pdf2md/tests/test_split.py

"""--split, --from-md, YAML-Kopf und --jpeg-quality."""

import zipfile

import pymupdf

import pytest

import pdf2md as p
from conftest import needs_pandoc
from test_e2e import run

BOOK = """# Das Buch

## Inhaltsverzeichnis

Teil I

1 Anfang

## Teil I: Der Anfang

Einleitung zum Teil.

### 1 Anfang

Text eins.

```
# kein Kapitel, nur ein Kommentar
```

### 2 Weiter

Text zwei.

## Impressum

Alle Rechte vorbehalten.
"""


def test_frontmatter_round_trip():
    meta = {"title": "Ein \"Buch\": Teil 1", "author": "Max Muster", "chapter": 3, "leer": ""}
    text = p.make_frontmatter(meta) + "# Text\n"
    parsed, body = p.split_frontmatter(text)
    assert parsed == {"title": "Ein \"Buch\": Teil 1", "author": "Max Muster", "chapter": 3}
    assert body == "# Text\n"


def test_text_without_frontmatter_is_unchanged():
    assert p.split_frontmatter("---\n\nText\n") == ({}, "---\n\nText\n")


def test_slugify():
    assert p.slugify("2 Ein Brief – eine Liste – ein Vermächtnis") == "2-ein-brief-eine-liste-ein-vermaechtnis"
    assert p.slugify("Größe & Maß") == "groesse-mass"
    assert p.slugify("!!!") == "abschnitt"


def test_split_by_chapters():
    sections = p.split_markdown(BOOK, "kapitel")
    assert [s["title"] for s in sections] == [
        "Vorspann", "Teil I: Der Anfang", "1 Anfang", "2 Weiter", "Impressum",
    ]
    assert sections[0]["body"].startswith("# Das Buch") and "## Inhaltsverzeichnis" in sections[0]["body"]
    assert sections[2]["part"] == "Teil I: Der Anfang" and sections[2]["chapter"] == 1
    assert "# kein Kapitel" in sections[2]["body"]  # Codeblock wird nicht geschnitten
    assert "".join(s["body"] for s in sections).replace("\n", "") == BOOK.replace("\n", "")


def test_split_by_parts():
    sections = p.split_markdown(BOOK, "teile")
    assert [s["title"] for s in sections] == ["Vorspann", "Teil I: Der Anfang", "Impressum"]
    assert "### 2 Weiter" in sections[1]["body"]


def test_write_split_and_read_back(tmp_path):
    sections = p.split_markdown(BOOK, "kapitel")
    files = p.write_split(tmp_path / "buch", sections, {"title": "Das Buch", "author": "Max"})
    assert [f.name for f in files][:3] == ["00_vorspann.md", "01_teil-i-der-anfang.md", "02_1-anfang.md"]
    meta, body, base = p.read_markdown_source(tmp_path / "buch")
    assert meta["title"] == "Das Buch" and base == tmp_path / "buch"
    assert body.replace("\n", "") == BOOK.replace("\n", "")
    head, _ = p.split_frontmatter(files[2].read_text())
    assert head == {"title": "1 Anfang", "part": "Teil I: Der Anfang", "chapter": 1,
                    "order": 2, "book": "Das Buch", "author": "Max"}


def test_missing_index_entry_is_reported(tmp_path):
    folder = tmp_path / "buch"
    p.write_split(folder, p.split_markdown(BOOK, "teile"), {"title": "X"})
    (folder / "01_teil-i-der-anfang.md").unlink()
    with pytest.raises(RuntimeError, match="nicht vorhanden"):
        p.read_markdown_source(folder)


# --- Ende zu Ende ------------------------------------------------------------------

def test_cli_split_writes_folder(pdfs, tmp_path):
    run(pdfs["digital"], tmp_path / "buch.md", "--split", "kapitel", "--images")
    folder = tmp_path / "buch"
    names = sorted(f.name for f in folder.glob("*.md"))
    assert names[0] == "00_vorspann.md" and "index.md" in names
    assert any(n.endswith("_1-der-notar.md") for n in names)
    assert (folder / "images" / "cover.jpg").is_file()
    text = "".join(f.read_text() for f in folder.glob("*.md"))
    assert "](images/" in text and "_images/" not in text


@needs_pandoc
def test_from_md_folder_follows_index_order(pdfs, tmp_path):
    run(pdfs["digital"], tmp_path / "buch.md", "--split", "kapitel", "--epub")
    index = tmp_path / "buch" / "index.md"
    lines = index.read_text().split("\n")
    i = next(n for n, line in enumerate(lines) if "1-der-notar" in line)
    j = next(n for n, line in enumerate(lines) if "2-ein-brief" in line)
    lines[i], lines[j] = lines[j], lines[i]
    index.write_text("\n".join(lines))

    run("--from-md", tmp_path / "buch", "--epub-file", tmp_path / "neu.epub")
    with zipfile.ZipFile(tmp_path / "neu.epub") as z:
        nav = z.read("EPUB/nav.xhtml").decode()
        opf = z.read("EPUB/content.opf").decode()
    assert nav.index("2 Ein Brief") < nav.index("1 Der Notar")
    assert "Erika Muster" in opf and "cover" in opf


@needs_pandoc
def test_from_md_file_uses_frontmatter_and_overrides(pdfs, tmp_path):
    run(pdfs["digital"], tmp_path / "buch.md", "--images")
    md = tmp_path / "buch.md"
    md.write_text(md.read_text().replace("Der Regen", "Der starke Regen"))
    run("--from-md", md, "--title", "Neuer Titel")
    with zipfile.ZipFile(tmp_path / "buch.epub") as z:
        opf = z.read("EPUB/content.opf").decode()
        text = "".join(z.read(n).decode() for n in z.namelist() if n.endswith(".xhtml"))
    assert ">Neuer Titel<" in opf and "Erika Muster" in opf and "cover" in opf
    assert "Der starke Regen" in text and "title:" not in text


def test_jpeg_quality_changes_image_size(pdfs, tmp_path):
    def size(quality):
        out = tmp_path / f"q{quality}" / "buch.md"
        run(pdfs["digital"], out, "--images", "--jpeg-quality", str(quality))
        return sum(f.stat().st_size for f in (out.parent / "buch_images").glob("*.jpg"))

    assert size(30) < size(95)


# --- Seitenzahlen, Teilen nach Länge -----------------------------------------------

def _book_pdf(outline: bool):
    doc = pymupdf.open()
    texts = [
        "Inhalt\n1 Anfang\n2 Weiter\n3 Schluss",
        "1 Anfang\nText eins.",
        "Noch mehr Text\nzum Anfang,\nüber mehrere\nZeilen.\n2 Weiter\nText zwei.",
        "3 Schluss\nText drei.",
    ]
    for text in texts:
        doc.new_page().insert_text((72, 72), text, fontsize=11)
    if outline:
        doc.set_toc([[1, "1 Anfang", 2], [1, "2 Weiter", 3], [1, "3 Schluss", 4]])
    return doc


@pytest.mark.parametrize("outline", [True, False])
def test_source_pages(outline):
    sections = [{"title": t} for t in ("Vorspann", "1 Anfang", "2 Weiter", "3 Schluss")]
    p.add_source_pages(_book_pdf(outline), sections)
    assert [s["source_pages"] for s in sections] == [[1, 1], [2, 3], [3, 3], [4, 4]]


def test_split_without_headings_by_length():
    paragraphs = [f"Absatz {n} " + " ".join(["Text"] * 40) for n in range(30)]
    code = "```\n" + "\n\n".join(f"zeile_{n} = {n}" for n in range(20)) + "\n```"
    markdown = "\n\n".join(paragraphs[:15] + [code] + paragraphs[15:]) + "\n"
    assert p.split_markdown(markdown, "kapitel")[0]["title"] == "Abschnitt 1"
    sections = p.split_by_length(markdown, size=1500)
    assert len(sections) > 3
    assert [s["title"] for s in sections[:2]] == ["Abschnitt 1", "Abschnitt 2"]
    assert all(s["body"].count("```") in (0, 2) for s in sections)  # Code bleibt ganz
    assert "".join(s["body"] for s in sections).replace("\n", "") == markdown.replace("\n", "")


def test_cli_split_writes_source_pages(pdfs, tmp_path):
    run(pdfs["digital"], tmp_path / "buch.md", "--split", "kapitel")
    files = {f.name: p.split_frontmatter(f.read_text())[0] for f in (tmp_path / "buch").glob("0*.md")}
    notar = next(meta for name, meta in files.items() if name.endswith("_1-der-notar.md"))
    assert notar["source_pages"] == [2, 3]


# --- Links mit eigenem Text --------------------------------------------------------

def _link_pdf(path):
    doc = pymupdf.open()
    page = doc.new_page()
    page.insert_text((72, 72), "Den ganzen Bericht finden Sie hier im Netz.", fontsize=11)
    page.insert_text((72, 90), "Siehe auch www.example.org/info zum Thema.", fontsize=11)
    for word, uri in (("hier", "https://beispiel.de/bericht?utm_source=x"),
                      ("www.example.org/info", "https://www.example.org/info")):
        rect = page.search_for(word)[0]
        page.insert_link({"kind": pymupdf.LINK_URI, "from": rect, "uri": uri})
    doc.save(path)
    return path


def test_load_link_anchors_skips_url_texts(tmp_path):
    doc = pymupdf.open(_link_pdf(tmp_path / "links.pdf"))
    assert p.load_link_anchors(doc) == [
        ("https://beispiel.de/bericht", ["Bericht", "finden", "Sie"], ["hier"])
    ]


def test_link_anchors_uses_context_and_skips_existing_links():
    anchors = [("https://a.de/x", ["finden", "Sie"], ["hier"])]
    md = "Hier steht hier. Den Bericht finden Sie hier im Netz.\n"
    assert p.link_anchors(md, anchors) == (
        "Hier steht hier. Den Bericht finden Sie [hier](https://a.de/x) im Netz.\n"
    )
    # Fußnotenziffer und Trennung zwischen Kontext und Linktext stören nicht
    anchors = [("https://b.de/", ["überprüfen.81"], ["Landes-", "medienanstalten"])]
    md = "Die Landesmedienanstalten.\n\nzu überprüfen.[81 ] Die Landesmedienanstalten prüfen.\n"
    anchors = [("https://b.de/", ["überprüfen.81", "Die"], ["Landes-", "medienanstalten"])]
    assert "Die [Landesmedienanstalten](https://b.de/) prüfen" in p.link_anchors(md, anchors)


def test_cli_sets_anchor_links(tmp_path):
    out = tmp_path / "links.md"
    run(_link_pdf(tmp_path / "links.pdf"), out)
    md = out.read_text()
    assert "[hier](https://beispiel.de/bericht)" in md
    assert "<https://www.example.org/info>" in md or "[www.example.org/info]" in md