"""Regenerate article/sim_numerics.tex (and the PDF) from the canonical article/sim_numerics.md.

The .md is the text; the .tex is a build product and is never edited by hand.  Title, author
and abstract become pandoc metadata; the body is converted as it stands (section numbers are
part of the headings, so LaTeX numbering is off).  Unicode that pdflatex does not know is
declared in the header.  Run on the server (pandoc 3.1.3, pdflatex):

    python -u build_tex.py            -> article/sim_numerics.tex, article/sim_numerics.pdf
"""
from __future__ import annotations

import json
import os
import subprocess
import sys

MD = "article/sim_numerics.md"
TEX = "article/sim_numerics.tex"

HEADER = r"""
\usepackage{amsmath,amssymb}
\usepackage{textcomp}
\DeclareUnicodeCharacter{2032}{\ensuremath{'}}
\DeclareUnicodeCharacter{2212}{\ensuremath{-}}
\DeclareUnicodeCharacter{00D7}{\ensuremath{\times}}
\DeclareUnicodeCharacter{00B5}{\textmu}
\DeclareUnicodeCharacter{2192}{\ensuremath{\rightarrow}}
\DeclareUnicodeCharacter{0394}{\ensuremath{\Delta}}
\DeclareUnicodeCharacter{2082}{\textsubscript{2}}
\usepackage[margin=2.5cm]{geometry}
\usepackage[htt]{hyphenat}
\setlength{\emergencystretch}{3em}
\usepackage{etoolbox}
\AtBeginEnvironment{longtable}{\small}
"""

BANNER = ("%% GENERATED from article/sim_numerics.md by build_tex.py - do not edit by hand.\n"
          "%% The .md is the canonical text; every number in it is checked by verify_article.py.\n")


def split_front(md):
    lines = md.split("\n")
    title = lines[0].lstrip("# ").strip()
    rest = lines[1:]
    author = next(ln.strip() for ln in rest if ln.strip())
    text = "\n".join(rest)
    head, _, body = text.partition("## Abstract")
    abstract, _, body = body.partition("\n---\n")
    return title, author, abstract.strip(), body.lstrip("\n")


def main():
    with open(MD, encoding="utf8") as fh:
        title, author, abstract, body = split_front(fh.read())
    meta = {"title": title, "author": author, "abstract": abstract,
            "header-includes": HEADER, "documentclass": "article",
            "classoption": ["11pt", "a4paper"]}
    os.makedirs("build", exist_ok=True)
    with open("build/meta.json", "w", encoding="utf8") as fh:
        json.dump(meta, fh)
    with open("build/body.md", "w", encoding="utf8") as fh:
        fh.write(body)
    subprocess.run(["pandoc", "build/body.md", "--metadata-file=build/meta.json", "-s",
                    "-f", "markdown+tex_math_dollars+pipe_tables", "-t", "latex",
                    "-o", "build/body.tex"], check=True)
    with open("build/body.tex", encoding="utf8") as fh:
        tex = fh.read()
    with open(TEX, "w", encoding="utf8") as fh:
        fh.write(BANNER + tex)
    for _ in range(2):
        r = subprocess.run(["pdflatex", "-interaction=nonstopmode", "-halt-on-error",
                            "-output-directory", "article", TEX],
                           capture_output=True, text=True)
    log = open("article/sim_numerics.log", encoding="utf8", errors="replace").read()
    errors = [ln for ln in log.splitlines() if ln.startswith("!")]
    # only undefined references / citations count; 'Font shape ... undefined' is a font
    # substitution warning, not a defect of the text
    undefined = [ln for ln in log.splitlines()
                 if "undefined" in ln.lower() and ("reference" in ln.lower()
                                                   or "citation" in ln.lower())]
    overfull = sum(1 for ln in log.splitlines() if ln.startswith("Overfull"))
    pages = next((ln for ln in log.splitlines() if "Output written" in ln), "no output")
    print(f"pdflatex exit {r.returncode}; errors {len(errors)}; undefined {len(undefined)}; "
          f"overfull boxes {overfull}")
    for ln in errors[:10] + undefined[:10]:
        print("  ", ln)
    print(pages)
    print("build done")
    sys.exit(0 if r.returncode == 0 and not errors else 1)


if __name__ == "__main__":
    main()