#!/usr/bin/env python3
"""
build-reference-docx.py
=======================

Baut die templates/reference.docx fuer die Pandoc-DOCX-Pipeline aus der
Pandoc-Default-Reference, mit gezielten Anpassungen.

Iteration B1 + B1.5 + B2 + B3 + B4 (aktuell):
  B1   - Theme-Schriften (majorFont und minorFont) beide auf Calibri.
  B1   - Direkte Schriftnamen-Referenzen in styles.xml auf Calibri
         (Code-Schriften wie Consolas bleiben).
  B1   - Tabellen-Default-Stil "Table" mit tblBorders=none.
  B1.5 - Body-DocDefault 11pt, Heading 1/2/3 auf 15/13/12 pt.
  B2   - Header (Name links, "Lebenslauf" rechts) ab Seite 2; Seite 1 mit
         leerem Header (titlePg-Mechanik). Footer (rechts: Seite n / m) auf
         allen Seiten inkl. Seite 1. Page-Setup explizit: A4, Raender
         analog PDF (top/bottom 2.2 cm, left/right 2.5 cm).
  B3   - DocDefault widowControl. Heading 1/2/3 mit keepNext + keepLines.
         Zusaetzlich 'FirstParagraph' (Pandoc-Stil fuer den ersten Absatz
         nach einem Heading) - deckt die fett formatierten Kenntnisse-
         Subsection-Labels ab. Hinweis: Listen-Bullet-Schutz (3-3-Regel)
         passiert nicht hier, sondern im Post-Processing
         (build/post-process-docx.py), das auf das fertige DOCX angewendet
         wird - ein Stil kann keine Per-Bullet-Logik abbilden.
  B4   - Heading 1/2/3 in destengsblue (3C68AE) gefaerbt (themeColor
         entfernt, damit die Farbe nicht aus dem Word-Theme kommt).
         Wichtig (S10-Fix): NICHT NUR die Paragraph-Styles
         Heading1/2/3, sondern AUCH die Linked Character Styles
         Heading1Char/2Char/3Char muessen gefaerbt werden.
         Pandoc-3.x-Default-Reference enthaelt diese Char-Styles
         mit eigener color val=0F4761 themeColor=accent1
         themeShade=BF (Aptos-Petrol). Char-Styles haben in Word
         Vorrang vor Para-Styles bei Run-Eigenschaften (Schrift,
         Farbe), d.h. ohne Char-Style-Aenderung gewinnt das
         Theme-Aptos-Blau gegenueber dem Para-Style-destengsblue.
  B4-Bold (S10) - Heading 1 und 2 fett (Para- UND Char-Stil),
         Heading 3 bleibt normal. Pandoc 3.x Default-Reference
         hat Headings nicht fett. Char-Stil-Vorrangs-Logik
         identisch wie bei der Farbe.
         Hinweis (S08): die zwischenzeitlich eingebauten Heading-
         Trennlinien (Bottom-Border + Indent-Trick) wurden zurueck-
         gerollt, weil sie in Word linksbuendig statt zentriert
         gerendert wurden (Word-Border folgt bei hanging-Indent der
         visuellen Absatz-Position, nicht den Indent-Werten).

Geplant in Folge-Iterationen:
  C  - Foto-Einbindung
  D  - Hyphenation-Feintuning fuer PDF
"""

from __future__ import annotations

import re
import subprocess
import sys
import tempfile
import zipfile
from pathlib import Path
from xml.etree import ElementTree as ET

SCRIPT_DIR    = Path(__file__).resolve().parent
BASE_DIR      = SCRIPT_DIR.parent
TEMPLATES_DIR = BASE_DIR / "templates"
OUTPUT_FILE   = TEMPLATES_DIR / "reference.docx"

NS = {
    "w":   "http://schemas.openxmlformats.org/wordprocessingml/2006/main",
    "a":   "http://schemas.openxmlformats.org/drawingml/2006/main",
    "r":   "http://schemas.openxmlformats.org/officeDocument/2006/relationships",
    "rel": "http://schemas.openxmlformats.org/package/2006/relationships",
    "ct":  "http://schemas.openxmlformats.org/package/2006/content-types",
}
for prefix, uri in NS.items():
    ET.register_namespace(prefix, uri)
ET.register_namespace("", NS["rel"])

W = "{%s}" % NS["w"]
A = "{%s}" % NS["a"]

CODE_FONTS = {"consolas", "courier", "courier new", "liberation mono",
              "monaco", "menlo", "fira mono", "fira code"}
TARGET_FONT = "Calibri"

SIZE_BODY     = 22
SIZE_HEADING1 = 30
SIZE_HEADING2 = 26
SIZE_HEADING3 = 24
HEADING_SIZES = {"Heading1": SIZE_HEADING1,
                 "Heading2": SIZE_HEADING2,
                 "Heading3": SIZE_HEADING3}
# Compact NICHT mehr in dieser Liste - Listen-Bullet-Schutz uebernimmt das
# Post-Processing-Skript pro-Bullet.
KEEP_STYLES = ("Heading1", "Heading2", "Heading3", "FirstParagraph")

# B4 - Heading-Farben (Trennlinien wurden in S08 zurueckgerollt, siehe
# Modul-Docstring). Bleibt: Heading 1/2/3 in destengsblue, themeColor entfernt.
HEADING_COLOR        = "3C68AE"   # destengsblue (analog template.tex)
HEADING_COLOR_STYLES = ("Heading1", "Heading2", "Heading3",
                        "Heading1Char", "Heading2Char", "Heading3Char")
# B4-Bold (S10) - Heading 1+2 fett, Heading 3 bleibt normal.
# Auch hier: Para- UND Char-Style modifizieren, weil Char-Style
# in Word Run-Properties dominiert (analog Farb-Fix).
HEADING_BOLD_STYLES  = ("Heading1", "Heading2",
                        "Heading1Char", "Heading2Char")

PAGE_W       = 11906
PAGE_H       = 16838
MARGIN_TOP   = 1247
MARGIN_BOT   = 1247
MARGIN_LEFT  = 1417
MARGIN_RIGHT = 1417
HEADER_POS   = 720
FOOTER_POS   = 720
HEADER_RIGHT_TAB = PAGE_W - MARGIN_LEFT - MARGIN_RIGHT

HEADER_LEFT  = "Dr.-Ing. Thomas Langer"
HEADER_RIGHT = "Lebenslauf"

def log(msg):
    print(f"[build-reference-docx] {msg}", flush=True)

XML_DECL = b'<?xml version="1.0" encoding="UTF-8" standalone="yes"?>\n'

def write_xml(tree, dest):
    body = ET.tostring(tree.getroot(), encoding="utf-8")
    dest.write_bytes(XML_DECL + body)

def write_xml_bytes(content, dest):
    dest.write_bytes(XML_DECL + content)

def fetch_pandoc_default(dest):
    log("Pandoc-Default-Reference extrahieren ...")
    result = subprocess.run(
        ["pandoc", "--print-default-data-file", "reference.docx"],
        capture_output=True, check=False,
    )
    if result.returncode != 0:
        sys.stderr.write(result.stderr.decode("utf-8", errors="replace"))
        raise SystemExit(f"pandoc liefert Exit-Code {result.returncode}")
    dest.write_bytes(result.stdout)
    log(f"  -> {dest} ({dest.stat().st_size} Bytes)")

def unpack_docx(src, dest_dir):
    with zipfile.ZipFile(src, "r") as z:
        z.extractall(dest_dir)

def repack_docx(src_dir, dest):
    files = []
    for path in src_dir.rglob("*"):
        if path.is_file():
            arcname = path.relative_to(src_dir).as_posix()
            files.append((path, arcname))
    files.sort(key=lambda t: (0 if t[1] == "[Content_Types].xml" else 1, t[1]))
    with zipfile.ZipFile(dest, "w", zipfile.ZIP_DEFLATED) as z:
        for path, arcname in files:
            z.write(path, arcname)

def is_code_font(name):
    return (name or "").strip().lower() in CODE_FONTS

# --- B1: Schriften ---------------------------------------------------------

def set_theme_fonts_to_calibri(theme_xml):
    tree = ET.parse(theme_xml)
    root = tree.getroot()
    for kind in ("majorFont", "minorFont"):
        font = root.find(f".//{A}{kind}")
        if font is None:
            raise RuntimeError(f"{kind}-Element nicht im Theme")
        latin = font.find(f"{A}latin")
        if latin is None:
            raise RuntimeError(f"{kind}/latin-Element nicht gefunden")
        old = latin.get("typeface")
        latin.set("typeface", TARGET_FONT)
        log(f"  Theme {kind}/latin: {old!r} -> {TARGET_FONT!r}")
    write_xml(tree, theme_xml)

def replace_direct_fonts_in_styles(styles_xml):
    tree = ET.parse(styles_xml)
    root = tree.getroot()
    changed = 0
    skipped = 0
    for rfonts in root.iter(f"{W}rFonts"):
        for attr in (f"{W}ascii", f"{W}hAnsi", f"{W}cs", f"{W}eastAsia"):
            val = rfonts.get(attr)
            if val is None:
                continue
            if is_code_font(val):
                skipped += 1
                continue
            if val != TARGET_FONT:
                rfonts.set(attr, TARGET_FONT)
                changed += 1
    log(f"  styles.xml: {changed} direkte Font-Attribute auf {TARGET_FONT!r}"
        f" gesetzt (Code-Fonts unangetastet: {skipped})")
    write_xml(tree, styles_xml)

def set_table_borders_none(styles_xml):
    tree = ET.parse(styles_xml)
    root = tree.getroot()
    style = next((s for s in root.findall(f"{W}style")
                  if s.get(f"{W}styleId") == "Table"), None)
    if style is None:
        raise RuntimeError("Style 'Table' nicht in styles.xml")
    tbl_pr = style.find(f"{W}tblPr")
    if tbl_pr is None:
        tbl_pr = ET.SubElement(style, f"{W}tblPr")
    existing = tbl_pr.find(f"{W}tblBorders")
    if existing is not None:
        tbl_pr.remove(existing)
    borders = ET.SubElement(tbl_pr, f"{W}tblBorders")
    for side in ("top", "left", "bottom", "right", "insideH", "insideV"):
        e = ET.SubElement(borders, f"{W}{side}")
        e.set(f"{W}val", "none")
        e.set(f"{W}sz", "0")
        e.set(f"{W}space", "0")
        e.set(f"{W}color", "auto")
    log("  Style 'Table': tblBorders=none auf allen Sides")
    write_xml(tree, styles_xml)

def set_default_body_size(styles_xml):
    tree = ET.parse(styles_xml)
    root = tree.getroot()
    docDefaults = root.find(f"{W}docDefaults")
    if docDefaults is None:
        docDefaults = ET.SubElement(root, f"{W}docDefaults")
    rPrDefault = docDefaults.find(f"{W}rPrDefault")
    if rPrDefault is None:
        rPrDefault = ET.SubElement(docDefaults, f"{W}rPrDefault")
    rPr = rPrDefault.find(f"{W}rPr")
    if rPr is None:
        rPr = ET.SubElement(rPrDefault, f"{W}rPr")
    for tag in (f"{W}sz", f"{W}szCs"):
        elem = rPr.find(tag)
        if elem is None:
            elem = ET.SubElement(rPr, tag)
        elem.set(f"{W}val", str(SIZE_BODY))
    log(f"  DocDefault Body-Schriftgroesse: {SIZE_BODY/2} pt")
    write_xml(tree, styles_xml)

def set_heading_sizes(styles_xml):
    tree = ET.parse(styles_xml)
    root = tree.getroot()
    for style in root.findall(f"{W}style"):
        sid = style.get(f"{W}styleId")
        if sid not in HEADING_SIZES:
            continue
        target = HEADING_SIZES[sid]
        rPr = style.find(f"{W}rPr")
        if rPr is None:
            rPr = ET.SubElement(style, f"{W}rPr")
        for tag in (f"{W}sz", f"{W}szCs"):
            elem = rPr.find(tag)
            if elem is None:
                elem = ET.SubElement(rPr, tag)
            elem.set(f"{W}val", str(target))
        log(f"  Stil {sid!r}: Schriftgroesse {target/2} pt")
    write_xml(tree, styles_xml)

def set_widow_control_default(styles_xml):
    tree = ET.parse(styles_xml)
    root = tree.getroot()
    docDefaults = root.find(f"{W}docDefaults")
    if docDefaults is None:
        docDefaults = ET.SubElement(root, f"{W}docDefaults")
    pPrDefault = docDefaults.find(f"{W}pPrDefault")
    if pPrDefault is None:
        pPrDefault = ET.SubElement(docDefaults, f"{W}pPrDefault")
    pPr = pPrDefault.find(f"{W}pPr")
    if pPr is None:
        pPr = ET.SubElement(pPrDefault, f"{W}pPr")
    if pPr.find(f"{W}widowControl") is None:
        ET.SubElement(pPr, f"{W}widowControl")
    log("  pPrDefault: widowControl aktiviert")
    write_xml(tree, styles_xml)

def set_keep_next_styles(styles_xml):
    tree = ET.parse(styles_xml)
    root = tree.getroot()
    seen = set()
    for style in root.findall(f"{W}style"):
        sid = style.get(f"{W}styleId")
        if sid not in KEEP_STYLES:
            continue
        pPr = style.find(f"{W}pPr")
        if pPr is None:
            pPr = ET.SubElement(style, f"{W}pPr")
        for tag in (f"{W}keepNext", f"{W}keepLines"):
            if pPr.find(tag) is None:
                ET.SubElement(pPr, tag)
        log(f"  Stil {sid!r}: keepNext + keepLines")
        seen.add(sid)
    missing = set(KEEP_STYLES) - seen
    if missing:
        log(f"  Hinweis: Stil(e) {sorted(missing)!r} nicht gefunden, uebersprungen")
    write_xml(tree, styles_xml)

# --- B4: Heading-Farben ----------------------------------------------------

def set_heading_colors(styles_xml):
    tree = ET.parse(styles_xml)
    root = tree.getroot()
    for style in root.findall(f"{W}style"):
        sid = style.get(f"{W}styleId")
        if sid not in HEADING_COLOR_STYLES:
            continue
        rPr = style.find(f"{W}rPr")
        if rPr is None:
            rPr = ET.SubElement(style, f"{W}rPr")
        color = rPr.find(f"{W}color")
        if color is None:
            color = ET.SubElement(rPr, f"{W}color")
        # Theme-Color-Attribute entfernen, damit die Farbe nicht aus dem
        # Word-Theme abgeleitet wird (Pandoc-Default: themeColor accent1).
        for attr in (f"{W}themeColor", f"{W}themeTint", f"{W}themeShade"):
            if attr in color.attrib:
                del color.attrib[attr]
        color.set(f"{W}val", HEADING_COLOR)
        log(f"  Stil {sid!r}: color={HEADING_COLOR} (themeColor entfernt)")
    write_xml(tree, styles_xml)

def set_heading_bold(styles_xml):
    """Setzt <w:b/> und <w:bCs/> auf Heading1 + Heading2 (Para- und Char-Stil).
    Pandoc 3.x Default-Reference hat weder Para- noch Char-Stil fett gesetzt.
    Char-Styles werden mit modifiziert, weil sie in Word fuer Run-Properties
    Vorrang vor Para-Styles haben (analog Farb-Fix in set_heading_colors)."""
    tree = ET.parse(styles_xml)
    root = tree.getroot()
    seen = set()
    for style in root.findall(f"{W}style"):
        sid = style.get(f"{W}styleId")
        if sid not in HEADING_BOLD_STYLES:
            continue
        rPr = style.find(f"{W}rPr")
        if rPr is None:
            rPr = ET.SubElement(style, f"{W}rPr")
        for tag in (f"{W}b", f"{W}bCs"):
            if rPr.find(tag) is None:
                ET.SubElement(rPr, tag)
        log(f"  Stil {sid!r}: bold (b + bCs)")
        seen.add(sid)
    missing = set(HEADING_BOLD_STYLES) - seen
    if missing:
        log(f"  Hinweis: Stil(e) {sorted(missing)!r} nicht gefunden, uebersprungen")
    write_xml(tree, styles_xml)

def header_default_xml():
    return (
        b'<w:hdr xmlns:w="' + NS["w"].encode() + b'">\n'
        b'  <w:p>\n'
        b'    <w:pPr>\n'
        b'      <w:tabs>\n'
        b'        <w:tab w:val="right" w:pos="' + str(HEADER_RIGHT_TAB).encode() + b'"/>\n'
        b'      </w:tabs>\n'
        b'    </w:pPr>\n'
        b'    <w:r><w:t xml:space="preserve">' + HEADER_LEFT.encode() + b'</w:t></w:r>\n'
        b'    <w:r><w:tab/><w:t xml:space="preserve">' + HEADER_RIGHT.encode() + b'</w:t></w:r>\n'
        b'  </w:p>\n'
        b'</w:hdr>\n'
    )

def header_first_blank_xml():
    return (
        b'<w:hdr xmlns:w="' + NS["w"].encode() + b'">\n'
        b'  <w:p/>\n'
        b'</w:hdr>\n'
    )

def footer_default_xml():
    return (
        b'<w:ftr xmlns:w="' + NS["w"].encode() + b'">\n'
        b'  <w:p>\n'
        b'    <w:pPr>\n'
        b'      <w:tabs>\n'
        b'        <w:tab w:val="right" w:pos="' + str(HEADER_RIGHT_TAB).encode() + b'"/>\n'
        b'      </w:tabs>\n'
        b'    </w:pPr>\n'
        b'    <w:r><w:tab/><w:t xml:space="preserve">Seite </w:t></w:r>\n'
        b'    <w:fldSimple w:instr="PAGE">\n'
        b'      <w:r><w:t>1</w:t></w:r>\n'
        b'    </w:fldSimple>\n'
        b'    <w:r><w:t xml:space="preserve"> / </w:t></w:r>\n'
        b'    <w:fldSimple w:instr="NUMPAGES">\n'
        b'      <w:r><w:t>1</w:t></w:r>\n'
        b'    </w:fldSimple>\n'
        b'  </w:p>\n'
        b'</w:ftr>\n'
    )

REL_HEADER = "http://schemas.openxmlformats.org/officeDocument/2006/relationships/header"
REL_FOOTER = "http://schemas.openxmlformats.org/officeDocument/2006/relationships/footer"
CT_HEADER  = "application/vnd.openxmlformats-officedocument.wordprocessingml.header+xml"
CT_FOOTER  = "application/vnd.openxmlformats-officedocument.wordprocessingml.footer+xml"

def next_free_rel_id(rels_xml):
    text = rels_xml.read_text(encoding="utf-8")
    ids = [int(m.group(1)) for m in re.finditer(r'Id="rId(\d+)"', text)]
    return (max(ids) + 1) if ids else 1

def add_relationship(rels_xml, rid, rtype, target):
    text = rels_xml.read_text(encoding="utf-8")
    new_rel = f'<Relationship Type="{rtype}" Id="{rid}" Target="{target}" />'
    if new_rel in text:
        return
    text = text.replace("</Relationships>", new_rel + "</Relationships>")
    rels_xml.write_text(text, encoding="utf-8")

def add_content_type_override(ct_xml, part_name, ct):
    text = ct_xml.read_text(encoding="utf-8")
    new_override = f'<Override PartName="{part_name}" ContentType="{ct}"/>'
    if part_name in text:
        return
    text = text.replace("</Types>", new_override + "</Types>")
    ct_xml.write_text(text, encoding="utf-8")

def update_sectpr_with_headers(document_xml, header_default_rid, header_first_rid, footer_default_rid):
    text = document_xml.read_text(encoding="utf-8")
    new_sectpr = (
        f'<w:sectPr>'
        f'<w:headerReference w:type="default" r:id="{header_default_rid}"/>'
        f'<w:headerReference w:type="first" r:id="{header_first_rid}"/>'
        f'<w:footerReference w:type="default" r:id="{footer_default_rid}"/>'
        f'<w:footerReference w:type="first" r:id="{footer_default_rid}"/>'
        f'<w:pgSz w:w="{PAGE_W}" w:h="{PAGE_H}"/>'
        f'<w:pgMar w:top="{MARGIN_TOP}" w:right="{MARGIN_RIGHT}"'
        f' w:bottom="{MARGIN_BOT}" w:left="{MARGIN_LEFT}"'
        f' w:header="{HEADER_POS}" w:footer="{FOOTER_POS}" w:gutter="0"/>'
        f'<w:titlePg/>'
        f'</w:sectPr>'
    )
    new_text, n = re.subn(
        r'<w:sectPr\s*/>|<w:sectPr>.*?</w:sectPr>',
        new_sectpr, text, flags=re.DOTALL,
    )
    if n == 0:
        new_text = text.replace("</w:body>", new_sectpr + "</w:body>")
    document_xml.write_text(new_text, encoding="utf-8")
    log(f"  document.xml sectPr: pgSz/pgMar (A4, 2.2/2.5cm Raender), Header"
        f" default+first, Footer default+first auf gleicher rId, titlePg")

def add_header_footer(unpacked):
    word_dir = unpacked / "word"
    rels_xml = word_dir / "_rels" / "document.xml.rels"
    ct_xml   = unpacked / "[Content_Types].xml"
    doc_xml  = word_dir / "document.xml"

    write_xml_bytes(header_default_xml(),     word_dir / "header1.xml")
    write_xml_bytes(header_first_blank_xml(), word_dir / "header2.xml")
    write_xml_bytes(footer_default_xml(),     word_dir / "footer1.xml")
    log("  word/header1.xml (default), header2.xml (first blank),"
        " footer1.xml geschrieben")

    next_id = next_free_rel_id(rels_xml)
    rid_h_def, rid_h_first, rid_f_def = (f"rId{next_id+i}" for i in range(3))
    add_relationship(rels_xml, rid_h_def,   REL_HEADER, "header1.xml")
    add_relationship(rels_xml, rid_h_first, REL_HEADER, "header2.xml")
    add_relationship(rels_xml, rid_f_def,   REL_FOOTER, "footer1.xml")
    log(f"  Beziehungen: {rid_h_def}=header1, {rid_h_first}=header2,"
        f" {rid_f_def}=footer1")

    add_content_type_override(ct_xml, "/word/header1.xml", CT_HEADER)
    add_content_type_override(ct_xml, "/word/header2.xml", CT_HEADER)
    add_content_type_override(ct_xml, "/word/footer1.xml", CT_FOOTER)
    log("  [Content_Types].xml: Override-Eintraege fuer header1/2 und footer1")

    update_sectpr_with_headers(doc_xml, rid_h_def, rid_h_first, rid_f_def)

def main():
    log(f"Ziel: {OUTPUT_FILE}")
    TEMPLATES_DIR.mkdir(parents=True, exist_ok=True)

    with tempfile.TemporaryDirectory(prefix="refdocx-") as tmp:
        tmp_dir = Path(tmp)
        default_docx = tmp_dir / "pandoc-default.docx"
        unpacked     = tmp_dir / "unpacked"

        fetch_pandoc_default(default_docx)
        unpacked.mkdir()
        unpack_docx(default_docx, unpacked)

        theme_xml  = unpacked / "word" / "theme" / "theme1.xml"
        styles_xml = unpacked / "word" / "styles.xml"

        log("Anpassung: Theme major+minor auf Calibri")
        set_theme_fonts_to_calibri(theme_xml)
        log("Anpassung: Direkte Font-Referenzen in styles.xml -> Calibri")
        replace_direct_fonts_in_styles(styles_xml)
        log("Anpassung: Tabellen-Default ohne Rahmen")
        set_table_borders_none(styles_xml)
        log("Anpassung: Body-Schriftgroesse 11 pt (DocDefault)")
        set_default_body_size(styles_xml)
        log("Anpassung: Heading-Schriftgroessen 15/13/12 pt")
        set_heading_sizes(styles_xml)
        log("Anpassung: Widow/Orphan-Control im DocDefault (B3)")
        set_widow_control_default(styles_xml)
        log("Anpassung: keepNext + keepLines auf Heading 1/2/3 + FirstParagraph (B3)")
        set_keep_next_styles(styles_xml)
        log("Anpassung: Heading 1/2/3 in destengsblue (B4)")
        set_heading_colors(styles_xml)
        log("Anpassung: Heading 1+2 fett (B4-Bold, S10)")
        set_heading_bold(styles_xml)
        log("Anpassung: Header und Footer einbauen (B2)")
        add_header_footer(unpacked)

        log("Repack als reference.docx")
        repack_docx(unpacked, OUTPUT_FILE)
        log(f"  -> {OUTPUT_FILE} ({OUTPUT_FILE.stat().st_size} Bytes)")

    log("Fertig.")
    return 0

if __name__ == "__main__":
    sys.exit(main())