11from __future__ import annotations
22import argparse , zipfile , hashlib , os
33from pathlib import Path
4+ from typing import Any
5+
6+ try :
7+ import tomllib as _toml_loader
8+ except ModuleNotFoundError :
9+ try :
10+ import tomli as _toml_loader
11+ except ModuleNotFoundError :
12+ _toml_loader = None
13+
414from .core import Config , generate_markdown_report
515from . import __version__
616
717# Load .env file if it exists (for Pro license and configuration)
818def load_env_file ():
9- # Try current directory first, then parent directories
1019 current = Path .cwd ()
1120 for parent in [current ] + list (current .parents ):
1221 env_file = parent / '.env'
@@ -17,17 +26,44 @@ def load_env_file():
1726 if line and not line .startswith ('#' ) and '=' in line :
1827 key , value = line .split ('=' , 1 )
1928 os .environ [key .strip ()] = value .strip ()
20- break # Stop after first .env file found
29+ break
2130 except Exception :
22- pass # Silently ignore .env file errors
31+ pass
2332
24- # Load environment configuration on import
2533load_env_file ()
2634
35+ DEFAULT_OUTPUT = "PROJECT_BLUEPRINT.md"
2736DEFAULT_EXCLUDES = [
2837 ".git" , "__pycache__" , "node_modules" , ".venv" ,
2938 "build" , "dist" , "*.pyc" , ".DS_Store" ,
3039]
40+ _CONFIG_KEYS = {
41+ "path" ,
42+ "output" ,
43+ "preset" ,
44+ "llm_mode" ,
45+ "budget_tokens" ,
46+ "max_file_tokens" ,
47+ "dedup" ,
48+ "sample_head" ,
49+ "sample_tail" ,
50+ "explain" ,
51+ "include_glob" ,
52+ "exclude_glob" ,
53+ "omit_glob" ,
54+ "only_ext" ,
55+ "respect_gitignore" ,
56+ "follow_symlinks" ,
57+ "max_bytes" ,
58+ "max_lines" ,
59+ "emit_manifest" ,
60+ "stats" ,
61+ "capsule" ,
62+ "dry_run" ,
63+ "no_timestamp" ,
64+ "masking" ,
65+ }
66+
3167
3268def positive_int (v : str ) -> int :
3369 try :
@@ -38,75 +74,139 @@ def positive_int(v: str) -> int:
3874 raise argparse .ArgumentTypeError ("Only positive integers are allowed." )
3975 return iv
4076
77+
78+ def _load_pyproject_config () -> dict [str , Any ]:
79+ if _toml_loader is None :
80+ return {}
81+
82+ decode_error = getattr (_toml_loader , "TOMLDecodeError" , ValueError )
83+ current_dir = Path .cwd ()
84+ for path in [current_dir ] + list (current_dir .parents ):
85+ pyproject_path = path / "pyproject.toml"
86+ if not pyproject_path .is_file ():
87+ continue
88+ try :
89+ with pyproject_path .open ("rb" ) as handle :
90+ data = _toml_loader .load (handle )
91+ except (decode_error , OSError ):
92+ return {}
93+
94+ tool_config = data .get ("tool" , {}).get ("dir2md" )
95+ if not isinstance (tool_config , dict ):
96+ return {}
97+
98+ sanitized : dict [str , Any ] = {}
99+ for raw_key , value in tool_config .items ():
100+ key = raw_key .replace ('-' , '_' )
101+ if key not in _CONFIG_KEYS :
102+ continue
103+ if key in {"include_glob" , "exclude_glob" , "omit_glob" }:
104+ if value is None :
105+ continue
106+ if isinstance (value , list ):
107+ sanitized [key ] = [str (item ) for item in value ]
108+ else :
109+ sanitized [key ] = [str (value )]
110+ continue
111+ if key == "only_ext" :
112+ if value is None :
113+ continue
114+ if isinstance (value , list ):
115+ sanitized [key ] = "," .join (str (item ) for item in value )
116+ else :
117+ sanitized [key ] = str (value )
118+ continue
119+ if key in {"budget_tokens" , "max_file_tokens" , "dedup" , "sample_head" , "sample_tail" , "max_bytes" , "max_lines" }:
120+ try :
121+ sanitized [key ] = int (value )
122+ except (TypeError , ValueError ):
123+ continue
124+ continue
125+ if key in {"respect_gitignore" , "follow_symlinks" , "emit_manifest" , "stats" , "capsule" , "dry_run" , "no_timestamp" , "explain" }:
126+ sanitized [key ] = bool (value )
127+ continue
128+ sanitized [key ] = value
129+ return sanitized
130+ return {}
131+
132+
41133def main (argv : list [str ] | None = None ) -> int :
42- ap = argparse .ArgumentParser (prog = "dir2md" , description = "Directory → Markdown exporter with LLM optimization" )
134+ config_from_file = _load_pyproject_config ()
135+
136+ ap = argparse .ArgumentParser (prog = "dir2md" , description = "Directory -> Markdown exporter with LLM optimization" )
43137 ap .add_argument ("path" , nargs = "?" , default = "." )
44- ap .add_argument ("-o" , "--output" , default = "PROJECT_BLUEPRINT.md" )
45-
46- # Preset options
47- ap .add_argument ("--preset" , default = "raw" , choices = ["iceberg" ,"pro" ,"raw" ], help = "Preset mode: iceberg/pro/raw" )
48-
49- # Token and selection control
50- ap .add_argument ("--llm-mode" , choices = ["off" ,"ref" ,"summary" ,"inline" ], default = None )
51- ap .add_argument ("--budget-tokens" , type = int , default = 6000 )
52- ap .add_argument ("--max-file-tokens" , type = int , default = 1200 )
53- ap .add_argument ("--dedup" , type = int , default = 16 )
54- ap .add_argument ("--sample-head" , type = int , default = 120 )
55- ap .add_argument ("--sample-tail" , type = int , default = 40 )
138+ ap .add_argument ("-o" , "--output" )
139+
140+ ap .add_argument ("--preset" , choices = ["iceberg" , "pro" , "raw" ], help = "Preset mode: iceberg/pro/raw" )
141+
142+ ap .add_argument ("--llm-mode" , choices = ["off" , "ref" , "summary" , "inline" ])
143+ ap .add_argument ("--budget-tokens" , type = int )
144+ ap .add_argument ("--max-file-tokens" , type = int )
145+ ap .add_argument ("--dedup" , type = int )
146+ ap .add_argument ("--sample-head" , type = int )
147+ ap .add_argument ("--sample-tail" , type = int )
56148 ap .add_argument ("--explain" , action = "store_true" , help = "Include selection rationale and drift_score in capsule comments" )
57149
58- # Filtering and safety controls
59- ap .add_argument ("--include-glob" , action = "append" , default = [])
60- ap .add_argument ("--exclude-glob" , action = "append" , default = [])
61- ap .add_argument ("--omit-glob" , action = "append" , default = [])
62- ap .add_argument ("--only-ext" , default = "" )
150+ ap .add_argument ("--include-glob" , action = "append" , help = "Gitignore-style include pattern (gitwildmatch syntax)" )
151+ ap .add_argument ("--exclude-glob" , action = "append" , help = "Gitignore-style exclude pattern" )
152+ ap .add_argument ("--omit-glob" , action = "append" , help = "Gitignore-style omit pattern (skips content)" )
153+ ap .add_argument ("--only-ext" , help = "Comma-separated extension list (e.g. py,md)" )
63154 ap .add_argument ("--respect-gitignore" , action = "store_true" )
64155 ap .add_argument ("--follow-symlinks" , action = "store_true" )
65- ap .add_argument ("--max-bytes" , type = positive_int , default = 200_000 )
66- ap .add_argument ("--max-lines" , type = positive_int , default = 2000 )
156+ ap .add_argument ("--max-bytes" , type = positive_int )
157+ ap .add_argument ("--max-lines" , type = positive_int )
67158
68- # Output options
69- ap .add_argument ("--emit-manifest" , action = "store_true" )
159+ ap .add_argument ("--emit-manifest" , action = "store_true" , help = "Write JSON manifest (raw preset overrides to off)" )
70160 ap .add_argument ("--stats" , action = "store_true" )
71161 ap .add_argument ("--capsule" , action = "store_true" , help = "Package md+manifest into zip" )
72162 ap .add_argument ("--dry-run" , action = "store_true" )
73163 ap .add_argument ("--no-timestamp" , action = "store_true" , help = "Omit timestamp for reproducible output" )
74- ap .add_argument ("--masking" , choices = ["off" , "basic" , "advanced" ], default = "off" , help = "Secret masking mode (advanced requires Pro license)" )
164+ ap .add_argument ("--masking" , choices = ["off" , "basic" , "advanced" ], help = "Secret masking mode (advanced requires Pro license)" )
75165
76166 ap .add_argument ("-V" , "--version" , action = "version" , version = f"dir2md { __version__ } )
77167
168+ if config_from_file :
169+ ap .set_defaults (** config_from_file )
170+
78171 ns = ap .parse_args (argv )
79172
80173 root = Path (ns .path ).resolve ()
81- output = Path (ns .output )
82- only_ext = {e .strip ().lstrip ('.' ) for e in ns .only_ext .split (',' ) if e .strip ()} or None
174+
175+ if ns .output :
176+ output = Path (ns .output )
177+ else :
178+ if root .is_dir ():
179+ output = root / f"{ root .name }
180+ else :
181+ output = Path (DEFAULT_OUTPUT ).resolve ()
182+ only_ext = {e .strip ().lstrip ('.' ) for e in (ns .only_ext or "" ).split (',' ) if e .strip ()} or None
83183
84184 cfg = Config (
85185 root = root ,
86186 output = output ,
87- include_globs = list (ns .include_glob ),
88- exclude_globs = list (ns .exclude_glob ) + DEFAULT_EXCLUDES ,
89- omit_globs = list (ns .omit_glob ),
90- respect_gitignore = bool (ns .respect_gitignore ),
91- follow_symlinks = bool (ns .follow_symlinks ),
92- max_bytes = int (ns .max_bytes ) if ns .max_bytes else None ,
93- max_lines = int (ns .max_lines ) if ns .max_lines else None ,
187+ include_globs = list (ns .include_glob or [] ),
188+ exclude_globs = list (ns .exclude_glob or [] ) + DEFAULT_EXCLUDES ,
189+ omit_globs = list (ns .omit_glob or [] ),
190+ respect_gitignore = bool (ns .respect_gitignore or False ),
191+ follow_symlinks = bool (ns .follow_symlinks or False ),
192+ max_bytes = int (ns .max_bytes ) if ns .max_bytes is not None else 200_000 ,
193+ max_lines = int (ns .max_lines ) if ns .max_lines is not None else 2000 ,
94194 include_contents = True ,
95195 only_ext = only_ext ,
96- add_stats = bool (ns .stats ),
196+ add_stats = bool (ns .stats or False ),
97197 add_toc = False ,
98198 llm_mode = (ns .llm_mode or "ref" ),
99- budget_tokens = int (ns .budget_tokens ),
100- max_file_tokens = int (ns .max_file_tokens ),
101- dedup_bits = int (ns .dedup ),
102- sample_head = int (ns .sample_head ),
103- sample_tail = int (ns .sample_tail ),
199+ budget_tokens = int (ns .budget_tokens ) if ns . budget_tokens is not None else 6000 ,
200+ max_file_tokens = int (ns .max_file_tokens ) if ns . max_file_tokens is not None else 1200 ,
201+ dedup_bits = int (ns .dedup ) if ns . dedup is not None else 16 ,
202+ sample_head = int (ns .sample_head ) if ns . sample_head is not None else 120 ,
203+ sample_tail = int (ns .sample_tail ) if ns . sample_tail is not None else 40 ,
104204 strip_comments = False ,
105- emit_manifest = bool (ns .emit_manifest ),
106- preset = str (ns .preset ),
107- explain_capsule = bool (ns .explain ),
108- no_timestamp = bool (ns .no_timestamp ),
109- masking_mode = str (ns .masking ),
205+ emit_manifest = bool (ns .emit_manifest if ns . emit_manifest is not None else True ),
206+ preset = str (ns .preset or "raw" ),
207+ explain_capsule = bool (ns .explain or False ),
208+ no_timestamp = bool (ns .no_timestamp or False ),
209+ masking_mode = str (ns .masking or "off" ),
110210 )
111211
112212 md = generate_markdown_report (cfg )
@@ -125,8 +225,12 @@ def main(argv: list[str] | None = None) -> int:
125225 try :
126226 print (f"[dir2md] Wrote: { output } )
127227 except UnicodeEncodeError :
128- print (f "[dir2md] Wrote: (File path contains unprintable characters, but the file was likely created successfully)"
228+ print ("[dir2md] Wrote: (File path contains unprintable characters, but the file was likely created successfully)" )
129229 return 0
130230
231+
131232if __name__ == "__main__" :
132233 raise SystemExit (main ())
234+
235+
236+
0 commit comments