217 lines
7.3 KiB
Python
217 lines
7.3 KiB
Python
#!/usr/bin/env python3
|
|
# file: scripts/audit_markdown_tables.py
|
|
# version: 2
|
|
|
|
"""Validate source formatting of Markdown tables in explicitly selected paths."""
|
|
|
|
from __future__ import annotations
|
|
|
|
import argparse
|
|
import pathlib
|
|
import re
|
|
import sys
|
|
|
|
_SEPARATOR = re.compile(r"^:?-{3,}:?$")
|
|
_IGNORED_DIRECTORIES = frozenset({".git", ".idea", ".venv", "__pycache__", "dist", "node_modules", "target", "builds"})
|
|
_MARKDOWN_FENCE_LANGUAGES = frozenset({"markdown", "md"})
|
|
|
|
|
|
def _selected_files(raw_paths: list[str]) -> tuple[list[pathlib.Path], list[str]]:
|
|
files: list[pathlib.Path] = []
|
|
errors: list[str] = []
|
|
|
|
for raw_path in raw_paths:
|
|
path = pathlib.Path(raw_path)
|
|
if path.is_dir():
|
|
files.extend(
|
|
candidate
|
|
for candidate in path.rglob("*.md")
|
|
if candidate.is_file() and not any(part in _IGNORED_DIRECTORIES for part in candidate.parts)
|
|
)
|
|
elif path.is_file() and path.suffix.lower() == ".md":
|
|
files.append(path)
|
|
else:
|
|
errors.append(f"Markdown table audit: unsupported or missing path: {path}")
|
|
|
|
return sorted(set(files)), errors
|
|
|
|
|
|
def _is_row(line: str) -> bool:
|
|
return line.startswith("|") and line.endswith("|")
|
|
|
|
|
|
def _cells(line: str) -> list[str]:
|
|
return line.split("|")[1:-1]
|
|
|
|
|
|
def _separator_marker(cell: str) -> str | None:
|
|
marker = cell.strip()
|
|
return marker if _SEPARATOR.fullmatch(marker) is not None else None
|
|
|
|
|
|
def _is_separator_row(line: str) -> bool:
|
|
if not _is_row(line):
|
|
return False
|
|
cells = _cells(line)
|
|
return bool(cells) and all(_separator_marker(cell) is not None for cell in cells)
|
|
|
|
|
|
def _alignment(marker: str) -> str:
|
|
if marker.startswith(":") and marker.endswith(":"):
|
|
return "center"
|
|
if marker.endswith(":"):
|
|
return "right"
|
|
return "left"
|
|
|
|
|
|
def _padding(cell: str) -> tuple[int, int]:
|
|
return len(cell) - len(cell.lstrip(" ")), len(cell) - len(cell.rstrip(" "))
|
|
|
|
|
|
def _validate_table(path: pathlib.Path, start_line: int, rows: list[str]) -> list[str]:
|
|
errors: list[str] = []
|
|
|
|
if any("\\|" in row for row in rows):
|
|
errors.append(f"{path}:{start_line}: escaped pipe is forbidden inside a table cell")
|
|
|
|
parsed = [_cells(row) for row in rows]
|
|
column_count = len(parsed[0])
|
|
|
|
for offset, row in enumerate(parsed):
|
|
if len(row) != column_count:
|
|
errors.append(f"{path}:{start_line + offset}: table row has {len(row)} columns; expected {column_count}")
|
|
return errors
|
|
|
|
for column in range(column_count):
|
|
raw_cells = [row[column] for row in parsed]
|
|
widths = [len(cell) for cell in raw_cells]
|
|
expected_width = widths[0]
|
|
|
|
if any(width != expected_width for width in widths):
|
|
errors.append(f"{path}:{start_line}: column {column + 1} is not vertically aligned; raw widths are {widths}")
|
|
continue
|
|
|
|
marker = _separator_marker(raw_cells[1])
|
|
if marker is None:
|
|
errors.append(f"{path}:{start_line + 1}: separator for column {column + 1} is malformed")
|
|
continue
|
|
|
|
content_cells = [cell for index, cell in enumerate(raw_cells) if index != 1]
|
|
longest_content = max(len(cell.strip()) for cell in content_cells)
|
|
required_width = max(longest_content + 2, 3 + marker.count(":"))
|
|
|
|
if expected_width != required_width:
|
|
errors.append(
|
|
f"{path}:{start_line}: column {column + 1} width is {expected_width}; expected {required_width}"
|
|
)
|
|
continue
|
|
|
|
alignment = _alignment(marker)
|
|
for cell in content_cells:
|
|
content_width = len(cell.strip())
|
|
left, right = _padding(cell)
|
|
expected_padding = expected_width - content_width
|
|
|
|
if content_width + left + right != len(cell) or left < 1 or right < 1:
|
|
errors.append(f"{path}:{start_line}: column {column + 1} must use spaces only for outer padding")
|
|
break
|
|
if alignment == "left" and (left != 1 or right != expected_padding - 1):
|
|
errors.append(f"{path}:{start_line}: column {column + 1} must be source-aligned to the left")
|
|
break
|
|
if alignment == "right" and (right != 1 or left != expected_padding - 1):
|
|
errors.append(f"{path}:{start_line}: column {column + 1} must be source-aligned to the right")
|
|
break
|
|
if alignment == "center" and abs(left - right) > 1:
|
|
errors.append(f"{path}:{start_line}: column {column + 1} must be source-aligned to the center")
|
|
break
|
|
|
|
if len(raw_cells[1]) != expected_width or _SEPARATOR.fullmatch(raw_cells[1]) is None:
|
|
errors.append(f"{path}:{start_line + 1}: separator for column {column + 1} must fill the column width")
|
|
|
|
return errors
|
|
|
|
|
|
def _fence_language(stripped: str) -> str | None:
|
|
info = stripped[3:].strip()
|
|
if not info:
|
|
return None
|
|
return info.split(None, 1)[0].lower()
|
|
|
|
|
|
def _audit_file(path: pathlib.Path) -> tuple[int, list[str]]:
|
|
lines = path.read_text(encoding="utf-8").splitlines()
|
|
errors: list[str] = []
|
|
tables = 0
|
|
fence: str | None = None
|
|
audit_fenced_markdown = False
|
|
index = 0
|
|
|
|
while index < len(lines):
|
|
stripped = lines[index].lstrip()
|
|
if stripped.startswith("```") or stripped.startswith("~~~"):
|
|
marker = stripped[:3]
|
|
if fence is None:
|
|
fence = marker
|
|
audit_fenced_markdown = _fence_language(stripped) in _MARKDOWN_FENCE_LANGUAGES
|
|
elif fence == marker:
|
|
fence = None
|
|
audit_fenced_markdown = False
|
|
index += 1
|
|
continue
|
|
|
|
if fence is not None and not audit_fenced_markdown:
|
|
index += 1
|
|
continue
|
|
|
|
if index + 1 < len(lines) and _is_row(lines[index]) and _is_separator_row(lines[index + 1]):
|
|
start = index
|
|
rows = [lines[index], lines[index + 1]]
|
|
index += 2
|
|
while index < len(lines) and _is_row(lines[index]):
|
|
rows.append(lines[index])
|
|
index += 1
|
|
tables += 1
|
|
errors.extend(_validate_table(path, start + 1, rows))
|
|
continue
|
|
|
|
index += 1
|
|
|
|
return tables, errors
|
|
|
|
|
|
def main() -> int:
|
|
"""Run the read-only Markdown table audit."""
|
|
|
|
parser = argparse.ArgumentParser()
|
|
parser.add_argument("paths", nargs="+", help="Markdown files or directories to audit")
|
|
arguments = parser.parse_args()
|
|
|
|
files, selection_errors = _selected_files(arguments.paths)
|
|
if selection_errors:
|
|
for error in selection_errors:
|
|
print(error, file=sys.stderr)
|
|
return 2
|
|
if not files:
|
|
print("Markdown table audit: no Markdown files selected", file=sys.stderr)
|
|
return 2
|
|
|
|
table_count = 0
|
|
errors: list[str] = []
|
|
for path in files:
|
|
current_count, current_errors = _audit_file(path)
|
|
table_count += current_count
|
|
errors.extend(current_errors)
|
|
|
|
if errors:
|
|
for error in errors:
|
|
print(error, file=sys.stderr)
|
|
print(f"Markdown table audit: {len(errors)} error(s) across {len(files)} file(s)", file=sys.stderr)
|
|
return 1
|
|
|
|
print(f"Markdown table audit: clean ({table_count} table(s), {len(files)} file(s))")
|
|
return 0
|
|
|
|
|
|
if __name__ == "__main__":
|
|
raise SystemExit(main())
|