validate_doctests.py

#!/usr/bin/env python
# Validate the code_begin lines in the Zig Language Reference.

import argparse
import sys

CODE_INDENT = 6


def validate(file, verbose=False):
    seen = set()
    doctests = 0
    tuples = []
    for i, line in enumerate(file):
        idx = line.find("{#code_begin|")
        if idx == -1:
            continue

        # Check that the code block is indented consistently.
        if idx != CODE_INDENT:
            debug(f"line: {i}: invalid indentation for code block")

        line = line.strip()
        line = line.removeprefix("{#")
        line = line.removesuffix("#}")
        fields = line.split("|")[1:]

        if len(fields) < 2:
            debug(f"line: {i}: missing doctest name")

        kind = fields[0]
        name = fields[1]

        if kind in ("test_err", "test_safety", "obj_err") and len(fields) < 3:
            debug(f"line: {i}: missing doctest error_str: {kind}|{name}")

        if name in seen:
            debug(f"line: {i}: duplicate doctest name: {name}")
        seen.add(name)
        doctests += 1

        if verbose:
            tuples.append((kind, name))

    if verbose:
        tuples.sort()
        for kind, name in tuples:
            print(kind + "|" + name)

    print(f"{doctests} doctests")


def debug(msg):
    sys.stderr.write(msg + "\n")


def main(path):
    parser = argparse.ArgumentParser()
    parser.add_argument("-v", "--verbose", action="store_true")
    args = parser.parse_args()

    with open(path) as file:
        validate(file, args.verbose)


if __name__ == "__main__":
    main("doc/langref.html.in")