This commit introduces a complete structural and architectural refactor of

Analysis-Ready Code (ARC). The project is now fully migrated to a modern
src/-based Python package layout, with proper packaging via pyproject.toml,
a clean Nix flake, and improved CLI entry points.

Major changes:

• Add `src/arc/` package with clean module structure:
  - arc/__init__.py now contains the main() dispatcher and clipboard helpers
  - arc/__main__.py provides a proper `python -m arc` entry point
  - arc/cli.py rewritten with full argparse-based interface
  - arc/code_processor.py modernized and relocated
  - arc/directory_handler.py rewritten with output_stream support
  - arc/tee.py added for multi-stream output (stdout + buffer)

• Remove legacy top-level modules:
  - cli.py
  - directory_handler.py
  - main.py

• Introduce fully PEP-517 compliant pyproject.toml with console script:
  - arc = arc.__main__:main

• Add Nix flake (`flake.nix`) providing:
  - buildPythonApplication package `arc`
  - `nix run .#arc` app
  - development shell with Python + xclip

• Add Makefile overhaul:
  - automatic detection of Nix vs Python installation
  - unified install/uninstall targets
  - Nix wrapper installation into ~/.local/bin
  - improved help text and shell safety

• Add GitHub CI pipelines:
  - ci-python.yml for Python builds + Makefile tests + arc --help
  - ci-nix.yml for Nix builds, flake checks, dev-shell tests, and `nix run .#arc`

• Refactor and extend unit tests:
  - test_arc.py updated for src/ imports
  - new tests: test_cli.py, test_main.py, test_tee.py
  - improved CodeProcessor and DirectoryHandler tests

• Add egg-info metadata for local builds

• Add build/lib/ tree for compatibility with setuptools (generated)

Overall, this commit modernizes ARC into a clean, robust, and fully packaged
Python/Nix hybrid tool, enabling reproducible builds, solid CLI behavior,
testable architecture, and CI automation.

https://chatgpt.com/share/693933a0-e280-800f-9cf0-26036d15be04
This commit is contained in:
2025-12-10 09:47:19 +01:00
parent b55576beb2
commit 039481d3a9
19 changed files with 965 additions and 186 deletions

285
src/arc/code_processor.py Normal file
View File

@@ -0,0 +1,285 @@
import io
import re
import tokenize
import zlib
from dataclasses import dataclass
from typing import Dict, Tuple, Pattern, Optional
@dataclass(frozen=True)
class LanguageSpec:
"""Holds compiled comment patterns for a language."""
patterns: Tuple[Pattern, ...]
class CodeProcessor:
"""
Utilities to strip comments and (de)compress code.
- Python: tokenize-based (safe) with precise docstring removal.
- C/CPP/JS: state-machine comment stripper that respects string/char literals.
- Shell/YAML: remove full-line hash comments only.
- Jinja: remove {# ... #} blocks.
"""
# File extensions (normalized to lowercase)
EXT_TO_LANG: Dict[str, str] = {
".py": "python",
".js": "cstyle",
".c": "cstyle",
".cpp": "cstyle",
".h": "cstyle",
".sh": "hash",
".bash": "hash",
".yml": "hash",
".yaml": "hash",
".j2": "jinja",
".jinja": "jinja",
".jinja2": "jinja",
".tpl": "jinja",
}
# Regex-based specs for hash and jinja
_HASH = LanguageSpec(patterns=(
re.compile(r"^\s*#.*$", flags=re.MULTILINE), # only full-line comments
))
_JINJA = LanguageSpec(patterns=(
re.compile(r"\{#.*?#\}", flags=re.DOTALL), # {# ... #} across lines
))
LANG_SPECS: Dict[str, LanguageSpec] = {
"hash": _HASH,
"jinja": _JINJA,
# "cstyle" handled by a state machine, not regex
# "python" handled by tokenize, not regex
}
@classmethod
def _lang_from_ext(cls, file_type: str) -> Optional[str]:
"""Map an extension like '.py' to an internal language key."""
ext = file_type.lower().strip()
return cls.EXT_TO_LANG.get(ext)
# -----------------------------
# Python stripping via tokenize
# -----------------------------
@staticmethod
def _strip_python_comments_tokenize(content: str) -> str:
"""
Remove comments and docstrings safely using tokenize.
Rules:
- Drop COMMENT tokens.
- Drop module docstring only if it's the very first statement at col 0.
- Drop the first STRING statement in a suite immediately after 'def'/'class'
header (':' NEWLINE INDENT).
"""
tokens = tokenize.generate_tokens(io.StringIO(content).readline)
out_tokens = []
indent_level = 0
module_docstring_candidate = True # until we see first real stmt at module level
expect_suite_docstring = False # just entered a suite after def/class
last_was_colon = False
seen_nontrivial_in_line = False # guards module docstring (start of logical line)
for tok_type, tok_str, start, end, line in tokens:
# Track indentation
if tok_type == tokenize.INDENT:
indent_level += 1
elif tok_type == tokenize.DEDENT:
indent_level = max(0, indent_level - 1)
# New logical line: reset guard
if tok_type in (tokenize.NEWLINE, tokenize.NL):
seen_nontrivial_in_line = False
out_tokens.append((tok_type, tok_str))
continue
# Comments are dropped
if tok_type == tokenize.COMMENT:
continue
# Detect ':' ending a def/class header
if tok_type == tokenize.OP and tok_str == ":":
last_was_colon = True
out_tokens.append((tok_type, tok_str))
continue
# After ':' + NEWLINE + INDENT comes a suite start -> allow docstring removal
if tok_type == tokenize.INDENT and last_was_colon:
expect_suite_docstring = True
last_was_colon = False
out_tokens.append((tok_type, tok_str))
continue
# Any non-INDENT token clears the last_was_colon flag
if tok_type != tokenize.NL:
last_was_colon = False
# STRING handling
if tok_type == tokenize.STRING:
at_line_start = (start[1] == 0) and not seen_nontrivial_in_line
if indent_level == 0:
# Potential module docstring only if first statement at col 0
if module_docstring_candidate and at_line_start:
module_docstring_candidate = False
# drop it
continue
# Any other top-level string is normal
module_docstring_candidate = False
out_tokens.append((tok_type, tok_str))
seen_nontrivial_in_line = True
continue
else:
# In a suite: if it's the first statement after def/class, drop regardless of column
if expect_suite_docstring:
expect_suite_docstring = False
# drop it
continue
expect_suite_docstring = False
out_tokens.append((tok_type, tok_str))
seen_nontrivial_in_line = True
continue
# Any other significant token disables module-docstring candidacy
if tok_type not in (tokenize.INDENT, tokenize.DEDENT):
if indent_level == 0:
module_docstring_candidate = False
# Mark we've seen something on this line
if tok_type not in (tokenize.NL, tokenize.NEWLINE):
seen_nontrivial_in_line = True
out_tokens.append((tok_type, tok_str))
return tokenize.untokenize(out_tokens)
# ---------------------------------
# C-style stripping via state machine
# ---------------------------------
@staticmethod
def _strip_cstyle_comments(content: str) -> str:
"""
Remove // line comments and /* ... */ block comments while preserving
string ("...") and char ('...') literals and their escape sequences.
"""
i = 0
n = len(content)
out = []
in_line_comment = False
in_block_comment = False
in_string = False
in_char = False
escape = False
while i < n:
c = content[i]
nxt = content[i + 1] if i + 1 < n else ""
# If inside line comment: consume until newline
if in_line_comment:
if c == "\n":
in_line_comment = False
out.append(c)
i += 1
continue
# If inside block comment: consume until '*/'
if in_block_comment:
if c == "*" and nxt == "/":
in_block_comment = False
i += 2
else:
i += 1
continue
# If inside string literal
if in_string:
out.append(c)
if escape:
escape = False
else:
if c == "\\":
escape = True
elif c == '"':
in_string = False
i += 1
continue
# If inside char literal
if in_char:
out.append(c)
if escape:
escape = False
else:
if c == "\\":
escape = True
elif c == "'":
in_char = False
i += 1
continue
# Not in any special state:
# Check for start of comments
if c == "/" and nxt == "/":
in_line_comment = True
i += 2
continue
if c == "/" and nxt == "*":
in_block_comment = True
i += 2
continue
# Check for start of string/char literals
if c == '"':
in_string = True
out.append(c)
i += 1
continue
if c == "'":
in_char = True
out.append(c)
i += 1
continue
# Normal character
out.append(c)
i += 1
return "".join(out)
# -------------------
# Public API
# -------------------
@classmethod
def remove_comments(cls, content: str, file_type: str) -> str:
"""
Remove comments based on file type/extension.
- Python: tokenize-based
- C/CPP/JS: state-machine
- Hash (sh/yaml): regex full-line
- Jinja: regex {# ... #}
"""
lang = cls._lang_from_ext(file_type)
if lang is None:
return content.strip()
if lang == "python":
return cls._strip_python_comments_tokenize(content).strip()
if lang == "cstyle":
return cls._strip_cstyle_comments(content).strip()
spec = cls.LANG_SPECS.get(lang)
if not spec:
return content.strip()
cleaned = content
for pat in spec.patterns:
cleaned = pat.sub("", cleaned)
return cleaned.strip()
@staticmethod
def compress(content: str, level: int = 9) -> bytes:
"""Compress code using zlib. Returns bytes."""
return zlib.compress(content.encode("utf-8"), level)
@staticmethod
def decompress(blob: bytes) -> str:
"""Decompress zlib-compressed code back to text."""
return zlib.decompress(blob).decode("utf-8")