Merge pull request #9685 from ClickHouse/backport-script-fixes

Update script for backporting.

utils/github/__main__.py

@@ -49,7 +49,7 @@ parser.add_argument('--token', type=str, required=True,
     help='token for Github access')
 parser.add_argument('--login', type=str,
     help='filter authorship by login')
-parser.add_argument('--auto-label', action='store_true', dest='autolabel',
+parser.add_argument('--auto-label', action='store_true', dest='autolabel', default=True,
     help='try to automatically parse PR description and put labels')
 
 args = parser.parse_args()
@@ -80,6 +80,8 @@ for i in reversed(range(len(stables))):
 
 members = set(github.get_members("ClickHouse", "ClickHouse"))
 def print_responsible(pull_request):
+    if "author" not in pull_request or pull_request["author"] is None:
+        return "No author"
     if pull_request["author"]["login"] in members:
         return colored(pull_request["author"]["login"], 'green')
     elif pull_request["mergedBy"]["login"] in members:

utils/github/parser.py

@@ -10,6 +10,10 @@ class Description:
         'Performance Improvement': 'pr-performance',
         # 'Backward Incompatible Change': doesn't match anything
         'Build/Testing/Packaging Improvement': 'pr-build',
+        'Non-significant (changelog entry is not needed)': 'pr-non-significant',
+        'Non-significant (changelog entry is not required)': 'pr-non-significant',
+        'Non-significant': 'pr-non-significant',
+        'Documentation (changelog entry is not required)': 'pr-documentation',
         # 'Other': doesn't match anything
     }
 
@@ -37,8 +41,20 @@ class Description:
             if stripped == 'I hereby agree to the terms of the CLA available at: https://yandex.ru/legal/cla/?lang=en':
                 self.legal = True
 
-            if stripped == 'Category (leave one):':
+            category_headers = (
+                'Category (leave one):',
+                'Changelog category (leave one):',
+                'Changelog category:',
+                'Category:'
+            )
+
+            if stripped in category_headers:
                 next_category = True
 
         if category in Description.MAP_CATEGORY_TO_LABEL:
             self.label_name = Description.MAP_CATEGORY_TO_LABEL[category]
+        else:
+            if not category:
+                print('Cannot find category in pr description')
+            else:
+                print('Unknown category: ' + category)

utils/github/query.py

 # -*- coding: utf-8 -*-
 
 import requests
+import time
 
 
 class Query:
@@ -394,20 +395,28 @@ class Query:
                 }}
             }}
             '''
-        request = requests_retry_session().post('https://api.github.com/graphql', json={'query': query}, headers=headers)
-        if request.status_code == 200:
-            result = request.json()
-            if 'errors' in result:
-                raise Exception(f'Errors occured: {result["errors"]}')
-
-            if not is_mutation:
-                import inspect
-                caller = inspect.getouterframes(inspect.currentframe(), 2)[1][3]
-                if caller not in self.api_costs.keys():
-                    self.api_costs[caller] = 0
-                self.api_costs[caller] += result['data']['rateLimit']['cost']
-
-            return result['data']
-        else:
-            import json
-            raise Exception(f'Query failed with code {request.status_code}:\n{json.dumps(request.json(), indent=4)}')
+
+        while True:
+            request = requests_retry_session().post('https://api.github.com/graphql', json={'query': query}, headers=headers)
+            if request.status_code == 200:
+                result = request.json()
+                if 'errors' in result:
+                    raise Exception(f'Errors occured: {result["errors"]}')
+
+                if not is_mutation:
+                    import inspect
+                    caller = inspect.getouterframes(inspect.currentframe(), 2)[1][3]
+                    if caller not in self.api_costs.keys():
+                        self.api_costs[caller] = 0
+                    self.api_costs[caller] += result['data']['rateLimit']['cost']
+
+                return result['data']
+            else:
+                import json
+                resp = request.json()
+                if resp and len(resp) > 0 and resp[0] and 'type' in resp[0] and resp[0]['type'] == 'RATE_LIMITED':
+                    print("API rate limit exceeded. Waiting for 1 second.")
+                    time.sleep(1)
+                    continue
+
+                raise Exception(f'Query failed with code {request.status_code}:\n{json.dumps(resp, indent=4)}')