' . $html . '
'); + + return array_values(unpack('N*', mb_convert_encoding($mpdf->drawnText[0], 'UTF-32BE', 'UTF-8'))); + } + + public function dataRuns() + { + return [ + 'a word ending in a mark, where the lookahead runs out' => [ + [self::BEH, self::BEH, self::FATHA], + [self::FATHA, self::BEH, self::BEH], + ], + 'a word starting with a mark, where the backtrack runs out' => [ + [self::FATHA, self::BEH, self::BEH], + [self::BEH, self::BEH, self::FATHA], + ], + 'the lookahead met past the mark' => [ + [self::BEH, self::BEH, self::FATHA, self::LOW_ALEF], + [self::LOW_ALEF, self::FATHA, self::DOTLESS_BEH_FINAL, self::BEH], + ], + 'the backtrack met past the mark' => [ + [self::LOW_ALEF, self::FATHA, self::BEH, self::BEH], + [self::BEH, self::DOTLESS_BEH_INITIAL, self::FATHA, self::LOW_ALEF], + ], + ]; + } + + /** + * @dataProvider dataRuns + */ + public function testAFormsContextIsMatchedUpToTheEdgeOfTheRun($codepoints, $expected) + { + $this->assertSame($expected, $this->drawn($codepoints)); + } + +} diff --git a/tests/Mpdf/Fixtures/arabic-shape.php b/tests/Mpdf/Fixtures/arabic-shape.php new file mode 100644 index 000000000..4d267e322 --- /dev/null +++ b/tests/Mpdf/Fixtures/arabic-shape.php @@ -0,0 +1,34 @@ + + */ + +require __DIR__ . '/../../../vendor/autoload.php'; + +set_time_limit(5); +error_reporting(E_ERROR); + +list($runs, $usetags, $marks) = json_decode(base64_decode($argv[1]), true); + +$forms = []; +foreach ($runs as $name => $run) { + list($hexes, $glyphs) = $run; + + $info = []; + foreach ($hexes as $hex) { + $info[] = ['hex' => $hex, 'uni' => hexdec($hex)]; + } + + \Mpdf\Shaper\Arabic::shape($info, $glyphs, $marks, $usetags, 'arab'); + + foreach ($info as $char) { + $forms[$name][] = [$char['hex'], $char['form']]; + } +} + +echo json_encode($forms); diff --git a/tests/Mpdf/Shaper/ArabicTest.php b/tests/Mpdf/Shaper/ArabicTest.php index 513cd289a..097cad861 100644 --- a/tests/Mpdf/Shaper/ArabicTest.php +++ b/tests/Mpdf/Shaper/ArabicTest.php @@ -512,6 +512,93 @@ public function dataContextNamingAPlaneSixteenGlyph() ]; } + /** + * The walk over the glyphs a chained rule's lookup ignores read past the edge of the run: a notice + * for each glyph on PHP 7, and from PHP 8 a walk that never returned (#204). Here the notice is what + * fails, so a regression cannot hang the suite. + * + * @dataProvider dataContextWalkedToTheEdgeOfTheRun + */ + public function testAFormsContextWalkReadsNothingPastTheEdgeOfTheRun($hexes, $glyphs, $expected) + { + set_error_handler(function ($number, $message, $file, $line) { + throw new \ErrorException($message, 0, $number, $file, $line); + }); + + try { + $forms = $this->shape($hexes, self::ALL_FORMS, 'arab', self::FATHA, $glyphs); + } finally { + restore_error_handler(); + } + + $this->assertSame($expected, $forms); + } + + /** + * The same cases, all in one child process under a time limit, for a regression that reads past the + * edge without a notice. The limit is the child's own rather than a `timeout` around it, so it holds + * on Windows too, and the child reports nothing short of a fatal error, so a walk warning on every + * step cannot fill the stderr pipe and leave it blocked instead of timed out. + */ + public function testAFormsContextWalkReturnsAtTheEdgeOfTheRun() + { + $runs = []; + $expected = []; + foreach ($this->dataContextWalkedToTheEdgeOfTheRun() as $name => $case) { + $runs[$name] = [$case[0], $case[1]]; + $expected[$name] = $case[2]; + } + + // base64, because Windows argument quoting does not survive the JSON's double quotes + $arg = base64_encode(json_encode([$runs, self::ALL_FORMS, ' ' . self::FATHA])); + $command = escapeshellarg(PHP_BINARY) . ' -d display_errors=stderr ' + . escapeshellarg(__DIR__ . '/../Fixtures/arabic-shape.php') . ' ' . $arg; + + $process = proc_open($command, [1 => ['pipe', 'w'], 2 => ['pipe', 'w']], $pipes, null, null, ['bypass_shell' => true]); + $output = stream_get_contents($pipes[1]); + $errors = stream_get_contents($pipes[2]); + fclose($pipes[1]); + fclose($pipes[2]); + $this->assertSame(0, proc_close($process), $errors); + + $this->assertSame($expected, json_decode($output, true)); + } + + public function dataContextWalkedToTheEdgeOfTheRun() + { + // The forms are the presentation forms rather than names, because the run is left holding them + // and shape() reads each back as hex + $ignoreFatha = '((?:(?: ' . self::FATHA . '))*)'; + + return [ + 'a backtrack that runs out at the start of the run' => [ + [self::FATHA, self::DAL], + [self::DAL => ['0FEA9', '0FEAA', 'prel' => [0 => [self::BEH]], 'ignore' => [0 => $ignoreFatha]]], + [[self::FATHA, 0], [self::DAL, 0]], + ], + 'a lookahead that runs out at the end of the run' => [ + [self::BEH, self::FATHA], + [self::BEH => ['0FE8F', 'postl' => [0 => [self::DAL]], 'ignore' => [0 => $ignoreFatha]]], + [[self::BEH, 0], [self::FATHA, 0]], + ], + 'a lookahead whose second position runs out after the first is met' => [ + [self::BEH, self::FATHA, self::DAL, self::FATHA], + [self::BEH => ['0FE8F', '0FE90', '0FE91', 'postl' => [2 => [self::DAL, self::DAL]], 'ignore' => [2 => $ignoreFatha]]], + [[self::BEH, 0], [self::FATHA, 0], [self::DAL, 0], [self::FATHA, 0]], + ], + 'a backtrack met past the ignored glyphs' => [ + [self::BEH, self::FATHA, self::DAL], + [self::DAL => ['0FEA9', '0FEAA', 'prel' => [1 => [self::BEH]], 'ignore' => [1 => $ignoreFatha]]], + [[self::BEH, 0], [self::FATHA, 0], ['0FEAA', 1]], + ], + 'a lookahead met past the ignored glyphs' => [ + [self::BEH, self::FATHA, self::DAL], + [self::BEH => ['0FE8F', '0FE90', '0FE91', 'postl' => [2 => [self::DAL]], 'ignore' => [2 => $ignoreFatha]]], + [['0FE91', 2], [self::FATHA, 0], [self::DAL, 0]], + ], + ]; + } + /** * @return array one [hex, form] pair per character, in logical order */ diff --git a/tests/data/fontcache/NotoSansArabic-ContextEdge-Synthetic.json b/tests/data/fontcache/NotoSansArabic-ContextEdge-Synthetic.json new file mode 100644 index 000000000..07d46f738 --- /dev/null +++ b/tests/data/fontcache/NotoSansArabic-ContextEdge-Synthetic.json @@ -0,0 +1,319 @@ +{ + "_": "Generated by composer fontcache:update. See tests/Mpdf/Fonts/ParserGoldenMaster.php.", + "fullName": "NotoSansArabicContextEdgeSynthetic-Regular", + "mtx": { + "GSUBScriptLang": { + "DFLT": "DFLT ", + "arab": "DFLT ", + "cyrl": "DFLT ", + "dev2": "DFLT ", + "grek": "DFLT ", + "latn": "DFLT " + }, + "GSUBFeatures": { + "DFLT": { + "DFLT": { + "init": [ + 0 + ], + "medi": [ + 1 + ], + "fina": [ + 2 + ] + } + }, + "arab": { + "DFLT": { + "init": [ + 0 + ], + "medi": [ + 1 + ], + "fina": [ + 2 + ] + } + }, + "cyrl": { + "DFLT": { + "init": [ + 0 + ], + "medi": [ + 1 + ], + "fina": [ + 2 + ] + } + }, + "dev2": { + "DFLT": { + "init": [ + 0 + ], + "medi": [ + 1 + ], + "fina": [ + 2 + ] + } + }, + "grek": { + "DFLT": { + "init": [ + 0 + ], + "medi": [ + 1 + ], + "fina": [ + 2 + ] + } + }, + "latn": { + "DFLT": { + "init": [ + 0 + ], + "medi": [ + 1 + ], + "fina": [ + 2 + ] + } + } + }, + "GSUBLookups": [ + { + "Type": 6, + "Flag": 8, + "SubtableCount": 1, + "Subtables": [ + 122 + ], + "MarkFilteringSet": "" + }, + { + "Type": 1, + "Flag": 0, + "SubtableCount": 1, + "Subtables": [ + 148 + ], + "MarkFilteringSet": "" + }, + { + "Type": 6, + "Flag": 8, + "SubtableCount": 1, + "Subtables": [ + 162 + ], + "MarkFilteringSet": "" + }, + { + "Type": 1, + "Flag": 0, + "SubtableCount": 1, + "Subtables": [ + 194 + ], + "MarkFilteringSet": "" + }, + { + "Type": 1, + "Flag": 0, + "SubtableCount": 1, + "Subtables": [ + 208 + ], + "MarkFilteringSet": "" + } + ], + "GPOSScriptLang": { + "DFLT": "DFLT ", + "arab": "DFLT ", + "cyrl": "DFLT ", + "dev2": "DFLT ", + "grek": "DFLT ", + "latn": "DFLT " + }, + "GPOSFeatures": { + "DFLT": { + "DFLT": { + "mark": [ + 0 + ], + "mkmk": [ + 1 + ] + } + }, + "arab": { + "DFLT": { + "mark": [ + 0 + ], + "mkmk": [ + 1 + ] + } + }, + "cyrl": { + "DFLT": { + "mark": [ + 0 + ], + "mkmk": [ + 1 + ] + } + }, + "dev2": { + "DFLT": { + "mark": [ + 0 + ], + "mkmk": [ + 1 + ] + } + }, + "grek": { + "DFLT": { + "mark": [ + 0 + ], + "mkmk": [ + 1 + ] + } + }, + "latn": { + "DFLT": { + "mark": [ + 0 + ], + "mkmk": [ + 1 + ] + } + } + }, + "GPOSLookups": [ + { + "Type": 4, + "Flag": 0, + "SubtableCount": 1, + "Subtables": [ + 102 + ], + "MarkFilteringSet": "" + }, + { + "Type": 6, + "Flag": 16, + "SubtableCount": 1, + "Subtables": [ + 194 + ], + "MarkFilteringSet": 0 + } + ], + "MarkGlyphSets": [ + " 0E005" + ], + "rtlPUAstr": "\\x{0E002}-\\x{0E004}", + "haskernGPOS": false, + "hassmallcapsGSUB": false + }, + "cache": { + "GDEFdata.json": { + "GlyphClassBases": " 00020| 00627| 008AD| 0E000| 0E001| 0E002| 0E003| 0E004", + "GlyphClassMarks": " 0064E| 0E005", + "GlyphClassLigatures": "", + "GlyphClassComponents": "", + "MarkGlyphSets": [ + " 0E005" + ], + "MarkAttachmentType": [] + }, + "GPOS.dat": "234 bytes, sha256 0f363351c5ea276d72ca75127977045024304d1f2b19b04e8cfded715c737a13", + "GPOSdata.json": [ + [ + { + "57349": 0 + } + ], + [ + { + "57349": 0 + } + ] + ], + "GSUB.arab.DFLT.json": { + "rtlSUB": { + "00628": { + "2": "0E004", + "prel": { + "2": [ + "008AD" + ] + }, + "ignore": { + "2": "((?:(?: 0064E| 0E005))*)", + "1": "((?:(?: 0064E| 0E005))*)" + }, + "3": "0E003", + "1": "0E002", + "postl": { + "1": [ + "008AD" + ] + } + } + }, + "finals": "0E002 ", + "rphf": [], + "half": [], + "pref": [], + "blwf": [], + "pstf": [] + }, + "GSUB.dat": "220 bytes, sha256 3ac7cb0c04aead3a9fc0c5f6c607b54931c16a81c5d5edb41654ceb5d58d593e", + "GSUBdata.json": [ + [ + { + "1576": 0 + } + ], + [ + { + "1576": 0 + } + ], + [ + { + "1576": 0 + } + ], + [ + { + "1576": 0 + } + ], + [ + { + "1576": 0 + } + ] + ] + } +} diff --git a/tests/data/otldump/NotoSansArabic-ContextEdge-Synthetic.txt b/tests/data/otldump/NotoSansArabic-ContextEdge-Synthetic.txt new file mode 100644 index 000000000..2388457a5 --- /dev/null +++ b/tests/data/otldump/NotoSansArabic-ContextEdge-Synthetic.txt @@ -0,0 +1,27 @@ +