Fix translation doc finder

agent/handler.py

@@ -8,7 +8,6 @@ import gradio as gr
 
 from agent.workflow import (
     report_translation_target_files,
-    report_in_translation_status_files,
     translate_docs_interactive,
     generate_github_pr,
 )
@@ -73,11 +72,8 @@ def process_file_search_handler(lang: str, k: int, history: list) -> tuple:
     state.step = "find_files"
 
     status_report, files_list = report_translation_target_files(lang, k)
-    in_progress_status_report, in_progress_docs = report_in_translation_status_files(
-        lang
-    )
     state.files_to_translate = (
-        [file[0] for file in files_list if file[0] not in in_progress_docs]
+        [file[0] for file in files_list]
         if files_list
         else []
     )
@@ -86,7 +82,7 @@ def process_file_search_handler(lang: str, k: int, history: list) -> tuple:
 
 **Status Report:**
 {status_report}
-{in_progress_status_report}
+
 **📁 Found first {len(state.files_to_translate)} files to translate:**
 """
 

agent/toctree_handler.py

@@ -90,26 +90,7 @@ Korean title:"""
                 'local': local_file_path,
                 'title': en_title
             }
-    
-    def update_local_toctree_file(self, new_entries: List[Dict[str, str]]):
-        """Update or create local _toctree.yml file"""
-        toctree_path = os.path.join(self.local_docs_path, "_toctree.yml")
-        
-        os.makedirs(self.local_docs_path, exist_ok=True)
-        
-        if os.path.exists(toctree_path):
-            with open(toctree_path, 'r', encoding='utf-8') as f:
-                existing_data = yaml.safe_load(f) or []
-        else:
-            existing_data = []
-        
-        for entry in new_entries:
-            if entry not in existing_data:
-                existing_data.append(entry)
-        
-        with open(toctree_path, 'w', encoding='utf-8') as f:
-            yaml.dump(existing_data, f, allow_unicode=True, default_flow_style=False, sort_keys=False)
-    
+   
     def create_updated_toctree_with_llm(self, en_toctree_yaml: str, ko_toctree_yaml: str, target_local: str) -> dict:
         """Use LLM to create updated Korean toctree with new entry at correct position"""
         try:
@@ -177,7 +158,7 @@ Example: If English entry is at position [1]['sections'][0] (1st item in section
             print(f"Error using LLM to create updated toctree: {e}")
             return None
 
-    def process_pr_commit(self, en_titles: List[str], local_paths: List[str], filepath: str):
+    def process_pr_commit(self, filepath: str):
         """Process PR commit by using LLM to create complete updated Korean toctree"""
         # Get filepath without prefix
         filepath_without_prefix = filepath.replace("docs/source/en/", "").replace(".md", "")
@@ -194,16 +175,12 @@ Example: If English entry is at position [1]['sections'][0] (1st item in section
         
         if not updated_ko_toctree:
             print(f"Failed to create updated Korean toctree for local: {filepath_without_prefix}")
-            return []
+            return
         
         print(f"LLM successfully updated Korean toctree")
         
         # Store the updated toctree for commit
         self.updated_ko_toctree = updated_ko_toctree
-        
-        print(f"Updated Korean toctree has {len(updated_ko_toctree)} items")
-        
-        return []
 
     def commit_and_push_toctree(self, pr_agent, owner: str, repo_name: str, branch_name: str):
         """Commit and push toctree updates as a separate commit"""
@@ -219,7 +196,7 @@ Example: If English entry is at position [1]['sections'][0] (1st item in section
             toctree_content = yaml.dump(ko_data, allow_unicode=True, default_flow_style=False, sort_keys=False)
             
             # Create toctree commit message
-            commit_message = "docs: update Korean documentation table of contents - test"
+            commit_message = "docs: update Korean documentation table of contents"
             
             # Commit toctree file
             file_result = pr_agent.create_or_update_file(
@@ -252,7 +229,6 @@ Example: If English entry is at position [1]['sections'][0] (1st item in section
     def update_toctree_after_translation(
         self,
         translation_result: dict, 
-        en_title: str, 
         filepath: str, 
         pr_agent, 
         github_config: dict
@@ -261,7 +237,6 @@ Example: If English entry is at position [1]['sections'][0] (1st item in section
         
         Args:
             translation_result: Result from translation PR workflow
-            en_title: English title for toctree mapping
             filepath: Original file path
             pr_agent: GitHub PR agent instance
             github_config: GitHub configuration dictionary
@@ -269,28 +244,22 @@ Example: If English entry is at position [1]['sections'][0] (1st item in section
         Returns:
             Dictionary with toctree update result
         """
-        if translation_result["status"] == "error" or not en_title:
+        if translation_result["status"] == "error":
             return None
             
         try:
-            local_path = filepath.split("/")[-1].replace(".md", "")
-            
-            # Create new toctree entries
-            new_entries = self.process_pr_commit([en_title], [local_path], filepath)
-            print("self.updated_ko_toctree = updated_ko_toctree:", self.updated_ko_toctree)
+            # Process toctree update with LLM
+            self.process_pr_commit(filepath)
             # Commit toctree as separate commit
-            return self.commit_and_push_toctree(
-                pr_agent=pr_agent,
-                owner=github_config["owner"],
-                repo_name=github_config["repo_name"],
-                branch_name=translation_result["branch"]
-            )
-            # return {
-            #     'status': 'success', 
-            #     'message': 'Toctree committed successfully: SUCCESS: File updated - docs/source/ko/_toctree.yml', 
-            #     'commit_message': 'docs: update Korean documentation table of contents'
-            #     }
-            
+            print("self.updated_ko_toctree:", self.updated_ko_toctree:)
+            if self.updated_ko_toctree:
+                return self.commit_and_push_toctree(
+                    pr_agent=pr_agent,
+                    owner=github_config["owner"],
+                    repo_name=github_config["repo_name"],
+                    branch_name=translation_result["branch"]
+                )
+
         except Exception as e:
             return {
                 "status": "error",

agent/workflow.py

@@ -28,27 +28,34 @@ except ImportError as e:
 def report_translation_target_files(
     translate_lang: str, top_k: int = 1
 ) -> tuple[str, list[list[str]]]:
-    """Return the top-k files that need translation.
+    """Return the top-k files that need translation, excluding files already in progress.
 
     Args:
         translate_lang: Target language to translate
         top_k: Number of top-first files to return for translation. (Default 1)
     """
-    status_report, filepath_list = report(translate_lang, top_k)
-    return status_report, [[file] for file in filepath_list]
+    # Get files in progress
+    docs_in_progress, pr_info_list = get_github_issue_open_pr(translate_lang)
+
+    # Get all available files for translation
+    all_status_report, all_filepath_list = report(translate_lang, top_k * 2)  # Get more to account for filtering
 
+    # Filter out files that are already in progress
+    available_files = [f for f in all_filepath_list if f not in docs_in_progress]
 
-def report_in_translation_status_files(translate_lang: str) -> tuple[str, list[str]]:
-    docs, pr_info_list = get_github_issue_open_pr(translate_lang)
+    # Take only the requested number
+    filepath_list = available_files[:top_k]
+    
+    # Build combined status report
+    status_report = all_status_report
 
-    status_report = ""
-    if docs:
-        status_report = f"""\n🤖 Found {len(docs)} in progress for translation.
-        """
-        for i, file in enumerate(docs):
+    if docs_in_progress:
+        status_report += f"\n\n🤖 Found {len(docs_in_progress)} files in progress for translation:"
+        for i, file in enumerate(docs_in_progress):
             status_report += f"\n{i+1}. `{file}`: {pr_info_list[i]}"
-        status_report += "\n"
-    return status_report, docs
+        status_report += f"\n\n📋 Showing {len(filepath_list)} available files (excluding in-progress):"
+
+    return status_report, [[file] for file in filepath_list]
 
 
 def translate_docs(lang: str, file_path: str, additional_instruction: str = "") -> tuple[str, str]:
@@ -58,14 +65,14 @@ def translate_docs(lang: str, file_path: str, additional_instruction: str = "")
         Path(__file__).resolve().parent.parent
         / f"translation_result/{file_path}"
     )
-    
+
     if translation_file_path.exists():
         print(f"📄 Found existing translation: {translation_file_path}")
         with open(translation_file_path, "r", encoding="utf-8") as f:
             existing_content = f.read()
         if existing_content.strip():
             return "Existing translation loaded (no tokens used)", existing_content
-    
+
     # step 1. Get content from file path
     content = get_content(file_path)
     to_translate = preprocess_content(content)
@@ -195,9 +202,8 @@ def generate_github_pr(
             from agent.toctree_handler import TocTreeHandler
             toctree_handler = TocTreeHandler()
             toctree_result = toctree_handler.update_toctree_after_translation(
-                result, en_title, filepath, agent, github_config
+                result, filepath, agent, github_config
             )
-            print("toctree_result:", toctree_result)
 
         # Process result
         # Generate toctree status message (shared for both success and partial_success)

pr_generator/agent.py

@@ -94,10 +94,17 @@ class GitHubPRAgent:
             if existing_pr:
                 return f"ERROR: {existing_pr}"
 
-            # 3. Verify head branch exists
+            # 3. Verify head and base branches exist
             repo = self.github_client.get_repo(f"{owner}/{repo_name}")
             try:
-                head_branch = repo.get_branch(head)
+                # For fork-to-upstream PR, head format is "fork_owner:branch_name"
+                if ":" in head:
+                    fork_owner, branch_name = head.split(":", 1)
+                    fork_repo = self.github_client.get_repo(f"{fork_owner}/{repo_name}")
+                    head_branch = fork_repo.get_branch(branch_name)
+                else:
+                    head_branch = repo.get_branch(head)
+
                 base_branch = repo.get_branch(base)
 
                 # 4. Check if head and base branches point to the same commit
@@ -159,7 +166,9 @@ class GitHubPRAgent:
         """Check if there's an existing PR with the same head and base."""
         try:
             repo = self.github_client.get_repo(f"{owner}/{repo_name}")
-            pulls = repo.get_pulls(state="open", head=f"{owner}:{head}", base=base)
+            # For head parameter, use exactly what was passed (could be "fork_owner:branch" or just "branch")
+            search_head = head if ":" in head else f"{owner}:{head}"
+            pulls = repo.get_pulls(state="open", head=search_head, base=base)
             for pr in pulls:
                 return f"Existing PR found: {pr.html_url}"
             return None
@@ -448,12 +457,12 @@ Please return only the commit message. No other explanation is needed."""
                 pr_analysis["head_branch"], target_language, file_name
             )
 
-            # 3. Get main branch SHA and create branch
-            repo = self.github_client.get_repo(f"{owner}/{repo_name}")
-            main_branch = repo.get_branch(base_branch)
+            # 3. Get main branch SHA from upstream and create branch in fork
+            upstream_repo = self.github_client.get_repo(f"huggingface/{repo_name}")
+            main_branch = upstream_repo.get_branch(base_branch)
             main_sha = main_branch.commit.sha
 
-            print(f"🌿 Creating branch: {branch_name}")
+            print(f"🌿 Creating branch: {branch_name} in fork repository")
             branch_result = self.create_branch(owner, repo_name, branch_name, main_sha)
 
             # Check branch creation result
@@ -466,8 +475,11 @@ Please return only the commit message. No other explanation is needed."""
             elif branch_result.startswith("WARNING"):
                 print(f"⚠️ {branch_result}")
                 # Continue if branch already exists
+            elif branch_result.startswith("SUCCESS"):
+                print(f"✅ {branch_result}")
             else:
-                print(f"{branch_result}")
+                print(f"⚠️ Unexpected branch creation result: {branch_result}")
+                # Continue anyway, might still work
 
             # 4. Generate commit message and save file
             commit_messages = [commit["message"] for commit in pr_analysis["commits"]]
@@ -506,10 +518,11 @@ Please return only the commit message. No other explanation is needed."""
             )
 
             print(f"🔄 Creating PR: {pr_title}")
-            print(f"   Head: {branch_name} → Base: {base_branch}")
+            print(f"   Head: {owner}:{branch_name} → Base: huggingface:{base_branch}")
 
+            # Create PR from fork to upstream repository
             pr_result = self.create_pull_request(
-                owner, repo_name, pr_title, branch_name, base_branch, pr_body
+                "huggingface", "transformers", pr_title, f"{owner}:{branch_name}", base_branch, pr_body
             )
 
             if pr_result.startswith("ERROR"):