# -*- coding: utf-8 -*-
"""Assemble the draft sections into one paper, resolving generated tables.

    python assemble.py            writes draft/paper.md

Sections are `draft/sNN-*.md`, in filename order. A line containing only
`{{table:name}}` is replaced by the contents of `draft/tables/name.md`, which is
written by `tables.py` straight from the session logs.

That indirection is the point. The recurring failure in this programme is not
arithmetic, it is a correct number printed on the wrong page: three earlier papers
shipped with another paper's cover statistics and every audit passed, because each
number was right somewhere. A number that cannot be typed into the prose cannot be
typed into the wrong prose.

Two gates, both fatal:

  * an unresolved `{{table:...}}` marker, so a renamed table cannot silently leave
    a hole;
  * a house-rule violation: an em dash, an en dash, a spaced hyphen used as
    punctuation, or an emoji.
"""
import glob
import io
import os
import re
import sys

HERE = os.path.dirname(os.path.abspath(__file__))
DRAFT = os.path.abspath(os.path.join(HERE, '..', 'draft'))
TABLES = os.path.join(DRAFT, 'tables')
OUT = os.path.join(DRAFT, 'paper.md')

MARKER = re.compile(r'^\{\{table:([a-z0-9\-]+)\}\}\s*$', re.M)
BANNED = {
    '—': 'em dash',
    '–': 'en dash',
    '‒': 'figure dash',
    '−': 'minus sign used as a dash',
}
EMOJI = re.compile('[\U0001F300-\U0001FAFF☀-➿️]')


def resolve(body, missing):
    def sub(m):
        name = m.group(1)
        path = os.path.join(TABLES, name + '.md')
        if not os.path.exists(path):
            missing.append(name)
            return m.group(0)
        with io.open(path, encoding='utf-8') as f:
            return f.read().rstrip() + '\n'
    return MARKER.sub(sub, body)


def house_rules(text):
    problems = []
    for ch, label in BANNED.items():
        n = text.count(ch)
        if n:
            problems.append('%d %s(s)' % (n, label))
    spaced = len(re.findall(r'\S \- \S', text))
    if spaced:
        problems.append('%d spaced hyphen(s) used as punctuation' % spaced)
    emo = EMOJI.findall(text)
    if emo:
        problems.append('%d emoji' % len(emo))
    return problems


def main():
    sections = sorted(glob.glob(os.path.join(DRAFT, 's[0-9][0-9]-*.md')))
    if not sections:
        print('no sections in %s' % DRAFT)
        return 1
    missing = []
    parts = []
    for path in sections:
        with io.open(path, encoding='utf-8') as f:
            parts.append(resolve(f.read().rstrip(), missing))
    text = '\n\n---\n\n'.join(parts) + '\n'

    problems = house_rules(text)
    if missing:
        problems.append('unresolved table marker(s): %s' % ', '.join(sorted(set(missing))))
    print('assembled %d section(s), %d words' % (len(sections), len(text.split())))
    for p in problems:
        print('  !!', p)
    if problems:
        print('refusing to write paper.md')
        return 1
    with io.open(OUT, 'w', encoding='utf-8') as f:
        f.write(text)
    print('wrote %s' % OUT)
    return 0


if __name__ == '__main__':
    raise SystemExit(main())