[APIView Copilot] Include full comment thread in Report Issue context (#15524)

* Include full comment thread in Report Issue context

get_comment_with_context now also fetches sibling comments sharing the anchor's ThreadId, ordered by CreatedOn, and exposes them as 'thread_comments'. _lookup_comment_context forwards the list (id, text, source, author, created_on) to the prompt context, and _format_comment_context_for_prompt renders them as a chronological 'Thread:' transcript when there is more than one comment so the LLM sees the full conversation, not just the anchor comment. Single-comment threads keep the existing single-line 'Comment:' rendering.

* Potential fix for pull request finding

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

* Address Copilot review feedback on PR #15524

1. Avoid GitHub mention spam in fallback issue body: _format_comment_context_for_prompt now accepts escape_mentions=True (used by _build_fallback_body) to wrap thread-author tokens in backticks (e.g. `@alice`) so the deterministic fallback body can never accidentally @-notify real users. The LLM prompt path still gets bare @author so the model sees authorship clearly.

2. Push the IsDeleted filter into the Cosmos thread query (AND (NOT IS_DEFINED(c.IsDeleted) OR c.IsDeleted = false)) so we no longer transfer/sort tombstoned comments client-side.

---------

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Author: helen229

packages/python-packages/apiview-copilot/prompts/report_issue/generate_issue.prompty

@@ -37,7 +37,10 @@ You receive these inputs:
 * **language** — optional language hint (e.g. `python`, `C#`). May be empty.
 * **comment_context** — optional structured snapshot of the comment thread
   the user was viewing (source, language, comment text, code snippet,
-  element id). Use it to infer what the user is talking about.
+  element id). When the thread has more than one comment, an additional
+  `Thread:` block lists every comment in chronological order as
+  `@author (timestamp): text` so you can see the full conversation, not
+  just the anchor comment. Use it to infer what the user is talking about.
 
 You must determine the **category** yourself. Allowed categories:
 

packages/python-packages/apiview-copilot/src/_apiview.py

@@ -1466,7 +1466,10 @@ def get_comment_with_context(comment_id: str, environment: str = "production") -
 
     Returns:
         A dict containing:
-        - comment: The full comment object from the database
+        - comment: The full comment object from the database (the anchor comment)
+        - thread_comments: List of non-deleted comments sharing the same
+          ThreadId, in chronological order. Deleted comments, including the
+          anchor comment, may be excluded from this list.
         - language: The pretty language name (e.g., "Python")
         - package_name: The package name from the review
         - code: The API code from the revision (if available)
@@ -1501,6 +1504,34 @@ def get_comment_with_context(comment_id: str, environment: str = "production") -
         comment = results[0]
         review_id = comment.get("ReviewId")
         revision_id = comment.get("APIRevisionId")
+        thread_id = comment.get("ThreadId")
+
+        # Fetch sibling comments in the same thread (chronological order) so
+        # callers can render the full conversation, not just the anchor comment.
+        # Filter out tombstoned (IsDeleted) entries server-side so we only
+        # transfer / sort live comments.
+        thread_comments: list[dict] = [comment]
+        if thread_id:
+            thread_query = """
+                SELECT c.id, c.CommentText, c.CommentSource, c.CreatedBy, c.CreatedOn,
+                       c.IsResolved, c.IsDeleted, c.ThreadId
+                FROM c
+                WHERE c.ThreadId = @thread_id
+                  AND (NOT IS_DEFINED(c.IsDeleted) OR c.IsDeleted = false)
+                ORDER BY c.CreatedOn ASC
+            """
+            try:
+                thread_results = list(
+                    comments_container.query_items(
+                        query=thread_query,
+                        parameters=[{"name": "@thread_id", "value": thread_id}],
+                        enable_cross_partition_query=True,
+                    )
+                )
+                thread_comments = thread_results or [comment]
+            except Exception as e:
+                print(f"Warning: Could not fetch thread siblings for {thread_id}: {e}")
+                thread_comments = [comment]
 
         # Get language and package name from Reviews container
         language = None
@@ -1568,6 +1599,7 @@ def get_comment_with_context(comment_id: str, environment: str = "production") -
 
         return {
             "comment": comment,
+            "thread_comments": thread_comments,
             "language": language,
             "package_name": package_name,
             "code": code,

packages/python-packages/apiview-copilot/src/_report_issue.py

@@ -67,8 +67,14 @@ def _build_labels(category: str, language: Optional[str]) -> list[str]:
     return ["APIView"]
 
 
-def _format_comment_context_for_prompt(ctx: Optional[dict]) -> str:
-    """Render the optional comment context as plain text for the LLM."""
+def _format_comment_context_for_prompt(ctx: Optional[dict], *, escape_mentions: bool = False) -> str:
+    """Render the optional comment context as plain text.
+
+    When ``escape_mentions`` is True (use this for content that will be
+    posted to GitHub, e.g. the deterministic fallback issue body), author
+    names in the rendered thread transcript are wrapped so they cannot
+    trigger GitHub ``@mention`` notifications.
+    """
     if not ctx:
         return ""
     parts: list[str] = []
@@ -82,6 +88,27 @@ def _format_comment_context_for_prompt(ctx: Optional[dict]) -> str:
         value = ctx.get(key)
         if value:
             parts.append(f"{label}: {value}")
+    thread = ctx.get("thread_comments") or []
+    # Only render a transcript when there is more than the anchor comment;
+    # the single-comment case is already covered by the "Comment" line above.
+    if len(thread) > 1:
+        transcript_lines: list[str] = []
+        for entry in thread:
+            author = entry.get("created_by") or "unknown"
+            created_on = entry.get("created_on") or ""
+            text = (entry.get("comment_text") or "").strip()
+            if not text:
+                continue
+            # In the GitHub issue body we wrap the author in backticks so
+            # the literal ``@author`` text does not turn into a real mention
+            # / notification. The LLM prompt keeps the bare ``@author``
+            # form so the model can still see authorship clearly.
+            header = f"`@{author}`" if escape_mentions else f"@{author}"
+            if created_on:
+                header = f"{header} ({created_on})"
+            transcript_lines.append(f"{header}: {text}")
+        if transcript_lines:
+            parts.append("Thread:\n" + "\n".join(transcript_lines))
     return "\n".join(parts)
 
 
@@ -91,7 +118,7 @@ def _build_fallback_body(description: str, review_link: Optional[str], comment_c
     if review_link:
         sections.append(f"## Review Link\n\n{review_link}")
     sections.append(f"## Description\n\n{description}")
-    ctx_text = _format_comment_context_for_prompt(comment_context)
+    ctx_text = _format_comment_context_for_prompt(comment_context, escape_mentions=True)
     if ctx_text:
         sections.append("## Comment Context\n\n" + ctx_text)
     sections.append("---\n*Reported via APIView*")
@@ -127,6 +154,17 @@ def _lookup_comment_context(comment_id: str) -> Optional[dict]:
     if not ctx:
         return None
     comment_obj = ctx.get("comment") or {}
+    thread_raw = ctx.get("thread_comments") or []
+    thread_comments = [
+        {
+            "id": entry.get("id"),
+            "comment_text": entry.get("CommentText"),
+            "comment_source": entry.get("CommentSource"),
+            "created_by": entry.get("CreatedBy"),
+            "created_on": entry.get("CreatedOn"),
+        }
+        for entry in thread_raw
+    ]
     return {
         "comment_text": comment_obj.get("CommentText"),
         "comment_source": comment_obj.get("CommentSource"),
@@ -135,6 +173,7 @@ def _lookup_comment_context(comment_id: str) -> Optional[dict]:
         "element_id": comment_obj.get("ElementId"),
         "review_id": comment_obj.get("ReviewId"),
         "revision_id": comment_obj.get("APIRevisionId"),
+        "thread_comments": thread_comments,
     }
 
 

packages/python-packages/apiview-copilot/tests/report_issue_test.py

@@ -101,6 +101,45 @@ def test_empty_dict(self):
     def test_none(self):
         assert _format_comment_context_for_prompt(None) == ""
 
+    def test_single_comment_thread_does_not_emit_transcript(self):
+        text = _format_comment_context_for_prompt({
+            "comment_text": "lone comment",
+            "thread_comments": [
+                {"comment_text": "lone comment", "created_by": "alice", "created_on": "2026-05-01T00:00:00Z"},
+            ],
+        })
+        assert "Thread:" not in text
+        assert "Comment: lone comment" in text
+
+    def test_multi_comment_thread_emits_transcript(self):
+        text = _format_comment_context_for_prompt({
+            "comment_text": "first",
+            "thread_comments": [
+                {"comment_text": "first", "created_by": "alice", "created_on": "2026-05-01T00:00:00Z"},
+                {"comment_text": "second", "created_by": "bob", "created_on": "2026-05-02T00:00:00Z"},
+            ],
+        })
+        assert "Thread:" in text
+        assert "@alice (2026-05-01T00:00:00Z): first" in text
+        assert "@bob (2026-05-02T00:00:00Z): second" in text
+
+    def test_escape_mentions_wraps_authors_in_backticks(self):
+        text = _format_comment_context_for_prompt(
+            {
+                "comment_text": "first",
+                "thread_comments": [
+                    {"comment_text": "first", "created_by": "alice", "created_on": "2026-05-01T00:00:00Z"},
+                    {"comment_text": "second", "created_by": "bob", "created_on": "2026-05-02T00:00:00Z"},
+                ],
+            },
+            escape_mentions=True,
+        )
+        assert "`@alice` (2026-05-01T00:00:00Z): first" in text
+        assert "`@bob` (2026-05-02T00:00:00Z): second" in text
+        # No bare @author tokens that GitHub would turn into mentions.
+        assert "@alice" not in text.replace("`@alice`", "")
+        assert "@bob" not in text.replace("`@bob`", "")
+
 
 class TestBuildFallbackTitleSnippet:
     def test_short(self):
@@ -280,6 +319,22 @@ def test_maps_db_payload_to_comment_context(self, mock_get):
                 "ReviewId": "r-1",
                 "APIRevisionId": "rev-2",
             },
+            "thread_comments": [
+                {
+                    "id": "c1",
+                    "CommentText": "remove async",
+                    "CommentSource": "copilot",
+                    "CreatedBy": "azure-sdk",
+                    "CreatedOn": "2026-05-01T00:00:00Z",
+                },
+                {
+                    "id": "c2",
+                    "CommentText": "actually it should stay async",
+                    "CommentSource": "UserGenerated",
+                    "CreatedBy": "alice",
+                    "CreatedOn": "2026-05-02T00:00:00Z",
+                },
+            ],
             "code": "async def upload_blob(self, name: str, data: bytes) -> None: ...",
             "language": "Python",
             "package_name": "azure-storage-blob",
@@ -293,6 +348,22 @@ def test_maps_db_payload_to_comment_context(self, mock_get):
             "element_id": "AsyncBlobClient.upload_blob",
             "review_id": "r-1",
             "revision_id": "rev-2",
+            "thread_comments": [
+                {
+                    "id": "c1",
+                    "comment_text": "remove async",
+                    "comment_source": "copilot",
+                    "created_by": "azure-sdk",
+                    "created_on": "2026-05-01T00:00:00Z",
+                },
+                {
+                    "id": "c2",
+                    "comment_text": "actually it should stay async",
+                    "comment_source": "UserGenerated",
+                    "created_by": "alice",
+                    "created_on": "2026-05-02T00:00:00Z",
+                },
+            ],
         }
 
     @patch("src._report_issue.os.getenv")