feat: Tooling fixes

src/codegen/extensions/langchain/tools.py

@@ -493,8 +493,7 @@ class GithubCreatePRReviewCommentInput(BaseModel):
     body: str = Field(..., description="The comment text")
     commit_sha: str = Field(..., description="The commit SHA to attach the comment to")
     path: str = Field(..., description="The file path to comment on")
-    line: int | None = Field(None, description="The line number to comment on")
-    side: str | None = Field(None, description="Which version of the file to comment on ('LEFT' or 'RIGHT')")
+    line: int = Field(..., description="The line number to comment on use the indices from the diff")
     start_line: int | None = Field(None, description="For multi-line comments, the starting line")
 
 
@@ -515,8 +514,7 @@ def _run(
         body: str,
         commit_sha: str,
         path: str,
-        line: int | None = None,
-        side: str | None = None,
+        line: int,
         start_line: int | None = None,
     ) -> str:
         result = create_pr_review_comment(
@@ -526,8 +524,6 @@ def _run(
             commit_sha=commit_sha,
             path=path,
             line=line,
-            side=side,
-            start_line=start_line,
         )
         return result.render()
 

src/codegen/extensions/tools/github/create_pr_review_comment.py

@@ -15,17 +15,18 @@ class PRReviewCommentObservation(Observation):
     pr_number: int = Field(
         description="PR number the comment was added to",
     )
+    body: str = Field(
+        description="Content of the comment",
+    )
+    commit_sha: str = Field(
+        description="Commit SHA the comment was added to",
+    )
     path: str = Field(
         description="File path the comment was added to",
     )
-    line: Optional[int] = Field(
-        default=None,
+    line: int = Field(
         description="Line number the comment was added to",
     )
-    body: str = Field(
-        description="Content of the comment",
-    )
-
     str_template: ClassVar[str] = "Added review comment to PR #{pr_number} at {path}:{line}"
 
 
@@ -35,8 +36,7 @@ def create_pr_review_comment(
     body: str,
     commit_sha: str,
     path: str,
-    line: Optional[int] = None,
-    side: Optional[str] = None,
+    line: int,
     start_line: Optional[int] = None,
 ) -> PRReviewCommentObservation:
     """Create an inline review comment on a specific line in a pull request.
@@ -48,8 +48,6 @@ def create_pr_review_comment(
         commit_sha: The commit SHA to attach the comment to
         path: The file path to comment on
         line: The line number to comment on
-        side: Which version of the file to comment on ('LEFT' or 'RIGHT')
-        start_line: For multi-line comments, the starting line
     """
     try:
         codebase.create_pr_review_comment(
@@ -58,15 +56,15 @@ def create_pr_review_comment(
             commit_sha=commit_sha,
             path=path,
             line=line,
-            side=side,
-            start_line=start_line,
+            side="RIGHT",
         )
         return PRReviewCommentObservation(
             status="success",
             pr_number=pr_number,
             path=path,
             line=line,
             body=body,
+            commit_sha=commit_sha,
         )
     except Exception as e:
         return PRReviewCommentObservation(
@@ -76,4 +74,5 @@ def create_pr_review_comment(
             path=path,
             line=line,
             body=body,
+            commit_sha=commit_sha,
         )

src/codegen/extensions/tools/github/view_pr.py

@@ -18,6 +18,9 @@ class ViewPRObservation(Observation):
     patch: str = Field(
         description="The PR's patch/diff content",
     )
+    file_commit_sha: dict[str, str] = Field(
+        description="Commit SHAs for each file in the PR",
+    )
 
     str_template: ClassVar[str] = "PR #{pr_id}"
 
@@ -30,12 +33,13 @@ def view_pr(codebase: Codebase, pr_id: int) -> ViewPRObservation:
         pr_id: Number of the PR to get the contents for
     """
     try:
-        modified_symbols, patch = codebase.get_modified_symbols_in_pr(pr_id)
+        patch, file_commit_sha = codebase.get_modified_symbols_in_pr(pr_id)
 
         return ViewPRObservation(
             status="success",
             pr_id=pr_id,
             patch=patch,
+            file_commit_sha=file_commit_sha,
         )
 
     except Exception as e:
@@ -44,4 +48,5 @@ def view_pr(codebase: Codebase, pr_id: int) -> ViewPRObservation:
             error=f"Failed to view PR: {e!s}",
             pr_id=pr_id,
             patch="",
+            file_commit_sha={},
         )

src/codegen/git/repo_operator/repo_operator.py

@@ -142,7 +142,7 @@
         email_level = None
         levels = ["system", "global", "user", "repository"]
         for level in levels:
             with git_cli.config_reader(level) as reader:
                 if reader.has_option("user", "name") and not username:
                     username = reader.get("user", "name")
                     user_level = level
@@ -554,7 +554,7 @@
                     return content
             except UnicodeDecodeError:
                 print(f"Warning: Unable to decode file {file_path}. Skipping.")
                 return None
 
     def write_file(self, relpath: str, content: str) -> None:
         """Writes file content to disk"""
@@ -621,7 +621,7 @@
         filepaths = self.get_filepaths_for_repo(ignore_list)
         # Iterate through files and yield contents
         for rel_filepath in filepaths:
             rel_filepath: str
             filepath = os.path.join(self.repo_path, rel_filepath)
 
             # Filter by subdirectory (includes full filenames)
@@ -652,7 +652,7 @@
         list_files = []
 
         for rel_filepath in self.git_cli.git.ls_files().split("\n"):
             rel_filepath: str
             if subdirs and not any(d in rel_filepath for d in subdirs):
                 continue
             if extensions is None or any(rel_filepath.endswith(e) for e in extensions):
@@ -676,7 +676,7 @@
 
     def get_modified_files_in_last_n_days(self, days: int = 1) -> tuple[list[str], list[str]]:
         """Returns a list of files modified and deleted in the last n days"""
         modified_files = []
         deleted_files = []
         allowed_extensions = [".py"]
 
@@ -692,9 +692,9 @@
                     if file in modified_files:
                         modified_files.remove(file)
                 else:
                     if file not in modified_files and file[-3:] in allowed_extensions:
                         modified_files.append(file)
         return modified_files, deleted_files
 
     @cached_property
     def base_url(self) -> str | None:
@@ -715,7 +715,7 @@
 
     def get_pr_data(self, pr_number: int) -> dict:
         """Returns the data associated with a PR"""
         return self.remote_git_repo.get_pr_data(pr_number)
 
     def create_pr_comment(self, pr_number: int, body: str) -> None:
         """Create a general comment on a pull request.
@@ -735,7 +735,7 @@
         commit_sha: str,
         path: str,
         line: int | None = None,
-        side: str | None = None,
+        side: str = "RIGHT",
         start_line: int | None = None,
     ) -> None:
         """Create an inline review comment on a specific line in a pull request.
@@ -758,9 +758,8 @@
                     body=body,
                     commit=commit,
                     path=path,
                     line=line,
                     side=side,
-                    start_line=start_line,
                 )
 
     def get_pull_request(self, pr_number: int) -> PullRequest | None:

src/codegen/git/utils/pr_review.py

@@ -9,7 +9,7 @@
 from codegen.git.repo_operator.repo_operator import RepoOperator
 
 if TYPE_CHECKING:
     from codegen.sdk.core.codebase import Codebase, Editable, File, Symbol
 
 
 def get_merge_base(git_repo_client: Repository, pull: PullRequest | PullRequestContext) -> str:
@@ -73,6 +73,45 @@
     return max(range1.start, range2.start) < min(range1.stop, range2.stop)
 
 
+def get_file_to_commit_sha(op: RepoOperator, pull: PullRequest) -> dict[str, str]:
+    """Gets a mapping of file paths to their latest commit SHA in the PR.
+
+    Args:
+        op (RepoOperator): The repository operator
+        pull (PullRequest): The pull request object
+
+    Returns:
+        dict[str, str]: A dictionary mapping file paths to their latest commit SHA
+    """
+    if not op.remote_git_repo:
+        msg = "GitHub API client is required to get PR commit information"
+        raise ValueError(msg)
+
+    file_to_commit = {}
+
+    # Get all commits in the PR
+    commits = list(pull.get_commits())
+
+    # Get all modified files
+    files = pull.get_files()
+
+    # For each file, find its latest commit
+    for file in files:
+        # Look through commits in reverse order to find the latest one that modified this file
+        for commit in reversed(commits):
+            # Get the files modified in this commit
+            files_in_commit = commit.files
+            if any(f.filename == file.filename for f in files_in_commit):
+                file_to_commit[file.filename] = commit.sha
+                break
+
+        # If we didn't find a commit (shouldn't happen), use the head SHA
+        if file.filename not in file_to_commit:
+            file_to_commit[file.filename] = pull.head.sha
+
+    return file_to_commit
+
+
 class CodegenPR:
     """Wrapper around PRs - enables codemods to interact with them"""
 
@@ -143,3 +182,11 @@
         else:
             # If diff_url not available, get the patch directly
             return self._gh_pr.get_patch()
+
+    def get_commit_sha(self) -> str:
+        """Get the commit SHA of the PR"""
+        return self._gh_pr.head.sha
+
+    def get_file_commit_shas(self) -> dict[str, str]:
+        """Get a mapping of file paths to their latest commit SHA in the PR"""
+        return get_file_to_commit_sha(op=self._op, pull=self._gh_pr)

src/codegen/sdk/core/codebase.py

@@ -1311,17 +1311,35 @@ def from_repo(
             logger.exception(f"Failed to initialize codebase: {e}")
             raise
 
-    def get_modified_symbols_in_pr(self, pr_id: int) -> tuple[list[Symbol], str]:
+    def get_modified_symbols_in_pr(self, pr_id: int) -> tuple[str, dict[str, str]]:
         """Get all modified symbols in a pull request"""
         pr = self._op.get_pull_request(pr_id)
         cg_pr = CodegenPR(self._op, self, pr)
         patch = cg_pr.get_pr_diff()
-        return cg_pr.modified_symbols, patch
+        commit_sha = cg_pr.get_file_commit_shas()
+        return patch, commit_sha
 
     def create_pr_comment(self, pr_number: int, body: str) -> None:
         """Create a comment on a pull request"""
         return self._op.create_pr_comment(pr_number, body)
 
+    def create_pr_review_comment(self, pr_number: int, body: str, commit_sha: str, path: str, line: int | None = None, side: str = "RIGHT", start_line: int | None = None) -> None:
+        """Create a review comment on a pull request.
+
+        Args:
+            pr_number: The number of the pull request
+            body: The body of the comment
+            commit_sha: The SHA of the commit to comment on
+            path: The path of the file to comment on
+            line: The line number to comment on
+            side: The side of the comment to create
+            start_line: The start line number to comment on
+
+        Returns:
+            None
+        """
+        return self._op.create_pr_review_comment(pr_number, body, commit_sha, path, line, side, start_line)
+
 
 # The last 2 lines of code are added to the runner. See codegen-backend/cli/generate/utils.py
 # Type Aliases