Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
82.19% |
360 / 438 |
|
62.50% |
25 / 40 |
CRAP | |
0.00% |
0 / 1 |
| PdfReader | |
82.19% |
360 / 438 |
|
62.50% |
25 / 40 |
397.03 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getParseWarnings | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| fromFile | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| fromString | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| fromFilePublicKey | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| fromStringPublicKey | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| fromStream | |
66.67% |
4 / 6 |
|
0.00% |
0 / 1 |
3.33 | |||
| getVersion | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getPdfVersion | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getEffectiveVersion | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
5 | |||
| validateVersion | |
63.41% |
26 / 41 |
|
0.00% |
0 / 1 |
66.39 | |||
| isLinearized | |
75.00% |
12 / 16 |
|
0.00% |
0 / 1 |
11.56 | |||
| getLinearizationParameters | |
90.48% |
19 / 21 |
|
0.00% |
0 / 1 |
8.06 | |||
| getPageOffsetHintTable | |
23.08% |
9 / 39 |
|
0.00% |
0 / 1 |
202.07 | |||
| getPageByteRange | |
50.00% |
3 / 6 |
|
0.00% |
0 / 1 |
4.12 | |||
| getTrailer | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getCatalog | |
88.24% |
15 / 17 |
|
0.00% |
0 / 1 |
7.08 | |||
| recoverCatalog | |
80.00% |
20 / 25 |
|
0.00% |
0 / 1 |
10.80 | |||
| getInfo | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
3 | |||
| getPageCount | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
4 | |||
| getPages | |
80.00% |
8 / 10 |
|
0.00% |
0 / 1 |
3.07 | |||
| getPage | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| getObject | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| resolveReference | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getResolver | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| extractText | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| extractAllText | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
2 | |||
| extractTextWithPositions | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| extractAllTextWithPositions | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
2 | |||
| getTypedCatalog | |
87.50% |
7 / 8 |
|
0.00% |
0 / 1 |
3.02 | |||
| getTypedPage | |
83.33% |
5 / 6 |
|
0.00% |
0 / 1 |
2.02 | |||
| getTypedPages | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
3 | |||
| getTypedObject | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
3 | |||
| build | |
98.02% |
99 / 101 |
|
0.00% |
0 / 1 |
25 | |||
| extractFileId | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
4 | |||
| parseXrefAt | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
2 | |||
| findStartxref | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
5 | |||
| reconstructXref | |
82.35% |
14 / 17 |
|
0.00% |
0 / 1 |
4.09 | |||
| findCatalogInScan | |
87.50% |
14 / 16 |
|
0.00% |
0 / 1 |
9.16 | |||
| collectPages | |
69.23% |
9 / 13 |
|
0.00% |
0 / 1 |
8.43 | |||
| 1 | <?php |
| 2 | |
| 3 | declare(strict_types=1); |
| 4 | |
| 5 | namespace Phpdftk\Pdf\Reader; |
| 6 | |
| 7 | use Phpdftk\Pdf\Core\Document\Catalog; |
| 8 | use Phpdftk\Pdf\Core\Document\Page; |
| 9 | use Phpdftk\Pdf\Core\File\PdfHydrator; |
| 10 | use Phpdftk\Pdf\Core\PdfArray; |
| 11 | use Phpdftk\Pdf\Core\PdfDictionary; |
| 12 | use Phpdftk\Pdf\Core\PdfName; |
| 13 | use Phpdftk\Pdf\Core\PdfNumber; |
| 14 | use Phpdftk\Pdf\Core\PdfObject; |
| 15 | use Phpdftk\Pdf\Core\PdfReference; |
| 16 | use Phpdftk\Pdf\Core\PdfString; |
| 17 | use Phpdftk\Pdf\Core\PdfVersion; |
| 18 | use Phpdftk\Pdf\Core\Serializable; |
| 19 | use Phpdftk\Pdf\Reader\Exception\InvalidPdfException; |
| 20 | use Phpdftk\Pdf\Reader\Parser\HintTableParser; |
| 21 | use Phpdftk\Pdf\Reader\Parser\ObjectParser; |
| 22 | use Phpdftk\Pdf\Reader\Parser\ObjectScanner; |
| 23 | use Phpdftk\Pdf\Reader\Parser\PageOffsetHintTable; |
| 24 | use Phpdftk\Pdf\Reader\Parser\StreamParser; |
| 25 | use Phpdftk\Pdf\Reader\Parser\XrefParser; |
| 26 | use Phpdftk\Pdf\Reader\Parser\XrefStreamParser; |
| 27 | use Phpdftk\Pdf\Reader\Tokenizer\FileSource; |
| 28 | use Phpdftk\Pdf\Reader\Tokenizer\Source; |
| 29 | use Phpdftk\Pdf\Reader\Tokenizer\StringSource; |
| 30 | use Phpdftk\Pdf\Reader\Tokenizer\Tokenizer; |
| 31 | |
| 32 | /** |
| 33 | * PDF reader — parses existing PDF files into the phpdftk object model. |
| 34 | * |
| 35 | * Phase 1 supports unencrypted PDFs with classic cross-reference tables. |
| 36 | * Returns raw `PdfDictionary` objects; typed hydration (into `Catalog`, |
| 37 | * `Page`, etc.) is a future phase. |
| 38 | * |
| 39 | * Three factory methods mirror the writer's output modes: |
| 40 | * |
| 41 | * ```php |
| 42 | * $pdf = PdfReader::fromFile('/path/to/document.pdf'); |
| 43 | * $pdf = PdfReader::fromString($bytes); |
| 44 | * $pdf = PdfReader::fromStream(fopen('php://stdin', 'rb')); |
| 45 | * ``` |
| 46 | * |
| 47 | * @api |
| 48 | */ |
| 49 | final class PdfReader |
| 50 | { |
| 51 | /** @var list<string> */ |
| 52 | private array $parseWarnings = []; |
| 53 | |
| 54 | private bool $strict = true; |
| 55 | |
| 56 | private function __construct( |
| 57 | private readonly string $version, |
| 58 | private readonly PdfDictionary $trailer, |
| 59 | private readonly ObjectResolver $resolver, |
| 60 | ) {} |
| 61 | |
| 62 | /** |
| 63 | * Return warnings accumulated during parsing. |
| 64 | * |
| 65 | * @return list<string> |
| 66 | */ |
| 67 | public function getParseWarnings(): array |
| 68 | { |
| 69 | return $this->parseWarnings; |
| 70 | } |
| 71 | |
| 72 | // ----------------------------------------------------------------------- |
| 73 | // Factory methods |
| 74 | // ----------------------------------------------------------------------- |
| 75 | |
| 76 | public static function fromFile(string $path, string $password = '', bool $strict = true): self |
| 77 | { |
| 78 | return self::build(new FileSource($path), $password, $strict); |
| 79 | } |
| 80 | |
| 81 | public static function fromString(string $content, string $password = '', bool $strict = true): self |
| 82 | { |
| 83 | return self::build(new StringSource($content), $password, $strict); |
| 84 | } |
| 85 | |
| 86 | /** |
| 87 | * Read a public-key (certificate-based) encrypted PDF from a file. |
| 88 | */ |
| 89 | public static function fromFilePublicKey( |
| 90 | string $path, |
| 91 | string $certificate, |
| 92 | string $privateKey, |
| 93 | bool $strict = true, |
| 94 | ): self { |
| 95 | return self::build(new FileSource($path), '', $strict, $certificate, $privateKey); |
| 96 | } |
| 97 | |
| 98 | /** |
| 99 | * Read a public-key (certificate-based) encrypted PDF from a string. |
| 100 | */ |
| 101 | public static function fromStringPublicKey( |
| 102 | string $content, |
| 103 | string $certificate, |
| 104 | string $privateKey, |
| 105 | bool $strict = true, |
| 106 | ): self { |
| 107 | return self::build(new StringSource($content), '', $strict, $certificate, $privateKey); |
| 108 | } |
| 109 | |
| 110 | /** @param resource $stream */ |
| 111 | public static function fromStream($stream, string $password = '', bool $strict = true): self |
| 112 | { |
| 113 | if (!is_resource($stream)) { |
| 114 | throw new \InvalidArgumentException('Expected a stream resource'); |
| 115 | } |
| 116 | $content = stream_get_contents($stream); |
| 117 | if ($content === false) { |
| 118 | throw new \RuntimeException('Failed to read stream'); |
| 119 | } |
| 120 | return self::fromString($content, $password, $strict); |
| 121 | } |
| 122 | |
| 123 | // ----------------------------------------------------------------------- |
| 124 | // Public API |
| 125 | // ----------------------------------------------------------------------- |
| 126 | |
| 127 | /** PDF version string, e.g. "1.7". */ |
| 128 | public function getVersion(): string |
| 129 | { |
| 130 | return $this->version; |
| 131 | } |
| 132 | |
| 133 | /** Typed PDF version from the file header. */ |
| 134 | public function getPdfVersion(): PdfVersion |
| 135 | { |
| 136 | return PdfVersion::tryFrom($this->version) ?? PdfVersion::V1_7; |
| 137 | } |
| 138 | |
| 139 | /** |
| 140 | * Effective PDF version — max(header, catalog /Version). |
| 141 | * |
| 142 | * Per ISO 32000 §7.2.2, the catalog /Version entry (PDF 1.4+) |
| 143 | * overrides the header version if it is higher. |
| 144 | */ |
| 145 | public function getEffectiveVersion(): PdfVersion |
| 146 | { |
| 147 | $headerVersion = $this->getPdfVersion(); |
| 148 | $catalog = $this->getCatalog(); |
| 149 | |
| 150 | if ($catalog instanceof PdfDictionary && $catalog->has('Version')) { |
| 151 | $catVersion = $catalog->get('Version'); |
| 152 | if ($catVersion instanceof PdfName) { |
| 153 | $catPdfVersion = PdfVersion::tryFrom($catVersion->value); |
| 154 | if ($catPdfVersion !== null) { |
| 155 | return $headerVersion->max($catPdfVersion); |
| 156 | } |
| 157 | } |
| 158 | } |
| 159 | |
| 160 | return $headerVersion; |
| 161 | } |
| 162 | |
| 163 | /** |
| 164 | * Scan the document for structural features inconsistent with the |
| 165 | * declared version. Returns a list of warning strings. |
| 166 | * |
| 167 | * Checks top-level indicators that can be detected from raw |
| 168 | * dictionaries without full object hydration. |
| 169 | * |
| 170 | * @return list<string> |
| 171 | */ |
| 172 | public function validateVersion(): array |
| 173 | { |
| 174 | $warnings = []; |
| 175 | $version = $this->getEffectiveVersion(); |
| 176 | |
| 177 | // Xref stream → requires 1.5 |
| 178 | $trailerType = $this->trailer->get('Type'); |
| 179 | if ($trailerType instanceof PdfName && $trailerType->value === 'XRef') { |
| 180 | if (!$version->isAtLeast(PdfVersion::V1_5)) { |
| 181 | $warnings[] = "Cross-reference stream requires PDF 1.5, but document declares {$version->value}"; |
| 182 | } |
| 183 | } |
| 184 | |
| 185 | // Encryption version |
| 186 | $encrypt = $this->trailer->get('Encrypt'); |
| 187 | if ($encrypt instanceof PdfReference) { |
| 188 | $encDict = $this->resolver->resolveReference($encrypt); |
| 189 | if ($encDict instanceof PdfDictionary) { |
| 190 | $v = $encDict->get('V'); |
| 191 | $vVal = $v instanceof PdfNumber ? (int) $v->toPdf() : 0; |
| 192 | $required = match (true) { |
| 193 | $vVal >= 5 => PdfVersion::V2_0, |
| 194 | $vVal >= 4 => PdfVersion::V1_6, |
| 195 | $vVal >= 2 => PdfVersion::V1_4, |
| 196 | default => PdfVersion::V1_0, |
| 197 | }; |
| 198 | if ($required->isGreaterThan($version)) { |
| 199 | $warnings[] = "Encryption V={$vVal} requires PDF {$required->value}, but document declares {$version->value}"; |
| 200 | } |
| 201 | } |
| 202 | } |
| 203 | |
| 204 | // Catalog-level structural checks |
| 205 | try { |
| 206 | $catalog = $this->getCatalog(); |
| 207 | |
| 208 | if ($catalog->has('OCProperties') && !$version->isAtLeast(PdfVersion::V1_5)) { |
| 209 | $warnings[] = "Optional content (/OCProperties) requires PDF 1.5, but document declares {$version->value}"; |
| 210 | } |
| 211 | if ($catalog->has('Collection') && !$version->isAtLeast(PdfVersion::V1_7)) { |
| 212 | $warnings[] = "PDF Portfolio (/Collection) requires PDF 1.7, but document declares {$version->value}"; |
| 213 | } |
| 214 | if ($catalog->has('DPartRoot') && !$version->isAtLeast(PdfVersion::V2_0)) { |
| 215 | $warnings[] = "Document parts (/DPartRoot) requires PDF 2.0, but document declares {$version->value}"; |
| 216 | } |
| 217 | if ($catalog->has('DSS') && !$version->isAtLeast(PdfVersion::V2_0)) { |
| 218 | $warnings[] = "Document security store (/DSS) requires PDF 2.0, but document declares {$version->value}"; |
| 219 | } |
| 220 | if ($catalog->has('AF') && !$version->isAtLeast(PdfVersion::V2_0)) { |
| 221 | $warnings[] = "Associated files (/AF) requires PDF 2.0, but document declares {$version->value}"; |
| 222 | } |
| 223 | if ($catalog->has('Requirements') && !$version->isAtLeast(PdfVersion::V1_7)) { |
| 224 | $warnings[] = "Requirements (/Requirements) requires PDF 1.7, but document declares {$version->value}"; |
| 225 | } |
| 226 | } catch (InvalidPdfException) { |
| 227 | // Can't resolve catalog — skip structural checks |
| 228 | } |
| 229 | |
| 230 | // Linearization integrity checks |
| 231 | $linParams = $this->getLinearizationParameters(); |
| 232 | if ($linParams !== null) { |
| 233 | if ($linParams['pageCount'] > 0 && $linParams['pageCount'] !== $this->getPageCount()) { |
| 234 | $warnings[] = sprintf( |
| 235 | 'Linearization /N (%d) does not match actual page count (%d)', |
| 236 | $linParams['pageCount'], |
| 237 | $this->getPageCount(), |
| 238 | ); |
| 239 | } |
| 240 | } |
| 241 | |
| 242 | return $warnings; |
| 243 | } |
| 244 | |
| 245 | /** |
| 246 | * Check whether this PDF is linearized (web-optimized). |
| 247 | * |
| 248 | * A linearized PDF has a LinearizationParameters dictionary as the |
| 249 | * very first indirect object, containing a /Linearized key. The |
| 250 | * reader handles linearized PDFs correctly (via startxref), but |
| 251 | * does not use the hint tables for progressive loading. |
| 252 | */ |
| 253 | public function isLinearized(): bool |
| 254 | { |
| 255 | // Per ISO 32000-2 §F.2, the linearization dict is the first indirect |
| 256 | // object in the file. Most generators assign it object number 1 or 2, |
| 257 | // but the spec doesn't require a specific number. Check the first |
| 258 | // few objects by number, then fall back to scanning the raw bytes. |
| 259 | foreach ([1, 2, 3] as $objNum) { |
| 260 | try { |
| 261 | $obj = $this->resolver->resolve($objNum); |
| 262 | } catch (\Throwable) { |
| 263 | continue; |
| 264 | } |
| 265 | if ($obj instanceof PdfDictionary && $obj->get('Linearized') !== null) { |
| 266 | return true; |
| 267 | } |
| 268 | } |
| 269 | |
| 270 | // Fallback: check all resolved objects for /Linearized key. |
| 271 | // The linearization dict can have any object number. |
| 272 | $trailerSize = $this->trailer->get('Size'); |
| 273 | $maxCheck = min(50, (int) ($trailerSize instanceof PdfNumber |
| 274 | ? $trailerSize->toPdf() : 50)); |
| 275 | for ($i = 4; $i <= $maxCheck; $i++) { |
| 276 | try { |
| 277 | $obj = $this->resolver->resolve($i); |
| 278 | } catch (\Throwable) { |
| 279 | continue; |
| 280 | } |
| 281 | if ($obj instanceof PdfDictionary && $obj->get('Linearized') !== null) { |
| 282 | return true; |
| 283 | } |
| 284 | } |
| 285 | |
| 286 | return false; |
| 287 | } |
| 288 | |
| 289 | /** |
| 290 | * Get linearization parameters if the PDF is linearized. |
| 291 | * |
| 292 | * @return array{linearized: float, fileLength: int, firstPageObj: int, firstPageEnd: int, pageCount: int, xrefOffset: int}|null |
| 293 | */ |
| 294 | public function getLinearizationParameters(): ?array |
| 295 | { |
| 296 | $maxCheck = min(10, (int) ($this->trailer->get('Size') instanceof PdfNumber |
| 297 | ? $this->trailer->get('Size')->toPdf() : 10)); |
| 298 | |
| 299 | for ($objNum = 1; $objNum <= $maxCheck; $objNum++) { |
| 300 | try { |
| 301 | $obj = $this->resolver->resolve($objNum); |
| 302 | } catch (\Throwable) { |
| 303 | continue; |
| 304 | } |
| 305 | if (!$obj instanceof PdfDictionary || $obj->get('Linearized') === null) { |
| 306 | continue; |
| 307 | } |
| 308 | |
| 309 | $getInt = static fn(string $key): int => |
| 310 | ($v = $obj->get($key)) instanceof PdfNumber ? (int) $v->toPdf() : 0; |
| 311 | $getFloat = static fn(string $key): float => |
| 312 | ($v = $obj->get($key)) instanceof PdfNumber ? (float) $v->toPdf() : 0.0; |
| 313 | |
| 314 | return [ |
| 315 | 'linearized' => $getFloat('Linearized'), |
| 316 | 'fileLength' => $getInt('L'), |
| 317 | 'firstPageObj' => $getInt('O'), |
| 318 | 'firstPageEnd' => $getInt('E'), |
| 319 | 'pageCount' => $getInt('N'), |
| 320 | 'xrefOffset' => $getInt('T'), |
| 321 | ]; |
| 322 | } |
| 323 | |
| 324 | return null; |
| 325 | } |
| 326 | |
| 327 | /** |
| 328 | * Parse the page offset hint table from a linearized PDF. |
| 329 | * |
| 330 | * Returns null if the PDF is not linearized or the hint stream |
| 331 | * cannot be located/parsed. |
| 332 | */ |
| 333 | public function getPageOffsetHintTable(): ?PageOffsetHintTable |
| 334 | { |
| 335 | $params = $this->getLinearizationParameters(); |
| 336 | if ($params === null) { |
| 337 | return null; |
| 338 | } |
| 339 | |
| 340 | // Find the /H array from the linearization dict |
| 341 | foreach ([1, 2] as $objNum) { |
| 342 | try { |
| 343 | $obj = $this->resolver->resolve($objNum); |
| 344 | } catch (\Throwable) { |
| 345 | continue; |
| 346 | } |
| 347 | if (!$obj instanceof PdfDictionary || $obj->get('Linearized') === null) { |
| 348 | continue; |
| 349 | } |
| 350 | |
| 351 | $hArray = $obj->get('H'); |
| 352 | if (!$hArray instanceof PdfArray || count($hArray->items) < 2) { |
| 353 | return null; |
| 354 | } |
| 355 | |
| 356 | $hintOffset = $hArray->items[0] instanceof PdfNumber |
| 357 | ? (int) $hArray->items[0]->toPdf() : 0; |
| 358 | $hintLength = $hArray->items[1] instanceof PdfNumber |
| 359 | ? (int) $hArray->items[1]->toPdf() : 0; |
| 360 | |
| 361 | if ($hintOffset <= 0 || $hintLength <= 0) { |
| 362 | return null; |
| 363 | } |
| 364 | |
| 365 | // Find the hint stream object — it's at the given byte offset. |
| 366 | // Look through resolved objects to find one at that offset. |
| 367 | // The hint stream is typically a regular indirect object we can resolve. |
| 368 | // Try to find it by scanning known objects near the offset. |
| 369 | try { |
| 370 | // The hint stream object might be identifiable by iterating objects |
| 371 | // or by directly parsing at the offset. For now, iterate objects |
| 372 | // and find the stream near the linearization dict. |
| 373 | $hintData = null; |
| 374 | $hintDict = null; |
| 375 | |
| 376 | // Try objects 2-10 (hint stream is typically early in the file) |
| 377 | for ($n = 1; $n <= min(20, $params['pageCount'] + 10); $n++) { |
| 378 | try { |
| 379 | $candidate = $this->resolver->resolve($n); |
| 380 | } catch (\Throwable) { |
| 381 | continue; |
| 382 | } |
| 383 | if ( |
| 384 | $candidate instanceof PdfDictionary |
| 385 | && $candidate->has('S') |
| 386 | && ($candidate->get('S') instanceof PdfNumber) |
| 387 | ) { |
| 388 | // This looks like a hint stream dict (has /S for shared obj table offset) |
| 389 | // Check if it's a stream by looking for data |
| 390 | $hintDict = $candidate; |
| 391 | break; |
| 392 | } |
| 393 | } |
| 394 | |
| 395 | // If we found the dict but no stream data, we can't parse hints |
| 396 | if ($hintDict === null) { |
| 397 | return null; |
| 398 | } |
| 399 | |
| 400 | // Get the page offset table offset (usually 0 within the hint data) |
| 401 | $pageTableOffset = 0; // /P offset, default 0 |
| 402 | $pVal = $hintDict->get('P'); |
| 403 | if ($pVal instanceof PdfNumber) { |
| 404 | $pageTableOffset = (int) $pVal->toPdf(); |
| 405 | } |
| 406 | |
| 407 | // For now, return null if we can't get the raw stream data |
| 408 | // (full implementation would parse the stream bytes directly) |
| 409 | return null; |
| 410 | } catch (\Throwable) { |
| 411 | return null; |
| 412 | } |
| 413 | } |
| 414 | |
| 415 | return null; |
| 416 | } |
| 417 | |
| 418 | /** |
| 419 | * Calculate the byte range for a specific page in a linearized PDF. |
| 420 | * |
| 421 | * Returns an associative array with 'offset' and 'length' keys, |
| 422 | * or null if the PDF is not linearized or hints are unavailable. |
| 423 | * |
| 424 | * @return array{offset: int, length: int}|null |
| 425 | */ |
| 426 | public function getPageByteRange(int $pageIndex): ?array |
| 427 | { |
| 428 | $hintTable = $this->getPageOffsetHintTable(); |
| 429 | if ($hintTable === null) { |
| 430 | return null; |
| 431 | } |
| 432 | |
| 433 | try { |
| 434 | return $hintTable->getPageByteRange($pageIndex); |
| 435 | } catch (\OutOfRangeException) { |
| 436 | return null; |
| 437 | } |
| 438 | } |
| 439 | |
| 440 | /** The raw trailer dictionary. */ |
| 441 | public function getTrailer(): PdfDictionary |
| 442 | { |
| 443 | return $this->trailer; |
| 444 | } |
| 445 | |
| 446 | /** Resolve /Root from the trailer — returns the Catalog dictionary. */ |
| 447 | public function getCatalog(): PdfDictionary |
| 448 | { |
| 449 | $root = $this->trailer->get('Root'); |
| 450 | if ($root instanceof PdfReference) { |
| 451 | try { |
| 452 | $obj = $this->resolver->resolveReference($root); |
| 453 | } catch (\Throwable $e) { |
| 454 | if ($this->strict) { |
| 455 | throw $e; |
| 456 | } |
| 457 | $obj = null; |
| 458 | $this->parseWarnings[] = 'Failed to resolve /Root: ' . $e->getMessage(); |
| 459 | } |
| 460 | if ($obj instanceof PdfDictionary) { |
| 461 | return $obj; |
| 462 | } |
| 463 | } |
| 464 | |
| 465 | if (!$this->strict) { |
| 466 | $recovered = $this->recoverCatalog(); |
| 467 | if ($recovered !== null) { |
| 468 | return $recovered; |
| 469 | } |
| 470 | // Last-resort: return an empty dict so callers can keep going. |
| 471 | // The parse warnings record that the document had no usable |
| 472 | // catalog (the test message whitelist is unaffected because |
| 473 | // we no longer throw). |
| 474 | $this->parseWarnings[] = 'No usable /Type /Catalog found; returning empty catalog'; |
| 475 | return new PdfDictionary(); |
| 476 | } |
| 477 | |
| 478 | throw new InvalidPdfException('Unable to resolve /Root catalog'); |
| 479 | } |
| 480 | |
| 481 | /** |
| 482 | * Lenient-mode catalog recovery: scan the file for an object whose |
| 483 | * body contains `/Type /Catalog` (or, failing that, an object that |
| 484 | * looks like a page-tree root referenced as `/Pages`). |
| 485 | */ |
| 486 | private function recoverCatalog(): ?PdfDictionary |
| 487 | { |
| 488 | $map = $this->resolver->scanObjectMap(); |
| 489 | if ($map === []) { |
| 490 | return null; |
| 491 | } |
| 492 | |
| 493 | // Pass 1: find an object whose first ~512 bytes contain `/Type /Catalog`. |
| 494 | foreach ($map as $objNum => $offset) { |
| 495 | $peek = $this->resolver->readRaw($offset, 512); |
| 496 | if (preg_match('#/Type\s*/Catalog\b#', $peek)) { |
| 497 | try { |
| 498 | $obj = $this->resolver->resolve($objNum); |
| 499 | } catch (\Throwable) { |
| 500 | continue; |
| 501 | } |
| 502 | if ($obj instanceof PdfDictionary) { |
| 503 | $this->parseWarnings[] = "Recovered /Root catalog by scanning (object $objNum)"; |
| 504 | return $obj; |
| 505 | } |
| 506 | } |
| 507 | } |
| 508 | |
| 509 | // Pass 2: find an object whose body looks like a page-tree root |
| 510 | // (`/Type /Pages`) and synthesise a minimal catalog pointing at it. |
| 511 | foreach ($map as $objNum => $offset) { |
| 512 | $peek = $this->resolver->readRaw($offset, 512); |
| 513 | if (preg_match('#/Type\s*/Pages\b#', $peek)) { |
| 514 | try { |
| 515 | $obj = $this->resolver->resolve($objNum); |
| 516 | } catch (\Throwable) { |
| 517 | continue; |
| 518 | } |
| 519 | if ($obj instanceof PdfDictionary) { |
| 520 | $synthetic = new PdfDictionary(); |
| 521 | $synthetic->set('Type', new PdfName('Catalog')); |
| 522 | $synthetic->set('Pages', new PdfReference($objNum, 0)); |
| 523 | $this->parseWarnings[] = "Synthesised catalog from /Pages object $objNum"; |
| 524 | return $synthetic; |
| 525 | } |
| 526 | } |
| 527 | } |
| 528 | |
| 529 | return null; |
| 530 | } |
| 531 | |
| 532 | /** Resolve /Info from the trailer. */ |
| 533 | public function getInfo(): ?PdfDictionary |
| 534 | { |
| 535 | $info = $this->trailer->get('Info'); |
| 536 | if ($info instanceof PdfReference) { |
| 537 | $obj = $this->resolver->resolveReference($info); |
| 538 | if ($obj instanceof PdfDictionary) { |
| 539 | return $obj; |
| 540 | } |
| 541 | } |
| 542 | return null; |
| 543 | } |
| 544 | |
| 545 | /** Get the total page count from /Pages -> /Count. */ |
| 546 | public function getPageCount(): int |
| 547 | { |
| 548 | $catalog = $this->getCatalog(); |
| 549 | $pagesRef = $catalog->get('Pages'); |
| 550 | if ($pagesRef instanceof PdfReference) { |
| 551 | $pages = $this->resolver->resolveReference($pagesRef); |
| 552 | if ($pages instanceof PdfDictionary) { |
| 553 | $count = $pages->get('Count'); |
| 554 | if ($count instanceof PdfNumber) { |
| 555 | return (int) $count->toPdf(); |
| 556 | } |
| 557 | } |
| 558 | } |
| 559 | return 0; |
| 560 | } |
| 561 | |
| 562 | /** |
| 563 | * Get all Page dictionaries by traversing the page tree. |
| 564 | * |
| 565 | * @return list<PdfDictionary> |
| 566 | */ |
| 567 | public function getPages(): array |
| 568 | { |
| 569 | $catalog = $this->getCatalog(); |
| 570 | $pagesRef = $catalog->get('Pages'); |
| 571 | if (!$pagesRef instanceof PdfReference) { |
| 572 | return []; |
| 573 | } |
| 574 | $pagesDict = $this->resolver->resolveReference($pagesRef); |
| 575 | if (!$pagesDict instanceof PdfDictionary) { |
| 576 | return []; |
| 577 | } |
| 578 | $result = []; |
| 579 | $this->collectPages($pagesDict, $result); |
| 580 | return $result; |
| 581 | } |
| 582 | |
| 583 | /** Get a specific page by zero-based index. */ |
| 584 | public function getPage(int $index): PdfDictionary |
| 585 | { |
| 586 | $pages = $this->getPages(); |
| 587 | if (!isset($pages[$index])) { |
| 588 | throw new \OutOfRangeException("Page index $index out of range (0.." . (count($pages) - 1) . ')'); |
| 589 | } |
| 590 | return $pages[$index]; |
| 591 | } |
| 592 | |
| 593 | /** Resolve any object by number. */ |
| 594 | public function getObject(int $objNum): Serializable |
| 595 | { |
| 596 | return $this->resolver->resolve($objNum); |
| 597 | } |
| 598 | |
| 599 | /** Resolve an indirect reference to its target. */ |
| 600 | public function resolveReference(PdfReference $ref): Serializable |
| 601 | { |
| 602 | return $this->resolver->resolveReference($ref); |
| 603 | } |
| 604 | |
| 605 | /** The underlying object resolver. */ |
| 606 | public function getResolver(): ObjectResolver |
| 607 | { |
| 608 | return $this->resolver; |
| 609 | } |
| 610 | |
| 611 | // ----------------------------------------------------------------------- |
| 612 | // Text extraction |
| 613 | // ----------------------------------------------------------------------- |
| 614 | |
| 615 | /** |
| 616 | * Extract text from a page by index (zero-based). |
| 617 | * |
| 618 | * Interprets content stream operators, resolves font encodings |
| 619 | * (ToUnicode CMap, /Encoding + /Differences, WinAnsi fallback), |
| 620 | * and infers spacing from text positioning operators. |
| 621 | */ |
| 622 | public function extractText(int $pageIndex): string |
| 623 | { |
| 624 | $page = $this->getPage($pageIndex); |
| 625 | $extractor = new TextExtractor($this->resolver); |
| 626 | return $extractor->extractFromPage($page); |
| 627 | } |
| 628 | |
| 629 | /** |
| 630 | * Extract text from all pages, concatenated with page separators. |
| 631 | * |
| 632 | * @param string $separator Separator between pages (default: newline) |
| 633 | */ |
| 634 | public function extractAllText(string $separator = "\n"): string |
| 635 | { |
| 636 | $pages = $this->getPages(); |
| 637 | $texts = []; |
| 638 | $extractor = new TextExtractor($this->resolver); |
| 639 | foreach ($pages as $page) { |
| 640 | $texts[] = $extractor->extractFromPage($page); |
| 641 | } |
| 642 | return implode($separator, $texts); |
| 643 | } |
| 644 | |
| 645 | /** |
| 646 | * Extract text with precise positioning from a page by index (zero-based). |
| 647 | * |
| 648 | * Returns a list of TextSpan objects, each containing the text content, |
| 649 | * position (x, y in user space), dimensions (width, height), font size, |
| 650 | * and font name. |
| 651 | * |
| 652 | * @return list<TextSpan> |
| 653 | */ |
| 654 | public function extractTextWithPositions(int $pageIndex): array |
| 655 | { |
| 656 | $page = $this->getPage($pageIndex); |
| 657 | $extractor = new PositionedTextExtractor($this->resolver); |
| 658 | return $extractor->extractFromPage($page); |
| 659 | } |
| 660 | |
| 661 | /** |
| 662 | * Extract text with precise positioning from all pages. |
| 663 | * |
| 664 | * @return array<int, list<TextSpan>> Zero-based page index => spans |
| 665 | */ |
| 666 | public function extractAllTextWithPositions(): array |
| 667 | { |
| 668 | $pages = $this->getPages(); |
| 669 | $result = []; |
| 670 | $extractor = new PositionedTextExtractor($this->resolver); |
| 671 | foreach ($pages as $index => $page) { |
| 672 | $result[$index] = $extractor->extractFromPage($page); |
| 673 | } |
| 674 | return $result; |
| 675 | } |
| 676 | |
| 677 | // ----------------------------------------------------------------------- |
| 678 | // Hydration — typed object access |
| 679 | // ----------------------------------------------------------------------- |
| 680 | |
| 681 | /** |
| 682 | * Return the document catalog as a typed Catalog object. |
| 683 | */ |
| 684 | public function getTypedCatalog(): Catalog |
| 685 | { |
| 686 | PdfHydrator::registerDefaults(); |
| 687 | $dict = $this->getCatalog(); |
| 688 | $root = $this->trailer->get('Root'); |
| 689 | $objNum = $root instanceof PdfReference ? $root->objectNumber : 0; |
| 690 | |
| 691 | $result = PdfHydrator::hydrate($dict, $objNum); |
| 692 | if ($result instanceof Catalog) { |
| 693 | return $result; |
| 694 | } |
| 695 | |
| 696 | throw new Exception\InvalidPdfException('Failed to hydrate /Root as Catalog'); |
| 697 | } |
| 698 | |
| 699 | /** |
| 700 | * Return a specific page as a typed Page object. |
| 701 | */ |
| 702 | public function getTypedPage(int $index): Page |
| 703 | { |
| 704 | PdfHydrator::registerDefaults(); |
| 705 | $dict = $this->getPage($index); |
| 706 | |
| 707 | $result = PdfHydrator::hydrate($dict); |
| 708 | if ($result instanceof Page) { |
| 709 | return $result; |
| 710 | } |
| 711 | |
| 712 | throw new Exception\InvalidPdfException("Failed to hydrate page $index as Page"); |
| 713 | } |
| 714 | |
| 715 | /** |
| 716 | * Return all pages as typed Page objects. |
| 717 | * |
| 718 | * @return list<Page> |
| 719 | */ |
| 720 | public function getTypedPages(): array |
| 721 | { |
| 722 | PdfHydrator::registerDefaults(); |
| 723 | $pages = []; |
| 724 | foreach ($this->getPages() as $dict) { |
| 725 | $result = PdfHydrator::hydrate($dict); |
| 726 | if ($result instanceof Page) { |
| 727 | $pages[] = $result; |
| 728 | } |
| 729 | } |
| 730 | return $pages; |
| 731 | } |
| 732 | |
| 733 | /** |
| 734 | * Hydrate any resolved object by object number. |
| 735 | */ |
| 736 | public function getTypedObject(int $objNum): PdfObject|PdfDictionary |
| 737 | { |
| 738 | PdfHydrator::registerDefaults(); |
| 739 | $obj = $this->resolver->resolve($objNum); |
| 740 | if ($obj instanceof PdfDictionary) { |
| 741 | return PdfHydrator::hydrate($obj, $objNum); |
| 742 | } |
| 743 | if ($obj instanceof PdfObject) { |
| 744 | return $obj; |
| 745 | } |
| 746 | return new PdfDictionary(); |
| 747 | } |
| 748 | |
| 749 | // ----------------------------------------------------------------------- |
| 750 | // Internal |
| 751 | // ----------------------------------------------------------------------- |
| 752 | |
| 753 | private static function build( |
| 754 | Source $source, |
| 755 | string $password = '', |
| 756 | bool $strict = true, |
| 757 | ?string $certificate = null, |
| 758 | ?string $privateKey = null, |
| 759 | ): self { |
| 760 | $warnings = []; |
| 761 | |
| 762 | // 1. Validate header — check first 20 bytes, then scan up to 1024 in lenient mode |
| 763 | $header = $source->read(20); |
| 764 | if (preg_match('/^%PDF-(\d+\.\d+)/', $header, $m)) { |
| 765 | $version = $m[1]; |
| 766 | } else { |
| 767 | // Header not at byte 0 — scan first 1024 bytes |
| 768 | $source->seek(0); |
| 769 | $headerBlock = $source->read(min(1024, $source->size())); |
| 770 | if (preg_match('/%PDF-(\d+\.\d+)/', $headerBlock, $m)) { |
| 771 | if ($strict) { |
| 772 | throw new InvalidPdfException('Not a PDF file (missing %PDF- header)'); |
| 773 | } |
| 774 | $version = $m[1]; |
| 775 | $warnings[] = 'PDF header not at byte 0; found at offset ' . strpos($headerBlock, '%PDF-'); |
| 776 | } else { |
| 777 | throw new InvalidPdfException('Not a PDF file (missing %PDF- header)'); |
| 778 | } |
| 779 | } |
| 780 | |
| 781 | // 2. Build parser chain |
| 782 | $tokenizer = new Tokenizer($source); |
| 783 | $objectParser = new ObjectParser($tokenizer, $source); |
| 784 | $streamParser = new StreamParser(); |
| 785 | $xrefParser = new XrefParser($tokenizer, $source, $objectParser); |
| 786 | $xrefStreamParser = new XrefStreamParser($tokenizer, $source, $objectParser, $streamParser); |
| 787 | |
| 788 | // 3. Find startxref + parse xref + trailer — with reconstruction fallback |
| 789 | $entries = null; |
| 790 | $trailer = null; |
| 791 | $reconstructed = false; |
| 792 | |
| 793 | try { |
| 794 | $startxrefOffset = self::findStartxref($source, $strict); |
| 795 | |
| 796 | if ($startxrefOffset !== null) { |
| 797 | // 4. Parse xref + trailer — auto-detect classic vs stream |
| 798 | [$entries, $trailer] = self::parseXrefAt( |
| 799 | $source, |
| 800 | $startxrefOffset, |
| 801 | $xrefParser, |
| 802 | $xrefStreamParser, |
| 803 | $strict, |
| 804 | $warnings, |
| 805 | ); |
| 806 | } |
| 807 | } catch (\Throwable $e) { |
| 808 | if ($strict) { |
| 809 | throw $e instanceof InvalidPdfException ? $e : new InvalidPdfException($e->getMessage(), 0, $e); |
| 810 | } |
| 811 | $warnings[] = 'xref parsing failed: ' . $e->getMessage(); |
| 812 | // Fall through to reconstruction |
| 813 | } |
| 814 | |
| 815 | if ($entries === null || $trailer === null) { |
| 816 | if ($strict) { |
| 817 | throw new InvalidPdfException('Cannot parse xref table or trailer'); |
| 818 | } |
| 819 | [$entries, $trailer] = self::reconstructXref($source); |
| 820 | $warnings[] = 'xref table reconstructed from object scan'; |
| 821 | $reconstructed = true; |
| 822 | } |
| 823 | |
| 824 | // 5. Set up decryptor if /Encrypt is present |
| 825 | $decryptor = null; |
| 826 | $encrypt = $trailer->get('Encrypt'); |
| 827 | if ($encrypt instanceof PdfReference) { |
| 828 | // /Encrypt might be an indirect reference — resolve it |
| 829 | $tempResolver = new ObjectResolver($entries, $tokenizer, $source, $objectParser, $streamParser); |
| 830 | $resolved = $tempResolver->resolveReference($encrypt); |
| 831 | if ($resolved instanceof PdfDictionary) { |
| 832 | $encrypt = $resolved; |
| 833 | } |
| 834 | } |
| 835 | if ($encrypt instanceof PdfDictionary) { |
| 836 | $fileId = self::extractFileId($trailer); |
| 837 | $filter = $encrypt->get('Filter'); |
| 838 | $isPublicKey = $filter instanceof PdfName && $filter->value === 'Adobe.PubSec'; |
| 839 | |
| 840 | if ($isPublicKey && $certificate !== null && $privateKey !== null) { |
| 841 | $decryptor = PdfDecryptor::fromEncryptDictPublicKey( |
| 842 | $encrypt, |
| 843 | $certificate, |
| 844 | $privateKey, |
| 845 | $fileId, |
| 846 | ); |
| 847 | } elseif (!$isPublicKey) { |
| 848 | $decryptor = PdfDecryptor::fromEncryptDict($encrypt, $password, $fileId); |
| 849 | } else { |
| 850 | throw new InvalidPdfException( |
| 851 | 'PDF uses public-key encryption; use fromFilePublicKey() or fromStringPublicKey() with certificate and private key', |
| 852 | ); |
| 853 | } |
| 854 | } |
| 855 | |
| 856 | // 6. Build resolver (with optional decryptor) |
| 857 | $resolver = new ObjectResolver( |
| 858 | $entries, |
| 859 | $tokenizer, |
| 860 | $source, |
| 861 | $objectParser, |
| 862 | $streamParser, |
| 863 | $decryptor, |
| 864 | ); |
| 865 | $resolver->setStrict($strict); |
| 866 | |
| 867 | // Wire resolver into stream parser for resolving indirect /DecodeParms |
| 868 | $streamParser->setResolver($resolver); |
| 869 | |
| 870 | // 7. Follow /Prev chain for incremental updates (skip if reconstructed) |
| 871 | if (!$reconstructed) { |
| 872 | $prev = $trailer->get('Prev'); |
| 873 | $seenPrevOffsets = []; |
| 874 | while ($prev instanceof PdfNumber) { |
| 875 | $prevOffset = (int) $prev->toPdf(); |
| 876 | |
| 877 | // Detect circular /Prev chains |
| 878 | if (isset($seenPrevOffsets[$prevOffset])) { |
| 879 | $warnings[] = "Circular /Prev chain detected at offset $prevOffset"; |
| 880 | break; |
| 881 | } |
| 882 | $seenPrevOffsets[$prevOffset] = true; |
| 883 | |
| 884 | try { |
| 885 | [$olderEntries, $olderTrailer] = self::parseXrefAt( |
| 886 | $source, |
| 887 | $prevOffset, |
| 888 | $xrefParser, |
| 889 | $xrefStreamParser, |
| 890 | $strict, |
| 891 | $warnings, |
| 892 | ); |
| 893 | $resolver->mergeOlderEntries($olderEntries); |
| 894 | $prev = $olderTrailer->get('Prev'); |
| 895 | } catch (\Throwable $e) { |
| 896 | if ($strict) { |
| 897 | throw $e instanceof InvalidPdfException ? $e : new InvalidPdfException($e->getMessage(), 0, $e); |
| 898 | } |
| 899 | $warnings[] = "/Prev chain parsing failed at offset $prevOffset: " . $e->getMessage(); |
| 900 | break; |
| 901 | } |
| 902 | } |
| 903 | } |
| 904 | |
| 905 | $reader = new self($version, $trailer, $resolver); |
| 906 | $reader->parseWarnings = $warnings; |
| 907 | $reader->strict = $strict; |
| 908 | return $reader; |
| 909 | } |
| 910 | |
| 911 | /** |
| 912 | * Extract the first element of the /ID array from the trailer. |
| 913 | */ |
| 914 | private static function extractFileId(PdfDictionary $trailer): string |
| 915 | { |
| 916 | $id = $trailer->get('ID'); |
| 917 | if ($id instanceof PdfArray && isset($id->items[0])) { |
| 918 | $first = $id->items[0]; |
| 919 | if ($first instanceof PdfString) { |
| 920 | return $first->value; |
| 921 | } |
| 922 | } |
| 923 | return ''; |
| 924 | } |
| 925 | |
| 926 | /** |
| 927 | * Auto-detect classic xref table vs cross-reference stream at the |
| 928 | * given offset and parse accordingly. |
| 929 | * |
| 930 | * @return array{0: array<int, XrefEntry>, 1: PdfDictionary} |
| 931 | */ |
| 932 | /** |
| 933 | * @param list<string> $warnings |
| 934 | * @return array{0: array<int, XrefEntry>, 1: PdfDictionary} |
| 935 | */ |
| 936 | private static function parseXrefAt( |
| 937 | Source $source, |
| 938 | int $offset, |
| 939 | XrefParser $classicParser, |
| 940 | XrefStreamParser $streamParser, |
| 941 | bool $strict = true, |
| 942 | array &$warnings = [], |
| 943 | ): array { |
| 944 | // Peek at the bytes at the offset to decide which parser to use. |
| 945 | $source->seek($offset); |
| 946 | $peek = $source->peek(4); |
| 947 | if (str_starts_with($peek, 'xref')) { |
| 948 | return $classicParser->parseClassicXref($offset, $strict, $warnings); |
| 949 | } |
| 950 | // Otherwise assume it's a cross-reference stream (starts with "N M obj") |
| 951 | return $streamParser->parseXrefStream($offset); |
| 952 | } |
| 953 | |
| 954 | /** |
| 955 | * Scan backward from EOF to find the `startxref` byte offset. |
| 956 | * |
| 957 | * Returns null if startxref is missing or corrupted (allows lenient |
| 958 | * fallback to xref reconstruction). |
| 959 | */ |
| 960 | private static function findStartxref(Source $source, bool $strict = true): ?int |
| 961 | { |
| 962 | $size = $source->size(); |
| 963 | |
| 964 | // Try progressively larger tail sizes: 1024, 8192, 65536 |
| 965 | foreach ([1024, 8192, 65536] as $tryLength) { |
| 966 | $tailLength = min($tryLength, $size); |
| 967 | $source->seek($size - $tailLength); |
| 968 | $tail = $source->read($tailLength); |
| 969 | |
| 970 | $pos = strrpos($tail, 'startxref'); |
| 971 | if ($pos !== false) { |
| 972 | $after = substr($tail, $pos + strlen('startxref')); |
| 973 | if (preg_match('/\s+(\d+)/', $after, $m)) { |
| 974 | return (int) $m[1]; |
| 975 | } |
| 976 | } |
| 977 | } |
| 978 | |
| 979 | if ($strict) { |
| 980 | throw new InvalidPdfException('Cannot find startxref'); |
| 981 | } |
| 982 | return null; |
| 983 | } |
| 984 | |
| 985 | /** |
| 986 | * Reconstruct xref entries and trailer by scanning for object definitions. |
| 987 | * |
| 988 | * Used as a fallback when the normal xref/trailer parsing fails in lenient mode. |
| 989 | * |
| 990 | * @return array{0: array<int, XrefEntry>, 1: PdfDictionary} |
| 991 | */ |
| 992 | private static function reconstructXref(Source $source): array |
| 993 | { |
| 994 | $source->seek(0); |
| 995 | $allBytes = $source->read($source->size()); |
| 996 | |
| 997 | $objectMap = ObjectScanner::scan($allBytes); |
| 998 | |
| 999 | if ($objectMap === []) { |
| 1000 | // No object headers found. Return an empty xref + synthetic |
| 1001 | // empty catalog so the caller can still emit a parseable |
| 1002 | // (if useless) document. Higher layers should treat this as |
| 1003 | // an irrecoverable file. |
| 1004 | $trailer = new PdfDictionary(); |
| 1005 | $trailer->set('Size', new PdfNumber(1)); |
| 1006 | return [[], $trailer]; |
| 1007 | } |
| 1008 | |
| 1009 | // Build xref entries |
| 1010 | $entries = []; |
| 1011 | foreach ($objectMap as $objNum => $offset) { |
| 1012 | $entries[$objNum] = new XrefEntry(XrefEntry::TYPE_IN_USE, $offset, 0); |
| 1013 | } |
| 1014 | |
| 1015 | $catalogObjNum = self::findCatalogInScan($objectMap, $allBytes); |
| 1016 | |
| 1017 | $maxObjNum = max(array_keys($objectMap)); |
| 1018 | $trailer = new PdfDictionary(); |
| 1019 | if ($catalogObjNum !== null) { |
| 1020 | $trailer->set('Root', new PdfReference($catalogObjNum, 0)); |
| 1021 | } |
| 1022 | $trailer->set('Size', new PdfNumber($maxObjNum + 1)); |
| 1023 | |
| 1024 | return [$entries, $trailer]; |
| 1025 | } |
| 1026 | |
| 1027 | /** |
| 1028 | * Identify which scanned object is the catalog using progressively |
| 1029 | * looser heuristics. Returns the object number or null if no |
| 1030 | * reasonable candidate is found. |
| 1031 | * |
| 1032 | * @param array<int, int> $objectMap |
| 1033 | */ |
| 1034 | private static function findCatalogInScan(array $objectMap, string $allBytes): ?int |
| 1035 | { |
| 1036 | $bytesLen = strlen($allBytes); |
| 1037 | $peek = static function (int $offset) use ($allBytes, $bytesLen): string { |
| 1038 | $peekLength = min(1024, $bytesLen - $offset); |
| 1039 | return $peekLength > 0 ? substr($allBytes, $offset, $peekLength) : ''; |
| 1040 | }; |
| 1041 | |
| 1042 | // Pass 1: explicit /Type /Catalog |
| 1043 | foreach ($objectMap as $objNum => $offset) { |
| 1044 | if (preg_match('#/Type\s*/Catalog\b#', $peek($offset))) { |
| 1045 | return $objNum; |
| 1046 | } |
| 1047 | } |
| 1048 | |
| 1049 | // Pass 2: a dict that has /Pages but no /Parent (heuristic for |
| 1050 | // catalog-with-missing-/Type, e.g. qpdf's bad8/bad11 stripped |
| 1051 | // catalogs). |
| 1052 | foreach ($objectMap as $objNum => $offset) { |
| 1053 | $body = $peek($offset); |
| 1054 | if (preg_match('#/Pages\s+\d+\s+\d+\s+R#', $body) && !preg_match('#/Parent\b#', $body)) { |
| 1055 | return $objNum; |
| 1056 | } |
| 1057 | } |
| 1058 | |
| 1059 | // Pass 3: object body literally contains the word "Catalog" |
| 1060 | // somewhere (covers PDFs like bug_454695 where /Type /Catalog is |
| 1061 | // formatted oddly or appears inside a hex string comment). |
| 1062 | foreach ($objectMap as $objNum => $offset) { |
| 1063 | if (str_contains($peek($offset), 'Catalog')) { |
| 1064 | return $objNum; |
| 1065 | } |
| 1066 | } |
| 1067 | |
| 1068 | return null; |
| 1069 | } |
| 1070 | |
| 1071 | /** |
| 1072 | * Recursively collect Page dicts from a Pages tree node. |
| 1073 | * |
| 1074 | * @param list<PdfDictionary> $result |
| 1075 | */ |
| 1076 | private function collectPages(PdfDictionary $node, array &$result): void |
| 1077 | { |
| 1078 | $kids = $node->get('Kids'); |
| 1079 | if (!$kids instanceof PdfArray) { |
| 1080 | return; |
| 1081 | } |
| 1082 | foreach ($kids->items as $kidRef) { |
| 1083 | if (!$kidRef instanceof PdfReference) { |
| 1084 | continue; |
| 1085 | } |
| 1086 | $kid = $this->resolver->resolveReference($kidRef); |
| 1087 | if (!$kid instanceof PdfDictionary) { |
| 1088 | continue; |
| 1089 | } |
| 1090 | $type = $kid->get('Type'); |
| 1091 | if ($type instanceof PdfName && $type->value === 'Pages') { |
| 1092 | $this->collectPages($kid, $result); |
| 1093 | } else { |
| 1094 | $result[] = $kid; |
| 1095 | } |
| 1096 | } |
| 1097 | } |
| 1098 | } |