A Python script showing multiply defined and unused labels in *.tex files

analysisLatexLabels.py

#!/usr/bin/env python3

"""
Analysis *.tex files in current directory or given <path>, show info about
multiply defined and unused labels.

Usage:
    python3 analysisLatexLabels.py
    python3 analysisLatexLabels.py <path>

For example,
    python3 analysisLatexLabels.py <path/to/pgfplots>/doc
"""

from os import PathLike
from pathlib import Path, PosixPath, WindowsPath
import re
import sys
from typing import Dict, List, Tuple, Union

# declare types
XPath = Union[PosixPath, WindowsPath, Path, PathLike]

# constant
if len(sys.argv) > 1:
    DIR = sys.argv[1]
else:
    DIR = '.'

# global variables
# list of tex files
tex_files: List[XPath] = list(Path(DIR).glob('**/*.tex'))

# defined labels, list of tuple (label name, tex file, line number)
labels: List[Tuple[str, str, int]] = []
# used refs, dict of { label name : [(tex file, line number), (tex file, line number), ...] }
refs: Dict[str, List[Tuple[str, int]]] = {}


# get all labels and references
re_label = re.compile(r'\\label{(.*?)}')
re_ref = re.compile(r'\\(|page|name|auto|autopage)ref\*?{(.*?)}')

for tex in tex_files:
    with open(tex) as f:
        # For better diagnostic info, read file by lines.
        # This assumes that every label or ref resides in single line.
        f_lines = f.readlines()
        
        curr_tex = str(tex)
        for lineno, line in enumerate(f_lines):
            lineno += 1
            # get labels
            for label in re_label.findall(line):
                labels.append((label, curr_tex, lineno))
            # get references
            for ref in re_ref.findall(line):
                label = ref[1]
                if label in refs:
                    refs[label].append((curr_tex, lineno))
                else:
                    refs[label] = [(curr_tex, lineno)]


labels.sort()


# check multiply defined label(s)
multiply_defined = []
for i in range(len(labels) - 1):
    # label_curr = labels[i]
    # label_next = labels[i+1]
    if labels[i][0] == labels[i+1][0]:
        error = f'Label "{labels[i][0]}" is multiply defined in'
        # Use format "<relative path to file>:<line number>" so that clicking "<file>:<line>" opens <file>
        # right at <line>. This is the feature of some shells, for example iTerm2.app for macOS.
        error += f'\n\t{labels[i][1]}:{labels[i][2]} and'
        error += f'\n\t{labels[i+1][1]}:{labels[i+1][2]}'
        multiply_defined.append(error)


# check unused label(s)
unused = []
for label in labels:
    if label[0] not in refs:
        error = f'Label "{label[0]}" unused:'
        error += f'\n\t{label[1]}:{label[2]}'
        unused.append(error)


# print results
print(*multiply_defined, sep='\n')
print(*unused, sep='\n')