build_logic.py

import os
import re
import tempfile
import shutil
import logging
from pathlib import Path

from huggingface_hub import (
    create_repo,
    upload_folder,
    list_repo_files,
    whoami,
    hf_hub_download,
    delete_file as hf_delete_file,
    HfApi
)
from huggingface_hub.hf_api import CommitOperationDelete, CommitOperationAdd, CommitOperation
# Import the general HTTP error from huggingface_hub.utils
from huggingface_hub.utils import HfHubHTTPError # For catching specific HF HTTP errors

# Setup basic logging
logging.basicConfig(
    level=logging.INFO,
    format="%(asctime)s - %(name)s - %(levelname)s - %(message)s"
)
logger = logging.getLogger(__name__)

# --- Helper Function to Get API Token ---
def _get_api_token(ui_token_from_textbox=None):
    env_token = os.getenv('HF_TOKEN')
    if env_token:
        logger.info("Using HF_TOKEN from environment variable.")
        return env_token, None
    if ui_token_from_textbox:
        logger.info("Using HF_TOKEN from UI textbox.")
        return ui_token_from_textbox.strip(), None
    logger.warning("Hugging Face API token not provided in UI or HF_TOKEN env var.")
    return None, "Error: Hugging Face API token not provided in UI or HF_TOKEN env var."

# --- Helper Function to Determine Repo ID ---
def _determine_repo_id(ui_api_token_from_textbox, space_name_ui, owner_ui):
    if not space_name_ui: return None, "Error: Space Name cannot be empty."
    if "/" in space_name_ui: return None, "Error: Space Name should not contain '/'. Use Owner field for the owner part."

    final_owner = owner_ui
    error_message = None

    if not final_owner:
        logger.info("Owner not specified, attempting to auto-detect from token.")
        resolved_api_token, token_err = _get_api_token(ui_api_token_from_textbox)
        if token_err: return None, token_err
        if not resolved_api_token: return None, "Error: API token required for auto owner determination if Owner field is empty."
        try:
            user_info = whoami(token=resolved_api_token)
            if user_info and 'name' in user_info:
                final_owner = user_info['name']
                logger.info(f"Auto-detected owner: {final_owner}")
            else:
                error_message = "Error: Could not retrieve username from token. Check token permissions or specify Owner."
                logger.error(error_message)
        except Exception as e:
            error_message = f"Error retrieving username from token: {str(e)}. Specify Owner or check token."
            logger.exception("Error retrieving username from token:")
        if error_message: return None, error_message

    if not final_owner: return None, "Error: Owner could not be determined. Please specify it in the Owner field."
    repo_id = f"{final_owner}/{space_name_ui}"
    logger.info(f"Determined repo_id: {repo_id}")
    return repo_id, None


# --- Corrected Markdown Parsing ---
# This function remains mostly the same as its purpose is just to parse the *AI's output format*
# into a structured format, not necessarily to represent the *current state* of a Space.
# The app.py logic will use this output and combine it with the current Space state.
def parse_markdown(markdown_input):
    space_info = {"repo_name_md": "", "owner_md": "", "files": []}
    current_file_path = None
    current_file_content_lines = []
    in_file_definition = False
    in_code_block = False
    file_parsing_errors = [] # To collect potential parsing issues

    lines = markdown_input.strip().split("\n")

    # Clean up potential leading '#' added by Gradio's Markdown sometimes
    cleaned_lines = []
    for line_content_orig in lines:
         if line_content_orig.strip().startswith("# "):
             # Only strip leading # if it looks like a Markdown heading related to our format
             if line_content_orig.strip().startswith("# ### File:") or \
                line_content_orig.strip().startswith("# ## File Structure") or \
                line_content_orig.strip().startswith("# # Space:"):
                 cleaned_lines.append(line_content_orig.strip()[2:])
             else:
                 cleaned_lines.append(line_content_orig)
         else:
              cleaned_lines.append(line_content_orig)

    lines = cleaned_lines


    for i, line_content_orig in enumerate(lines):
        line_content_stripped = line_content_orig.strip()
        line_num = i + 1

        # Check for file header
        file_match = re.match(r"### File:\s*(?P<filename_line>[^\n]+)", line_content_stripped)
        if file_match:
            # Before processing a new file, save the content of the previous one
            if current_file_path is not None and in_file_definition: # Check if we were inside a file definition
                # Remove leading/trailing blank lines from content lines
                content_to_save = "\n".join(current_file_content_lines).strip()
                space_info["files"].append({"path": current_file_path, "content": content_to_save})

            filename_line = file_match.group("filename_line").strip()
            current_file_path = filename_line
            # Clean up potential trailing descriptions like "(main application)"
            current_file_path = re.split(r'\s*\(', current_file_path, 1)[0].strip()
            # Clean up potential backticks around the filename
            current_file_path = current_file_path.strip('`\'"').strip() # Add more chars to strip

            if not current_file_path:
                file_parsing_errors.append(f"Line {line_num}: Found '### File:' but filename is empty or invalid.")
                current_file_path = None # Invalidate current file path if parsing failed
                in_file_definition = False # Exit file definition mode until a valid one is found
                continue # Move to next line

            current_file_content_lines = []
            in_file_definition = True
            in_code_block = False # Reset code block flag for the new file
            logger.debug(f"Parsed file header: {current_file_path}")
            continue # Move to next line

        # If not a file header, check for other top-level structures *before* file definitions start
        if not in_file_definition:
            if line_content_stripped.startswith("# Space:"):
                full_space_name_md = line_content_stripped.replace("# Space:", "").strip()
                if "/" in full_space_name_md:
                    parts = full_space_name_md.split("/", 1)
                    if len(parts) == 2:
                         space_info["owner_md"], space_info["repo_name_md"] = parts[0].strip(), parts[1].strip()
                    else:
                         space_info["repo_name_md"] = full_space_name_md # Handle case like "user/repo/"
                else:
                    space_info["repo_name_md"] = full_space_name_md
                logger.debug(f"Parsed space header: {space_info['owner_md']}/{space_info['repo_name_md']}")
                continue
            # Ignore ## File Structure headers and their code blocks, as they are not file content
            if line_content_stripped.startswith("## File Structure"):
                 # Need to consume the following code block if it exists
                 structure_block_start = i + 1
                 while structure_block_start < len(lines) and not lines[structure_block_start].strip().startswith("```"):
                      structure_block_start += 1
                 if structure_block_start < len(lines) and lines[structure_block_start].strip().startswith("```"):
                      # Found opening ```, look for closing ```
                      structure_block_end = structure_block_start + 1
                      while structure_block_end < len(lines) and not lines[structure_block_end].strip().startswith("```"):
                           structure_block_end += 1
                      if structure_block_end < len(lines) and lines[structure_block_end].strip().startswith("```"):
                           # Found closing ```, skip all these lines
                           logger.debug(f"Skipping File Structure block from line {i+1} to {structure_block_end+1}")
                           i = structure_block_end # Adjust loop counter (outer loop will increment)
                           continue
            # Ignore other lines outside a file block definition
            continue

        # If we are inside a file definition block (in_file_definition is True)
        if in_file_definition:
            # Check for code block start/end
            if line_content_stripped.startswith("```"):
                # Toggle code block status
                in_code_block = not in_code_block
                # We consume the ``` line(s), do not add to content
                logger.debug(f"Toggled code block to {in_code_block} at line {line_num}")
                continue # Do not add the ``` line to content

            # If inside a code block, add the line as-is (original content, including leading/trailing whitespace)
            if in_code_block:
                current_file_content_lines.append(line_content_orig)
            # If not inside a code block, check for binary file marker or error messages
            elif line_content_stripped.startswith("[Binary file") or line_content_stripped.startswith("[Error loading content:") or line_content_stripped.startswith("[Binary or Skipped file]"):
                 # Handle binary file markers or error messages as content if not in code block
                 current_file_content_lines.append(line_content_orig)
                 logger.debug(f"Parsed binary/error marker for {current_file_path} at line {line_num}")
            # Any other lines outside code blocks within a file definition are ignored (e.g., descriptions, blank lines)
            # This assumes all code/content *must* be within ``` blocks or be a specific marker line.
            else:
                 # Optionally log ignored lines within a file block if debugging parsing
                 # logger.debug(f"Ignoring line {line_num} within file {current_file_path}: '{line_content_orig}'")
                 pass


    # After the loop, save the content of the last file if we were inside a file definition
    if current_file_path is not None and in_file_definition:
        content_to_save = "\n".join(current_file_content_lines).strip()
        space_info["files"].append({"path": current_file_path, "content": content_to_save})


    # Ensure all file paths are valid and clean up empty files if necessary (based on content parsing)
    # The parsing logic above should handle stripping content, but this is a final check
    space_info["files"] = [f for f in space_info["files"] if f.get("path")] # Ensure path exists

    # Clean up owner/repo names from potential whitespace
    space_info["owner_md"] = space_info["owner_md"].strip()
    space_info["repo_name_md"] = space_info["repo_name_md"].strip()

    if file_parsing_errors:
         logger.warning(f"Markdown parsing encountered errors: {file_parsing_errors}")
         # You might want to return the errors or include them in the space_info dict
         # For now, we just log them.

    logger.info(f"Parsed markdown. Found {len(space_info['files'])} files.")
    return space_info


# --- Function to Get Space SDK and Files ---
def get_space_repository_info(ui_api_token_from_textbox, space_name_ui, owner_ui):
    repo_id_for_error_logging = f"{owner_ui}/{space_name_ui}" if owner_ui else space_name_ui
    sdk = None
    files = []
    error = None
    repo_id = None # Define repo_id here to ensure it's available for error logging after _determine_repo_id

    logger.info(f"Attempting to get repo info for {repo_id_for_error_logging}")

    try:
        resolved_api_token, token_err = _get_api_token(ui_api_token_from_textbox)
        if token_err: return None, [], token_err

        repo_id, err_repo_id = _determine_repo_id(ui_api_token_from_textbox, space_name_ui, owner_ui)
        if err_repo_id: return None, [], err_repo_id
        repo_id_for_error_logging = repo_id # Update logging name

        api = HfApi(token=resolved_api_token)
        # Use repo_info endpoint as it's more robust and gives SDK
        repo_info_obj = api.repo_info(repo_id=repo_id, repo_type="space", timeout=20) # Added timeout, increased slightly
        sdk = repo_info_obj.sdk
        files = [sibling.rfilename for sibling in repo_info_obj.siblings if sibling.rfilename]

        if not files and repo_info_obj.siblings:
            logger.warning(f"Repo {repo_id} has siblings but no rfilenames extracted. Total siblings: {len(repo_info_obj.siblings)}")
            # Sometimes empty repos exist, or listing might fail partially.
            # Continue, files list is just empty.

        logger.info(f"Successfully got repo info for {repo_id}. SDK: {sdk}, Files found: {len(files)}")


    except HfHubHTTPError as e_http: # Catch specific HF HTTP errors first
        logger.error(f"HTTP error getting repo info for {repo_id_for_error_logging or 'unknown repo'}: {e_http}")
        error_message = str(e_http)
        status_code = e_http.response.status_code if e_http.response is not None else None

        if status_code == 404:
            error = f"Space '{repo_id_for_error_logging or 'unknown repo'}' not found (404)."
        elif status_code in (401,403):
            error = f"Access denied for '{repo_id_for_error_logging or 'unknown repo'}' ({status_code}). Check token permissions."
        else:
            error = f"HTTP Error {status_code or 'unknown'} for '{repo_id_for_error_logging or 'unknown repo'}': {error_message}"

    except Exception as e: # Catch other general exceptions
        # If repo_info failed, try listing files as a fallback
        logger.warning(f"Could not get full repo_info for {repo_id_for_error_logging or 'unknown repo'}, attempting list_repo_files fallback: {e}")
        error = f"Error retrieving Space info for `{repo_id_for_error_logging or 'unknown repo'}`: {str(e)}. Attempting file list fallback." # Set a warning message

        try:
            # Re-determine repo_id and get token for fallback
            resolved_api_token_fb, token_err_fb = _get_api_token(ui_api_token_from_textbox)
            if token_err_fb: return None, [], f"{error}\nAPI Token Error during fallback: {token_err_fb}" # Propagate token error
            repo_id_fb, err_repo_id_fb = _determine_repo_id(ui_api_token_from_textbox, space_name_ui, owner_ui)
            if err_repo_id_fb: return None, [], f"{error}\nRepo ID Error during fallback: {err_repo_id_fb}" # Propagate repo ID error

            # Attempt to list files
            files = list_repo_files(repo_id=repo_id_fb, token=resolved_api_token_fb, repo_type="space", timeout=20) # Added timeout
            # If fallback is successful, update error message to a warning about repo_info
            error = f"Warning: Could not fetch full Space info (SDK etc.) for `{repo_id_for_error_logging or 'unknown repo'}`: {str(e)}. File list loaded via fallback."
            logger.info(f"Fallback list_repo_files successful for {repo_id_fb}. Files found: {len(files)}")

        except HfHubHTTPError as e2_http:
            logger.error(f"HTTP error during fallback list_repo_files for {repo_id_for_error_logging or 'unknown repo'}: {e2_http}")
            error_message_fb = str(e2_http)
            status_code_fb = e2_http.response.status_code if e2_http.response is not None else None
            if status_code_fb == 404:
                error = f"Space '{repo_id_for_error_logging or 'unknown repo'}' not found during fallback (404)."
            else:
                error = f"HTTP Error {status_code_fb or 'unknown'} for '{repo_id_for_error_logging or 'unknown repo'}' during fallback: {error_message_fb}"
            files = [] # Ensure files list is empty on fallback error

        except Exception as e2:
            logger.exception(f"Error listing files for {repo_id_for_error_logging or 'unknown repo'} during fallback: {e2}")
            error = f"{error}\nError listing files during fallback for `{repo_id_for_error_logging or 'unknown repo'}`: {str(e2)}"
            files = [] # Ensure files list is empty on fallback error


    # Final check: if files are still empty and there's no specific error, provide a generic "no files" message
    # or if a specific 404 error occurred.
    if not files and not error and (repo_id_for_error_logging is not None):
         error = f"No files found in Space `{repo_id_for_error_logging or 'unknown repo'}`."

    return sdk, files, error


# --- Function to list files ---
def list_space_files_for_browsing(ui_api_token_from_textbox, space_name_ui, owner_ui):
    files, err = get_space_repository_info(ui_api_token_from_textbox, space_name_ui, owner_ui)[1:]
    return files, err


# --- Function to Fetch File Content from Hub ---
def get_space_file_content(ui_api_token_from_textbox, space_name_ui, owner_ui, file_path_in_repo):
    repo_id_for_error_logging = f"{owner_ui}/{space_name_ui}" if owner_ui else space_name_ui
    repo_id = None
    logger.info(f"Attempting to get content for file '{file_path_in_repo}' from {repo_id_for_error_logging}")
    try:
        resolved_api_token, token_err = _get_api_token(ui_api_token_from_textbox)
        if token_err: return None, token_err
        repo_id, err_repo_id = _determine_repo_id(ui_api_token_from_textbox, space_name_ui, owner_ui)
        if err_repo_id: return None, err_repo_id
        repo_id_for_error_logging = repo_id # Update logging name

        if not file_path_in_repo: return None, "Error: File path cannot be empty."
        # Ensure file_path_in_repo uses forward slashes
        file_path_in_repo = file_path_in_repo.replace("\\", "/")

        # Use hf_hub_download first, which caches locally
        downloaded_file_path = hf_hub_download(
            repo_id=repo_id,
            filename=file_path_in_repo,
            repo_type="space",
            token=resolved_api_token,
            local_dir_use_symlinks=False, # Avoid symlinks issues
            cache_dir=None, # Use default cache dir
            timeout=20 # Added timeout
        )
        content = Path(downloaded_file_path).read_text(encoding="utf-8")
        logger.info(f"Successfully downloaded and read content for '{file_path_in_repo}'.")
        return content, None
    except FileNotFoundError:
         logger.error(f"FileNotFoundError for '{file_path_in_repo}' in {repo_id_for_error_logging or 'unknown'}")
         return None, f"Error: File '{file_path_in_repo}' not found in Space '{repo_id_for_error_logging or 'unknown repo'}' (404)." # Often gets translated from 404 by hf_hub_download
    except UnicodeDecodeError:
        # If read_text fails, it's likely binary or non-utf8 text
        logger.warning(f"UnicodeDecodeError for '{file_path_in_repo}'. Likely binary.")
        return None, f"Error: File '{file_path_in_repo}' is not valid UTF-8 text. Cannot display."
    except HfHubHTTPError as e_http:
        logger.error(f"HTTP error fetching file {file_path_in_repo} from {repo_id_for_error_logging or 'unknown repo'}: {e_http}")
        error_message = str(e_http)
        status_code = e_http.response.status_code if e_http.response is not None else None
        if status_code == 404:
            return None, f"Error: File '{file_path_in_repo}' not found in Space '{repo_id_for_error_logging or 'unknown repo'}' (404)."
        if status_code in (401, 403):
             return None, f"Error: Access denied or authentication required for '{repo_id_for_error_logging or 'unknown repo'}' ({status_code}). Check token permissions."
        return None, f"HTTP Error {status_code or 'unknown'} fetching file '{file_path_in_repo}': {error_message}"
    except Exception as e:
        logger.exception(f"Error fetching file content for {file_path_in_repo} from {repo_id_for_error_logging or 'unknown repo'}:")
        return None, f"Error fetching file content: {str(e)}"

# --- Create/Update Space from Staged Changes ---
# This function is modified to take a list of operations (changeset) instead of markdown
# It's designed to be called by handle_confirm_changes
def apply_staged_changes(ui_api_token_from_textbox, owner_ui, space_name_ui, changeset):
    repo_id_for_error_logging = f"{owner_ui}/{space_name_ui}" if owner_ui else space_name_ui
    repo_id = None
    status_messages = []
    operations = [] # List of CommitOperation objects

    logger.info(f"Attempting to apply {len(changeset)} staged changes to {repo_id_for_error_logging}")

    try:
        resolved_api_token, token_err = _get_api_token(ui_api_token_from_textbox)
        if token_err: return [token_err]

        repo_id, err_repo_id = _determine_repo_id(ui_api_token_from_textbox, owner_ui, space_name_ui)
        if err_repo_id: return [err_repo_id]
        repo_id_for_error_logging = repo_id

        api = HfApi(token=resolved_api_token)

        # First, handle Space Creation if it's part of the changeset
        create_space_op = next((c for c in changeset if c['type'] == 'CREATE_SPACE'), None)
        if create_space_op:
            # The repo_id here should match the one determined above, derived from owner_ui/space_name_ui
            # We assume the UI fields owner_ui and space_name_ui are set to the *target* space for creation
            # This is a slight divergence from the AI's CREATE_SPACE command which specifies repo_id
            # We'll prioritize the UI fields as the user's explicit target.
            # The AI command should ideally set the UI fields first.
            # TODO: Refine AI CREATE_SPACE to update UI fields first? Or make build_logic_create_space use the AI's repo_id?
            # Let's adjust `create_space` to take the repo_id directly from the AI command, but validate/use UI token/owner if needed.
            # But the *current* structure assumes owner_ui/space_name_ui are the target.
            # Let's stick to the current structure for now: CREATE_SPACE action is noted, but the target repo is from UI fields.
            # The `create_space` call below *will* create the repo specified by UI fields.
            # The file operations will then be applied to this new repo.

            logger.info(f"Detected CREATE_SPACE action for {create_space_op['repo_id']}. Proceeding with creation of {repo_id_for_error_logging} based on UI fields.")
            # Call the existing create_space logic, but perhaps without markdown?
            # The subsequent file operations will populate it.
            # We need a way to create an *empty* repo first. HfApi().create_repo handles this.
            try:
                 api.create_repo(repo_id=repo_id, repo_type="space", space_sdk=create_space_op.get('sdk', 'gradio'), private=create_space_op.get('private', False), exist_ok=True)
                 status_messages.append(f"CREATE_SPACE: Successfully created or ensured space [{repo_id}](https://huggingface.co/spaces/{repo_id}) exists with SDK '{create_space_op.get('sdk', 'gradio')}' and private={create_space_op.get('private', False)}.")
                 logger.info(f"Successfully created or ensured space {repo_id} exists.")
            except Exception as e:
                 status_messages.append(f"CREATE_SPACE Error: {e}")
                 logger.error(f"Error creating space {repo_id}: {e}")
                 # If space creation fails, subsequent operations will also fail.
                 # Should we stop here? Let's add a check.
                 # We could try to proceed with file operations if `exist_ok=True` succeeded partially.
                 # For simplicity, let's just report the error and continue, hoping upload_folder is resilient.
                 # A better approach might stop if create_repo fails definitively (e.g., 401/403/409).
                 pass # Continue attempting other operations

        # Prepare commit operations for file changes (Add/Update/Delete)
        temp_dir = None
        operations = [] # Reset operations list for the file commit
        paths_to_upload = {} # Map of local temp path -> path in repo for Add/Update

        try:
            temp_dir = tempfile.TemporaryDirectory()
            repo_staging_path = Path(temp_dir.name) / "repo_staging_content"
            repo_staging_path.mkdir(exist_ok=True)

            # Always stage .gitattributes to ensure consistent line endings
            gitattributes_path_local = repo_staging_path / ".gitattributes"
            with open(gitattributes_path_local, "w", encoding="utf-8") as f:
                f.write("* text=auto eol=lf\n")
            paths_to_upload[str(gitattributes_path_local)] = ".gitattributes"


            for change in changeset:
                if change['type'] == 'UPDATE_FILE' or change['type'] == 'CREATE_FILE':
                    file_path_in_repo = change['path'].lstrip('/').replace(os.sep, '/')
                    if not file_path_in_repo:
                         status_messages.append(f"Skipping {change['type']} operation: empty path.")
                         continue

                    content_to_write = change.get('content', '')
                    # Skip files that were marked as binary/error during loading
                    if content_to_write.startswith("[Binary file") or content_to_write.startswith("[Error loading content:") or content_to_write.startswith("[Binary or Skipped file]"):
                         status_messages.append(f"Skipping {change['type']} for '{file_path_in_repo}': Content is a binary/error placeholder.")
                         logger.warning(f"Skipping {change['type']} operation for '{file_path_in_repo}': Content is binary/error placeholder.")
                         continue

                    file_path_local = repo_staging_path / file_path_in_repo
                    file_path_local.parent.mkdir(parents=True, exist_ok=True) # Create parent directories

                    try:
                        with open(file_path_local, "w", encoding="utf-8") as f:
                             f.write(content_to_write)
                        paths_to_upload[str(file_path_local)] = file_path_in_repo
                        logger.info(f"Staged file for {change['type']}: {file_path_in_repo}")
                    except Exception as file_write_error:
                        status_messages.append(f"Error staging file {file_path_in_repo} for {change['type']}: {file_write_error}")
                        logger.error(f"Error writing file {file_path_in_repo} during staging for {change['type']}: {file_write_error}")


                elif change['type'] == 'DELETE_FILE':
                    file_path_in_repo = change['path'].lstrip('/').replace(os.sep, '/')
                    if not file_path_in_repo:
                        status_messages.append(f"Skipping DELETE_FILE operation: empty path.")
                        continue
                    operations.append(CommitOperationDelete(path_in_repo=file_path_in_repo))
                    logger.info(f"Added DELETE_FILE operation for: {file_path_in_repo}")

                # SET_PRIVACY and DELETE_SPACE are handled separately below after the commit

            # Add Add/Update operations from staged files
            for local_path, repo_path in paths_to_upload.items():
                # upload_folder/upload_file is simpler than CommitOperationAdd for content
                # Let's use upload_folder approach if there are files to upload
                pass # Just stage paths, upload_folder handles the ops


            # Perform the combined file commit (uploads and deletes)
            if paths_to_upload or operations: # Check if there's anything to commit
                logger.info(f"Committing file changes to {repo_id_for_error_logging}. Uploads: {len(paths_to_upload)}, Deletes: {len(operations)}")

                # upload_folder automatically handles Add/Update operations based on the local folder state
                # Deletes need to be handled via CommitOperations or a separate delete call.
                # The simplest approach might be:
                # 1. Perform deletes using create_commit with CommitOperationDelete.
                # 2. Perform adds/updates using upload_folder.
                # This requires two commits if there are both types of operations.

                delete_operations = [op for op in operations if isinstance(op, CommitOperationDelete)]
                if delete_operations:
                     try:
                          commit_message_delete = f"AI Space Builder: Deleted {len(delete_operations)} files."
                          logger.info(f"Performing delete commit for {repo_id_for_error_logging}: {commit_message_delete}")
                          api.create_commit(
                             repo_id=repo_id,
                             repo_type="space",
                             operations=delete_operations,
                             commit_message=commit_message_delete
                          )
                          status_messages.append(f"File Deletions: Successfully committed {len(delete_operations)} deletions.")
                          logger.info("Delete commit successful.")
                     except HfHubHTTPError as e_http:
                          status_messages.append(f"File Deletion Error ({e_http.response.status_code if e_http.response else 'N/A'}): {e_http.response.text if e_http.response else str(e_http)}. Check logs.")
                          logger.error(f"HTTP error during delete commit for {repo_id}: {e_http}")
                     except Exception as e_delete_commit:
                          status_messages.append(f"File Deletion Error: {str(e_delete_commit)}. Check logs.")
                          logger.exception(f"Error during delete commit for {repo_id}:")


                if paths_to_upload: # If there are files to upload/update (including .gitattributes)
                     try:
                          commit_message_upload = f"AI Space Builder: Updated Space content for {repo_id}"
                          logger.info(f"Uploading staged files from {str(repo_staging_path)} to {repo_id}...")
                          upload_folder(
                             repo_id=repo_id,
                             folder_path=str(repo_staging_path),
                             path_in_repo=".", # Upload to the root of the repository
                             token=resolved_api_token,
                             repo_type="space",
                             commit_message=commit_message_upload,
                             allow_patterns=["*"], # Ensure all staged files are considered
                             # Use force_patterns to ensure specific files are uploaded even if git thinks they are binary
                             # force_patterns=[f.get("path").replace("\\", "/") for f in space_info["files"] if f.get("path")] # This requires knowing the original markdown files
                             # Let's rely on .gitattributes text=auto instead for now.
                          )
                          status_messages.append(f"File Uploads/Updates: Successfully uploaded/updated {len(paths_to_upload)} files.")
                          logger.info("Upload/Update commit successful.")
                     except HfHubHTTPError as e_http:
                          status_messages.append(f"File Upload/Update Error ({e_http.response.status_code if e_http.response else 'N/A'}): {e_http.response.text if e_http.response else str(e_http)}. Check logs.")
                          logger.error(f"HTTP error during upload_folder for {repo_id}: {e_http}")
                     except Exception as e_upload:
                          status_messages.append(f"File Upload/Update Error: {str(e_upload)}. Check logs.")
                          logger.exception(f"Error during upload_folder for {repo_id}:")

            else:
                status_messages.append("No file changes (create/update/delete) to commit.")
                logger.info("No file changes to commit.")


        finally:
            # Clean up temporary directory
            if temp_dir:
                try:
                    temp_dir.cleanup()
                    logger.info("Cleaned up temporary staging directory.")
                except Exception as e:
                    logger.error(f"Error cleaning up temp dir: {e}")


        # Handle Space Privacy and Delete actions *after* the file commit (if any)
        for change in changeset:
            if change['type'] == 'SET_PRIVACY':
                 try:
                      target_repo_id = change.get('repo_id', repo_id) # Use specified repo_id or current
                      if not target_repo_id:
                           status_messages.append("SET_PRIVACY Error: Target repo_id not specified.")
                           continue
                      api.update_repo_visibility(repo_id=target_repo_id, private=change['private'], repo_type='space')
                      status_messages.append(f"SET_PRIVACY: Successfully set `{target_repo_id}` to `private={change['private']}`.")
                      logger.info(f"Successfully set privacy for {target_repo_id} to {change['private']}.")
                 except HfHubHTTPError as e_http:
                      status_messages.append(f"SET_PRIVACY Error ({e_http.response.status_code if e_http.response else 'N/A'}): {e_http.response.text if e_http.response else str(e_http)}. Check token/permissions.")
                      logger.error(f"HTTP error setting privacy for {target_repo_id}: {e_http}")
                 except Exception as e:
                      status_messages.append(f"SET_PRIVACY Error: {str(e)}. Check logs.")
                      logger.exception(f"Error setting privacy for {target_repo_id}:")

            elif change['type'] == 'DELETE_SPACE':
                 # This is destructive and typically requires the exact owner/space_name
                 delete_owner = change.get('owner') or owner_ui # Use specified owner or current UI owner
                 delete_space = change.get('space_name') or space_name_ui # Use specified space_name or current UI space name
                 delete_repo_id = f"{delete_owner}/{delete_space}" if delete_owner and delete_space else repo_id

                 if not delete_repo_id:
                      status_messages.append("DELETE_SPACE Error: Target repo_id not specified.")
                      continue

                 # Add an extra safeguard: Only delete the *currently loaded* space unless AI specifies otherwise AND it matches the current UI fields?
                 # Or strictly use the repo_id from the action? Let's strictly use the action's specified repo_id,
                 # falling back to UI fields only if the action didn't provide them.
                 # The action format is `DELETE_SPACE` (implies current UI space) or maybe `DELETE_SPACE owner/repo`?
                 # The prompt defined `DELETE_SPACE` only, implying the current space. Let's enforce that.
                 # The change object should have owner/space_name populated by generate_and_stage_changes based on current UI.
                 if delete_repo_id != repo_id:
                      status_messages.append(f"DELETE_SPACE Error: AI requested deletion of '{delete_repo_id}', but this action is only permitted for the currently loaded space '{repo_id}'. Action blocked.")
                      logger.warning(f"Blocked DELETE_SPACE action: requested '{delete_repo_id}', current '{repo_id}'.")
                      continue # Block deletion if not the currently loaded space

                 logger.warning(f"Attempting DESTRUCTIVE DELETE_SPACE action for {delete_repo_id}")
                 try:
                      api.delete_repo(repo_id=delete_repo_id, repo_type='space')
                      status_messages.append(f"DELETE_SPACE: Successfully deleted space `{delete_repo_id}`.")
                      logger.info(f"Successfully deleted space {delete_repo_id}.")
                 except HfHubHTTPError as e_http:
                      status_messages.append(f"DELETE_SPACE Error ({e_http.response.status_code if e_http.response else 'N/A'}): {e_http.response.text if e_http.response else str(e_http)}. Check token/permissions.")
                      logger.error(f"HTTP error deleting space {delete_repo_id}: {e_http}")
                 except Exception as e:
                      status_messages.append(f"DELETE_SPACE Error: {str(e)}. Check logs.")
                      logger.exception(f"Error deleting space {delete_repo_id}:")


    except HfHubHTTPError as e_http:
        logger.error(f"Top-level HTTP error during apply_staged_changes for {repo_id_for_error_logging or 'unknown repo'}: {e_http}")
        status_messages.append(f"API HTTP Error ({e_http.response.status_code if e_http.response else 'N/A'}): {e_http.response.text if e_http.response else str(e_http)}")
    except Exception as e:
        logger.exception(f"Top-level error during apply_staged_changes for {repo_id_for_error_logging or 'unknown repo'}:")
        status_messages.append(f"An unexpected error occurred: {str(e)}")

    # Format the final status message
    final_status = " | ".join(status_messages) if status_messages else "No operations were applied."
    logger.info(f"Finished applying staged changes. Final status: {final_status}")
    return final_status


# --- Delete Single File (Manual UI Trigger) ---
# This function remains for direct UI file deletion, distinct from the AI-driven workflow
def delete_space_file(ui_api_token_from_textbox, space_name_ui, owner_ui, file_path_in_repo, commit_message_ui=None):
    repo_id_for_error_logging = f"{owner_ui}/{space_name_ui}" if owner_ui else space_name_ui
    repo_id = None
    logger.info(f"Attempting manual file deletion for '{file_path_in_repo}' from {repo_id_for_error_logging}")
    try:
        resolved_api_token, token_err = _get_api_token(ui_api_token_from_textbox)
        if token_err: return f"API Token Error: {token_err}"
        repo_id, err_repo_id = _determine_repo_id(ui_api_token_from_textbox, space_name_ui, owner_ui)
        if err_repo_id: return f"Repo ID Error: {err_repo_id}"
        repo_id_for_error_logging = repo_id # Update logging name

        if not file_path_in_repo: return "Error: File path cannot be empty for deletion."
        file_path_in_repo = file_path_in_repo.lstrip('/').replace(os.sep, '/') # Clean path for Hub

        # Prevent deleting essential files like .gitattributes or README.md unless explicitly handled?
        # For now, allow deleting anything selected in the dropdown.

        effective_commit_message = commit_message_ui or f"Deleted file: {file_path_in_repo} via AI Space Editor UI"

        # Use hf_delete_file directly
        hf_delete_file(
            path_in_repo=file_path_in_repo,
            repo_id=repo_id,
            repo_type="space",
            token=resolved_api_token,
            commit_message=effective_commit_message,
            timeout=20 # Added timeout
        )
        logger.info(f"Successfully deleted file: {file_path_in_repo}")
        return f"Successfully deleted file: `{file_path_in_repo}`"

    except FileNotFoundError:
         logger.error(f"FileNotFoundError during manual delete for '{file_path_in_repo}' in {repo_id_for_error_logging or 'unknown'}")
         return f"Error: File '{file_path_in_repo}' not found in Space '{repo_id_for_error_logging or 'unknown repo'}' (404)." # hf_delete_file translates 404
    except HfHubHTTPError as e_http: # Catch specific HF HTTP errors
        logger.error(f"HTTP error deleting file {file_path_in_repo} from {repo_id_for_error_logging or 'unknown repo'}: {e_http}")
        error_message = str(e_http)
        status_code = e_http.response.status_code if e_http.response is not None else None

        if status_code == 404:
             return f"Error: File '{file_path_in_repo}' not found in Space '{repo_id_for_error_logging or 'unknown repo'}' for deletion (404)."
        if status_code in (401, 403):
             return f"Error: Access denied or authentication required for '{repo_id_for_error_logging or 'unknown repo'}' ({status_code}). Check token permissions."
        return f"HTTP Error {status_code or 'unknown'} deleting file '{file_path_in_repo}': {error_message}"
    except Exception as e:
        logger.exception(f"Error deleting file {file_path_in_repo} from {repo_id_for_error_logging or 'unknown repo'}:")
        return f"Error deleting file '{file_path_in_repo}': {str(e)}"

# --- Update Single File (Manual UI Trigger) ---
# This function remains for direct UI file editing, distinct from the AI-driven workflow
def update_space_file(ui_api_token_from_textbox, space_name_ui, owner_ui, file_path_in_repo, file_content, commit_message_ui):
    repo_id_for_error_logging = f"{owner_ui}/{space_name_ui}" if owner_ui else space_name_ui
    repo_id = None
    logger.info(f"Attempting manual file update for '{file_path_in_repo}' in {repo_id_for_error_logging}")
    try:
        resolved_api_token, token_err = _get_api_token(ui_api_token_from_textbox)
        if token_err: return token_err
        repo_id, err_repo_id = _determine_repo_id(ui_api_token_from_textbox, space_name_ui, owner_ui)
        if err_repo_id: return err_repo_id
        repo_id_for_error_logging = repo_id # Update logging name

        if not file_path_in_repo: return "Error: File Path to update cannot be empty."
        file_path_in_repo = file_path_in_repo.lstrip('/').replace(os.sep, '/') # Clean path for Hub
        commit_msg = commit_message_ui or f"Update {file_path_in_repo} via AI Space Editor UI"

        api = HfApi(token=resolved_api_token)

        # Use a temporary file to upload content safely
        tmp_file_path = None
        try:
            with tempfile.NamedTemporaryFile(mode='w', delete=False, encoding='utf-8') as tmp_file_obj:
                tmp_file_obj.write(file_content)
                tmp_file_path = tmp_file_obj.name

            # Upload the temporary file to the specified path in the repo
            api.upload_file(
                path_or_fileobj=tmp_file_path,
                path_in_repo=file_path_in_repo,
                repo_id=repo_id,
                repo_type="space",
                commit_message=commit_msg,
                timeout=20 # Added timeout
            )
            logger.info(f"Successfully updated file: {file_path_in_repo}")
            return f"Successfully updated `{file_path_in_repo}`"
        finally:
            # Ensure the temporary file is removed
            if tmp_file_path and os.path.exists(tmp_file_path):
                 os.remove(tmp_file_path)

    except FileNotFoundError:
         logger.error(f"FileNotFoundError during manual update for '{file_path_in_repo}' in {repo_id_for_error_logging or 'unknown'}")
         return f"Error: Local temporary file not found during upload for '{file_path_in_repo}'."
    except UnicodeDecodeError:
        # If read_text fails, it's likely binary or non-utf8 text
        logger.warning(f"UnicodeDecodeError for '{file_path_in_repo}' during manual update.")
        return f"Error: Content for '{file_path_in_repo}' is not valid UTF-8 text. Cannot edit this way."
    except HfHubHTTPError as e_http:
        logger.error(f"HTTP error in update_space_file for {repo_id_for_error_logging or 'unknown repo'}, file {file_path_in_repo}: {e_http}")
        error_message = str(e_http)
        status_code = e_http.response.status_code if e_http.response is not None else None
        if status_code == 404:
            return f"Error: Space '{repo_id_for_error_logging or 'unknown repo'}' or file '{file_path_in_repo}' not found (404)."
        if status_code in (401, 403):
             return f"Error: Access denied or authentication required for '{repo_id_for_error_logging or 'unknown repo'}' ({status_code}). Check token permissions."
        return f"HTTP Error {status_code or 'unknown'} updating file '{file_path_in_repo}': {error_message}"
    except Exception as e:
        logger.exception(f"Error in update_space_file for {repo_id_for_error_logging or 'unknown repo'}, file {file_path_in_repo}:")
        return f"Error updating file for `{repo_id_for_error_logging or 'unknown repo'}`: {str(e)}"


# --- Get Space Runtime Status ---
def get_space_runtime_status(ui_api_token_from_textbox, space_name_ui, owner_ui):
    repo_id_for_error_logging = f"{owner_ui}/{space_name_ui}" if owner_ui else space_name_ui
    repo_id = None
    logger.info(f"Fetching runtime status for Space: {repo_id_for_error_logging}")
    try:
        resolved_api_token, token_err = _get_api_token(ui_api_token_from_textbox)
        if token_err: return None, f"API Token Error: {token_err}"
        repo_id, err_repo_id = _determine_repo_id(ui_api_token_from_textbox, space_name_ui, owner_ui)
        if err_repo_id: return None, f"Repo ID Error: {err_repo_id}"
        repo_id_for_error_logging = repo_id # Update logging name

        api = HfApi(token=resolved_api_token)

        # Use get_space_runtime which provides details like stage, hardware, etc.
        # Added timeout for robustness
        runtime_info = api.get_space_runtime(repo_id=repo_id, timeout=20)
        logger.info(f"Received runtime info for {repo_id}. Stage: {runtime_info.stage}")


        # Structure the details for display
        status_details = {
            "stage": runtime_info.stage,
            "hardware": runtime_info.hardware,
            "requested_hardware": runtime_info.requested_hardware if hasattr(runtime_info, 'requested_hardware') else None, # requested_hardware might not always be present
            "error_message": None, # Populate this if stage is ERRORED
            "status": runtime_info.status if hasattr(runtime_info, 'status') else None, # Additional status field
            "full_log_link": f"https://huggingface.co/spaces/{repo_id}/logs" if repo_id else "#"
            # We can add more fields from runtime_info.raw if useful
        }

        # Check for specific error states or messages
        if runtime_info.stage == "ERRORED":
            error_content = None
            # Look for error details in various places within the raw data or the error attribute
            if hasattr(runtime_info, 'error') and runtime_info.error: error_content = str(runtime_info.error)
            # Check build/run specific error messages in raw data
            if 'build' in runtime_info.raw and isinstance(runtime_info.raw['build'], dict) and runtime_info.raw['build'].get('status') == 'error':
                 error_content = f"Build Error: {runtime_info.raw['build'].get('message', error_content or 'Unknown build error')}"
            elif 'run' in runtime_info.raw and isinstance(runtime_info.raw['run'], dict) and runtime_info.raw['run'].get('status') == 'error':
                 error_content = f"Runtime Error: {runtime_info.raw['run'].get('message', error_content or 'Unknown runtime error')}"
            elif 'message' in runtime_info.raw and isinstance(runtime_info.raw['message'], str) and ('error' in runtime_info.raw['message'].lower() or runtime_info.raw['message'].strip().endswith('!')): # Basic check for message indicative of error
                 error_content = runtime_info.raw['message'] # Prioritize messages from raw data if they look like errors

            status_details["error_message"] = error_content if error_content else "Space is in an errored state. Check logs for details."

        logger.info(f"Runtime status details for {repo_id}: {status_details}")
        return status_details, None

    except HfHubHTTPError as e_http: # Catch specific HF HTTP errors
        logger.error(f"HTTP error fetching runtime status for {repo_id_for_error_logging or 'unknown repo'}: {e_http}")
        error_message = str(e_http)
        status_code = e_http.response.status_code if e_http.response is not None else None

        if status_code == 404:
            # A 404 could mean the space doesn't exist or doesn't have an active runtime state recorded
            return None, f"Error: Space '{repo_id_for_error_logging or 'unknown repo'}' not found or has no active runtime status (404)."
        if status_code in (401, 403):
             return None, f"Error: Access denied or authentication required for '{repo_id_for_error_logging or 'unknown repo'}' ({status_code}). Check token permissions."
        return None, f"HTTP Error {status_code or 'unknown'} fetching runtime status for '{repo_id_for_error_logging or 'unknown repo'}': {error_message}"

    except Exception as e:
        logger.exception(f"Error fetching runtime status for {repo_id_for_error_logging or 'unknown repo'}:")
        return None, f"Error fetching runtime status: {str(e)}"

# --- Function to set space privacy ---
def build_logic_set_space_privacy(hf_api_key, repo_id, private: bool):
    """Sets the privacy of a Hugging Face Space."""
    logger.info(f"Attempting to set privacy for '{repo_id}' to {private}.")
    try:
        token, err = _get_api_token(hf_api_key)
        if err or not token:
            logger.error(f"Token error setting privacy: {err or 'Token not found'}")
            return f"Error getting token: {err or 'Token not found.'}"
        api = HfApi(token=token)
        api.update_repo_visibility(repo_id=repo_id, private=private, repo_type='space')
        logger.info(f"Successfully set privacy for {repo_id} to {private}.")
        return f"Successfully set privacy for `{repo_id}` to `{private}`."
    except HfHubHTTPError as e_http:
        logger.error(f"HTTP error setting privacy for {repo_id}: {e_http}")
        status_code = e_http.response.status_code if e_http.response else 'N/A'
        return f"HTTP Error ({status_code}) setting privacy for `{repo_id}`: {e_http.response.text if e_http.response else str(e_http)}"
    except Exception as e:
        logger.exception(f"Error setting privacy for {repo_id}:")
        return f"Error setting privacy for `{repo_id}`: {e}"

# --- Function to delete an entire space ---
def build_logic_delete_space(hf_api_key, owner, space_name):
    """Deletes an entire Hugging Face Space."""
    repo_id = f"{owner}/{space_name}"
    logger.warning(f"Attempting DESTRUCTIVE DELETE_SPACE action for '{repo_id}'.")
    try:
        token, err = _get_api_token(hf_api_key)
        if err or not token:
            logger.error(f"Token error deleting space: {err or 'Token not found'}")
            return f"Error getting token: {err or 'Token not found.'}"
        api = HfApi(token=token)
        api.delete_repo(repo_id=repo_id, repo_type='space')
        logger.warning(f"Successfully deleted space {repo_id}.")
        return f"Successfully deleted space `{repo_id}`."
    except HfHubHTTPError as e_http:
        logger.error(f"HTTP error deleting space {repo_id}: {e_http}")
        status_code = e_http.response.status_code if e_http.response else 'N/A'
        return f"HTTP Error ({status_code}) deleting space `{repo_id}`: {e_http.response.text if e_http.response else str(e_http)}"
    except Exception as e:
        logger.exception(f"Error deleting space {repo_id}:")
        return f"Error deleting space `{repo_id}`: {e}"