usability.UNKNOWN: 'unknown',
+            Usability.USER: 'user',
+            Usability.BOT: 'bot',
+            Usability.BOTH: 'both',
+        }[self]
+
 
 class MethodInfo:
     def __init__(self, name, usability, errors, friendly):

telethon_generator/syncerrors.py

@@ -0,0 +1,53 @@
+# Should be fed with the JSON obtained from https://core.telegram.org/api/errors#error-database
+import re
+import csv
+import sys
+import json
+from pathlib import Path
+
+sys.path.insert(0, '..')
+
+from telethon_generator.parsers.errors import parse_errors, Error
+from telethon_generator.parsers.methods import parse_methods, MethodInfo
+
+ERRORS = Path('data/errors.csv')
+METHODS = Path('data/methods.csv')
+FRIENDLY = Path('data/friendly.csv')
+
+
+def main():
+    new_errors = []
+    new_methods = []
+
+    self_errors = {e.str_code: e for e in parse_errors(ERRORS)}
+    self_methods = {m.name: m for m in parse_methods(METHODS, FRIENDLY, self_errors)}
+
+    tg_data = json.load(sys.stdin)
+
+    def get_desc(code):
+        return re.sub(r'\s*&\w+;\s*', '', (tg_data['descriptions'].get(code) or '').rstrip('.'))
+
+    for int_code, errors in tg_data['errors'].items():
+        int_code = int(int_code)  # json does not support non-string keys
+        for code, methods in errors.items():
+            str_code = code.replace('%d', 'X')
+            if error := self_errors.get(str_code):
+                error.int_codes.append(int_code)  # de-duplicated once later
+                if not error.description:  # prefer our descriptions
+                    if not error.has_captures:  # need descriptions with specific text if error has captures
+                        error.description = get_desc(code)
+            else:
+                self_errors[str_code] = Error([int_code], str_code, get_desc(code))
+
+    new_errors.extend((e.str_code, ' '.join(map(str, sorted(set(e.int_codes)))), e.description) for e in self_errors.values())
+    new_methods.extend((m.name, m.usability.key, ' '.join(sorted(e.str_code for e in m.errors))) for m in self_methods.values())
+
+    csv.register_dialect('plain', lineterminator='\n')
+    with ERRORS.open('w', encoding='utf-8', newline='') as fd:
+        csv.writer(fd, 'plain').writerows((('name', 'codes', 'description'), *sorted(new_errors)))
+    with METHODS.open('w', encoding='utf-8', newline='') as fd:
+        csv.writer(fd, 'plain').writerows((('method', 'usability', 'errors'), *sorted(new_methods)))
+
+
+if __name__ == '__main__':
+    main()