Merge pull request #3 from Jwaminju/update-translator

.gitignore

@@ -1,2 +1,3 @@
 .env
-*/__pycache__/
+*/__pycache__/
+pr_success.log

agent/handler.py

@@ -150,12 +150,6 @@ def start_translation_process():
             ""
             f"{original_file_link}\n"
             "**🌐 Translated Content:**\n"
-            # f"\n```\n\n{_extract_content_for_display(translated)}\n```"
-            # "\n```\n\n"
-            # f"\n{translated}\n"
-            # f"```"
-            # f"{status}\n"
-            # "✅ Translation completed. The code block will be added when generating PR."
         )
         return response, translated
 

agent/toctree_handler.py

@@ -265,7 +265,6 @@ Korean title:"""
             # Process toctree update with LLM
             self.process_pr_commit(filepath)
             # Commit toctree as separate commit
-            print("self.updated_ko_toctree:", self.updated_ko_toctree)
             if self.updated_ko_toctree:
                 return self.commit_and_push_toctree(
                     pr_agent=pr_agent,

agent/workflow.py

@@ -12,7 +12,6 @@ from translator.content import (
     preprocess_content,
 )
 from translator.retriever import report, get_github_issue_open_pr
-
 # GitHub PR Agent import
 try:
     from pr_generator.agent import GitHubPRAgent
@@ -22,7 +21,8 @@ except ImportError as e:
     print(f"⚠️ GitHub PR Agent is not available: {e}")
     GITHUB_PR_AVAILABLE = False
 
-# GitHub configuration - must be provided by user or environment variables
+import json
+from logger.github_logger import GitHubLogger
 
 
 def report_translation_target_files(
@@ -71,7 +71,7 @@ def translate_docs(lang: str, file_path: str, additional_instruction: str = "")
         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
+            return "Existing translation loaded (no tokens used). If you want to translate again, please restart the gradio app.", existing_content
 
     # step 1. Get content from file path
     content = get_content(file_path)
@@ -115,6 +115,7 @@ def translate_docs_interactive(
     callback_result, translated_content = translate_docs(translate_lang, current_file, additional_instruction)
     status += f"💰 Used token and cost: \n```\n{callback_result}\n```"
 
+    print(callback_result)
     print(status)
 
     return translated_content
@@ -190,10 +191,11 @@ def generate_github_pr(
             repo_name=github_config["repo_name"],
             base_branch=github_config.get("base_branch", "main"),
         )
+        # TEST CODE
         # result = {
-        #     'status': 'partial_success', 
-        #     'branch': 'ko-attention_interface', 
-        #     'file_path': 'docs/source/ko/attention_interface.md', 
+        #     'status': 'partial_success',
+        #     'branch': 'ko-attention_interface',
+        #     'file_path': 'docs/source/ko/attention_interface.md',
         #     'message': 'File was saved and commit was successful.\nPR creation failed: ERROR: Existing PR found: https://github.com/Jwaminju/transformers/pull/1', 'error_details': 'ERROR: Existing PR found: https://github.com/Jwaminju/transformers/pull/1'
         #     }
         # Process toctree update after successful translation PR
@@ -213,19 +215,19 @@ def generate_github_pr(
                 toctree_status = f"\n📋 **Toctree Updated:** ✅ {toctree_result['message']}"
             else:
                 toctree_status = f"\n📋 **Toctree Update Failed:** ❌ {toctree_result['message']}"
-        
+
+        # Append full result JSON to dedicated GitHub logging repository (always)
+        try:
+            log_data = result.copy()
+            if toctree_result:
+                log_data["toctree_result"] = toctree_result
+            log_entry = json.dumps(log_data, ensure_ascii=False) + "\n"
+            log_res = GitHubLogger().append_jsonl(log_entry)
+            print(f"📝 Log append result: {log_res}")
+        except Exception as e:
+            print(f"❌ Failed to append PR log via GitHub API: {e}")
+
         if result["status"] == "success":
-            # Log successful PR to file
-            import datetime
-            pr_url = result.get('pr_url', 'NO_PR_URL')
-            log_entry = f"[{datetime.datetime.now().isoformat()}] {result['file_path']} -> {pr_url} ({result['status']})\n"
-            try:
-                with open("pr_success.log", "a", encoding="utf-8") as f:
-                    f.write(log_entry)
-                print(f"✅ Logged PR result: {log_entry.strip()}")
-            except Exception as e:
-                print(f"❌ Failed to log PR result: {e}")
-            
             return f"""✅ **GitHub PR Creation Successful!**
 
 🔗 **PR URL:** {result.get('pr_url', 'NO_PR_URL')}
@@ -235,17 +237,6 @@ def generate_github_pr(
 {result["message"]}"""
 
         elif result["status"] == "partial_success":
-            # Log partial success to file
-            import datetime
-            pr_url = result.get('pr_url', 'NO_PR_URL')
-            log_entry = f"[{datetime.datetime.now().isoformat()}] {result['file_path']} -> {pr_url} ({result['status']})\n"
-            try:
-                with open("pr_success.log", "a", encoding="utf-8") as f:
-                    f.write(log_entry)
-                print(f"✅ Logged PR result: {log_entry.strip()}")
-            except Exception as e:
-                print(f"❌ Failed to log PR result: {e}")
-                
             return f"""⚠️ **Partial Success**
 
 🌿 **Branch:** {result["branch"]}

example.env

@@ -4,4 +4,15 @@ ANTHROPIC_API_KEY=<your api key>
 GITHUB_TOKEN=<your github token>
 GITHUB_OWNER=<your github username>
 GITHUB_REPO=<your repository name>
-REFERENCE_PR_URL=<reference pr url for style analysis>
+REFERENCE_PR_URL=<reference pr url for style analysis>
+
+#  Secrets for deployment to HF space
+HF_TOKEN=
+HF_USERNAME=
+HF_SPACE_NAME=
+
+# Secrets for logging to Github
+LOG_REPO=
+LOG_GITHUB_TOKEN=
+LOG_BRANCH=
+LOG_FILE_PATH=

logger/github_logger.py

@@ -0,0 +1,71 @@
+import os
+import base64
+from typing import Optional
+
+try:
+    from github import Github, GithubException
+    LIBS_OK = True
+except ImportError:
+    LIBS_OK = False
+
+class GitHubLogger:
+    """Dedicated logger that appends JSONL entries to a GitHub repo/branch/file.
+
+    Env vars:
+      - LOG_GITHUB_TOKEN (fallback: GITHUB_TOKEN)
+      - LOG_REPO (format: owner/repo)
+      - LOG_BRANCH (default: 'log_event')
+      - LOG_FILE_PATH (default: 'pr_success.log')
+    """
+
+    def __init__(self):
+        if not LIBS_OK:
+            raise ImportError("PyGithub not installed. Please install PyGithub.")
+        token = os.environ.get("LOG_GITHUB_TOKEN") or os.environ.get("GITHUB_TOKEN")
+        if not token:
+            raise ValueError("Missing LOG_GITHUB_TOKEN or GITHUB_TOKEN for logging")
+        self._client = Github(token)
+
+        repo_spec = os.environ.get("LOG_REPO")
+        if not repo_spec or "/" not in repo_spec:
+            raise ValueError("Missing or invalid LOG_REPO. Expected 'owner/repo'.")
+        self.owner, self.repo_name = repo_spec.split("/", 1)
+
+        self.branch = os.environ.get("LOG_BRANCH", "log_event")
+        self.path = os.environ.get("LOG_FILE_PATH", "pr_success.log")
+
+    def _ensure_branch(self, repo):
+        try:
+            repo.get_branch(self.branch)
+        except GithubException as e:
+            if e.status == 404:
+                base = repo.get_branch(repo.default_branch)
+                repo.create_git_ref(ref=f"refs/heads/{self.branch}", sha=base.commit.sha)
+            else:
+                raise
+
+    def append_jsonl(self, jsonl_line: str, commit_message: str = "chore(log): append entry") -> str:
+        repo = self._client.get_repo(f"{self.owner}/{self.repo_name}")
+        self._ensure_branch(repo)
+        try:
+            existing = repo.get_contents(self.path, ref=self.branch)
+            existing_content = base64.b64decode(existing.content).decode("utf-8")
+            new_content = existing_content + jsonl_line
+            repo.update_file(
+                path=self.path,
+                message=commit_message,
+                content=new_content,
+                sha=existing.sha,
+                branch=self.branch,
+            )
+            return "SUCCESS: Log appended"
+        except GithubException as e:
+            if e.status == 404:
+                repo.create_file(
+                    path=self.path,
+                    message=commit_message,
+                    content=jsonl_line,
+                    branch=self.branch,
+                )
+                return "SUCCESS: Log file created and first entry appended"
+            raise

pr_success.log

@@ -1,2 +0,0 @@
-# PR Success Log
-# Format: [timestamp] file_path -> pr_url