custom_nodes/ComfyUI-Manager/scanner.py

import ast
import re
import os
import json
import threading
from collections import defaultdict
from git import Repo
import concurrent
import datetime
import concurrent.futures
import requests
import warnings
import argparse

builtin_nodes = set()

import sys

from urllib.parse import urlparse
from urllib3.util.retry import Retry
from github import Github, Auth
from pathlib import Path
from typing import Set, Dict, Optional

# Scanner version for cache invalidation
SCANNER_VERSION = "2.0.13"  # Add fallback for dynamic v3 node_id

# Cache for extract_nodes and extract_nodes_enhanced results
_extract_nodes_cache: Dict[str, Set[str]] = {}
_extract_nodes_enhanced_cache: Dict[str, Set[str]] = {}
_file_mtime_cache: Dict[Path, float] = {}


def _get_repo_root(file_path: Path) -> Optional[Path]:
    """Find the repository root directory containing .git"""
    current = file_path if file_path.is_dir() else file_path.parent
    while current != current.parent:
        if (current / ".git").exists():
            return current
        current = current.parent
    return None


def _get_repo_hash(repo_path: Path) -> str:
    """Get git commit hash or fallback identifier"""
    git_dir = repo_path / ".git"
    if not git_dir.exists():
        return ""

    try:
        # Read HEAD to get current commit
        head_file = git_dir / "HEAD"
        if head_file.exists():
            head_content = head_file.read_text().strip()
            if head_content.startswith("ref:"):
                # HEAD points to a ref
                ref_path = git_dir / head_content[5:].strip()
                if ref_path.exists():
                    commit_hash = ref_path.read_text().strip()
                    return commit_hash[:16]  # First 16 chars
            else:
                # Detached HEAD
                return head_content[:16]
    except:
        pass

    return ""


def _load_per_repo_cache(repo_path: Path) -> Optional[tuple]:
    """Load nodes and metadata from per-repo cache

    Returns:
        tuple: (nodes_set, metadata_dict) or None if cache invalid
    """
    cache_file = repo_path / ".git" / "nodecache.json"

    if not cache_file.exists():
        return None

    try:
        with open(cache_file, 'r') as f:
            cache_data = json.load(f)

        # Verify scanner version
        if cache_data.get('scanner_version') != SCANNER_VERSION:
            return None

        # Verify git hash
        current_hash = _get_repo_hash(repo_path)
        if cache_data.get('git_hash') != current_hash:
            return None

        # Return nodes and metadata
        nodes = cache_data.get('nodes', [])
        metadata = cache_data.get('metadata', {})
        return (set(nodes) if nodes else set(), metadata)

    except:
        return None


def _save_per_repo_cache(repo_path: Path, all_nodes: Set[str], metadata: dict = None):
    """Save nodes and metadata to per-repo cache"""
    cache_file = repo_path / ".git" / "nodecache.json"

    if not cache_file.parent.exists():
        return

    git_hash = _get_repo_hash(repo_path)
    cache_data = {
        "scanner_version": SCANNER_VERSION,
        "git_hash": git_hash,
        "scanned_at": datetime.datetime.now().isoformat(),
        "nodes": sorted(list(all_nodes)),
        "metadata": metadata if metadata else {}
    }

    try:
        with open(cache_file, 'w') as f:
            json.dump(cache_data, f, indent=2)
    except:
        pass  # Silently fail - cache is optional


def download_url(url, dest_folder, filename=None):
    # Ensure the destination folder exists
    if not os.path.exists(dest_folder):
        os.makedirs(dest_folder)

    # Extract filename from URL if not provided
    if filename is None:
        filename = os.path.basename(url)

    # Full path to save the file
    dest_path = os.path.join(dest_folder, filename)

    # Download the file
    response = requests.get(url, stream=True)
    if response.status_code == 200:
        with open(dest_path, 'wb') as file:
            for chunk in response.iter_content(chunk_size=1024):
                if chunk:
                    file.write(chunk)
    else:
        raise Exception(f"Failed to download file from {url}")


def parse_arguments():
    """Parse command-line arguments"""
    parser = argparse.ArgumentParser(
        description='ComfyUI Manager Node Scanner',
        formatter_class=argparse.RawDescriptionHelpFormatter,
        epilog='''
Examples:
  # Standard mode
  python3 scanner.py
  python3 scanner.py --skip-update
  python3 scanner.py --skip-all --force-rescan

  # Scan-only mode
  python3 scanner.py --scan-only temp-urls-clean.list
  python3 scanner.py --scan-only urls.list --temp-dir /custom/temp
  python3 scanner.py --scan-only urls.list --skip-update --force-rescan
        '''
    )

    parser.add_argument('--scan-only', type=str, metavar='URL_LIST_FILE',
                       help='Scan-only mode: provide URL list file (one URL per line)')
    parser.add_argument('--temp-dir', type=str, metavar='DIR',
                       help='Temporary directory for cloned repositories')
    parser.add_argument('--skip-update', action='store_true',
                       help='Skip git clone/pull operations')
    parser.add_argument('--skip-stat-update', action='store_true',
                       help='Skip GitHub stats collection')
    parser.add_argument('--skip-all', action='store_true',
                       help='Skip all update operations')
    parser.add_argument('--force-rescan', action='store_true',
                       help='Force rescan all nodes (ignore cache)')

    # Backward compatibility: positional argument for temp_dir
    parser.add_argument('temp_dir_positional', nargs='?', metavar='TEMP_DIR',
                       help='(Legacy) Temporary directory path')

    args = parser.parse_args()
    return args


# Module-level variables (will be set in main if running as script)
args = None
scan_only_mode = False
url_list_file = None
temp_dir = None
skip_update = False
skip_stat_update = True
g = None


parse_cnt = 0

# Thread-safe git error state
_git_error_lock = threading.Lock()
_git_errors: defaultdict = defaultdict(list)  # category -> list[{'repo': str, 'op': str, 'msg': str}]

# Ordered categories: (key, display label, compiled regex). First match wins.
# Single source of truth — add new categories here only.
_GIT_ERROR_CATEGORIES = [
    ('repository_not_found', 'Repository Not Found', re.compile(
        r'repository\s+not\s+found|does\s+not\s+exist|\b404\b|remote:\s*repository\s+not\s+found',
        re.IGNORECASE
    )),
    ('divergent_branch', 'Divergent Branch', re.compile(
        r'divergent\s+branches|need\s+to\s+specify\s+how\s+to\s+reconcile\s+divergent\s+branches',
        re.IGNORECASE
    )),
    ('auth_failed', 'Authentication Failed', re.compile(
        r'authentication\s+failed|could\s+not\s+read\s+username|invalid\s+username|invalid\s+password|auth\s+failed',
        re.IGNORECASE
    )),
    ('network_error', 'Network Error', re.compile(
        r'could\s+not\s+resolve\s+host|connection\s+refused|timed?\s*out|failed\s+to\s+connect|'
        r'network\s+is\s+unreachable|temporary\s+failure\s+in\s+name\s+resolution',
        re.IGNORECASE
    )),
    ('merge_conflict', 'Merge Conflict', re.compile(
        r'merge\s+conflict|\bCONFLICT\b|automatic\s+merge\s+failed',
        re.IGNORECASE
    )),
    ('permission_denied', 'Permission Denied', re.compile(
        r'permission\s+denied|access\s+denied|operation\s+not\s+permitted|publickey',
        re.IGNORECASE
    )),
]


def _categorize_git_error(error_str: str) -> str:
    """Classify a git error string into a category. First match wins."""
    for category, _label, pattern in _GIT_ERROR_CATEGORIES:
        if pattern.search(error_str):
            return category
    return 'other'


def _record_git_error(repo_name: str, op: str, error: Exception) -> None:
    """Record a git error in the thread-safe collector."""
    category = _categorize_git_error(str(error))
    with _git_error_lock:
        _git_errors[category].append({'repo': repo_name, 'op': op, 'msg': str(error)})


def _report_git_errors() -> None:
    """Print a grouped summary of git errors by category."""
    if not _git_errors:
        return

    total = sum(len(v) for v in _git_errors.values())
    print(f"\n{'='*60}")
    print(f"Git Operation Errors Summary: {total} failure(s)")
    print(f"{'='*60}")

    for category, label, _pattern in _GIT_ERROR_CATEGORIES:
        entries = _git_errors.get(category, [])
        if not entries:
            continue
        print(f"\n[{label}] ({len(entries)} repo(s))")
        for entry in entries:
            print(f"  • {entry['repo']} ({entry['op']}): {entry['msg']}")

    other_entries = _git_errors.get('other', [])
    if other_entries:
        print(f"\n[Other] ({len(other_entries)} repo(s))")
        for entry in other_entries:
            print(f"  • {entry['repo']} ({entry['op']}): {entry['msg']}")

    print(f"{'='*60}\n")


def extract_nodes(code_text):
    global parse_cnt

    # Check cache first
    cache_key = hash(code_text)
    if cache_key in _extract_nodes_cache:
        return _extract_nodes_cache[cache_key].copy()

    try:
        if parse_cnt % 100 == 0:
            print(".", end="", flush=True)
        parse_cnt += 1

        code_text = re.sub(r'\\[^"\']', '', code_text)
        with warnings.catch_warnings():
            warnings.filterwarnings('ignore', category=SyntaxWarning)
            warnings.filterwarnings('ignore', category=DeprecationWarning)
            parsed_code = ast.parse(code_text)

        # Support both ast.Assign and ast.AnnAssign (for type-annotated assignments)
        assignments = (node for node in parsed_code.body if isinstance(node, (ast.Assign, ast.AnnAssign)))

        for assignment in assignments:
            # Handle ast.AnnAssign (e.g., NODE_CLASS_MAPPINGS: Type = {...})
            if isinstance(assignment, ast.AnnAssign):
                if isinstance(assignment.target, ast.Name) and assignment.target.id in ['NODE_CONFIG', 'NODE_CLASS_MAPPINGS']:
                    node_class_mappings = assignment.value
                    break
            # Handle ast.Assign (e.g., NODE_CLASS_MAPPINGS = {...})
            elif isinstance(assignment.targets[0], ast.Name) and assignment.targets[0].id in ['NODE_CONFIG', 'NODE_CLASS_MAPPINGS']:
                node_class_mappings = assignment.value
                break
        else:
            node_class_mappings = None

        if node_class_mappings:
            s = set()

            for key in node_class_mappings.keys:
                    if key is not None and isinstance(key.value, str):
                        s.add(key.value.strip())

            # Cache the result
            _extract_nodes_cache[cache_key] = s
            return s
        else:
            # Cache empty result
            _extract_nodes_cache[cache_key] = set()
            return set()
    except:
        # Cache empty result on error
        _extract_nodes_cache[cache_key] = set()
        return set()

def extract_nodes_from_repo(repo_path: Path, verbose: bool = False, force_rescan: bool = False) -> tuple:
    """
    Extract all nodes and metadata from a repository with per-repo caching.

    Automatically caches results in .git/nodecache.json.
    Cache is invalidated when:
    - Git commit hash changes
    - Scanner version changes
    - force_rescan flag is True

    Args:
        repo_path: Path to repository root
        verbose: If True, print UI-only extension detection messages
        force_rescan: If True, ignore cache and force fresh scan

    Returns:
        tuple: (nodes_set, metadata_dict)
    """
    # Ensure path is absolute
    repo_path = repo_path.resolve()

    # Check per-repo cache first (unless force_rescan is True)
    if not force_rescan:
        cached_result = _load_per_repo_cache(repo_path)
        if cached_result is not None:
            return cached_result

    # Cache miss - scan all .py files
    all_nodes = set()
    all_metadata = {}
    py_files = list(repo_path.rglob("*.py"))

    # Filter out __pycache__, .git, and other hidden directories
    filtered_files = []
    for f in py_files:
        try:
            rel_path = f.relative_to(repo_path)
            # Skip __pycache__, .git, and any directory starting with .
            if '__pycache__' not in str(rel_path) and not any(part.startswith('.') for part in rel_path.parts):
                filtered_files.append(f)
        except:
            continue
    py_files = filtered_files

    for py_file in py_files:
        try:
            # Read file with proper encoding
            with open(py_file, 'r', encoding='utf-8', errors='ignore') as f:
                code = f.read()

            if code:
                # Extract nodes using SAME logic as scan_in_file
                # V1 nodes (enhanced with fallback patterns)
                nodes = extract_nodes_enhanced(code, py_file, visited=set(), verbose=verbose)
                all_nodes.update(nodes)

                # V3 nodes detection
                v3_nodes = extract_v3_nodes(code)
                all_nodes.update(v3_nodes)

                # Dict parsing - exclude commented NODE_CLASS_MAPPINGS lines
                pattern = r"_CLASS_MAPPINGS\s*(?::\s*\w+\s*)?=\s*(?:\\\s*)?{([^}]*)}"
                regex = re.compile(pattern, re.MULTILINE | re.DOTALL)

                for match_obj in regex.finditer(code):
                    # Get the line where NODE_CLASS_MAPPINGS is defined
                    match_start = match_obj.start()
                    line_start = code.rfind('\n', 0, match_start) + 1
                    line_end = code.find('\n', match_start)
                    if line_end == -1:
                        line_end = len(code)
                    line = code[line_start:line_end]

                    # Skip if line starts with # (commented)
                    if re.match(r'^\s*#', line):
                        continue

                    match = match_obj.group(1)

                    # Filter out commented lines from dict content
                    match_lines = match.split('\n')
                    match_filtered = '\n'.join(
                        line for line in match_lines
                        if not re.match(r'^\s*#', line)
                    )

                    # Extract key-value pairs with double quotes
                    key_value_pairs = re.findall(r"\"([^\"]*)\"\s*:\s*([^,\n]*)", match_filtered)
                    for key, value in key_value_pairs:
                        all_nodes.add(key.strip())

                    # Extract key-value pairs with single quotes
                    key_value_pairs = re.findall(r"'([^']*)'\s*:\s*([^,\n]*)", match_filtered)
                    for key, value in key_value_pairs:
                        all_nodes.add(key.strip())

                # Handle .update() pattern (AFTER comment removal)
                code_cleaned = re.sub(r'^#.*?$', '', code, flags=re.MULTILINE)

                update_pattern = r"_CLASS_MAPPINGS\.update\s*\(\s*{([^}]*)}\s*\)"
                update_match = re.search(update_pattern, code_cleaned, re.DOTALL)
                if update_match:
                    update_dict_text = update_match.group(1)
                    # Extract key-value pairs (double quotes)
                    update_pairs = re.findall(r'"([^"]*)"\s*:\s*([^,\n]*)', update_dict_text)
                    for key, value in update_pairs:
                        all_nodes.add(key.strip())
                    # Extract key-value pairs (single quotes)
                    update_pairs_single = re.findall(r"'([^']*)'\s*:\s*([^,\n]*)", update_dict_text)
                    for key, value in update_pairs_single:
                        all_nodes.add(key.strip())

                # Additional regex patterns (AFTER comment removal)
                patterns = [
                    r'^[^=]*_CLASS_MAPPINGS\["(.*?)"\]',
                    r'^[^=]*_CLASS_MAPPINGS\[\'(.*?)\'\]',
                    r'@register_node\("(.+)",\s*\".+"\)',
                    r'"(\w+)"\s*:\s*{"class":\s*\w+\s*'
                ]

                for pattern in patterns:
                    keys = re.findall(pattern, code_cleaned)
                    all_nodes.update(key.strip() for key in keys)

                # Extract metadata from this file
                metadata = extract_metadata_only(str(py_file))
                all_metadata.update(metadata)
        except Exception:
            # Silently skip files that can't be read
            continue

    # Save to per-repo cache
    _save_per_repo_cache(repo_path, all_nodes, all_metadata)

    return (all_nodes, all_metadata)


def _verify_class_exists(node_name: str, code_text: str, file_path: Optional[Path] = None) -> tuple[bool, Optional[str], Optional[int]]:
    """
    Verify that a node class exists and has ComfyUI node structure.

    Returns: (exists: bool, file_path: str, line_number: int)

    A valid ComfyUI node must have:
    - Class definition (not commented)
    - At least one of: INPUT_TYPES, RETURN_TYPES, FUNCTION method/attribute
    """
    try:
        with warnings.catch_warnings():
            warnings.filterwarnings('ignore', category=SyntaxWarning)
            tree = ast.parse(code_text)
    except:
        return (False, None, None)

    for node in ast.walk(tree):
        if isinstance(node, ast.ClassDef):
            if node.name == node_name or node.name.replace('_', '') == node_name.replace('_', ''):
                # Found class definition - check if it has ComfyUI interface
                has_input_types = False
                has_return_types = False
                has_function = False

                for item in node.body:
                    # Check for INPUT_TYPES method
                    if isinstance(item, ast.FunctionDef) and item.name == 'INPUT_TYPES':
                        has_input_types = True
                    # Check for RETURN_TYPES attribute
                    elif isinstance(item, ast.Assign):
                        for target in item.targets:
                            if isinstance(target, ast.Name):
                                if target.id == 'RETURN_TYPES':
                                    has_return_types = True
                                elif target.id == 'FUNCTION':
                                    has_function = True
                    # Check for FUNCTION method
                    elif isinstance(item, ast.FunctionDef):
                        has_function = True

                # Valid if has any ComfyUI signature
                if has_input_types or has_return_types or has_function:
                    file_str = str(file_path) if file_path else None
                    return (True, file_str, node.lineno)

    return (False, None, None)


def _extract_display_name_mappings(code_text: str) -> Set[str]:
    """
    Extract node names from NODE_DISPLAY_NAME_MAPPINGS.

    Pattern:
        NODE_DISPLAY_NAME_MAPPINGS = {
            "node_key": "Display Name",
            ...
        }

    Returns:
        Set of node keys from NODE_DISPLAY_NAME_MAPPINGS
    """
    try:
        with warnings.catch_warnings():
            warnings.filterwarnings('ignore', category=SyntaxWarning)
            tree = ast.parse(code_text)
    except:
        return set()

    nodes = set()

    for node in tree.body:
        if isinstance(node, ast.Assign):
            for target in node.targets:
                if isinstance(target, ast.Name) and target.id == 'NODE_DISPLAY_NAME_MAPPINGS':
                    if isinstance(node.value, ast.Dict):
                        for key in node.value.keys:
                            if isinstance(key, ast.Constant) and isinstance(key.value, str):
                                nodes.add(key.value.strip())

    return nodes


def extract_nodes_enhanced(
    code_text: str,
    file_path: Optional[Path] = None,
    visited: Optional[Set[Path]] = None,
    verbose: bool = False
) -> Set[str]:
    """
    Enhanced node extraction with multi-layer detection system.

    Scanner 2.0.11 - Comprehensive detection strategy:
    - Phase 1: NODE_CLASS_MAPPINGS dict literal
    - Phase 2: Class.NAME attribute access (e.g., FreeChat.NAME)
    - Phase 3: Item assignment (NODE_CLASS_MAPPINGS["key"] = value)
    - Phase 4: Class existence verification (detects active classes even if registration commented)
    - Phase 5: NODE_DISPLAY_NAME_MAPPINGS cross-reference
    - Phase 6: Empty dict detection (UI-only extensions, logging only)

    Fixed Bugs:
    - Scanner 2.0.9: Fallback cascade prevented Phase 3 execution
    - Scanner 2.0.10: Missed active classes with commented registrations (15 false negatives)

    Args:
        code_text: Python source code
        file_path: Path to file (for logging and caching)
        visited: Visited paths (for circular import prevention)
        verbose: If True, print UI-only extension detection messages

    Returns:
        Set of node names (union of all detected patterns)
    """
    # Check file-based cache if file_path provided
    if file_path is not None:
        try:
            file_path_obj = Path(file_path) if not isinstance(file_path, Path) else file_path
            if file_path_obj.exists():
                current_mtime = file_path_obj.stat().st_mtime

                # Check if we have cached result with matching mtime and scanner version
                if file_path_obj in _file_mtime_cache:
                    cached_mtime = _file_mtime_cache[file_path_obj]
                    cache_key = (str(file_path_obj), cached_mtime, SCANNER_VERSION)

                    if current_mtime == cached_mtime and cache_key in _extract_nodes_enhanced_cache:
                        return _extract_nodes_enhanced_cache[cache_key].copy()
        except:
            pass  # Ignore cache errors, proceed with normal execution

    # Suppress warnings from AST parsing
    with warnings.catch_warnings():
        warnings.filterwarnings('ignore', category=SyntaxWarning)
        warnings.filterwarnings('ignore', category=DeprecationWarning)

        # Phase 1: Original extract_nodes() - dict literal
        phase1_nodes = extract_nodes(code_text)

    # Phase 2: Class.NAME pattern
    if visited is None:
        visited = set()
    phase2_nodes = _fallback_classname_resolver(code_text, file_path)

    # Phase 3: Item assignment pattern
    phase3_nodes = _fallback_item_assignment(code_text)

    # Phase 4: NODE_DISPLAY_NAME_MAPPINGS cross-reference (NEW in 2.0.11)
    # This catches nodes that are in display names but not in NODE_CLASS_MAPPINGS
    phase4_nodes = _extract_display_name_mappings(code_text)

    # Phase 5: Class existence verification ONLY for display name candidates (NEW in 2.0.11)
    # This phase is CONSERVATIVE - only verify classes that appear in display names
    # This catches the specific Scanner 2.0.10 bug pattern:
    #   - NODE_CLASS_MAPPINGS registration is commented
    #   - NODE_DISPLAY_NAME_MAPPINGS still has the entry
    #   - Class implementation exists
    # Example: Bjornulf_ollamaLoader in Bjornulf_custom_nodes
    phase5_nodes = set()
    for node_name in phase4_nodes:
        # Only check classes that appear in display names but not in registrations
        if node_name not in (phase1_nodes | phase2_nodes | phase3_nodes):
            exists, _, _ = _verify_class_exists(node_name, code_text, file_path)
            if exists:
                phase5_nodes.add(node_name)

    # Phase 6: Dict comprehension pattern (NEW in 2.0.12)
    # Detects: NODE_CLASS_MAPPINGS = {cls.__name__: cls for cls in to_export}
    # Example: TobiasGlaubach/ComfyUI-TG_PyCode
    phase6_nodes = _fallback_dict_comprehension(code_text, file_path)

    # Phase 7: Import-based class names for dict comprehension (NEW in 2.0.12)
    # Detects imported classes that are added to export lists
    phase7_nodes = _fallback_import_class_names(code_text, file_path)

    # Union all results (FIX: Scanner 2.0.9 bug + Scanner 2.0.10 bug + Scanner 2.0.12 dict comp)
    # 2.0.9: Used early return which missed Phase 3 nodes
    # 2.0.10: Only checked registrations, missed classes referenced in display names
    # 2.0.12: Added dict comprehension and import-based class detection
    all_nodes = phase1_nodes | phase2_nodes | phase3_nodes | phase4_nodes | phase5_nodes | phase6_nodes | phase7_nodes

    # Phase 8: Empty dict detector (logging only, doesn't add nodes)
    if not all_nodes:
        _fallback_empty_dict_detector(code_text, file_path, verbose)

    # Cache the result
    if file_path is not None:
        try:
            file_path_obj = Path(file_path) if not isinstance(file_path, Path) else file_path
            if file_path_obj.exists():
                current_mtime = file_path_obj.stat().st_mtime
                cache_key = (str(file_path_obj), current_mtime, SCANNER_VERSION)
                _extract_nodes_enhanced_cache[cache_key] = all_nodes
                _file_mtime_cache[file_path_obj] = current_mtime
        except:
            pass

    return all_nodes


def _fallback_classname_resolver(code_text: str, file_path: Optional[Path]) -> Set[str]:
    """
    Detect Class.NAME pattern in NODE_CLASS_MAPPINGS.

    Pattern:
        NODE_CLASS_MAPPINGS = {
            FreeChat.NAME: FreeChat,
            PaidChat.NAME: PaidChat
        }
    """
    try:
        with warnings.catch_warnings():
            warnings.filterwarnings('ignore', category=SyntaxWarning)
            parsed = ast.parse(code_text)
    except:
        return set()
    
    nodes = set()
    
    for node in parsed.body:
        if isinstance(node, ast.Assign):
            for target in node.targets:
                if isinstance(target, ast.Name) and target.id == 'NODE_CLASS_MAPPINGS':
                    if isinstance(node.value, ast.Dict):
                        for key in node.value.keys:
                            # Detect Class.NAME pattern
                            if isinstance(key, ast.Attribute):
                                if isinstance(key.value, ast.Name):
                                    # Use class name as node name
                                    nodes.add(key.value.id)
                            # Also handle literal strings
                            elif isinstance(key, ast.Constant) and isinstance(key.value, str):
                                nodes.add(key.value.strip())
    
    return nodes


def _fallback_item_assignment(code_text: str) -> Set[str]:
    """
    Detect item assignment pattern.

    Pattern:
        NODE_CLASS_MAPPINGS = {}
        NODE_CLASS_MAPPINGS["MyNode"] = MyNode
    """
    try:
        with warnings.catch_warnings():
            warnings.filterwarnings('ignore', category=SyntaxWarning)
            parsed = ast.parse(code_text)
    except:
        return set()

    nodes = set()

    for node in ast.walk(parsed):
        if isinstance(node, ast.Assign):
            for target in node.targets:
                if isinstance(target, ast.Subscript):
                    if (isinstance(target.value, ast.Name) and
                        target.value.id in ['NODE_CLASS_MAPPINGS', 'NODE_CONFIG']):
                        # Extract key
                        if isinstance(target.slice, ast.Constant):
                            if isinstance(target.slice.value, str):
                                nodes.add(target.slice.value)

    return nodes


def _fallback_dict_comprehension(code_text: str, file_path: Optional[Path] = None) -> Set[str]:
    """
    Detect dict comprehension pattern with __name__ attribute access.

    Pattern:
        NODE_CLASS_MAPPINGS = {cls.__name__: cls for cls in to_export}
        NODE_CLASS_MAPPINGS = {c.__name__: c for c in [ClassA, ClassB]}

    This function detects dict comprehension assignments to NODE_CLASS_MAPPINGS
    and extracts class names from the iterable (list literal or variable reference).

    Returns:
        Set of class names extracted from the dict comprehension
    """
    try:
        with warnings.catch_warnings():
            warnings.filterwarnings('ignore', category=SyntaxWarning)
            parsed = ast.parse(code_text)
    except:
        return set()

    nodes = set()
    export_lists = {}  # Track list variables and their contents

    # First pass: collect list assignments (to_export = [...], exports = [...])
    for node in ast.walk(parsed):
        if isinstance(node, ast.Assign):
            for target in node.targets:
                if isinstance(target, ast.Name):
                    var_name = target.id
                    # Check for list literal
                    if isinstance(node.value, ast.List):
                        class_names = set()
                        for elt in node.value.elts:
                            if isinstance(elt, ast.Name):
                                class_names.add(elt.id)
                        export_lists[var_name] = class_names

        # Handle augmented assignment: to_export += [...]
        elif isinstance(node, ast.AugAssign):
            if isinstance(node.target, ast.Name) and isinstance(node.op, ast.Add):
                var_name = node.target.id
                if isinstance(node.value, ast.List):
                    class_names = set()
                    for elt in node.value.elts:
                        if isinstance(elt, ast.Name):
                            class_names.add(elt.id)
                    if var_name in export_lists:
                        export_lists[var_name].update(class_names)
                    else:
                        export_lists[var_name] = class_names

    # Second pass: find NODE_CLASS_MAPPINGS dict comprehension
    for node in ast.walk(parsed):
        if isinstance(node, ast.Assign):
            for target in node.targets:
                if isinstance(target, ast.Name) and target.id in ['NODE_CLASS_MAPPINGS', 'NODE_CONFIG']:
                    # Check for dict comprehension
                    if isinstance(node.value, ast.DictComp):
                        dictcomp = node.value

                        # Check if key is cls.__name__ pattern
                        key = dictcomp.key
                        if isinstance(key, ast.Attribute) and key.attr == '__name__':
                            # Get the iterable from the first generator
                            for generator in dictcomp.generators:
                                iter_node = generator.iter

                                # Case 1: Inline list [ClassA, ClassB, ...]
                                if isinstance(iter_node, ast.List):
                                    for elt in iter_node.elts:
                                        if isinstance(elt, ast.Name):
                                            nodes.add(elt.id)

                                # Case 2: Variable reference (to_export, exports, etc.)
                                elif isinstance(iter_node, ast.Name):
                                    var_name = iter_node.id
                                    if var_name in export_lists:
                                        nodes.update(export_lists[var_name])

    return nodes


def _fallback_import_class_names(code_text: str, file_path: Optional[Path] = None) -> Set[str]:
    """
    Extract class names from imports that are added to export lists.

    Pattern:
        from .module import ClassA, ClassB
        to_export = [ClassA, ClassB]
        NODE_CLASS_MAPPINGS = {cls.__name__: cls for cls in to_export}

    This is a complementary fallback that works with _fallback_dict_comprehension
    to resolve import-based node registrations.

    Returns:
        Set of imported class names that appear in export-like contexts
    """
    try:
        with warnings.catch_warnings():
            warnings.filterwarnings('ignore', category=SyntaxWarning)
            parsed = ast.parse(code_text)
    except:
        return set()

    # Collect imported names
    imported_names = set()
    for node in ast.walk(parsed):
        if isinstance(node, ast.ImportFrom):
            for alias in node.names:
                name = alias.asname if alias.asname else alias.name
                imported_names.add(name)

    # Check if these names appear in list assignments that feed into NODE_CLASS_MAPPINGS
    export_candidates = set()
    has_dict_comp_mapping = False

    for node in ast.walk(parsed):
        # Check for dict comprehension NODE_CLASS_MAPPINGS
        if isinstance(node, ast.Assign):
            for target in node.targets:
                if isinstance(target, ast.Name) and target.id == 'NODE_CLASS_MAPPINGS':
                    if isinstance(node.value, ast.DictComp):
                        has_dict_comp_mapping = True

        # Collect list contents
        if isinstance(node, ast.Assign):
            if isinstance(node.value, ast.List):
                for elt in node.value.elts:
                    if isinstance(elt, ast.Name) and elt.id in imported_names:
                        export_candidates.add(elt.id)

        # Handle augmented assignment
        elif isinstance(node, ast.AugAssign):
            if isinstance(node.value, ast.List):
                for elt in node.value.elts:
                    if isinstance(elt, ast.Name) and elt.id in imported_names:
                        export_candidates.add(elt.id)

    # Only return if there's a dict comprehension mapping
    if has_dict_comp_mapping:
        return export_candidates

    return set()


def _extract_repo_name(file_path: Path) -> str:
    """
    Extract repository name from file path.

    Path structure: /home/rho/.tmp/analysis/temp/{author}_{reponame}/{path/to/file.py}
    Returns: {author}_{reponame} or filename if extraction fails
    """
    try:
        parts = file_path.parts
        # Find 'temp' directory in path
        if 'temp' in parts:
            temp_idx = parts.index('temp')
            if temp_idx + 1 < len(parts):
                # Next part after 'temp' is the repo directory
                return parts[temp_idx + 1]
    except (ValueError, IndexError):
        pass

    # Fallback to filename if extraction fails
    return file_path.name if hasattr(file_path, 'name') else str(file_path)


def _fallback_empty_dict_detector(code_text: str, file_path: Optional[Path], verbose: bool = False) -> None:
    """
    Detect empty NODE_CLASS_MAPPINGS (UI-only extensions).
    Logs for documentation purposes only (when verbose=True).

    Args:
        code_text: Python source code to analyze
        file_path: Path to the file being analyzed
        verbose: If True, print detection messages
    """
    empty_patterns = [
        'NODE_CLASS_MAPPINGS = {}',
        'NODE_CLASS_MAPPINGS={}',
    ]

    code_normalized = code_text.replace(' ', '').replace('\n', '')

    for pattern in empty_patterns:
        pattern_normalized = pattern.replace(' ', '')
        if pattern_normalized in code_normalized:
            if file_path and verbose:
                repo_name = _extract_repo_name(file_path)
                print(f"Info: UI-only extension (empty NODE_CLASS_MAPPINGS): {repo_name}")
            return

def has_comfy_node_base(class_node):
    """Check if class inherits from io.ComfyNode or ComfyNode"""
    for base in class_node.bases:
        # Case 1: ComfyNode
        if isinstance(base, ast.Name) and base.id == 'ComfyNode':
            return True
        # Case 2: io.ComfyNode
        elif isinstance(base, ast.Attribute):
            if base.attr == 'ComfyNode':
                return True
    return False


def extract_keyword_value(call_node, keyword):
    """
    Extract string value of keyword argument
    Schema(node_id="MyNode") -> "MyNode"
    """
    for kw in call_node.keywords:
        if kw.arg == keyword:
            # ast.Constant (Python 3.8+)
            if isinstance(kw.value, ast.Constant):
                if isinstance(kw.value.value, str):
                    return kw.value.value
            # ast.Str (Python 3.7-) - suppress deprecation warning
            else:
                with warnings.catch_warnings():
                    warnings.filterwarnings('ignore', category=DeprecationWarning)
                    if hasattr(ast, 'Str') and isinstance(kw.value, ast.Str):
                        return kw.value.s
    return None


def is_schema_call(call_node):
    """Check if ast.Call is io.Schema() or Schema()"""
    func = call_node.func
    if isinstance(func, ast.Name) and func.id == 'Schema':
        return True
    elif isinstance(func, ast.Attribute) and func.attr == 'Schema':
        return True
    return False


def extract_node_id_from_schema(class_node):
    """
    Extract node_id from define_schema() method
    """
    for item in class_node.body:
        if isinstance(item, ast.FunctionDef) and item.name == 'define_schema':
            # Walk through function body
            for stmt in ast.walk(item):
                if isinstance(stmt, ast.Call):
                    # Check if it's Schema() call
                    if is_schema_call(stmt):
                        node_id = extract_keyword_value(stmt, 'node_id')
                        if node_id:
                            return node_id
    return None


def extract_v3_nodes(code_text):
    """
    Extract V3 node IDs using AST parsing
    Returns: set of node_id strings
    """
    global parse_cnt

    try:
        if parse_cnt % 100 == 0:
            print(".", end="", flush=True)
        parse_cnt += 1

        with warnings.catch_warnings():
            warnings.filterwarnings('ignore', category=SyntaxWarning)
            warnings.filterwarnings('ignore', category=DeprecationWarning)
            tree = ast.parse(code_text)
    except (SyntaxError, UnicodeDecodeError):
        return set()

    nodes = set()

    # Find io.ComfyNode subclasses
    for node in ast.walk(tree):
        if isinstance(node, ast.ClassDef):
            # Check if inherits from ComfyNode
            if has_comfy_node_base(node):
                node_id = extract_node_id_from_schema(node)
                if node_id:
                    nodes.add(node_id)
                else:
                    # Fallback: use class name when node_id is dynamic/empty
                    nodes.add(node.name)

    return nodes


# scan
def extract_metadata_only(filename):
    """Extract only metadata (@author, @title, etc) without node scanning"""
    try:
        with open(filename, encoding='utf-8', errors='ignore') as file:
            code = file.read()

        metadata = {}
        lines = code.strip().split('\n')
        for line in lines:
            if line.startswith('@'):
                if line.startswith("@author:") or line.startswith("@title:") or line.startswith("@nickname:") or line.startswith("@description:"):
                    key, value = line[1:].strip().split(':', 1)
                    metadata[key.strip()] = value.strip()

        return metadata
    except:
        return {}


def scan_in_file(filename, is_builtin=False):
    global builtin_nodes

    with open(filename, encoding='utf-8', errors='ignore') as file:
        code = file.read()

    # Support type annotations (e.g., NODE_CLASS_MAPPINGS: Type = {...}) and line continuations (\)
    pattern = r"_CLASS_MAPPINGS\s*(?::\s*\w+\s*)?=\s*(?:\\\s*)?{([^}]*)}"
    regex = re.compile(pattern, re.MULTILINE | re.DOTALL)

    nodes = set()
    class_dict = {}

    # V1 nodes detection (enhanced with fallback patterns)
    nodes |= extract_nodes_enhanced(code, file_path=Path(filename), visited=set())

    # V3 nodes detection
    nodes |= extract_v3_nodes(code)
    code = re.sub(r'^#.*?$', '', code, flags=re.MULTILINE)

    def extract_keys(pattern, code):
        keys = re.findall(pattern, code)
        return {key.strip() for key in keys}

    def update_nodes(nodes, new_keys):
        nodes |= new_keys

    patterns = [
        r'^[^=]*_CLASS_MAPPINGS\["(.*?)"\]',
        r'^[^=]*_CLASS_MAPPINGS\[\'(.*?)\'\]',
        r'@register_node\("(.+)",\s*\".+"\)',
        r'"(\w+)"\s*:\s*{"class":\s*\w+\s*'
    ]

    with concurrent.futures.ThreadPoolExecutor() as executor:
        futures = {executor.submit(extract_keys, pattern, code): pattern for pattern in patterns}
        for future in concurrent.futures.as_completed(futures):
            update_nodes(nodes, future.result())

    matches = regex.findall(code)
    for match in matches:
        dict_text = match

        key_value_pairs = re.findall(r"\"([^\"]*)\"\s*:\s*([^,\n]*)", dict_text)
        for key, value in key_value_pairs:
            class_dict[key.strip()] = value.strip()

        key_value_pairs = re.findall(r"'([^']*)'\s*:\s*([^,\n]*)", dict_text)
        for key, value in key_value_pairs:
            class_dict[key.strip()] = value.strip()

        for key, value in class_dict.items():
            nodes.add(key.strip())

        update_pattern = r"_CLASS_MAPPINGS.update\s*\({([^}]*)}\)"
        update_match = re.search(update_pattern, code)
        if update_match:
            update_dict_text = update_match.group(1)
            update_key_value_pairs = re.findall(r"\"([^\"]*)\"\s*:\s*([^,\n]*)", update_dict_text)
            for key, value in update_key_value_pairs:
                class_dict[key.strip()] = value.strip()
                nodes.add(key.strip())

    metadata = {}
    lines = code.strip().split('\n')
    for line in lines:
        if line.startswith('@'):
            if line.startswith("@author:") or line.startswith("@title:") or line.startswith("@nickname:") or line.startswith("@description:"):
                key, value = line[1:].strip().split(':', 1)
                metadata[key.strip()] = value.strip()

    if is_builtin:
        builtin_nodes += set(nodes)
    else:
        for x in builtin_nodes:
            if x in nodes:
                nodes.remove(x)

    return nodes, metadata


def get_py_file_paths(dirname):
    file_paths = []
    
    for root, dirs, files in os.walk(dirname):
        if ".git" in root or "__pycache__" in root:
            continue

        for file in files:
            if file.endswith(".py"):
                file_path = os.path.join(root, file)
                file_paths.append(file_path)
    
    return file_paths


def get_nodes(target_dir):
    py_files = []
    directories = []
    
    for item in os.listdir(target_dir):
        if ".git" in item or "__pycache__" in item:
            continue

        path = os.path.abspath(os.path.join(target_dir, item))
        
        if os.path.isfile(path) and item.endswith(".py"):
            py_files.append(path)
        elif os.path.isdir(path):
            directories.append(path)
    
    return py_files, directories


def get_urls_from_list_file(list_file):
    """
    Read URLs from list file for scan-only mode

    Args:
        list_file (str): Path to URL list file (one URL per line)

    Returns:
        list of tuples: [(url, "", None, None), ...]
        Format: (url, title, preemptions, nodename_pattern)
        - title: Empty string
        - preemptions: None
        - nodename_pattern: None

    File format:
        https://github.com/owner/repo1
        https://github.com/owner/repo2
        # Comments starting with # are ignored

    Raises:
        FileNotFoundError: If list_file does not exist
    """
    if not os.path.exists(list_file):
        raise FileNotFoundError(f"URL list file not found: {list_file}")

    urls = []
    with open(list_file, 'r', encoding='utf-8') as f:
        for line_num, line in enumerate(f, 1):
            line = line.strip()

            # Skip empty lines and comments
            if not line or line.startswith('#'):
                continue

            # Validate URL format (basic check)
            if not (line.startswith('http://') or line.startswith('https://')):
                print(f"WARNING: Line {line_num} is not a valid URL: {line}")
                continue

            # Add URL with empty metadata
            # (url, title, preemptions, nodename_pattern)
            urls.append((line, "", None, None))

    print(f"Loaded {len(urls)} URLs from {list_file}")
    return urls


def get_git_urls_from_json(json_file):
    with open(json_file, encoding='utf-8') as file:
        data = json.load(file)

        custom_nodes = data.get('custom_nodes', [])
        git_clone_files = []
        for node in custom_nodes:
            if node.get('install_type') == 'git-clone':
                files = node.get('files', [])
                if files:
                    git_clone_files.append((files[0], node.get('title'), node.get('preemptions'), node.get('nodename_pattern')))

    git_clone_files.append(("https://github.com/comfyanonymous/ComfyUI", "ComfyUI", None, None))

    return git_clone_files


def get_py_urls_from_json(json_file):
    with open(json_file, encoding='utf-8') as file:
        data = json.load(file)

        custom_nodes = data.get('custom_nodes', [])
        py_files = []
        for node in custom_nodes:
            if node.get('install_type') == 'copy':
                files = node.get('files', [])
                if files:
                    py_files.append((files[0], node.get('title'), node.get('preemptions'), node.get('nodename_pattern')))

    return py_files


def clone_or_pull_git_repository(git_url):
    repo_name = git_url.split("/")[-1]
    if repo_name.endswith(".git"):
        repo_name = repo_name[:-4]

    repo_dir = os.path.join(temp_dir, repo_name)

    if os.path.exists(repo_dir):
        try:
            repo = Repo(repo_dir)
            origin = repo.remote(name="origin")
            origin.pull()
            repo.git.submodule('update', '--init', '--recursive')
            print(f"Pulling {repo_name}...")
        except Exception as e:
            print(f"Failed to pull '{repo_name}': {e}")
            _record_git_error(repo_name, 'pull', e)
    else:
        try:
            Repo.clone_from(git_url, repo_dir, recursive=True)
            print(f"Cloning {repo_name}...")
        except Exception as e:
            print(f"Failed to clone '{repo_name}': {e}")
            _record_git_error(repo_name, 'clone', e)


def update_custom_nodes(scan_only_mode=False, url_list_file=None):
    """
    Update custom nodes by cloning/pulling repositories

    Args:
        scan_only_mode (bool): If True, use URL list file instead of custom-node-list.json
        url_list_file (str): Path to URL list file (required if scan_only_mode=True)

    Returns:
        dict: node_info mapping {repo_name: (url, title, preemptions, node_pattern)}
    """
    if not os.path.exists(temp_dir):
        os.makedirs(temp_dir)

    node_info = {}

    # Select URL source based on mode
    if scan_only_mode:
        if not url_list_file:
            raise ValueError("url_list_file is required in scan-only mode")

        git_url_titles_preemptions = get_urls_from_list_file(url_list_file)
        print("\n[Scan-Only Mode]")
        print(f"  - URL source: {url_list_file}")
        print("  - GitHub stats: DISABLED")
        print(f"  - Git clone/pull: {'ENABLED' if not skip_update else 'DISABLED'}")
        print("  - Metadata: EMPTY")
    else:
        if not os.path.exists('custom-node-list.json'):
            raise FileNotFoundError("custom-node-list.json not found")

        git_url_titles_preemptions = get_git_urls_from_json('custom-node-list.json')
        print("\n[Standard Mode]")
        print("  - URL source: custom-node-list.json")
        print(f"  - GitHub stats: {'ENABLED' if not skip_stat_update else 'DISABLED'}")
        print(f"  - Git clone/pull: {'ENABLED' if not skip_update else 'DISABLED'}")
        print("  - Metadata: FULL")

    def process_git_url_title(url, title, preemptions, node_pattern):
        name = os.path.basename(url)
        if name.endswith(".git"):
            name = name[:-4]
        
        node_info[name] = (url, title, preemptions, node_pattern)
        if not skip_update:
            clone_or_pull_git_repository(url)

    def process_git_stats(git_url_titles_preemptions):
        GITHUB_STATS_CACHE_FILENAME = 'github-stats-cache.json'
        GITHUB_STATS_FILENAME = 'github-stats.json'

        github_stats = {}
        try:
            with open(GITHUB_STATS_CACHE_FILENAME, 'r', encoding='utf-8') as file:
                github_stats = json.load(file)
        except FileNotFoundError:
            pass

        def is_rate_limit_exceeded():
            return g.rate_limiting[0] <= 20

        if is_rate_limit_exceeded():
            print(f"GitHub API Rate Limit Exceeded: remained - {(g.rate_limiting_resettime - datetime.datetime.now().timestamp())/60:.2f} min")
        else:
            def renew_stat(url):
                if is_rate_limit_exceeded():
                    return

                if 'github.com' not in url:
                    return None

                print('.', end="")
                sys.stdout.flush()
                try:
                    # Parsing the URL
                    parsed_url = urlparse(url)
                    domain = parsed_url.netloc
                    path = parsed_url.path
                    path_parts = path.strip("/").split("/")
                    if len(path_parts) >= 2 and domain == "github.com":
                        owner_repo = "/".join(path_parts[-2:])
                        repo = g.get_repo(owner_repo)
                        owner = repo.owner
                        now = datetime.datetime.now(datetime.timezone.utc)
                        author_time_diff = now - owner.created_at
                        
                        last_update = repo.pushed_at.strftime("%Y-%m-%d %H:%M:%S") if repo.pushed_at else 'N/A'
                        item = {
                            "stars": repo.stargazers_count,
                            "last_update": last_update,
                            "cached_time": now.timestamp(),
                            "author_account_age_days": author_time_diff.days,
                        }
                        return url, item
                    else:
                        print(f"\nInvalid URL format for GitHub repository: {url}\n")
                except Exception as e:
                    print(f"\nERROR on {url}\n{e}")

                return None

            # resolve unresolved urls
            with concurrent.futures.ThreadPoolExecutor(11) as executor:
                futures = []
                for url, title, preemptions, node_pattern in git_url_titles_preemptions:
                    if url not in github_stats:
                        futures.append(executor.submit(renew_stat, url))

                for future in concurrent.futures.as_completed(futures):
                    url_item = future.result()
                    if url_item is not None:
                        url, item = url_item
                        github_stats[url] = item

            # renew outdated cache
            outdated_urls = []
            for k, v in github_stats.items():
                elapsed = (datetime.datetime.now().timestamp() - v['cached_time'])
                if elapsed > 60*60*12:  # 12 hours
                    outdated_urls.append(k)

            with concurrent.futures.ThreadPoolExecutor(11) as executor:
                for url in outdated_urls:
                    futures.append(executor.submit(renew_stat, url))

                for future in concurrent.futures.as_completed(futures):
                    url_item = future.result()
                    if url_item is not None:
                        url, item = url_item
                        github_stats[url] = item
                        
            with open('github-stats-cache.json', 'w', encoding='utf-8') as file:
                json.dump(github_stats, file, ensure_ascii=False, indent=4)

        with open(GITHUB_STATS_FILENAME, 'w', encoding='utf-8') as file:
            for v in github_stats.values():
                if "cached_time" in v:
                    del v["cached_time"]

            github_stats = dict(sorted(github_stats.items()))

            json.dump(github_stats, file, ensure_ascii=False, indent=4)

        print(f"Successfully written to {GITHUB_STATS_FILENAME}.")

    if not skip_stat_update:
        process_git_stats(git_url_titles_preemptions)

    # Reset error collector before this run
    with _git_error_lock:
        _git_errors.clear()

    # Git clone/pull for all repositories
    with concurrent.futures.ThreadPoolExecutor(11) as executor:
        for url, title, preemptions, node_pattern in git_url_titles_preemptions:
            executor.submit(process_git_url_title, url, title, preemptions, node_pattern)

    # Report any git errors grouped by category (after all workers complete)
    _report_git_errors()

    # .py file download (skip in scan-only mode - only process git repos)
    if not scan_only_mode:
        py_url_titles_and_pattern = get_py_urls_from_json('custom-node-list.json')

        def download_and_store_info(url_title_preemptions_and_pattern):
            url, title, preemptions, node_pattern = url_title_preemptions_and_pattern
            name = os.path.basename(url)
            if name.endswith(".py"):
                node_info[name] = (url, title, preemptions, node_pattern)

            try:
                download_url(url, temp_dir)
            except:
                print(f"[ERROR] Cannot download '{url}'")

        with concurrent.futures.ThreadPoolExecutor(10) as executor:
            executor.map(download_and_store_info, py_url_titles_and_pattern)

    return node_info


def gen_json(node_info, scan_only_mode=False, force_rescan=False):
    """
    Generate extension-node-map.json from scanned node information

    Args:
        node_info (dict): Repository metadata mapping
        scan_only_mode (bool): If True, exclude metadata from output
        force_rescan (bool): If True, ignore cache and force rescan all nodes
    """
    # scan from .py file
    node_files, node_dirs = get_nodes(temp_dir)

    comfyui_path = os.path.abspath(os.path.join(temp_dir, "ComfyUI"))
    # Only reorder if ComfyUI exists in the list
    if comfyui_path in node_dirs:
        node_dirs.remove(comfyui_path)
        node_dirs = [comfyui_path] + node_dirs

    data = {}
    for dirname in node_dirs:
        py_files = get_py_file_paths(dirname)
        metadata = {}

        # Use per-repo cache for node AND metadata extraction
        try:
            nodes, metadata = extract_nodes_from_repo(Path(dirname), verbose=False, force_rescan=force_rescan)
        except:
            # Fallback to file-by-file scanning if extract_nodes_from_repo fails
            nodes = set()
            for py in py_files:
                nodes_in_file, metadata_in_file = scan_in_file(py, dirname == "ComfyUI")
                nodes.update(nodes_in_file)
                metadata.update(metadata_in_file)

        dirname = os.path.basename(dirname)

        if 'Jovimetrix' in dirname:
            pass

        if len(nodes) > 0 or (dirname in node_info and node_info[dirname][3] is not None):
            nodes = list(nodes)
            nodes.sort()

            if dirname in node_info:
                git_url, title, preemptions, node_pattern = node_info[dirname]

                # Conditionally add metadata based on mode
                if not scan_only_mode:
                    # Standard mode: include all metadata
                    metadata['title_aux'] = title

                    if preemptions is not None:
                        metadata['preemptions'] = preemptions

                    if node_pattern is not None:
                        metadata['nodename_pattern'] = node_pattern
                # Scan-only mode: metadata remains empty

                data[git_url] = (nodes, metadata)
            else:
                # Scan-only mode: Repository not in node_info (expected behavior)
                # Construct URL from dirname (author_repo format)
                if '_' in dirname:
                    parts = dirname.split('_', 1)
                    git_url = f"https://github.com/{parts[0]}/{parts[1]}"
                    data[git_url] = (nodes, metadata)
                else:
                    print(f"WARN: {dirname} is removed from custom-node-list.json")

    for file in node_files:
        nodes, metadata = scan_in_file(file)

        if len(nodes) > 0 or (dirname in node_info and node_info[dirname][3] is not None):
            nodes = list(nodes)
            nodes.sort()

            file = os.path.basename(file)

            if file in node_info:
                url, title, preemptions, node_pattern = node_info[file]

                # Conditionally add metadata based on mode
                if not scan_only_mode:
                    metadata['title_aux'] = title

                    if preemptions is not None:
                        metadata['preemptions'] = preemptions

                    if node_pattern is not None:
                        metadata['nodename_pattern'] = node_pattern

                data[url] = (nodes, metadata)
            else:
                print(f"Missing info: {file}")

    # scan from node_list.json file
    extensions = [name for name in os.listdir(temp_dir) if os.path.isdir(os.path.join(temp_dir, name))]

    for extension in extensions:
        node_list_json_path = os.path.join(temp_dir, extension, 'node_list.json')
        if os.path.exists(node_list_json_path):
            # Skip if extension not in node_info (scan-only mode with limited URLs)
            if extension not in node_info:
                continue

            git_url, title, preemptions, node_pattern = node_info[extension]

            with open(node_list_json_path, 'r', encoding='utf-8') as f:
                try:
                    node_list_json = json.load(f)
                except Exception as e:
                    print(f"\nERROR: Invalid json format '{node_list_json_path}'")
                    print("------------------------------------------------------")
                    print(e)
                    print("------------------------------------------------------")
                    node_list_json = {}

            metadata_in_url = {}
            if git_url not in data:
                nodes = set()
            else:
                nodes_in_url, metadata_in_url = data[git_url]
                nodes = set(nodes_in_url)

            try:
                for x, desc in node_list_json.items():
                    nodes.add(x.strip())
            except Exception as e:
                print(f"\nERROR: Invalid json format '{node_list_json_path}'")
                print("------------------------------------------------------")
                print(e)
                print("------------------------------------------------------")
                node_list_json = {}

            # Conditionally add metadata based on mode
            if not scan_only_mode:
                metadata_in_url['title_aux'] = title

                if preemptions is not None:
                    metadata_in_url['preemptions'] = preemptions

                if node_pattern is not None:
                    metadata_in_url['nodename_pattern'] = node_pattern

            nodes = list(nodes)
            nodes.sort()
            data[git_url] = (nodes, metadata_in_url)

    json_path = "extension-node-map.json"
    with open(json_path, "w", encoding='utf-8') as file:
        json.dump(data, file, indent=4, sort_keys=True)


if __name__ == "__main__":
    # Parse arguments
    args = parse_arguments()

    # Determine mode
    scan_only_mode = args.scan_only is not None
    url_list_file = args.scan_only if scan_only_mode else None

    # Determine temp_dir
    if args.temp_dir:
        temp_dir = args.temp_dir
    elif args.temp_dir_positional:
        temp_dir = args.temp_dir_positional
    else:
        temp_dir = os.path.join(os.getcwd(), ".tmp")

    if not os.path.exists(temp_dir):
        os.makedirs(temp_dir)

    # Determine skip flags
    skip_update = args.skip_update or args.skip_all
    skip_stat_update = args.skip_stat_update or args.skip_all or scan_only_mode

    if not skip_stat_update:
        auth = Auth.Token(os.environ.get('GITHUB_TOKEN'))
        # Use a plain urllib3 Retry (NOT PyGithub's default GithubRetry) so that
        # a GitHub rate-limit response (403/429) raises RateLimitExceededException
        # IMMEDIATELY instead of sleeping until the rate-limit reset and retrying
        # up to 10x. The except-block in renew_stat() then catches it and skips
        # (returns None). 403/429 are intentionally NOT in status_forcelist, so
        # they fail fast; only transient 5xx errors are retried.
        g = Github(auth=auth, retry=Retry(total=2, backoff_factor=0.5,
                                          status_forcelist=[500, 502, 503, 504]))
    else:
        g = None

    print("### ComfyUI Manager Node Scanner ###")

    if scan_only_mode:
        print(f"\n# [Scan-Only Mode] Processing URL list: {url_list_file}\n")
    else:
        print("\n# [Standard Mode] Updating extensions\n")

    # Update/clone repositories and collect node info
    updated_node_info = update_custom_nodes(scan_only_mode, url_list_file)

    print("\n# Generating 'extension-node-map.json'...\n")

    # Generate extension-node-map.json
    force_rescan = args.force_rescan if hasattr(args, 'force_rescan') else False
    if force_rescan:
        print("⚠️  Force rescan enabled - ignoring all cached results\n")
    gen_json(updated_node_info, scan_only_mode, force_rescan)

    print("\n✅ DONE.\n")

    if scan_only_mode:
        print("Output: extension-node-map.json (node mappings only)")
    else:
        print("Output: extension-node-map.json (full metadata)")