xref: /dokuwiki/_test/tests/Parsing/ParserMode/GfmListblockTest.php (revision 685560eb3044321b3bdd0be40985871ced5f1d05)
1<?php
2
3namespace dokuwiki\test\Parsing\ParserMode;
4
5use dokuwiki\Parsing\Handler\GfmLists;
6use dokuwiki\Parsing\ModeRegistry;
7use dokuwiki\Parsing\ParserMode\GfmListblock;
8
9/**
10 * Tests for GFM list blocks.
11 *
12 * GfmListblock captures the entire list block via addSpecialPattern then
13 * sub-parses each item's body through ModeRegistry::getSubParser(), so the
14 * outer parser only needs gfm_listblock added; inline modes (emphasis,
15 * strong, etc.) and block modes (gfm_code) are picked up by the sub-parser.
16 */
17class GfmListblockTest extends ParserTestBase
18{
19    public function setUp(): void
20    {
21        parent::setUp();
22        global $conf;
23        $conf['syntax'] = 'markdown';
24        ModeRegistry::reset();
25    }
26
27    public function tearDown(): void
28    {
29        ModeRegistry::reset();
30        parent::tearDown();
31    }
32
33    public function testUnorderedDash()
34    {
35        // Each item's body is sub-parsed and wrapped in a `nest` call so
36        // the main handler's Block rewriter doesn't double-wrap multi-block
37        // content. See AbstractListsRewriter / Block / Nest interaction.
38        $this->P->addMode('gfm_listblock', new GfmListblock());
39        $this->P->parse("- A\n- B\n- C\n");
40
41        $expected = [
42            ['document_start', []],
43            ['listu_open', []],
44            ['listitem_open', [1]],
45            ['listcontent_open', []],
46            ['nest', [[ ['cdata', ['A']] ]]],
47            ['listcontent_close', []],
48            ['listitem_close', []],
49            ['listitem_open', [1]],
50            ['listcontent_open', []],
51            ['nest', [[ ['cdata', ['B']] ]]],
52            ['listcontent_close', []],
53            ['listitem_close', []],
54            ['listitem_open', [1]],
55            ['listcontent_open', []],
56            ['nest', [[ ['cdata', ['C']] ]]],
57            ['listcontent_close', []],
58            ['listitem_close', []],
59            ['listu_close', []],
60            ['document_end', []],
61        ];
62        $this->assertCalls($expected, $this->H->calls);
63    }
64
65    public function testUnorderedAsterisk()
66    {
67        $this->P->addMode('gfm_listblock', new GfmListblock());
68        $this->P->parse("* A\n* B\n");
69
70        $names = array_column($this->H->calls, 0);
71        $this->assertContains('listu_open', $names);
72        $this->assertNotContains('listo_open', $names);
73    }
74
75    public function testUnorderedPlus()
76    {
77        $this->P->addMode('gfm_listblock', new GfmListblock());
78        $this->P->parse("+ A\n+ B\n");
79
80        $names = array_column($this->H->calls, 0);
81        $this->assertContains('listu_open', $names);
82        $this->assertNotContains('listo_open', $names);
83    }
84
85    public function testOrderedDot()
86    {
87        $this->P->addMode('gfm_listblock', new GfmListblock());
88        $this->P->parse("1. A\n2. B\n3. C\n");
89
90        $names = array_column($this->H->calls, 0);
91        $this->assertContains('listo_open', $names);
92        $this->assertNotContains('listu_open', $names);
93    }
94
95    public function testOrderedParen()
96    {
97        $this->P->addMode('gfm_listblock', new GfmListblock());
98        $this->P->parse("1) A\n2) B\n");
99
100        $names = array_column($this->H->calls, 0);
101        $this->assertContains('listo_open', $names);
102    }
103
104    public function testOrderedStartNumber()
105    {
106        $this->P->addMode('gfm_listblock', new GfmListblock());
107        $this->P->parse("5. A\n6. B\n");
108
109        $opens = array_filter($this->H->calls, static fn($c) => $c[0] === 'listo_open');
110        $this->assertCount(1, $opens);
111        $open = array_values($opens)[0];
112        $this->assertSame([null, 5], $open[1], 'listo_open must carry the first item start number');
113    }
114
115    public function testOrderedDefaultStartNotEmittedSpecially()
116    {
117        // For start=1 the rewriter omits the start argument entirely (the
118        // renderer would suppress it anyway). The wire shape is bare [].
119        $this->P->addMode('gfm_listblock', new GfmListblock());
120        $this->P->parse("1. A\n2. B\n");
121
122        $opens = array_filter($this->H->calls, static fn($c) => $c[0] === 'listo_open');
123        $open = array_values($opens)[0];
124        $this->assertSame([], $open[1]);
125    }
126
127    public function testNestedTwoLevels()
128    {
129        $this->P->addMode('gfm_listblock', new GfmListblock());
130        $this->P->parse("- A\n  - B\n- C\n");
131
132        $expected = [
133            ['document_start', []],
134            ['listu_open', []],
135            ['listitem_open', [1, GfmLists::NODE]],
136            ['listcontent_open', []],
137            ['nest', [[ ['cdata', ['A']] ]]],
138            ['listcontent_close', []],
139            ['listu_open', []],
140            ['listitem_open', [2]],
141            ['listcontent_open', []],
142            ['nest', [[ ['cdata', ['B']] ]]],
143            ['listcontent_close', []],
144            ['listitem_close', []],
145            ['listu_close', []],
146            ['listitem_close', []],
147            ['listitem_open', [1]],
148            ['listcontent_open', []],
149            ['nest', [[ ['cdata', ['C']] ]]],
150            ['listcontent_close', []],
151            ['listitem_close', []],
152            ['listu_close', []],
153            ['document_end', []],
154        ];
155        $this->assertCalls($expected, $this->H->calls);
156    }
157
158    /**
159     * Flatten a call list, recursing into `nest` calls' inner content.
160     * Useful for tests that just want to verify a particular instruction
161     * appears somewhere in the rendered output regardless of nesting.
162     */
163    private function flatNames(array $calls): array
164    {
165        $names = [];
166        foreach ($calls as $call) {
167            $names[] = $call[0];
168            if ($call[0] === 'nest') {
169                $names = array_merge($names, $this->flatNames($call[1][0]));
170            }
171        }
172        return $names;
173    }
174
175    public function testNestedThreeLevels()
176    {
177        $this->P->addMode('gfm_listblock', new GfmListblock());
178        $this->P->parse("- A\n  - B\n    - C\n");
179
180        $itemOpens = array_filter($this->H->calls, static fn($c) => $c[0] === 'listitem_open');
181        $levels = array_map(static fn($c) => $c[1][0], array_values($itemOpens));
182        $this->assertSame([1, 2, 3], $levels);
183    }
184
185    public function testInlineFormatting()
186    {
187        $this->P->addMode('gfm_listblock', new GfmListblock());
188        $this->P->parse("- **bold** text\n");
189
190        $names = $this->flatNames($this->H->calls);
191        $this->assertContains('strong_open', $names, 'inline strong must be parsed inside item');
192        $this->assertContains('strong_close', $names);
193    }
194
195    public function testMarkerCharSwitchKeepsOneList()
196    {
197        // CommonMark: changing marker character (`-` → `+`) starts a new list.
198        // Our simpler model groups by type ('u' / 'o') only, so `-` and `+`
199        // share one <ul>. Deliberate simplification — the rewriter doesn't
200        // distinguish marker characters within the same type.
201        $this->P->addMode('gfm_listblock', new GfmListblock());
202        $this->P->parse("- A\n+ B\n");
203
204        $opens = array_filter($this->H->calls, static fn($c) => $c[0] === 'listu_open');
205        $this->assertCount(1, $opens, 'marker-character change does not split unordered lists');
206    }
207
208    public function testOrderedToUnorderedSplits()
209    {
210        // Type change (o → u) DOES split, since the rewriter does close/open
211        // when the type differs.
212        $this->P->addMode('gfm_listblock', new GfmListblock());
213        $this->P->parse("1. A\n- B\n");
214
215        $oOpens = array_filter($this->H->calls, static fn($c) => $c[0] === 'listo_open');
216        $uOpens = array_filter($this->H->calls, static fn($c) => $c[0] === 'listu_open');
217        $this->assertCount(1, $oOpens);
218        $this->assertCount(1, $uOpens);
219    }
220
221    public function testNotAListMidParagraph()
222    {
223        $this->P->addMode('gfm_listblock', new GfmListblock());
224        $this->P->parse("Foo - bar");
225
226        $names = array_column($this->H->calls, 0);
227        $this->assertNotContains('listu_open', $names);
228        $this->assertNotContains('listo_open', $names);
229    }
230
231    public function testEmptyMarkerEol()
232    {
233        $this->P->addMode('gfm_listblock', new GfmListblock());
234        $this->P->parse("-\n");
235
236        $names = array_column($this->H->calls, 0);
237        $this->assertContains('listu_open', $names, 'a bare marker still opens a list');
238        $this->assertContains('listitem_open', $names);
239    }
240
241    public function testHeaderRejectedInsideItem()
242    {
243        // Sub-parser excludes BASEONLY (gfm_header), so `# bar` inside an item
244        // body must NOT produce a header instruction.
245        $this->P->addMode('gfm_listblock', new GfmListblock());
246        $this->P->parse("- foo\n  # bar\n");
247
248        $names = $this->flatNames($this->H->calls);
249        $this->assertNotContains('header', $names);
250        $this->assertNotContains('section_open', $names);
251    }
252
253    public function testFencedCodeInsideItem()
254    {
255        // After the dedent step strips the 2-space prefix from the body,
256        // the fence sits at column 0 from the sub-parser's point of view
257        // and gfm_code matches it.
258        $this->P->addMode('gfm_listblock', new GfmListblock());
259        $this->P->parse("- foo\n  ```\n  hello\n  ```\n");
260
261        $names = $this->flatNames($this->H->calls);
262        $this->assertContains('code', $names, 'fenced code inside item must be parsed');
263    }
264
265    public function testMultiParagraphItemIsLoose()
266    {
267        $this->P->addMode('gfm_listblock', new GfmListblock());
268        $this->P->parse("- foo\n\n  bar\n");
269
270        // Loose item: the nest contains two p_open / p_close pairs (one per
271        // paragraph) since the outer-only stripping in filterSubCalls only
272        // collapses single-paragraph items.
273        $names = $this->flatNames($this->H->calls);
274        $pOpens = array_filter($names, static fn($n) => $n === 'p_open');
275        $this->assertGreaterThanOrEqual(2, count($pOpens),
276            'multi-paragraph items must keep both p_open calls');
277    }
278
279    public function testSortValue()
280    {
281        $mode = new GfmListblock();
282        $this->assertSame(10, $mode->getSort());
283    }
284
285    /**
286     * Regression: an item's sub-parsed content must reach the main handler
287     * inside a `nest` call. Without the wrap, the main handler's Block
288     * rewriter wraps the item content in another `<p>` (it already has
289     * its own `<p>` from the sub-parser), producing nested paragraph tags.
290     */
291    public function testItemContentIsWrappedInNest()
292    {
293        $this->P->addMode('gfm_listblock', new GfmListblock());
294        $this->P->parse("- foo\n");
295
296        $nests = array_filter($this->H->calls, static fn($c) => $c[0] === 'nest');
297        $this->assertCount(1, $nests, 'each item body should land in one nest call');
298    }
299
300    /**
301     * Regression: multiple consecutive blank lines inside a list block must
302     * NOT terminate the list. Spec example 242 (`- Foo\n\n      bar\n\n\n
303     * baz`) ends with a triple blank between two indented continuations and
304     * expects all three to remain inside one list item.
305     */
306    public function testTripleBlankBetweenContinuationsKeepsListOpen()
307    {
308        $this->P->addMode('gfm_listblock', new GfmListblock());
309        $this->P->parse("- Foo\n\n      bar\n\n\n      baz\n");
310
311        // The list should bracket all three indented lines: `- Foo`, `bar`,
312        // and `baz` all live inside a single `<ul>`. We assert there is
313        // exactly one listu_open / listu_close pair (no early termination
314        // splitting `baz` into a separate top-level block).
315        $opens  = array_filter($this->H->calls, static fn($c) => $c[0] === 'listu_open');
316        $closes = array_filter($this->H->calls, static fn($c) => $c[0] === 'listu_close');
317        $this->assertCount(1, $opens,
318            'triple blank line between continuations must not split the list');
319        $this->assertCount(1, $closes);
320    }
321
322    /**
323     * Regression: blank lines between items (any number) must not split the
324     * list. Spec example 270 stresses two-blank cases.
325     */
326    public function testMultipleBlanksBetweenItemsKeepsOneList()
327    {
328        $this->P->addMode('gfm_listblock', new GfmListblock());
329        $this->P->parse("- one\n\n\n- two\n");
330
331        $opens = array_filter($this->H->calls, static fn($c) => $c[0] === 'listu_open');
332        $this->assertCount(1, $opens, 'blank lines between items must stay inside the list');
333    }
334}
335