fix: improve file filtering, add new utility,

- Improved the speed of file filtering in `crawl_local_files.py` with folder-level exclusion
- Added `fix_yaml.py` utility for YAML indentation fixes
- Updated `nodes.py` to support up to 20 core abstractions
- add option for no cache.

.env.sample

@@ -1,2 +1,5 @@
 GEMINI_PROJECT_ID=<GEMINI_PROJECT_ID>
+GEMINI_API_KEY=<GEMINI_API_KEY>
 GITHUB_TOKEN=<GITHUB_TOKEN>
+OPENROUTER_API_KEY = <OPENROUTER_API_KEY>
+OPENROUTER_MODEL = <OPENROUTER_MODEL>

.gitignore

@@ -100,3 +100,10 @@ llm_cache.json
 
 # Output files
 output/
+
+# uv manage
+pyproject.toml
+uv.lock
+
+docs/*.pdf
+docs/design-cn.md

main.py

@@ -14,8 +14,10 @@ DEFAULT_INCLUDE_PATTERNS = {
 }
 
 DEFAULT_EXCLUDE_PATTERNS = {
+    "assets/*", "data/*", "examples/*", "images/*", "public/*", "static/*", "temp/*",
+    "docs/*", 
     "venv/*", ".venv/*", "*test*", "tests/*", "docs/*", "examples/*", "v1/*",
-    "dist/*", "build/*", "experimental/*", "deprecated/*",
+    "dist/*", "build/*", "experimental/*", "deprecated/*", "misc/*", 
     "legacy/*", ".git/*", ".github/*", ".next/*", ".vscode/*", "obj/*", "bin/*", "node_modules/*", "*.log"
 }
 
@@ -36,6 +38,8 @@ def main():
     parser.add_argument("-s", "--max-size", type=int, default=100000, help="Maximum file size in bytes (default: 100000, about 100KB).")
     # Add language parameter for multi-language support
     parser.add_argument("--language", default="english", help="Language for the generated tutorial (default: english)")
+    # Add use_cache parameter to control LLM caching
+    parser.add_argument("--no-cache", action="store_true", help="Disable LLM response caching (default: caching enabled)")
 
     args = parser.parse_args()
 
@@ -62,6 +66,9 @@ def main():
         # Add language for multi-language support
         "language": args.language,
         
+        # Add use_cache flag (inverse of no-cache flag)
+        "use_cache": not args.no_cache,
+
         # Outputs will be populated by the nodes
         "files": [],
         "abstractions": [],
@@ -73,6 +80,7 @@ def main():
 
     # Display starting message with repository/directory and language
     print(f"Starting tutorial generation for: {args.repo or args.dir} in {args.language.capitalize()} language")
+    print(f"LLM caching: {'Disabled' if args.no_cache else 'Enabled'}")
 
     # Create the flow instance
     tutorial_flow = create_tutorial_flow()

nodes.py

@@ -1,9 +1,12 @@
 import os
+import re
 import yaml
 from pocketflow import Node, BatchNode
 from utils.crawl_github_files import crawl_github_files
 from utils.call_llm import call_llm
 from utils.crawl_local_files import crawl_local_files
+from utils.fix_yaml import add_indentation
+
 
 # Helper to get content for specific file indices
 def get_content_for_indices(files_data, indices):
@@ -79,6 +82,7 @@ class IdentifyAbstractions(Node):
         files_data = shared["files"]
         project_name = shared["project_name"]  # Get project name
         language = shared.get("language", "english") # Get language
+        use_cache = shared.get("use_cache", True)  # Get use_cache flag, default to True
 
         # Helper to create context from files, respecting limits (basic example)
         def create_llm_context(files_data):
@@ -94,10 +98,10 @@ class IdentifyAbstractions(Node):
         context, file_info = create_llm_context(files_data)
         # Format file info for the prompt (comment is just a hint for LLM)
         file_listing_for_prompt = "\n".join([f"- {idx} # {path}" for idx, path in file_info])
-        return context, file_listing_for_prompt, len(files_data), project_name, language # Return language
+        return context, file_listing_for_prompt, len(files_data), project_name, language, use_cache # Return use_cache
 
     def exec(self, prep_res):
-        context, file_listing_for_prompt, file_count, project_name, language = prep_res  # Unpack project name and language
+        context, file_listing_for_prompt, file_count, project_name, language, use_cache = prep_res  # Unpack use_cache
         print(f"Identifying abstractions using LLM...")
 
         # Add language instruction and hints only if not English
@@ -117,7 +121,7 @@ Codebase Context:
 {context}
 
 {language_instruction}Analyze the codebase context.
-Identify the top 5-10 core most important abstractions to help those new to the codebase.
+Identify the top 5-20 core most important abstractions to help those new to the codebase.
 
 For each abstraction, provide:
 1. A concise `name`{name_lang_hint}.
@@ -144,12 +148,14 @@ Format the output as a YAML list of dictionaries:
     Another core concept, similar to a blueprint for objects.{desc_lang_hint}
   file_indices:
     - 5 # path/to/another.js
-# ... up to 10 abstractions
+# ... up to 20 abstractions
 ```"""
-        response = call_llm(prompt)
+        response = call_llm(prompt, use_cache=use_cache)  # Pass use_cache parameter
 
         # --- Validation ---
         yaml_str = response.strip().split("```yaml")[1].split("```")[0].strip()
+        # add whitespace to fix llm generation error(except -)
+        yaml_str = add_indentation(yaml_str)
         abstractions = yaml.safe_load(yaml_str)
 
         if not isinstance(abstractions, list):
@@ -203,6 +209,7 @@ class AnalyzeRelationships(Node):
         files_data = shared["files"]
         project_name = shared["project_name"]  # Get project name
         language = shared.get("language", "english") # Get language
+        use_cache = shared.get("use_cache", True)  # Get use_cache flag, default to True
 
         # Create context with abstraction names, indices, descriptions, and relevant file snippets
         context = "Identified Abstractions:\n"
@@ -230,10 +237,10 @@ class AnalyzeRelationships(Node):
         )
         context += file_context_str
 
-        return context, "\n".join(abstraction_info_for_prompt), project_name, language # Return language
+        return context, "\n".join(abstraction_info_for_prompt), project_name, language, use_cache # Return use_cache
 
     def exec(self, prep_res):
-        context, abstraction_listing, project_name, language = prep_res  # Unpack project name and language
+        context, abstraction_listing, project_name, language, use_cache = prep_res  # Unpack use_cache
         print(f"Analyzing relationships using LLM...")
 
         # Add language instruction and hints only if not English
@@ -339,6 +346,7 @@ class OrderChapters(Node):
         relationships = shared["relationships"] # Summary/label might be translated
         project_name = shared["project_name"]  # Get project name
         language = shared.get("language", "english") # Get language
+        use_cache = shared.get("use_cache", True)  # Get use_cache flag, default to True
 
         # Prepare context for the LLM
         abstraction_info_for_prompt = []
@@ -363,10 +371,10 @@ class OrderChapters(Node):
         if language.lower() != "english":
              list_lang_note = f" (Names might be in {language.capitalize()})"
 
-        return abstraction_listing, context, len(abstractions), project_name, list_lang_note
+        return abstraction_listing, context, len(abstractions), project_name, list_lang_note, use_cache  # Return use_cache
 
     def exec(self, prep_res):
-        abstraction_listing, context, num_abstractions, project_name, list_lang_note = prep_res
+        abstraction_listing, context, num_abstractions, project_name, list_lang_note, use_cache = prep_res  # Unpack use_cache
         print("Determining chapter order using LLM...")
         # No language variation needed here in prompt instructions, just ordering based on structure
         # The input names might be translated, hence the note.
@@ -437,9 +445,11 @@ Now, provide the YAML output:
 class WriteChapters(BatchNode):
     def prep(self, shared):
         chapter_order = shared["chapter_order"] # List of indices
-        abstractions = shared["abstractions"]   # List of dicts, name/desc potentially translated
-        files_data = shared["files"]
-        language = shared.get("language", "english") # Get language
+        abstractions = shared["abstractions"] # List of {"name": str, "description": str, "files": [int]}
+        files_data = shared["files"] # List of (path, content) tuples
+        project_name = shared["project_name"]
+        language = shared.get("language", "english")
+        use_cache = shared.get("use_cache", True)  # Get use_cache flag, default to True
         
         # Get already written chapters to provide context
         # We store them temporarily during the batch run, not in shared memory yet

utils/call_llm.py

@@ -3,23 +3,29 @@ import os
 import logging
 import json
 from datetime import datetime
+import requests
 
 # Configure logging
 log_directory = os.getenv("LOG_DIR", "logs")
 os.makedirs(log_directory, exist_ok=True)
-log_file = os.path.join(log_directory, f"llm_calls_{datetime.now().strftime('%Y%m%d')}.log")
+log_file = os.path.join(
+    log_directory, f"llm_calls_{datetime.now().strftime('%Y%m%d')}.log"
+)
 
 # Set up logger
 logger = logging.getLogger("llm_logger")
 logger.setLevel(logging.INFO)
 logger.propagate = False  # Prevent propagation to root logger
 file_handler = logging.FileHandler(log_file)
-file_handler.setFormatter(logging.Formatter('%(asctime)s - %(levelname)s - %(message)s'))
+file_handler.setFormatter(
+    logging.Formatter("%(asctime)s - %(levelname)s - %(message)s")
+)
 logger.addHandler(file_handler)
 
 # Simple cache configuration
 cache_file = "llm_cache.json"
 
+
 # By default, we Google Gemini 2.5 pro, as it shows great performance for code understanding
 def call_llm(prompt: str, use_cache: bool = True) -> str:
     # Log the prompt
@@ -31,7 +37,7 @@ def call_llm(prompt: str, use_cache: bool = True) -> str:
         cache = {}
         if os.path.exists(cache_file):
             try:
-                with open(cache_file, 'r') as f:
+                with open(cache_file, "r") as f:
                     cache = json.load(f)
             except:
                 logger.warning(f"Failed to load cache, starting with empty cache")
@@ -42,21 +48,21 @@ def call_llm(prompt: str, use_cache: bool = True) -> str:
             return cache[prompt]
 
     # Call the LLM if not in cache or cache disabled
-    client = genai.Client(
-        vertexai=True, 
-        # TODO: change to your own project id and location
-        project=os.getenv("GEMINI_PROJECT_ID", "your-project-id"),
-        location=os.getenv("GEMINI_LOCATION", "us-central1")
-    )
-    # You can comment the previous line and use the AI Studio key instead:
     # client = genai.Client(
-    #     api_key=os.getenv("GEMINI_API_KEY", "your-api_key"),
+    #     vertexai=True,
+    #     # TODO: change to your own project id and location
+    #     project=os.getenv("GEMINI_PROJECT_ID", "your-project-id"),
+    #     location=os.getenv("GEMINI_LOCATION", "us-central1")
     # )
-    model = os.getenv("GEMINI_MODEL", "gemini-2.5-pro-exp-03-25")
-    response = client.models.generate_content(
-        model=model,
-        contents=[prompt]
+
+    # You can comment the previous line and use the AI Studio key instead:
+    client = genai.Client(
+        api_key=os.getenv("GEMINI_API_KEY", ""),
     )
+    # model = os.getenv("GEMINI_MODEL", "gemini-2.5-pro-exp-03-25")
+    model = os.getenv("GEMINI_MODEL", "gemini-2.0-flash-exp")
+    
+    response = client.models.generate_content(model=model, contents=[prompt])
     response_text = response.text
 
     # Log the response
@@ -68,7 +74,7 @@ def call_llm(prompt: str, use_cache: bool = True) -> str:
         cache = {}
         if os.path.exists(cache_file):
             try:
-                with open(cache_file, 'r') as f:
+                with open(cache_file, "r") as f:
                     cache = json.load(f)
             except:
                 pass
@@ -76,13 +82,14 @@ def call_llm(prompt: str, use_cache: bool = True) -> str:
         # Add to cache and save
         cache[prompt] = response_text
         try:
-            with open(cache_file, 'w') as f:
+            with open(cache_file, "w") as f:
                 json.dump(cache, f)
         except Exception as e:
             logger.error(f"Failed to save cache: {e}")
 
     return response_text
 
+
 # # Use Anthropic Claude 3.7 Sonnet Extended Thinking
 # def call_llm(prompt, use_cache: bool = True):
 #     from anthropic import Anthropic
@@ -115,6 +122,82 @@ def call_llm(prompt: str, use_cache: bool = True) -> str:
 #     )
 #     return r.choices[0].message.content
 
+# Use OpenRouter API
+# def call_llm(prompt: str, use_cache: bool = True) -> str:
+#     # Log the prompt
+#     logger.info(f"PROMPT: {prompt}")
+
+#     # Check cache if enabled
+#     if use_cache:
+#         # Load cache from disk
+#         cache = {}
+#         if os.path.exists(cache_file):
+#             try:
+#                 with open(cache_file, "r") as f:
+#                     cache = json.load(f)
+#             except:
+#                 logger.warning(f"Failed to load cache, starting with empty cache")
+
+#         # Return from cache if exists
+#         if prompt in cache:
+#             logger.info(f"RESPONSE: {cache[prompt]}")
+#             return cache[prompt]
+
+#     # OpenRouter API configuration
+#     api_key = os.getenv("OPENROUTER_API_KEY", "")
+#     model = os.getenv("OPENROUTER_MODEL", "google/gemini-2.0-flash-exp:free")
+    
+#     headers = {
+#         "Authorization": f"Bearer {api_key}",
+#     }
+
+#     data = {
+#         "model": model,
+#         "messages": [{"role": "user", "content": prompt}]
+#     }
+
+#     response = requests.post(
+#         "https://openrouter.ai/api/v1/chat/completions",
+#         headers=headers,
+#         json=data
+#     )
+
+#     if response.status_code != 200:
+#         error_msg = f"OpenRouter API call failed with status {response.status_code}: {response.text}"
+#         logger.error(error_msg)
+#         raise Exception(error_msg)
+#     try:
+#         response_text = response.json()["choices"][0]["message"]["content"]
+#     except Exception as e:
+#         error_msg = f"Failed to parse OpenRouter response: {e}; Response: {response.text}"
+#         logger.error(error_msg)        
+#         raise Exception(error_msg)
+    
+
+#     # Log the response
+#     logger.info(f"RESPONSE: {response_text}")
+
+#     # Update cache if enabled
+#     if use_cache:
+#         # Load cache again to avoid overwrites
+#         cache = {}
+#         if os.path.exists(cache_file):
+#             try:
+#                 with open(cache_file, "r") as f:
+#                     cache = json.load(f)
+#             except:
+#                 pass
+
+#         # Add to cache and save
+#         cache[prompt] = response_text
+#         try:
+#             with open(cache_file, "w") as f:
+#                 json.dump(cache, f)
+#         except Exception as e:
+#             logger.error(f"Failed to save cache: {e}")
+
+#     return response_text
+
 if __name__ == "__main__":
     test_prompt = "Hello, how are you?"
 
@@ -122,4 +205,3 @@ if __name__ == "__main__":
     print("Making call...")
     response1 = call_llm(test_prompt, use_cache=False)
     print(f"Response: {response1}")
-    

utils/crawl_local_files.py

@@ -1,9 +1,12 @@
 import os
 import fnmatch
+from tqdm import tqdm
 
 def crawl_local_files(directory, include_patterns=None, exclude_patterns=None, max_file_size=None, use_relative_paths=True):
     """
     Crawl files in a local directory with similar interface as crawl_github_files.
+    Implements efficient folder-level filtering to skip entire directories that match exclude patterns,
+    significantly improving performance when excluding large directory trees.
     
     Args:
         directory (str): Path to local directory
@@ -20,7 +23,31 @@ def crawl_local_files(directory, include_patterns=None, exclude_patterns=None, m
         
     files_dict = {}
     
-    for root, _, files in os.walk(directory):
+    for root, dirs, files in os.walk(directory):
+        print(f"root: {root}")
+        # Check if current directory should be excluded
+        if exclude_patterns:
+            # Get path relative to directory if requested
+            rel_root = os.path.relpath(root, directory) if use_relative_paths else root
+            
+            # Handle the case where rel_root is the current directory
+            if rel_root == '.':
+                rel_root = ''
+            
+            # Check if directory matches any exclude pattern
+            for pattern in exclude_patterns:
+                # Normalize pattern to handle both forward and backward slashes
+                norm_pattern = pattern.replace("/", os.path.sep)
+                
+                # Check if the directory matches the pattern
+                if fnmatch.fnmatch(rel_root, norm_pattern) or \
+                   fnmatch.fnmatch(os.path.join(rel_root, ''), norm_pattern + os.path.sep):
+                    # Skip this directory and all subdirectories
+                    dirs[:] = []  # Clear dirs list to prevent further traversal
+                    print(f"Skipping directory: {rel_root} (matches pattern {pattern})")
+                    break
+            else:   
+            # print(root)
                 for filename in files:
                     filepath = os.path.join(root, filename)
                     
@@ -44,7 +71,8 @@ def crawl_local_files(directory, include_patterns=None, exclude_patterns=None, m
                     excluded = False
                     if exclude_patterns:
                         for pattern in exclude_patterns:
-                    if fnmatch.fnmatch(relpath, pattern):
+                            if fnmatch.fnmatch(relpath, pattern) or fnmatch.fnmatch(relpath, pattern.replace("/", "\\")):
+                                print(relpath, pattern)
                                 excluded = True
                                 break
                                 
@@ -66,7 +94,7 @@ def crawl_local_files(directory, include_patterns=None, exclude_patterns=None, m
 
 if __name__ == "__main__":
     print("--- Crawling parent directory ('..') ---")
-    files_data = crawl_local_files("..", exclude_patterns={"*.pyc", "__pycache__/*", ".git/*", "output/*"})
+    files_data = crawl_local_files("..", exclude_patterns={"*.pyc", "__pycache__/*",".venv/*", ".git/*","docs/*", "output/*"})
     print(f"Found {len(files_data['files'])} files:")
     for path in files_data["files"]:
         print(f"  {path}")

utils/fix_yaml.py

@@ -0,0 +1,20 @@
+import re
+
+def add_indentation(text):
+    # This pattern matches lines that don't start with a hyphen or whitespace
+    pattern = r'^(?![-\s])(.*)$'
+    
+    # Replace with 4 spaces followed by the captured content
+    result = re.sub(pattern, r'    \1', text, flags=re.MULTILINE)
+    
+    return result
+
+
+if __name__ == "__main__":
+    # Example usage
+    text = """This line will be indented
+    - This line won't be indented
+    This line won't be indented either
+    Another line that will be indented"""
+
+    indented_text = add_indentation(text)