"""After an editorial pass: every number in the new article that was not in the old one, with its
context - a transcription slip shows up here.  Numbers only removed are fine (verify_article.py
checks the data, and a removed statement claims nothing).

    python -u check_new_numbers.py OLD.md NEW.md
"""
import re
import sys

NUM = re.compile(r"(?<![\w.])\d+(?:\.\d+)?(?![\w])")


def nums(text):
    return set(NUM.findall(text))


def main(old_path, new_path):
    old = open(old_path, encoding="utf8").read()
    new = open(new_path, encoding="utf8").read()
    fresh = sorted(nums(new) - nums(old), key=lambda s: float(s))
    print(f"{len(fresh)} numbers in the new text that the old one did not contain")
    for n in fresh:
        for m in re.finditer(rf"(?<![\w.]){re.escape(n)}(?![\w])", new):
            print(f"  {n:>10}: ...{new[max(0, m.start() - 60):m.end() + 30]!r}...")
            break
    print("new numbers done")


if __name__ == "__main__":
    main(sys.argv[1], sys.argv[2])