` mit `
`-Zeilen und `div.*programlisting* > span.pre` werden zu Codeblöcken. Literaturlisten mit eigenen Labels (`[BryL06]`) werden nicht doppelt nummeriert. -**Tipp:** Hat ein Buch Teile als `h1` und Kapitel als `h2` (typisch bei Romanen), erzeugt -`--chapter-level 2` eine Datei pro Kapitel. Mit dem Standard `1` entsteht eine Datei pro Teil. -Ist eine Kapiteldatei größer als 30 000 Wörter und hat sie mindestens zwei `##`-Unterabschnitte, meldet -epub2md eine Warnung `chapter-large` mit der passenden `--chapter-level`-Empfehlung. +**Teile und Kapitel:** Hat ein Buch Teile als `h1` und Kapitel als `h2` (typisch bei Romanen), teilt der +Standard `auto` jeden Teil mit über 30 000 Wörtern und mindestens zwei Kapiteln in eine kurze Datei für den +Teil (Überschrift, Einleitung) und eine Datei pro Kapitel; kleinere Kapitel (Vorwort, Anhang) bleiben ganz. +Verzeichnisse (Register, Literatur, Glossar) werden dabei nie zerlegt. `--chapter-level 2` teilt dagegen +jedes Kapitel des Buches an `h2`. Ist bei einer festen Zahl eine Kapiteldatei größer als 30 000 Wörter und +hat sie mindestens zwei `##`-Unterabschnitte, meldet epub2md eine Warnung `chapter-large` mit der +passenden `--chapter-level`-Empfehlung. ## Math-Modi @@ -309,7 +312,8 @@ verwendet und das Verzeichnis gibt den Namen vor. Beim Zusammenfügen passiert Folgendes: - Die Frontmatter der einzelnen Kapitel wird entfernt. -- Jedes Kapitel erhält einen Startanker (``). +- Jedes Kapitel erhält einen Startanker (``). Er beginnt mit einem Buchstaben, + weil Pandocs EPUB-Ausgabe Links auf IDs mit führender Ziffer nicht auflöst. - Links zwischen Kapiteln (`003-x.md#anker`) werden zu Sprunglinks in der Gesamtdatei. - Anker, die in mehreren Kapiteln vorkommen (z. B. `Sec1` in jedem Kapitel eines Springer-Buchs), werden ab dem zweiten Vorkommen in `c012-Sec1` umbenannt, samt aller Links darauf. @@ -387,9 +391,8 @@ Die `README.md` eines konvertierten Buches mit Kästen enthält die Befehle bere **PDF immer über HTML erzeugen.** Beim direkten Weg Markdown → LaTeX verwirft Pandoc allen rohen HTML-Code. Dabei gehen HTML-Tabellen samt Inhalt verloren, und ``-Sprungziele fehlen, sodass interne Links im PDF ins Leere zeigen. Liest Pandoc dagegen die HTML-Fassung, macht es daraus echte Tabellen und Sprungziele. -Getestet mit dem C-Buch (405 Seiten, 34 Tabellen, 728 von 755 internen Links funktionieren; die übrigen -stehen in Codeblöcken innerhalb von Tabellen). Die `README.md` jedes konvertierten Buches enthält den -passenden Befehl. +Getestet mit dem C-Buch (322 Seiten, 34 Tabellen; alle 831 internen Links haben ein Ziel, kein Zeichen +fehlt). Die `README.md` jedes konvertierten Buches enthält den passenden Befehl. Alternativ direkt aus den Kapiteldateien: @@ -404,8 +407,10 @@ pandoc output/programmierbuch/chapters/*.md \ - Ausgaben mit `--math tex` (`\(…\)`) brauchen `-f markdown+tex_math_single_backslash`. - Das YAML-Frontmatter jeder Kapiteldatei wird von Pandoc als Metadaten gelesen. Wer die Kapiteldateien direkt zusammen verarbeitet, sollte mit `--no-frontmatter` konvertieren oder die Gesamtdatei verwenden. -- Einzelne TeX-Befehle aus EPUBs (z. B. `\O` oder `@`-Konstrukte) kennt Pandocs Formelparser nicht. Er - belässt sie als TeX, mit der Warnung „Could not convert TeX math“. +- TeX aus dem E-Book wird für Pandocs Formelparser angepasst, ohne den Inhalt zu ändern: `\"u` → `ü`, + `\O` (leere Menge) → `\emptyset`, keine `@{…}`-Spaltentrenner und kein `\arraycolsep` in `array`, kein + leeres `\displaystyle` vor einem Zellenende. Seltene Handkonstruktionen (z. B. `\mathinner{…\raise…}`) + bleiben als TeX stehen, mit der Warnung „Could not convert TeX math“ (Mathebuch: 1 von 4986 Formeln). ## Bericht und Diagnose diff --git a/epub2md/__init__.py b/epub2md/__init__.py index 6885324..292d78c 100644 --- a/epub2md/__init__.py +++ b/epub2md/__init__.py @@ -1,4 +1,4 @@ """epub2md – konvertiert DRM-freie EPUB-Dateien in kapitelweise Markdown-Sammlungen.""" -__version__ = "0.2.0" +__version__ = "0.3.0" TOOL_NAME = "epub2md" diff --git a/epub2md/chapters.py b/epub2md/chapters.py index 2384eb7..8716c3b 100644 --- a/epub2md/chapters.py +++ b/epub2md/chapters.py @@ -105,9 +105,11 @@ def semantic_types(el: Any) -> list[str]: class ChapterBuilder: def __init__(self, chapter_level: int, nav: list[NavEntry], diag: Diagnostics, - book_title: str = "", authors: list[str] | None = None) -> None: + book_title: str = "", authors: list[str] | None = None, deep_starts: set[Any] | None = None) -> None: self.level = chapter_level self.diag = diag + self.deep_starts = deep_starts or set() # chapter headings whose sections are split one level deeper + self._deep = False # current chapter is one of them self._special_level = 0 # level of a split-out special heading whose section is still open self._in_special = False # current chapter itself is a bibliography/index/glossary self.book_title = book_title @@ -129,6 +131,12 @@ class ChapterBuilder: if lvl and has_content(el): special = _special_heading(el) if lvl <= self.level: + if update: + self._special_level = 0 + self._in_special = classify_title(_heading_text(el)) in SPECIAL_KINDS + self._deep = el in self.deep_starts + return Marker(el, "heading", lvl) + if self._deep and lvl == self.level + 1: if update: self._special_level = 0 self._in_special = classify_title(_heading_text(el)) in SPECIAL_KINDS @@ -196,7 +204,7 @@ class ChapterBuilder: body_segments = segments[1:] if _meaningful(head_items): first = _first_meaningful(head_items) - if chapters and leading_heading_level(first) > self.level and not semantic_types(doc.body): + if chapters and leading_heading_level(first) > self.level + self._deep and not semantic_types(doc.body): chapters[-1].items.extend(pending + head_items) pending = [] if doc.path not in chapters[-1].docs: diff --git a/epub2md/cli.py b/epub2md/cli.py index 21865a3..8066c45 100644 --- a/epub2md/cli.py +++ b/epub2md/cli.py @@ -46,6 +46,14 @@ def code_language(value: str) -> str: return lang +def _chapter_level(value: str) -> int | str: + if value == "auto": + return value + if value.isdigit() and 1 <= int(value) <= 6: + return int(value) + raise argparse.ArgumentTypeError("auto oder eine Zahl zwischen 1 und 6 erwartet") + + def build_parser() -> argparse.ArgumentParser: p = _Parser( prog="epub2md.py", @@ -61,9 +69,10 @@ def build_parser() -> argparse.ArgumentParser: help="Bilder/Medien nach OUTPUT/media/ extrahieren (Standard: an)") p.add_argument("--copy-original", action=argparse.BooleanOptionalAction, default=False, help="Original-EPUB nach OUTPUT/source/ kopieren (Standard: aus)") - p.add_argument("--chapter-level", type=int, default=1, metavar="N", - help="Überschriftenebene 1–6, ab der neue Kapiteldateien entstehen (Standard: 1). " - "Für Bücher mit Teilen (h1) und Kapiteln (h2) ist 2 sinnvoll") + p.add_argument("--chapter-level", type=_chapter_level, default="auto", metavar="{auto,1-6}", + help="Überschriftenebene, ab der neue Kapiteldateien entstehen. auto (Standard): Ebene 1, " + "nur sehr große Kapitel (über 30000 Wörter, z. B. Teile eines Romans) werden an ihren " + "Unterkapiteln geteilt. Eine Zahl 1–6 gilt ausnahmslos für das ganze Buch") p.add_argument("--toc", action=argparse.BooleanOptionalAction, default=True, help="README.md mit Inhaltsverzeichnis erzeugen (Standard: an)") p.add_argument("--frontmatter", action=argparse.BooleanOptionalAction, default=True, diff --git a/epub2md/mathml.py b/epub2md/mathml.py index 4687944..fdee4d8 100644 --- a/epub2md/mathml.py +++ b/epub2md/mathml.py @@ -365,7 +365,54 @@ def strip_tex_delimiters(text: str) -> str: if t.startswith(o) and t.endswith(c) and len(t) >= len(o) + len(c): t = t[len(o) : len(t) - len(c)] break - return t.strip() + return portable_tex(t.strip()) + + +UMLAUTS = {"a": "ä", "o": "ö", "u": "ü", "A": "Ä", "O": "Ö", "U": "Ü"} + + +def portable_tex(tex: str) -> str: + """Publisher TeX → TeX that Pandoc's formula parser (texmath, MathJax, KaTeX) understands, still valid LaTeX: + text accents \\"u → ü, \\O (empty set in Springer books) → \\emptyset, no @{…} column separators in + array specifications, no \\arraycolsep assignments, no empty \\displaystyle before a cell end. Only changes + spacing, never content.""" + tex = re.sub(r'\\"(?:\{([aouAOU])\}|([aouAOU]))', lambda m: UMLAUTS[m.group(1) or m.group(2)], tex) + tex = re.sub(r"\\(?:text|mbox|mathrm)\s*\{\s*\\O\s*\}", r"\\emptyset", tex) + tex = re.sub(r"\\O(?![A-Za-z])", r"\\emptyset", tex) + tex = re.sub(r"\\arraycolsep\s*-?\d*\.?\d+\s*(?:pt|em|ex|mm|cm)", "", tex) + tex = re.sub(r"\\displaystyle\s*(?=\\\\|&|\\end\b)", "", tex) # no-op before a cell end; texmath fails + out: list[str] = [] + pos = 0 + for m in re.finditer(r"\\begin\{array\}\{", tex): + if m.start() < pos: + continue + end = _closing_brace(tex, m.end() - 1) + if end < 0: + break + spec = tex[m.end():end] + i = spec.find("@{") + while i >= 0: + close = _closing_brace(spec, i + 1) + if close < 0: + break + spec = spec[:i] + spec[close + 1:] + i = spec.find("@{") + out.append(tex[pos:m.end()] + spec) + pos = end + return "".join(out) + tex[pos:] + + +def _closing_brace(text: str, open_pos: int) -> int: + """Index of the '}' matching the '{' at ``open_pos`` (-1 if unbalanced).""" + depth = 0 + for i in range(open_pos, len(text)): + if text[i] == "{": + depth += 1 + elif text[i] == "}": + depth -= 1 + if depth == 0: + return i + return -1 def is_display_tex(text: str) -> bool: diff --git a/epub2md/merge.py b/epub2md/merge.py index 5a052f9..3652e4d 100644 --- a/epub2md/merge.py +++ b/epub2md/merge.py @@ -188,7 +188,7 @@ def assign_anchors(chapters: list[ChapterFile]) -> int: for m in TOKEN.finditer(seg): if m.group("attr") and m.group("aname") == "id" and m.group("avalue") not in ids: ids.append(m.group("avalue")) - start = stems[ch.rel] + start = "kap-" + stems[ch.rel] # ids must start with a letter (Pandoc's EPUB writer breaks '#039-…') while start in used or start in ids: start = "kapitel-" + start ch.start_anchor = start diff --git a/epub2md/pipeline.py b/epub2md/pipeline.py index 9f1f68e..45500bc 100644 --- a/epub2md/pipeline.py +++ b/epub2md/pipeline.py @@ -16,7 +16,7 @@ from typing import Any from urllib.parse import unquote from . import TOOL_NAME, __version__ -from .chapters import ChapterBuilder +from .chapters import SPECIAL_KINDS, ChapterBuilder, heading_level from .diagnostics import Diagnostics from .css import StyleSheet, base_style, emphasis from .epub_reader import EpubReader, decode_markup, parse_xml @@ -26,7 +26,8 @@ from .html_converter import RenderOptions, Renderer from .linkcheck import check_links from .media import MediaRegistry from .models import Chapter -from .normalize import Doc, book_uses_title_tables, is_html_toc_doc, make_doc, normalize_doc, remove_keep_tail +from .normalize import (Doc, book_uses_title_tables, has_content, is_html_toc_doc, make_doc, normalize_doc, + remove_keep_tail) from .references import LinkResolver from .utils import is_element, is_inside, iter_elements, local, resolve_href, sha256_file, slugify, text_content from . import writer @@ -40,7 +41,7 @@ class Options: format: str = "gfm" extract_media: bool = True copy_original: bool = False - chapter_level: int = 1 + chapter_level: int | str = "auto" # 1–6, or "auto": level 1 plus oversized chapters split one level deeper toc: bool = True frontmatter: bool = True footnotes: str = "end-of-chapter" @@ -81,8 +82,8 @@ def check_paths(input_path: Path, output_dir: Path, opts: Options) -> None: raise InputError(f"Eingabedatei nicht gefunden oder keine Datei: {input_path}") if not os.access(input_path, os.R_OK): raise InputError(f"Eingabedatei nicht lesbar: {input_path}") - if not 1 <= opts.chapter_level <= 6: - raise InputError("--chapter-level muss zwischen 1 und 6 liegen.") + if opts.chapter_level != "auto" and not (isinstance(opts.chapter_level, int) and 1 <= opts.chapter_level <= 6): + raise InputError("--chapter-level muss auto oder eine Zahl zwischen 1 und 6 sein.") if opts.dry_run: return if output_dir.exists() and not output_dir.is_dir(): @@ -214,10 +215,9 @@ def _prepare_docs(docs: list[Doc], package: Any, diag: Diagnostics, def _build_structure(docs: list[Doc], package: Any, opts: Options, diag: Diagnostics) -> tuple[Any, list[Chapter], LinkResolver]: """Notes, chapter units with file names, and the link resolver (incl. NAV/NCX targets).""" - md = package.metadata notes = collect_notes(docs, opts.footnotes, diag) diag.step(f"{len(notes.notes)} Anmerkungen erkannt") - chapters = ChapterBuilder(opts.chapter_level, package.toc, diag, md.title, md.authors).build(docs) + chapters = _build_chapters(docs, package, opts, diag) _assign_filenames(chapters, opts.slug_language) resolver = LinkResolver(docs, chapters, notes, "README.md" if opts.toc else None) nav_targets: list[tuple[str, str]] = [] @@ -296,7 +296,7 @@ def _render(reader: EpubReader, package: Any, chapters: list[Chapter], resolver: diag.step(f"Rendere {ch.filename}") ch.markdown = renderer.render_chapter(ch) ch.stats.word_count = _word_count(ch.markdown) - _check_chapter_size(ch, opts.chapter_level, diag) + _check_chapter_size(ch, _base_level(opts), diag) _report_wrap_markers(renderer, diag) if not opts.code_language: _suggest_code_language(renderer, diag) @@ -433,16 +433,52 @@ def _assign_filenames(chapters: list[Chapter], slug_language: str) -> None: ch.filename = f"chapters/{candidate}.md" +def _base_level(opts: Options) -> int: + return 1 if opts.chapter_level == "auto" else int(opts.chapter_level) + + +def _build_chapters(docs: list[Doc], package: Any, opts: Options, diag: Diagnostics) -> list[Chapter]: + """Chapter units at --chapter-level. With "auto", chapters over LARGE_CHAPTER_WORDS with at least two + sub-headings (parts of a novel, say) are split one level deeper; all other chapters stay whole.""" + md = package.metadata + level = _base_level(opts) + if opts.chapter_level == "auto" and level < 6: + probe = ChapterBuilder(level, package.toc, Diagnostics(), md.title, md.authors).build(docs) + deep = {ch.title_el for ch in probe if _oversized(ch, level)} + if deep: + titles = ", ".join(f"„{ch.title}“" for ch in probe if ch.title_el in deep) + diag.info("chapter-split", f"Sehr große Kapitel ({titles}) werden an ihren Überschriften der Ebene " + f"{level + 1} aufgeteilt (--chapter-level auto).") + return ChapterBuilder(level, package.toc, diag, md.title, md.authors, deep_starts=deep).build(docs) + return ChapterBuilder(level, package.toc, diag, md.title, md.authors).build(docs) + + +def _oversized(ch: Chapter, level: int) -> bool: + """Heading-started chapter (not glossary/index/…) with too many words and >= 2 next-level sub-headings.""" + if ch.title_el is None or ch.title_level > level or SPECIAL_KINDS & set(ch.types): + return False + words = 0 + sections = 0 + for _, it in ch.items: + if isinstance(it, str): + words += len(it.split()) + elif is_element(it): + words += len(text_content(it).split()) + sections += sum(1 for e in iter_elements(it) if heading_level(e) == level + 1 and has_content(e)) + return words > LARGE_CHAPTER_WORDS and sections >= 2 + + def _check_chapter_size(ch: Chapter, chapter_level: int, diag: Diagnostics) -> None: """Suggest a deeper --chapter-level for very large chapters with enough sub-headings.""" if ch.stats.word_count <= LARGE_CHAPTER_WORDS or chapter_level >= 6: return sections = len(re.findall(r"^## ", ch.markdown, re.M)) if sections >= 2: + deeper = max(chapter_level, ch.title_level or 1) + 1 diag.warn( "chapter-large", f"Kapitel „{ch.title}“ hat {ch.stats.word_count} Wörter und {sections} Unterabschnitte; " - f"--chapter-level {chapter_level + 1} würde sie als eigene Dateien ausgeben.", + f"--chapter-level {deeper} würde sie als eigene Dateien ausgeben.", ch.docs[0] if ch.docs else "", ch.filename, ) diff --git a/epub2md/writer.py b/epub2md/writer.py index cf4333b..061326e 100644 --- a/epub2md/writer.py +++ b/epub2md/writer.py @@ -157,7 +157,7 @@ def readme(md: BookMetadata, chapters: list[Chapter], endnotes: Chapter | None, f"python3 epub2md_merge.py {shlex.quote(output_dir)}", "# danach in diesem Ordner:", f"pandoc {md_} -s{math}{toc}{lua} -o {html_}", - f"pandoc {md_} --toc{lua} -o {shlex.quote(stem + '.epub')}", + f"pandoc {md_} --toc{math}{lua} -o {shlex.quote(stem + '.epub')}", ] + (latex + weasy if math else weasy + latex) out.append("## Weiterverarbeitung") out.append("```bash\n" + "\n".join(commands) + "\n```") @@ -166,7 +166,7 @@ def readme(md: BookMetadata, chapters: list[Chapter], endnotes: Chapter | None, out.append(f"Das Buch enthält {callouts} {noun} (`> [!TIP]` …). Der Filter `callouts.lua` macht daraus " "in HTML, EPUB und PDF gerahmte Kästen mit Titel; ohne ihn zeigt Pandoc „[!TIP]“ als Text.") if stats.get("formulas"): - out.append("Das Buch enthält Formeln: Für HTML `--mathml` (oder `--mathjax`) verwenden, sonst belässt " + out.append("Das Buch enthält Formeln: Für HTML und EPUB `--mathml` (bei HTML auch `--mathjax`) verwenden, sonst belässt " "Pandoc Matrizen und ähnliche Konstrukte als TeX-Text.") out.append("## Reproduktion") out.append("```bash\n" + " ".join(shlex.quote(c) for c in command) + "\n```") diff --git a/pandoc/book.lua b/pandoc/book.lua index 67b76b3..06a189c 100644 --- a/pandoc/book.lua +++ b/pandoc/book.lua @@ -11,7 +11,7 @@ This filter, for HTML output: * builds the table of contents at the placeholder (use it INSTEAD of --toc; --toc-depth is honoured), * drops Pandoc's title block when the book has its own title page, or moves it to the placeholder, * shows the cover (first chapter with nothing but one picture) without the caption "Cover", - * turns the chapter anchors intoso that print.css can + * turns the chapter anchors intoso that print.css can start every chapter – including cover, title page and imprint – on a new page. For LaTeX (PDF route md -> html -> pdf, filter in BOTH steps, no --toc) the table of contents becomes @@ -55,7 +55,7 @@ local function title_block(meta) return pandoc.RawBlock("html", table.concat(html, "\n")) end --- Paragraph that only holds chapter start anchors () → id of the first one. +-- Paragraph that only holds chapter start anchors () → id of the first one. local function chapter_anchor(block) if block.t ~= "Para" and block.t ~= "Plain" then return nil @@ -65,7 +65,8 @@ local function chapter_anchor(block) if inl.t ~= "RawInline" then return nil end - local found = inl.text:match('^$') or inl.text:match('^$') + local found = inl.text:match('^$') or inl.text:match('^$') + or inl.text:match('^$') or inl.text:match('^$') -- before 0.3.0 if found then id = id or found elseif inl.text ~= "" then @@ -143,6 +144,7 @@ local function latex_book(doc) blocks:insert(pandoc.RawBlock("latex", "\\clearpage")) blocks:insert(block) elseif block.t == "Div" and block.classes:includes("cover") then + blocks:insert(pandoc.RawBlock("latex", "\\thispagestyle{empty}")) -- no page number on the cover blocks:insert(block:walk({ Image = function(img) -- cover as large as the page allows img.attributes.width = nil img.attributes.height = "20cm" diff --git a/pandoc/print.css b/pandoc/print.css index e68f52d..2bea7b8 100644 --- a/pandoc/print.css +++ b/pandoc/print.css @@ -3,6 +3,7 @@ --pdf-engine=weasyprint -o buch.pdf book.lua places the table of contents after cover, title page and imprint and marks chapter starts. */ @page { size: A4; margin: 2cm 1.8cm; @bottom-center { content: counter(page); font-size: 9pt; } } +@page :first { @bottom-center { content: none; } } /* no page number on the cover / title page */ body { max-width: none; font-size: 10.5pt; line-height: 1.4; hyphens: auto; text-align: justify; font-family: "DejaVu Serif", "Liberation Serif", Georgia, serif; } #title-block-header, #title-block-header h1.title { text-align: center; margin-bottom: 2em; } diff --git a/tests/test_chapters.py b/tests/test_chapters.py index 75d71fd..befe5d6 100644 --- a/tests/test_chapters.py +++ b/tests/test_chapters.py @@ -127,7 +127,7 @@ def test_glossary_anchor_is_placed_before_bold_term(convert): def test_large_chapter_with_subheadings_suggests_deeper_chapter_level(convert): words = " ".join(["Wort"] * 11000) body = "Teil 1
" + "".join(f"Kapitel {i}
{words}
" for i in range(3)) - run = convert(Book(docs=[Doc("t.xhtml", body)])) + run = convert(Book(docs=[Doc("t.xhtml", body)]), "--chapter-level", "1") hints = [w for w in run.report["warnings"] if w["code"] == "chapter-large"] assert len(hints) == 1 assert "--chapter-level 2" in hints[0]["message"] and "3 Unterabschnitte" in hints[0]["message"] @@ -135,6 +135,33 @@ def test_large_chapter_with_subheadings_suggests_deeper_chapter_level(convert): assert not any(w["code"] == "chapter-large" for w in small.report["warnings"]) +def test_auto_level_splits_only_oversized_chapters(convert): + """Default --chapter-level auto: a part over 30000 words is split at its h2, other chapters stay whole (R2).""" + words = " ".join(["Wort"] * 11000) + docs = [ + Doc("v.xhtml", "Vorwort
kurz
Dank
danke
Hinweise
lesen
"), + Doc("t1.xhtml", "Teil 1
Einleitung
"), + *[Doc(f"k{i}.xhtml", f"Kapitel {i}
{words}
Abschnitt
x
") for i in range(1, 4)], + Doc("t2.xhtml", "Teil 2
Kapitel 4
kurz
Kapitel 5
kurz
"), + ] + run = convert(Book(docs=docs)) + assert [p.name for p in run.chapters()] == [ + "001-vorwort.md", "002-teil-1.md", "003-kapitel-1.md", "004-kapitel-2.md", "005-kapitel-3.md", "006-teil-2.md"] + assert "## Dank" in run.chapter("001") and "## Kapitel 5" in run.chapter("006") + assert "Einleitung" in run.chapter("002") and "## Abschnitt" in run.chapter("003") + assert run.chapter("003").count("\n# ") == 1 + assert not any(w["code"] == "chapter-large" for w in run.report["warnings"]) + assert any(i["code"] == "chapter-split" for i in run.report["infos"]) + assert run.report["options"]["chapter_level"] == "auto" + explicit = convert(Book(docs=docs), "--chapter-level", "1") + assert len(explicit.chapters()) == 3 + + +def test_chapter_level_rejects_invalid_values(convert): + for value in ("0", "7", "x"): + assert convert(Book(docs=[Doc("a.xhtml", "A
")]), "--chapter-level", value).rc == 1 + + def test_titles_derived_from_content_when_no_other_source(convert): docs = [ Doc("ccover.xhtml", ''), diff --git a/tests/test_math.py b/tests/test_math.py index a89df9c..1266d5a 100644 --- a/tests/test_math.py +++ b/tests/test_math.py @@ -2,7 +2,7 @@ import pytest from lxml import etree from conftest import PNG, SVG, Book, Doc -from epub2md.mathml import Unsupported, mathml_to_tex +from epub2md.mathml import Unsupported, mathml_to_tex, portable_tex M = 'xmlns="http://www.w3.org/1998/Math/MathML"' @@ -111,6 +111,24 @@ def test_book_readme_suggests_mathml_for_pandoc_only_with_formulas(convert): readme = (run.out / "README.md").read_text(encoding="utf-8") assert "## Weiterverarbeitung" in readme and "epub2md_merge.py" in readme assert "--mathml" in readme + pandoc_lines = [ln for ln in readme.splitlines() if ln.startswith("pandoc ")] + assert pandoc_lines and all("--mathml" in ln for ln in pandoc_lines) # EPUB too, else TeX stays text (QA3-04) plain = convert(Book(docs=[Doc("a.xhtml", "A
Text
")])) plain_readme = (plain.out / "README.md").read_text(encoding="utf-8") assert "## Weiterverarbeitung" in plain_readme and "--mathml" not in plain_readme + + +@pytest.mark.parametrize("tex,expected", [ + (r'\mbox{gr\"un}, \text{F\"{u}r}', r"\mbox{grün}, \text{Für}"), + (r'\text{ f{\"u}r } j', r"\text{ f{ü}r } j"), # braces stay balanced + (r"{\mathscr L}(A,b)={{\O }}", r"{\mathscr L}(A,b)={{\emptyset }}"), + (r"M\cap N=\text{\O }", r"M\cap N=\emptyset"), + (r"\Omega\neq\O", r"\Omega\neq\emptyset"), + (r"\begin{array}{@{}l@{\qquad \quad }l@{}} a & b\end{array}", r"\begin{array}{ll} a & b\end{array}"), + (r"\arraycolsep8pt\begin{array}{c@{\,}r} x\end{array}", r"\begin{array}{cr} x\end{array}"), + (r"a & =&\displaystyle b, \\ & \vdots &\displaystyle \\ c", r"a & =&\displaystyle b, \\ & \vdots &\\ c"), + (r"\frac{a}{b}+\{x\}", r"\frac{a}{b}+\{x\}"), +]) +def test_publisher_tex_is_made_portable(tex, expected): + """Pandoc's texmath rejects \\", \\O, @{…} and \\arraycolsep (Springer alt texts, QA3-06).""" + assert portable_tex(tex) == expected diff --git a/tests/test_merge.py b/tests/test_merge.py index 1a2d122..8277259 100644 --- a/tests/test_merge.py +++ b/tests/test_merge.py @@ -34,10 +34,10 @@ def test_merge_complete_book(convert, tmp_path): # links between chapters point into the merged file; colliding id "sec" renamed in chapter B assert '' in text and '' in text assert "[B](#c002-sec)" in text and "[A](#sec)" in text and "[hier](#sec)" in text - assert "[Kapitel B](#002-kapitel-b)" in text and '' in text + assert "[Kapitel B](#kap-002-kapitel-b)" in text and '' in text # media paths relative to the merged file, also inside image links assert "" in text - assert "[](#002-kapitel-b)" in text + assert "[](#kap-002-kapitel-b)" in text # code untouched assert '[kein](link.md) id="sec" `code`' in text # footnote defined in both chapters -> only once in the merged file @@ -63,8 +63,8 @@ def test_merge_to_other_location_rewrites_media_paths(convert, tmp_path): text = target.read_text(encoding="utf-8") assert not text.startswith("---") assert f"" in text - assert "1. [Kapitel A](#001-kapitel-a)" in text - assert "001-kapitel-a" in anchors_in(text) + assert "1. [Kapitel A](#kap-001-kapitel-a)" in text + assert "kap-001-kapitel-a" in anchors_in(text) def test_merge_invalid_directory(tmp_path): diff --git a/tests/test_pandoc.py b/tests/test_pandoc.py new file mode 100644 index 0000000..4f4e00c --- /dev/null +++ b/tests/test_pandoc.py @@ -0,0 +1,60 @@ +"""Pandoc path: merged file → EPUB/HTML with the filters from pandoc/ (skipped without pandoc).""" + +import re +import shutil +import subprocess +import zipfile +from pathlib import Path + +import pytest + +from conftest import Book, Doc +from epub2md.merge import main as merge_main + +pytestmark = pytest.mark.skipif(shutil.which("pandoc") is None, reason="pandoc nicht installiert") +FILTERS = Path(__file__).resolve().parent.parent / "pandoc" + + +def merged(convert, book: Book) -> Path: + run = convert(book) + assert merge_main([str(run.out)]) == 0 + return run.out / f"{run.report['input']['name'][:-5]}.md" + + +def pandoc(md: Path, *args: str) -> subprocess.CompletedProcess: + return subprocess.run(["pandoc", md.name, *args], cwd=md.parent, capture_output=True, text=True, check=True) + + +def links_and_ids(epub: Path) -> tuple[list[tuple[str, str]], dict[str, set[str]]]: + with zipfile.ZipFile(epub) as z: + docs = {n: z.read(n).decode("utf-8") for n in z.namelist() if n.endswith(".xhtml")} + ids = {n: set(re.findall(r'\bid="([^"]+)"', t)) for n, t in docs.items()} + links = [(n, h) for n, t in docs.items() for h in re.findall(r'href="([^"]+)"', t) + if not re.match(r"^[a-z]+:", h) and not h.endswith(".css")] + return links, ids + + +def test_epub_links_to_chapter_starts_resolve(convert): + """Chapter start anchors must begin with a letter: Pandoc's EPUB writer breaks '#039-…' (QA3-05).""" + docs = [Doc("a.xhtml", 'Kapitel A
'), + Doc("b.xhtml", "Kapitel B
Text B.
"), + Doc("c.xhtml", "Nur eine Bildseite ohne Überschrift.
")] + md = merged(convert, Book(docs=docs)) + pandoc(md, "--toc", "-o", "buch.epub") + links, ids = links_and_ids(md.parent / "buch.epub") + broken = [] + for doc, href in links: + file, _, frag = href.partition("#") + target = str(Path(doc).parent / file).replace("\\", "/") if file else doc + target = re.sub(r"[^/]+/\.\./", "", target) + if target not in ids or (frag and frag not in ids[target]): + broken.append(f"{doc}: {href}") + assert not broken + + +def test_book_filter_turns_chapter_anchors_into_chapter_starts(convert): + docs = [Doc("a.xhtml", "Kapitel A
a
"), Doc("b.xhtml", "Kapitel B
b
")] + md = merged(convert, Book(docs=docs)) + html = pandoc(md, "-s", "--lua-filter", str(FILTERS / "book.lua"), "-t", "html").stdout + assert html.count('class="chapter-start"') == 2 + assert 'id="TOC"' in html and "Kapitel B" in html
