#!/usr/bin/env python3
"""extract_demos.py - build the m2 gtk-demo catalog from the GTK C demo sources.

Reads the per-demo `.c` files under gtk/demos/gtk-demo and emits
`demos/DemoCatalog.def` / `demos/DemoCatalog.mod`: the metadata (name,
title, keywords, description) of every demo, so the Modula-2 browser can
list and search them.

Usage:
    tools/extract_demos.py [--src gtk/demos/gtk-demo] [--out demos]
"""
import argparse
import glob
import os
import re
import sys

MAX_NAME = 48
MAX_TITLE = 80
MAX_KEYWORDS = 256
MAX_DESCRIPTION = 512


def m2_str(text, limit):
    """Escape a Python string for a Modula-2 string literal of bounded size."""
    text = " ".join(text.split())
    if len(text) > limit:
        text = text[:limit]
    text = text.encode("ascii", "replace").decode("ascii")
    text = text.replace("\\", " ").replace('"', "'")
    return text


def parse(path):
    src = open(path, encoding="utf-8", errors="replace").read()
    name = os.path.basename(path)[:-2]
    title = ""
    keywords = ""
    description = ""
    m = re.match(r"\ufeff?\s*/\*(.*?)\*/", src, re.S)
    if m:
        lines = [l.strip(" *\t") for l in m.group(1).splitlines()]
        lines = [l for l in lines if l]
        if lines:
            title = lines[0]
        desc = []
        for l in lines[1:]:
            if l.startswith("#Keywords:"):
                keywords = l.split(":", 1)[1].strip()
            else:
                desc.append(l)
        description = " ".join(desc)
    # the do_<name> function defines the demo id used by the C registry
    ids = re.findall(r"\bdo_([a-z0-9_]+)\s*\(", src)
    if ids:
        name = ids[0]
    return name, title, keywords, description


def main():
    ap = argparse.ArgumentParser()
    ap.add_argument("--src", default="gtk/demos/gtk-demo")
    ap.add_argument("--out", default="demos")
    args = ap.parse_args()

    files = sorted(glob.glob(os.path.join(args.src, "*.c")))
    entries = []
    for f in files:
        name, title, kw, desc = parse(f)
        if name == "main":
            continue
        entries.append((name, title or name.replace("_", " "), kw, desc))

    entries.sort(key=lambda e: e[1].lower())

    os.makedirs(args.out, exist_ok=True)

    with open(os.path.join(args.out, "DemoCatalog.def"), "w") as f:
        f.write(
            'DEFINITION MODULE DemoCatalog ;\n\n'
            '(*\n'
            '   m2-GTK4 - the catalogue of gtk-demo demos, generated from the\n'
            '   GTK C demo sources by tools/extract_demos.py.  Provides the\n'
            '   title/keywords metadata used by the demo browser.\n'
            '*)\n\n'
            'EXPORT UNQUALIFIED DemoCatalogCount, DemoCatalogGet;\n\n'
            'PROCEDURE DemoCatalogCount () : CARDINAL;\n\n'
            '(* copy the metadata of demo i into the caller buffers *)\n'
            'PROCEDURE DemoCatalogGet (i: CARDINAL; VAR name, title, keywords,\n'
            '               description: ARRAY OF CHAR);\n\n'
            'END DemoCatalog.\n'
        )

    with open(os.path.join(args.out, "DemoCatalog.mod"), "w") as f:
        f.write("IMPLEMENTATION MODULE DemoCatalog ;\n\n")
        f.write("TYPE\n   Entry = RECORD\n")
        f.write("      name: ARRAY [0..%d] OF CHAR;\n" % (MAX_NAME - 1))
        f.write("      title: ARRAY [0..%d] OF CHAR;\n" % (MAX_TITLE - 1))
        f.write("      keywords: ARRAY [0..%d] OF CHAR;\n" % (MAX_KEYWORDS - 1))
        f.write("      description: ARRAY [0..%d] OF CHAR;\n" % (MAX_DESCRIPTION - 1))
        f.write("   END;\n\n")
        f.write("VAR\n   entries: ARRAY [0..%d] OF Entry;\n   n: CARDINAL;\n\n" % (len(entries) - 1))
        f.write(
            "PROCEDURE Copy (VAR dst: ARRAY OF CHAR; src: ARRAY OF CHAR);\n"
            "VAR i: CARDINAL;\n"
            "BEGIN\n"
            "   i := 0;\n"
            "   WHILE (i <= HIGH(dst)) AND (i <= HIGH(src)) AND (src[i] # 0C) DO\n"
            "      dst[i] := src[i]; INC(i)\n"
            "   END;\n"
            "   IF i <= HIGH(dst) THEN dst[i] := 0C END\n"
            "END Copy;\n\n"
            "PROCEDURE DemoCatalogCount () : CARDINAL;\nBEGIN\n   RETURN n\nEND DemoCatalogCount;\n\n"
            "PROCEDURE DemoCatalogGet (i: CARDINAL; VAR name, title, keywords,\n"
            "               description: ARRAY OF CHAR);\n"
            "BEGIN\n"
            "   IF i < n THEN\n"
            "      Copy(name, entries[i].name);\n"
            "      Copy(title, entries[i].title);\n"
            "      Copy(keywords, entries[i].keywords);\n"
            "      Copy(description, entries[i].description)\n"
            "   ELSE\n"
            "      name[0] := 0C; title[0] := 0C; keywords[0] := 0C;\n"
            "      description[0] := 0C\n"
            "   END\n"
            "END DemoCatalogGet;\n\n"
            "BEGIN\n"
            "   n := %d;\n" % len(entries)
        )
        for i, (name, title, kw, desc) in enumerate(entries):
            f.write('   entries[%d].name := "%s";\n' % (i, m2_str(name, MAX_NAME - 1)))
            f.write('   entries[%d].title := "%s";\n' % (i, m2_str(title, MAX_TITLE - 1)))
            f.write('   entries[%d].keywords := "%s";\n' % (i, m2_str(kw, MAX_KEYWORDS - 1)))
            f.write('   entries[%d].description := "%s";\n' % (i, m2_str(desc, MAX_DESCRIPTION - 1)))
        f.write("END DemoCatalog.\n")

    sys.stderr.write("extract_demos: %d demos -> %s/DemoCatalog.{def,mod}\n"
                     % (len(entries), args.out))


if __name__ == "__main__":
    main()