gwoe-antragspruefer/tests/test_redline_parser.py

"""Tests für _parse_redline_segments aus app.main.

Prüft alle Marker-Formate (§INS§/§DEL§ und **/**+~~), Edge-Cases und
gemischte Eingaben. Keine DB- oder HTTP-Abhängigkeiten.
"""
from __future__ import annotations

import pytest

from app.redline_utils import parse_redline_segments as _parse_redline_segments


# ─────────────────────────────────────────────────────────────────────────────
# Hilfsfunktionen
# ─────────────────────────────────────────────────────────────────────────────

def types(segments):
    return [s["type"] for s in segments]

def texts(segments):
    return [s["text"] for s in segments]


# ─────────────────────────────────────────────────────────────────────────────
# Basis-Parsing
# ─────────────────────────────────────────────────────────────────────────────

class TestBasicParsing:
    def test_empty_string_returns_empty(self):
        assert _parse_redline_segments("") == []

    def test_none_returns_empty(self):
        assert _parse_redline_segments(None) == []  # type: ignore[arg-type]

    def test_plain_text_is_ctx(self):
        segs = _parse_redline_segments("kein Marker hier")
        assert types(segs) == ["ctx"]
        assert texts(segs) == ["kein Marker hier"]

    def test_ins_marker_tag_format(self):
        segs = _parse_redline_segments("§INS§neuer Text§INS§")
        assert types(segs) == ["ins"]
        assert texts(segs) == ["neuer Text"]

    def test_del_marker_tag_format(self):
        segs = _parse_redline_segments("§DEL§alter Text§DEL§")
        assert types(segs) == ["del"]
        assert texts(segs) == ["alter Text"]

    def test_markdown_bold_becomes_ins(self):
        segs = _parse_redline_segments("**eingefügt**")
        assert types(segs) == ["ins"]
        assert texts(segs) == ["eingefügt"]

    def test_markdown_strikethrough_becomes_del(self):
        segs = _parse_redline_segments("~~gestrichen~~")
        assert types(segs) == ["del"]
        assert texts(segs) == ["gestrichen"]


# ─────────────────────────────────────────────────────────────────────────────
# Kontext + Marker gemischt
# ─────────────────────────────────────────────────────────────────────────────

class TestMixedContent:
    def test_ctx_ins_ctx(self):
        segs = _parse_redline_segments("§ 3 Abs. 2 §INS§verpflichtend§INS§ ab 2026")
        assert types(segs) == ["ctx", "ins", "ctx"]
        assert texts(segs)[1] == "verpflichtend"

    def test_ctx_del_ins(self):
        segs = _parse_redline_segments("Text §DEL§alt§DEL§§INS§neu§INS§ Ende")
        assert types(segs) == ["ctx", "del", "ins", "ctx"]

    def test_markdown_mixed(self):
        segs = _parse_redline_segments("Vor ~~weg~~ und **rein** nach")
        assert types(segs) == ["ctx", "del", "ctx", "ins", "ctx"]

    def test_both_formats_in_one_string(self):
        segs = _parse_redline_segments("§DEL§raus§DEL§ und **rein**")
        assert "del" in types(segs)
        assert "ins" in types(segs)


# ─────────────────────────────────────────────────────────────────────────────
# Edge-Cases
# ─────────────────────────────────────────────────────────────────────────────

class TestEdgeCases:
    def test_empty_ins_marker(self):
        segs = _parse_redline_segments("§INS§§INS§")
        # Leerer ins-Marker bleibt ein ins-Segment mit leerem Text
        assert any(s["type"] == "ins" for s in segs)

    def test_empty_del_marker(self):
        segs = _parse_redline_segments("§DEL§§DEL§")
        assert any(s["type"] == "del" for s in segs)

    def test_unbalanced_marker_treated_as_ctx(self):
        # Nur ein §INS§ ohne schließenden Partner → kein ins-Segment
        segs = _parse_redline_segments("§INS§unvollständig")
        assert all(s["type"] == "ctx" for s in segs)

    def test_marker_with_whitespace_only(self):
        segs = _parse_redline_segments("§INS§   §INS§")
        ins_segs = [s for s in segs if s["type"] == "ins"]
        assert len(ins_segs) == 1
        assert ins_segs[0]["text"].strip() == ""

    def test_multiple_ins_markers(self):
        segs = _parse_redline_segments("§INS§A§INS§ und §INS§B§INS§")
        ins_texts = [s["text"] for s in segs if s["type"] == "ins"]
        assert ins_texts == ["A", "B"]

    def test_multiple_del_markers(self):
        segs = _parse_redline_segments("§DEL§X§DEL§ und §DEL§Y§DEL§")
        del_texts = [s["text"] for s in segs if s["type"] == "del"]
        assert del_texts == ["X", "Y"]

    def test_no_empty_ctx_segments(self):
        # Leere ctx-Segmente sollen nicht in der Ergebnisliste auftauchen
        segs = _parse_redline_segments("§INS§nur Marker§INS§")
        ctx_segs = [s for s in segs if s["type"] == "ctx"]
        # Leere ctx-Strings dürfen nicht enthalten sein
        for seg in ctx_segs:
            assert seg["text"] != ""

    def test_newline_inside_marker(self):
        segs = _parse_redline_segments("§INS§Zeile 1\nZeile 2§INS§")
        ins_segs = [s for s in segs if s["type"] == "ins"]
        assert len(ins_segs) == 1
        assert "Zeile 1" in ins_segs[0]["text"]

    def test_long_realistic_redline(self):
        text = (
            "Die Gemeinde §DEL§soll§DEL§ §INS§muss§INS§ bis zum "
            "§DEL§31.12.2026§DEL§ §INS§30.06.2025§INS§ einen Plan vorlegen."
        )
        segs = _parse_redline_segments(text)
        del_texts = [s["text"] for s in segs if s["type"] == "del"]
        ins_texts = [s["text"] for s in segs if s["type"] == "ins"]
        assert "soll" in del_texts
        assert "muss" in ins_texts
        assert "31.12.2026" in del_texts
        assert "30.06.2025" in ins_texts