Skip to content

Commit 6a30d54

Browse files
JanSlabonJanSlabon
authored
Merge pull request #182 from Setasign/development
Development
2 parents f4ba73e + 4a4ca8d commit 6a30d54

12 files changed

Lines changed: 362 additions & 32 deletions

File tree

‎src/PdfParser/PdfParser.php‎

Lines changed: 68 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -25,6 +25,7 @@
2525
use setasign\Fpdi\PdfParser\Type\PdfString;
2626
use setasign\Fpdi\PdfParser\Type\PdfToken;
2727
use setasign\Fpdi\PdfParser\Type\PdfType;
28+
use setasign\Fpdi\PdfParser\Type\PdfTypeException;
2829

2930
/**
3031
* A PDF parser class
@@ -258,25 +259,25 @@ public function readValue($token = null, $expectedType = null)
258259
switch ($token) {
259260
case '(':
260261
$this->ensureExpectedType($token, $expectedType);
261-
return PdfString::parse($this->streamReader);
262+
return $this->parsePdfString();
262263

263264
case '<':
264265
if ($this->streamReader->getByte() === '<') {
265266
$this->ensureExpectedType('<<', $expectedType);
266267
$this->streamReader->addOffset(1);
267-
return PdfDictionary::parse($this->tokenizer, $this->streamReader, $this);
268+
return $this->parsePdfDictionary();
268269
}
269270

270271
$this->ensureExpectedType($token, $expectedType);
271-
return PdfHexString::parse($this->streamReader);
272+
return $this->parsePdfHexString();
272273

273274
case '/':
274275
$this->ensureExpectedType($token, $expectedType);
275-
return PdfName::parse($this->tokenizer, $this->streamReader);
276+
return $this->parsePdfName();
276277

277278
case '[':
278279
$this->ensureExpectedType($token, $expectedType);
279-
return PdfArray::parse($this->tokenizer, $this);
280+
return $this->parsePdfArray();
280281

281282
default:
282283
if (\is_numeric($token)) {
@@ -291,13 +292,7 @@ public function readValue($token = null, $expectedType = null)
291292
);
292293
}
293294

294-
return PdfIndirectObject::parse(
295-
(int) $token,
296-
(int) $token2,
297-
$this,
298-
$this->tokenizer,
299-
$this->streamReader
300-
);
295+
return $this->parsePdfIndirectObject((int)$token, (int)$token2);
301296
case 'R':
302297
if (
303298
$expectedType !== null &&
@@ -309,7 +304,7 @@ public function readValue($token = null, $expectedType = null)
309304
);
310305
}
311306

312-
return PdfIndirectObjectReference::create((int) $token, (int) $token2);
307+
return PdfIndirectObjectReference::create((int)$token, (int)$token2);
313308
}
314309

315310
$this->tokenizer->pushStack($token3);
@@ -351,6 +346,65 @@ public function readValue($token = null, $expectedType = null)
351346
}
352347
}
353348

349+
/**
350+
* @return PdfString
351+
*/
352+
protected function parsePdfString()
353+
{
354+
return PdfString::parse($this->streamReader);
355+
}
356+
357+
/**
358+
* @return false|PdfHexString
359+
*/
360+
protected function parsePdfHexString()
361+
{
362+
return PdfHexString::parse($this->streamReader);
363+
}
364+
365+
/**
366+
* @return bool|PdfDictionary
367+
* @throws PdfTypeException
368+
*/
369+
protected function parsePdfDictionary()
370+
{
371+
return PdfDictionary::parse($this->tokenizer, $this->streamReader, $this);
372+
}
373+
374+
/**
375+
* @return PdfName
376+
*/
377+
protected function parsePdfName()
378+
{
379+
return PdfName::parse($this->tokenizer, $this->streamReader);
380+
}
381+
382+
/**
383+
* @return false|PdfArray
384+
* @throws PdfTypeException
385+
*/
386+
protected function parsePdfArray()
387+
{
388+
return PdfArray::parse($this->tokenizer, $this);
389+
}
390+
391+
/**
392+
* @param int $objectNumber
393+
* @param int $generationNumber
394+
* @return false|PdfIndirectObject
395+
* @throws Type\PdfTypeException
396+
*/
397+
protected function parsePdfIndirectObject($objectNumber, $generationNumber)
398+
{
399+
return PdfIndirectObject::parse(
400+
$objectNumber,
401+
$generationNumber,
402+
$this,
403+
$this->tokenizer,
404+
$this->streamReader
405+
);
406+
}
407+
354408
/**
355409
* Ensures that the token will evaluate to an expected object type (or not).
356410
*
@@ -359,7 +413,7 @@ public function readValue($token = null, $expectedType = null)
359413
* @return bool
360414
* @throws Type\PdfTypeException
361415
*/
362-
private function ensureExpectedType($token, $expectedType)
416+
protected function ensureExpectedType($token, $expectedType)
363417
{
364418
static $mapping = [
365419
'(' => PdfString::class,

‎src/PdfParser/StreamReader.php‎

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -113,6 +113,12 @@ public function __construct($stream, $closeStream = false)
113113
);
114114
}
115115

116+
if (fseek($stream, 0) === -1) {
117+
throw new \InvalidArgumentException(
118+
'Given stream is not seekable!'
119+
);
120+
}
121+
116122
$this->stream = $stream;
117123
$this->closeStream = $closeStream;
118124
$this->reset();

‎src/PdfParser/Type/PdfArray.php‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -25,7 +25,7 @@ class PdfArray extends PdfType
2525
*
2626
* @param Tokenizer $tokenizer
2727
* @param PdfParser $parser
28-
* @return bool|self
28+
* @return false|self
2929
* @throws PdfTypeException
3030
*/
3131
public static function parse(Tokenizer $tokenizer, PdfParser $parser)

‎src/PdfParser/Type/PdfHexString.php‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -21,7 +21,7 @@ class PdfHexString extends PdfType
2121
* Parses a hexadecimal string object from the stream reader.
2222
*
2323
* @param StreamReader $streamReader
24-
* @return bool|self
24+
* @return false|self
2525
*/
2626
public static function parse(StreamReader $streamReader)
2727
{

‎src/PdfParser/Type/PdfIndirectObject.php‎

Lines changed: 7 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -22,17 +22,17 @@ class PdfIndirectObject extends PdfType
2222
/**
2323
* Parses an indirect object from a tokenizer, parser and stream-reader.
2424
*
25-
* @param int $objectNumberToken
26-
* @param int $objectGenerationNumberToken
25+
* @param int $objectNumber
26+
* @param int $objectGenerationNumber
2727
* @param PdfParser $parser
2828
* @param Tokenizer $tokenizer
2929
* @param StreamReader $reader
30-
* @return bool|self
30+
* @return self|false
3131
* @throws PdfTypeException
3232
*/
3333
public static function parse(
34-
$objectNumberToken,
35-
$objectGenerationNumberToken,
34+
$objectNumber,
35+
$objectGenerationNumber,
3636
PdfParser $parser,
3737
Tokenizer $tokenizer,
3838
StreamReader $reader
@@ -50,8 +50,8 @@ public static function parse(
5050
}
5151

5252
$v = new self();
53-
$v->objectNumber = (int) $objectNumberToken;
54-
$v->generationNumber = (int) $objectGenerationNumberToken;
53+
$v->objectNumber = (int) $objectNumber;
54+
$v->generationNumber = (int) $objectGenerationNumber;
5555
$v->value = $value;
5656

5757
return $v;

‎src/PdfParser/Type/PdfStream.php‎

Lines changed: 38 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -213,18 +213,16 @@ protected function extractStream()
213213
}
214214

215215
/**
216-
* Get the unfiltered stream data.
216+
* Get all filters defined for this stream.
217217
*
218-
* @return string
219-
* @throws FilterException
220-
* @throws PdfParserException
218+
* @return PdfType[]
219+
* @throws PdfTypeException
221220
*/
222-
public function getUnfilteredStream()
221+
public function getFilters()
223222
{
224-
$stream = $this->getStream();
225223
$filters = PdfDictionary::get($this->value, 'Filter');
226224
if ($filters instanceof PdfNull) {
227-
return $stream;
225+
return [];
228226
}
229227

230228
if ($filters instanceof PdfArray) {
@@ -233,6 +231,24 @@ public function getUnfilteredStream()
233231
$filters = [$filters];
234232
}
235233

234+
return $filters;
235+
}
236+
237+
/**
238+
* Get the unfiltered stream data.
239+
*
240+
* @return string
241+
* @throws FilterException
242+
* @throws PdfParserException
243+
*/
244+
public function getUnfilteredStream()
245+
{
246+
$stream = $this->getStream();
247+
$filters = $this->getFilters();
248+
if ($filters === []) {
249+
return $stream;
250+
}
251+
236252
$decodeParams = PdfDictionary::get($this->value, 'DecodeParms');
237253
if ($decodeParams instanceof PdfArray) {
238254
$decodeParams = $decodeParams->value;
@@ -308,6 +324,21 @@ public function getUnfilteredStream()
308324
$stream = $filterObject->decode($stream);
309325
break;
310326

327+
case 'Crypt':
328+
if (!$decodeParam instanceof PdfDictionary) {
329+
break;
330+
}
331+
// Filter is "Identity"
332+
$name = PdfDictionary::get($decodeParam, 'Name');
333+
if (!$name instanceof PdfName || $name->value !== 'Identity') {
334+
break;
335+
}
336+
337+
throw new FilterException(
338+
'Support for Crypt filters other than "Identity" is not implemented.',
339+
FilterException::UNSUPPORTED_FILTER
340+
);
341+
311342
default:
312343
throw new FilterException(
313344
\sprintf('Unsupported filter "%s".', $filter->value),

‎src/PdfParser/Type/PdfString.php‎

Lines changed: 30 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -78,6 +78,36 @@ public static function ensure($string)
7878
return PdfType::ensureType(self::class, $string, 'String value expected.');
7979
}
8080

81+
/**
82+
* Escapes sequences in a string according to the PDF specification.
83+
*
84+
* @param string $s
85+
* @return string
86+
*/
87+
public static function escape($s)
88+
{
89+
// Still a bit faster, than direct replacing
90+
if (
91+
\strpos($s, '\\') !== false ||
92+
\strpos($s, ')') !== false ||
93+
\strpos($s, '(') !== false ||
94+
\strpos($s, "\x0D") !== false ||
95+
\strpos($s, "\x0A") !== false ||
96+
\strpos($s, "\x09") !== false ||
97+
\strpos($s, "\x08") !== false ||
98+
\strpos($s, "\x0C") !== false
99+
) {
100+
// is faster than strtr(...)
101+
return \str_replace(
102+
['\\', ')', '(', "\x0D", "\x0A", "\x09", "\x08", "\x0C"],
103+
['\\\\', '\\)', '\\(', '\r', '\n', '\t', '\b', '\f'],
104+
$s
105+
);
106+
}
107+
108+
return $s;
109+
}
110+
81111
/**
82112
* Unescapes escaped sequences in a PDF string according to the PDF specification.
83113
*

‎src/Tcpdf/Fpdi.php‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -251,7 +251,7 @@ protected function writePdfType(PdfType $value)
251251
if ($value instanceof PdfString) {
252252
$string = PdfString::unescape($value->value);
253253
$string = $this->_encrypt_data($this->currentObjectNumber, $string);
254-
$value->value = \TCPDF_STATIC::_escape($string);
254+
$value->value = PdfString::escape($string);
255255
} elseif ($value instanceof PdfHexString) {
256256
$filter = new AsciiHex();
257257
$string = $filter->decode($value->value);

‎tests/functional/PdfParser/Type/PdfStreamTest.php‎

Lines changed: 21 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -423,4 +423,25 @@ public function testGetUnfilteredStream($file, $objectNumber, $expectedResult)
423423

424424
$this->assertEquals($expectedResult, $stream->getUnfilteredStream());
425425
}
426+
427+
public function testParseWithCryptFilter()
428+
{
429+
$in = "123 0 obj\n<</Filter /Crypt /Length 5>>\nstream\nHello\nendstream\nendobj";
430+
431+
$stream = StreamReader::createByString($in);
432+
433+
// set position and prepare dictionary (equals to result)
434+
$stream->setOffset(45);
435+
$this->assertSame("\n", $stream->getByte()); // this is the \n after the stream keyword
436+
437+
$dict = PdfDictionary::create([
438+
'Filter' => PdfName::create('Crypt'),
439+
'Length' => PdfNumeric::create(5)
440+
]);
441+
442+
$result = PdfStream::parse($dict, $stream);
443+
444+
$this->assertSame($dict, $result->value);
445+
$this->assertSame('Hello', $result->getUnfilteredStream());
446+
}
426447
}

‎tests/unit/PdfParser/DummyFaultyStreamWrapper.php‎

Lines changed: 52 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,52 @@
1+
<?php
2+
3+
namespace setasign\Fpdi\unit\PdfParser;
4+
5+
/**
6+
* This stream wrapper is created to simulate a stream wrapper that is not seekable (while its metadata report that
7+
* it is seekable).
8+
*/
9+
class DummyFaultyStreamWrapper
10+
{
11+
public $context;
12+
13+
function stream_open($path, $mode, $options, &$opened_path)
14+
{
15+
return true;
16+
}
17+
18+
function stream_read($count)
19+
{
20+
return '';
21+
}
22+
23+
function stream_write($data)
24+
{
25+
return 0;
26+
}
27+
28+
function stream_tell()
29+
{
30+
return 0;
31+
}
32+
33+
function stream_eof()
34+
{
35+
return false;
36+
}
37+
38+
function stream_seek($offset, $whence)
39+
{
40+
return false;
41+
}
42+
43+
function stream_metadata($path, $option, $var)
44+
{
45+
return true;
46+
}
47+
48+
function stream_stat()
49+
{
50+
return [];
51+
}
52+
}

0 commit comments

Comments
 (0)