[llvm] 1b18e98 - [Github] Improve formating of PR diffs in bot notifications (#66118)

[via llvm-commits]

Author: cor3ntin
Date: 2023-09-15T16:58:05+02:00
New Revision: 1b18e98642db3ef5b47d16dc556dd26e485ce5dd

URL: https://github.com/llvm/llvm-project/commit/1b18e98642db3ef5b47d16dc556dd26e485ce5dd
DIFF: https://github.com/llvm/llvm-project/commit/1b18e98642db3ef5b47d16dc556dd26e485ce5dd.diff

LOG: [Github] Improve formating of PR diffs in bot notifications (#66118)

* This avoid pinging folks on all issue when they got pinged on bugzilla
eons ago
* Avoid formatting bugs when there is html in the issue description
* Truncate the list of files and the diff independently of each other.
This avoids truncating cutting a file line in 2 and to cut in the middle
of html markup. This is a fringe case but it does happen when people
accidentally push weird branches conflicting on all the files.

Added: 
    

Modified: 
    llvm/utils/git/github-automation.py

Removed: 
    


################################################################################
diff  --git a/llvm/utils/git/github-automation.py b/llvm/utils/git/github-automation.py
index eac5816b5499f6a..5c90ca8bb71505d 100755
--- a/llvm/utils/git/github-automation.py
+++ b/llvm/utils/git/github-automation.py
@@ -47,6 +47,16 @@ def _get_curent_team(team_name, teams) -> Optional[github.Team.Team]:
     return None
 
 
+def escape_description(str):
+    # https://github.com/github/markup/issues/1168#issuecomment-494946168
+    str = html.escape(str, False)
+    # '@' followed by alphanum is a user name
+    str = re.sub("@(?=\w+)", "@<!-- -->", str)
+    # '#' followed by digits is considered an issue number
+    str = re.sub("#(?=\d+\s)", "#<!-- -->", str)
+    return str
+
+
 class IssueSubscriber:
     @property
     def team_name(self) -> str:
@@ -67,12 +77,15 @@ def run(self) -> bool:
         if team.slug == "issue-subscribers-good-first-issue":
             comment = "{}\n".format(beginner_comment)
 
-        comment = (
-            f"@llvm/{team.slug}"
-            + "\n\n<details>\n"
-            + f"{self.issue.body}\n"
-            + "</details>"
-        )
+        body = escape_description(self.issue.body)
+
+        comment = f"""
+ at llvm/{team.slug}
+
+<details>
+{body}
+</details>
+"""
 
         self.issue.create_comment(comment)
         return True
@@ -113,6 +126,11 @@ def run(self) -> bool:
             print(f"couldn't find team named {self.team_name}")
             return False
 
+        # GitHub limits comments to 65,536 characters, let's limit the 
diff 
+        # and the file list to 20kB each.
+        STAT_LIMIT = 20 * 1024
+        DIFF_LIMIT = 20 * 1024
+
         # Get statistics for each file
         
diff _stats = f"{self.pr.changed_files} Files Affected:\n\n"
         for file in self.pr.get_files():
@@ -125,35 +143,41 @@ def run(self) -> bool:
             if file.status == "renamed":
                 print(f"(from {file.previous_filename})")
             
diff _stats += "\n"
-        
diff _stats += "\n"
+            if len(
diff _stats) > STAT_LIMIT:
+                break
 
         # Get the 
diff 
         try:
-            patch = html.escape(requests.get(self.pr.
diff _url).text)
+            patch = requests.get(self.pr.
diff _url).text
         except:
             patch = ""
-        
diff _stats += "\n<pre>\n" + html.escape(patch)
 
-        # GitHub limits comments to 65,536 characters, let's limit the 
diff  to 20kB.
-        DIFF_LIMIT = 20 * 1024
         patch_link = f"Full 
diff : {self.pr.
diff _url}\n"
         if len(patch) > DIFF_LIMIT:
             patch_link = f"\nPatch is {human_readable_size(len(patch))}, truncated to {human_readable_size(DIFF_LIMIT)} below, full version: {self.pr.
diff _url}\n"
-            
diff _stats = 
diff _stats[0:DIFF_LIMIT] + "...\n<truncated>\n"
-        
diff _stats += "</pre>"
+            patch = patch[0:DIFF_LIMIT] + "...\n[truncated]\n"
         team_mention = "@llvm/{}".format(team.slug)
 
-        body = self.pr.body
+        body = escape_description(self.pr.body)
+        # Note: the comment is in markdown and the code below
+        # is sensible to line break
         comment = f"""
 {self.COMMENT_TAG}
 {team_mention}
-            
+
 <details>
 <summary>Changes</summary>
+
 {body}
---
+---
 {patch_link}
+
 {
diff _stats}
+
+``````````
diff 
+{patch}
+``````````
+
 </details>
 """