Add a method to retrieve a Phabricator revision ID from a commit

bugbug/repository.py

@@ -12,6 +12,7 @@ import logging
 import math
 import os
 import pickle
+import re
 import shelve
 import subprocess
 import sys
@@ -108,6 +109,10 @@ TYPES_TO_EXT = {**SOURCE_CODE_TYPES_TO_EXT, **OTHER_TYPES_TO_EXT}
 
 EXT_TO_TYPES = {ext: typ for typ, exts in TYPES_TO_EXT.items() for ext in exts}
 
+PHABRICATOR_REVISION_REGEX = re.compile(
+    "Differential Revision: (https://phabricator.services.mozilla.com/D([0-9]+))"
+)
+
 
 def get_type(path: str) -> str:
     file_name = os.path.basename(path)
@@ -290,6 +295,13 @@ def get_commits(
     )
 
 
+def get_revision_id(commit):
+    match = PHABRICATOR_REVISION_REGEX.search(commit["desc"])
+    if not match:
+        return None
+    return int(match.group(2))
+
+
 def _init_process(repo_dir):
     global HG, REPO_DIR
     REPO_DIR = repo_dir

tests/test_repository.py

@@ -1859,3 +1859,15 @@ def test_get_commits():
     }
     assert len(retrieved_commits) == 10
     assert included_commits.issubset({c["node"] for c in retrieved_commits})
+
+
+def test_get_revision_id():
+    commit = {
+        "desc": "My desc",
+    }
+    assert repository.get_revision_id(commit) is None
+
+    commit = {
+        "desc": "Bug 1667333: Remove unnecessary prefs for mime type checking r=necko-reviewers,evilpie,valentin\n\nDifferential Revision: https://phabricator.services.mozilla.com/D91406",
+    }
+    assert repository.get_revision_id(commit) == 91406