OLERead::getStream() - Code Metrics - Inspection of "Xls file threw exception during open by Xls reader" - PHPOffice/PhpSpreadsheet - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — develop ( ae9dd1...08b445 )

by Adrien

created 2018-10-07 07:50 UTC

OLERead::getStream() B

↳ Parent: OLERead

Complexity

Conditions	7
Paths	9

Size

Total Lines	40
Code Lines	22

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	22
CRAP Score	7.004

Importance

Changes

Metric	Value
cc	7
eloc	22
nc	9
nop	1
dl	0
loc	40
ccs	22
cts	23
cp	0.9565
crap	7.004
rs	8.6346
c	0
b	0
f	0

<?php

namespace PhpOffice\PhpSpreadsheet\Shared;

use PhpOffice\PhpSpreadsheet\Reader\Exception as ReaderException;

class OLERead
{
    private $data = '';

    // Size of a sector = 512 bytes
    const BIG_BLOCK_SIZE = 0x200;

    // Size of a short sector = 64 bytes
    const SMALL_BLOCK_SIZE = 0x40;

    // Size of a directory entry always = 128 bytes
    const PROPERTY_STORAGE_BLOCK_SIZE = 0x80;

    // Minimum size of a standard stream = 4096 bytes, streams smaller than this are stored as short streams
    const SMALL_BLOCK_THRESHOLD = 0x1000;

    // header offsets
    const NUM_BIG_BLOCK_DEPOT_BLOCKS_POS = 0x2c;
    const ROOT_START_BLOCK_POS = 0x30;
    const SMALL_BLOCK_DEPOT_BLOCK_POS = 0x3c;
    const EXTENSION_BLOCK_POS = 0x44;
    const NUM_EXTENSION_BLOCK_POS = 0x48;
    const BIG_BLOCK_DEPOT_BLOCKS_POS = 0x4c;

    // property storage offsets (directory offsets)
    const SIZE_OF_NAME_POS = 0x40;
    const TYPE_POS = 0x42;
    const START_BLOCK_POS = 0x74;
    const SIZE_POS = 0x78;

    public $wrkbook;

    public $summaryInformation;

    public $documentSummaryInformation;

    /**
     * @var int
     */
    private $numBigBlockDepotBlocks;

    /**
     * @var int
     */
    private $rootStartBlock;

    /**
     * @var int
     */
    private $sbdStartBlock;

    /**
     * @var int
     */
    private $extensionBlock;

    /**
     * @var int
     */
    private $numExtensionBlocks;

    /**
     * @var string
     */
    private $bigBlockChain;

    /**
     * @var string
     */
    private $smallBlockChain;

    /**
     * @var string
     */
    private $entry;

    /**
     * @var int
     */
    private $rootentry;

    /**
     * @var array
     */
    private $props = [];

    /**
     * Read the file.
     *
     * @param $pFilename string Filename
     *
     * @throws ReaderException
     */
    public function read($pFilename)
    {
        File::assertFile($pFilename);

        // Get the file identifier
        // Don't bother reading the whole file until we know it's a valid OLE file
        $this->data = file_get_contents($pFilename, false, null, 0, 8);

        // Check OLE identifier
        $identifierOle = pack('CCCCCCCC', 0xd0, 0xcf, 0x11, 0xe0, 0xa1, 0xb1, 0x1a, 0xe1);
        if ($this->data != $identifierOle) {
            throw new ReaderException('The filename ' . $pFilename . ' is not recognised as an OLE file');
        }

        // Get the file data
        $this->data = file_get_contents($pFilename);

        // Total number of sectors used for the SAT
        $this->numBigBlockDepotBlocks = self::getInt4d($this->data, self::NUM_BIG_BLOCK_DEPOT_BLOCKS_POS);

        // SecID of the first sector of the directory stream
        $this->rootStartBlock = self::getInt4d($this->data, self::ROOT_START_BLOCK_POS);

        // SecID of the first sector of the SSAT (or -2 if not extant)
        $this->sbdStartBlock = self::getInt4d($this->data, self::SMALL_BLOCK_DEPOT_BLOCK_POS);

        // SecID of the first sector of the MSAT (or -2 if no additional sectors are used)
        $this->extensionBlock = self::getInt4d($this->data, self::EXTENSION_BLOCK_POS);

        // Total number of sectors used by MSAT
        $this->numExtensionBlocks = self::getInt4d($this->data, self::NUM_EXTENSION_BLOCK_POS);

        $bigBlockDepotBlocks = [];
        $pos = self::BIG_BLOCK_DEPOT_BLOCKS_POS;

        $bbdBlocks = $this->numBigBlockDepotBlocks;

        if ($this->numExtensionBlocks != 0) {
            $bbdBlocks = (self::BIG_BLOCK_SIZE - self::BIG_BLOCK_DEPOT_BLOCKS_POS) / 4;
        }

        for ($i = 0; $i < $bbdBlocks; ++$i) {
            $bigBlockDepotBlocks[$i] = self::getInt4d($this->data, $pos);
            $pos += 4;
        }

        for ($j = 0; $j < $this->numExtensionBlocks; ++$j) {
            $pos = ($this->extensionBlock + 1) * self::BIG_BLOCK_SIZE;
            $blocksToRead = min($this->numBigBlockDepotBlocks - $bbdBlocks, self::BIG_BLOCK_SIZE / 4 - 1);

            for ($i = $bbdBlocks; $i < $bbdBlocks + $blocksToRead; ++$i) {
                $bigBlockDepotBlocks[$i] = self::getInt4d($this->data, $pos);
                $pos += 4;
            }

            $bbdBlocks += $blocksToRead;
            if ($bbdBlocks < $this->numBigBlockDepotBlocks) {
                $this->extensionBlock = self::getInt4d($this->data, $pos);
            }
        }

        $pos = 0;
        $this->bigBlockChain = '';
        $bbs = self::BIG_BLOCK_SIZE / 4;
        for ($i = 0; $i < $this->numBigBlockDepotBlocks; ++$i) {
            $pos = ($bigBlockDepotBlocks[$i] + 1) * self::BIG_BLOCK_SIZE;

            $this->bigBlockChain .= substr($this->data, $pos, 4 * $bbs);
            $pos += 4 * $bbs;
        }

        $pos = 0;

        $sbdBlock = $this->sbdStartBlock;
        $this->smallBlockChain = '';
        while ($sbdBlock != -2) {
            $pos = ($sbdBlock + 1) * self::BIG_BLOCK_SIZE;

            $this->smallBlockChain .= substr($this->data, $pos, 4 * $bbs);
            $pos += 4 * $bbs;

            $sbdBlock = self::getInt4d($this->bigBlockChain, $sbdBlock * 4);
        }

        // read the directory stream
        $block = $this->rootStartBlock;
        $this->entry = $this->_readData($block);

        $this->readPropertySets();
    }

    /**
     * Extract binary stream data.
     *
     * @param int $stream
     *
     * @return string
     */
    public function getStream($stream)
    {
        if ($stream === null) {

            return null;
        }

        $streamData = '';

        if ($this->props[$stream]['size'] < self::SMALL_BLOCK_THRESHOLD) {
            $rootdata = $this->_readData($this->props[$this->rootentry]['startBlock']);

            $block = $this->props[$stream]['startBlock'];

            while ($block != -2) {
                $pos = $block * self::SMALL_BLOCK_SIZE;
                $streamData .= substr($rootdata, $pos, self::SMALL_BLOCK_SIZE);

                $block = self::getInt4d($this->smallBlockChain, $block * 4);
            }

            return $streamData;
        }
        $numBlocks = $this->props[$stream]['size'] / self::BIG_BLOCK_SIZE;
        if ($this->props[$stream]['size'] % self::BIG_BLOCK_SIZE != 0) {
            ++$numBlocks;
        }

        if ($numBlocks == 0) {
            return '';
        }

        $block = $this->props[$stream]['startBlock'];

        while ($block != -2) {
            $pos = ($block + 1) * self::BIG_BLOCK_SIZE;
            $streamData .= substr($this->data, $pos, self::BIG_BLOCK_SIZE);
            $block = self::getInt4d($this->bigBlockChain, $block * 4);
        }

        return $streamData;
    }

    /**
     * Read a standard stream (by joining sectors using information from SAT).
     *
     * @param int $bl Sector ID where the stream starts
     *
     * @return string Data for standard stream
     */
    private function _readData($bl)
    {
        $block = $bl;
        $data = '';

        while ($block != -2) {
            $pos = ($block + 1) * self::BIG_BLOCK_SIZE;
            $data .= substr($this->data, $pos, self::BIG_BLOCK_SIZE);
            $block = self::getInt4d($this->bigBlockChain, $block * 4);
        }

        return $data;
    }

    /**
     * Read entries in the directory stream.
     */
    private function readPropertySets()
    {
        $offset = 0;

        // loop through entires, each entry is 128 bytes
        $entryLen = strlen($this->entry);
        while ($offset < $entryLen) {
            // entry data (128 bytes)
            $d = substr($this->entry, $offset, self::PROPERTY_STORAGE_BLOCK_SIZE);

            // size in bytes of name
            $nameSize = ord($d[self::SIZE_OF_NAME_POS]) | (ord($d[self::SIZE_OF_NAME_POS + 1]) << 8);

            // type of entry
            $type = ord($d[self::TYPE_POS]);

            // sectorID of first sector or short sector, if this entry refers to a stream (the case with workbook)
            // sectorID of first sector of the short-stream container stream, if this entry is root entry
            $startBlock = self::getInt4d($d, self::START_BLOCK_POS);

            $size = self::getInt4d($d, self::SIZE_POS);

            $name = str_replace("\x00", '', substr($d, 0, $nameSize));

            $this->props[] = [
                'name' => $name,
                'type' => $type,
                'startBlock' => $startBlock,
                'size' => $size,
            ];

            // tmp helper to simplify checks
            $upName = strtoupper($name);

            // Workbook directory entry (BIFF5 uses Book, BIFF8 uses Workbook)
            if (($upName === 'WORKBOOK') || ($upName === 'BOOK')) {
                $this->wrkbook = count($this->props) - 1;
            } elseif ($upName === 'ROOT ENTRY' || $upName === 'R') {
                // Root entry
                $this->rootentry = count($this->props) - 1;
            }

            // Summary information
            if ($name == chr(5) . 'SummaryInformation') {
                $this->summaryInformation = count($this->props) - 1;
            }

            // Additional Document Summary information
            if ($name == chr(5) . 'DocumentSummaryInformation') {
                $this->documentSummaryInformation = count($this->props) - 1;
            }

            $offset += self::PROPERTY_STORAGE_BLOCK_SIZE;
        }
    }

    /**
     * Read 4 bytes of data at specified position.
     *
     * @param string $data
     * @param int $pos
     *
     * @return int
     */
    private static function getInt4d($data, $pos)
    {
        if ($pos < 0) {
            // Invalid position
            throw new ReaderException('Parameter pos=' . $pos . ' is invalid.');
        }

        $len = strlen($data);
        if ($len < $pos + 4) {
            $data .= str_repeat("\0", $pos + 4 - $len);
        }

        // FIX: represent numbers correctly on 64-bit system
        // http://sourceforge.net/tracker/index.php?func=detail&aid=1487372&group_id=99160&atid=623334
        // Changed by Andreas Rehm 2006 to ensure correct result of the <<24 block on 32 and 64bit systems
        $_or_24 = ord($data[$pos + 3]);
        if ($_or_24 >= 128) {
            // negative number
            $_ord_24 = -abs((256 - $_or_24) << 24);
        } else {
            $_ord_24 = ($_or_24 & 127) << 24;
        }

        return ord($data[$pos]) | (ord($data[$pos + 1]) << 8) | (ord($data[$pos + 2]) << 16) | $_ord_24;
    }
}


1		<?php
2
3		namespace PhpOffice\PhpSpreadsheet\Shared;
4
5		use PhpOffice\PhpSpreadsheet\Reader\Exception as ReaderException;
6
7		class OLERead
8		{
9		private $data = '';
10
11		// Size of a sector = 512 bytes
12		const BIG_BLOCK_SIZE = 0x200;
13
14		// Size of a short sector = 64 bytes
15		const SMALL_BLOCK_SIZE = 0x40;
16
17		// Size of a directory entry always = 128 bytes
18		const PROPERTY_STORAGE_BLOCK_SIZE = 0x80;
19
20		// Minimum size of a standard stream = 4096 bytes, streams smaller than this are stored as short streams
21		const SMALL_BLOCK_THRESHOLD = 0x1000;
22
23		// header offsets
24		const NUM_BIG_BLOCK_DEPOT_BLOCKS_POS = 0x2c;
25		const ROOT_START_BLOCK_POS = 0x30;
26		const SMALL_BLOCK_DEPOT_BLOCK_POS = 0x3c;
27		const EXTENSION_BLOCK_POS = 0x44;
28		const NUM_EXTENSION_BLOCK_POS = 0x48;
29		const BIG_BLOCK_DEPOT_BLOCKS_POS = 0x4c;
30
31		// property storage offsets (directory offsets)
32		const SIZE_OF_NAME_POS = 0x40;
33		const TYPE_POS = 0x42;
34		const START_BLOCK_POS = 0x74;
35		const SIZE_POS = 0x78;
36
37		public $wrkbook;
38
39		public $summaryInformation;
40
41		public $documentSummaryInformation;
42
43		/**
44		* @var int
45		*/
46		private $numBigBlockDepotBlocks;
47
48		/**
49		* @var int
50		*/
51		private $rootStartBlock;
52
53		/**
54		* @var int
55		*/
56		private $sbdStartBlock;
57
58		/**
59		* @var int
60		*/
61		private $extensionBlock;
62
63		/**
64		* @var int
65		*/
66		private $numExtensionBlocks;
67
68		/**
69		* @var string
70		*/
71		private $bigBlockChain;
72
73		/**
74		* @var string
75		*/
76		private $smallBlockChain;
77
78		/**
79		* @var string
80		*/
81		private $entry;
82
83		/**
84		* @var int
85		*/
86		private $rootentry;
87
88		/**
89		* @var array
90		*/
91		private $props = [];
92
93		/**
94		* Read the file.
95		*
96		* @param $pFilename string Filename
97		*
98		* @throws ReaderException
99		*/
100	29	public function read($pFilename)
101		{
102	29	File::assertFile($pFilename);
103
104		// Get the file identifier
105		// Don't bother reading the whole file until we know it's a valid OLE file
106	29	$this->data = file_get_contents($pFilename, false, null, 0, 8);
107
108		// Check OLE identifier
109	29	$identifierOle = pack('CCCCCCCC', 0xd0, 0xcf, 0x11, 0xe0, 0xa1, 0xb1, 0x1a, 0xe1);
110	29	if ($this->data != $identifierOle) {
111		throw new ReaderException('The filename ' . $pFilename . ' is not recognised as an OLE file');
112		}
113
114		// Get the file data
115	29	$this->data = file_get_contents($pFilename);
116
117		// Total number of sectors used for the SAT
118	29	$this->numBigBlockDepotBlocks = self::getInt4d($this->data, self::NUM_BIG_BLOCK_DEPOT_BLOCKS_POS);
119
120		// SecID of the first sector of the directory stream
121	29	$this->rootStartBlock = self::getInt4d($this->data, self::ROOT_START_BLOCK_POS);
122
123		// SecID of the first sector of the SSAT (or -2 if not extant)
124	29	$this->sbdStartBlock = self::getInt4d($this->data, self::SMALL_BLOCK_DEPOT_BLOCK_POS);
125
126		// SecID of the first sector of the MSAT (or -2 if no additional sectors are used)
127	29	$this->extensionBlock = self::getInt4d($this->data, self::EXTENSION_BLOCK_POS);
128
129		// Total number of sectors used by MSAT
130	29	$this->numExtensionBlocks = self::getInt4d($this->data, self::NUM_EXTENSION_BLOCK_POS);
131
132	29	$bigBlockDepotBlocks = [];
133	29	$pos = self::BIG_BLOCK_DEPOT_BLOCKS_POS;
134
135	29	$bbdBlocks = $this->numBigBlockDepotBlocks;
136
137	29	if ($this->numExtensionBlocks != 0) {
138		$bbdBlocks = (self::BIG_BLOCK_SIZE - self::BIG_BLOCK_DEPOT_BLOCKS_POS) / 4;
139		}
140
141	29	for ($i = 0; $i < $bbdBlocks; ++$i) {
142	29	$bigBlockDepotBlocks[$i] = self::getInt4d($this->data, $pos);
143	29	$pos += 4;
144		}
145
146	29	for ($j = 0; $j < $this->numExtensionBlocks; ++$j) {
147		$pos = ($this->extensionBlock + 1) * self::BIG_BLOCK_SIZE;
148		$blocksToRead = min($this->numBigBlockDepotBlocks - $bbdBlocks, self::BIG_BLOCK_SIZE / 4 - 1);
149
150		for ($i = $bbdBlocks; $i < $bbdBlocks + $blocksToRead; ++$i) {
151		$bigBlockDepotBlocks[$i] = self::getInt4d($this->data, $pos);
152		$pos += 4;
153		}
154
155		$bbdBlocks += $blocksToRead;
156		if ($bbdBlocks < $this->numBigBlockDepotBlocks) {
157		$this->extensionBlock = self::getInt4d($this->data, $pos);
158		}
159		}
160
161	29	$pos = 0;
162	29	$this->bigBlockChain = '';
163	29	$bbs = self::BIG_BLOCK_SIZE / 4;
164	29	for ($i = 0; $i < $this->numBigBlockDepotBlocks; ++$i) {
165	29	$pos = ($bigBlockDepotBlocks[$i] + 1) * self::BIG_BLOCK_SIZE;
166
167	29	$this->bigBlockChain .= substr($this->data, $pos, 4 * $bbs);
168	29	$pos += 4 * $bbs;
169		}
170
171	29	$pos = 0;
		0 ignored issues – show Unused Code introduced 2017-12-03 09:09 UTC by Report Bug Copy Issue Report The assignment to `$pos` is dead and can be removed. Loading history...
172	29	$sbdBlock = $this->sbdStartBlock;
173	29	$this->smallBlockChain = '';
174	29	while ($sbdBlock != -2) {
175	12	$pos = ($sbdBlock + 1) * self::BIG_BLOCK_SIZE;
176
177	12	$this->smallBlockChain .= substr($this->data, $pos, 4 * $bbs);
178	12	$pos += 4 * $bbs;
179
180	12	$sbdBlock = self::getInt4d($this->bigBlockChain, $sbdBlock * 4);
181		}
182
183		// read the directory stream
184	29	$block = $this->rootStartBlock;
185	29	$this->entry = $this->_readData($block);
186
187	29	$this->readPropertySets();
188	29	}
189
190		/**
191		* Extract binary stream data.
192		*
193		* @param int $stream
194		*
195		* @return string
196		*/
197	27	public function getStream($stream)
198		{
199	27	if ($stream === null) {
		0 ignored issues – show introduced 2018-03-14 12:07 UTC by Report Bug Copy Issue Report The condition `$stream === null` is always `false`. Loading history...
200	2	return null;
201		}
202
203	27	$streamData = '';
204
205	27	if ($this->props[$stream]['size'] < self::SMALL_BLOCK_THRESHOLD) {
206	10	$rootdata = $this->_readData($this->props[$this->rootentry]['startBlock']);
207
208	10	$block = $this->props[$stream]['startBlock'];
209
210	10	while ($block != -2) {
211	10	$pos = $block * self::SMALL_BLOCK_SIZE;
212	10	$streamData .= substr($rootdata, $pos, self::SMALL_BLOCK_SIZE);
213
214	10	$block = self::getInt4d($this->smallBlockChain, $block * 4);
215		}
216
217	10	return $streamData;
218		}
219	21	$numBlocks = $this->props[$stream]['size'] / self::BIG_BLOCK_SIZE;
220	21	if ($this->props[$stream]['size'] % self::BIG_BLOCK_SIZE != 0) {
221	21	++$numBlocks;
222		}
223
224	21	if ($numBlocks == 0) {
225		return '';
226		}
227
228	21	$block = $this->props[$stream]['startBlock'];
229
230	21	while ($block != -2) {
231	21	$pos = ($block + 1) * self::BIG_BLOCK_SIZE;
232	21	$streamData .= substr($this->data, $pos, self::BIG_BLOCK_SIZE);
233	21	$block = self::getInt4d($this->bigBlockChain, $block * 4);
234		}
235
236	21	return $streamData;
237		}
238
239		/**
240		* Read a standard stream (by joining sectors using information from SAT).
241		*
242		* @param int $bl Sector ID where the stream starts
243		*
244		* @return string Data for standard stream
245		*/
246	29	private function _readData($bl)
247		{
248	29	$block = $bl;
249	29	$data = '';
250
251	29	while ($block != -2) {
252	29	$pos = ($block + 1) * self::BIG_BLOCK_SIZE;
253	29	$data .= substr($this->data, $pos, self::BIG_BLOCK_SIZE);
254	29	$block = self::getInt4d($this->bigBlockChain, $block * 4);
255		}
256
257	29	return $data;
258		}
259
260		/**
261		* Read entries in the directory stream.
262		*/
263	29	private function readPropertySets()
264		{
265	29	$offset = 0;
266
267		// loop through entires, each entry is 128 bytes
268	29	$entryLen = strlen($this->entry);
269	29	while ($offset < $entryLen) {
270		// entry data (128 bytes)
271	29	$d = substr($this->entry, $offset, self::PROPERTY_STORAGE_BLOCK_SIZE);
272
273		// size in bytes of name
274	29	$nameSize = ord($d[self::SIZE_OF_NAME_POS]) \| (ord($d[self::SIZE_OF_NAME_POS + 1]) << 8);
275
276		// type of entry
277	29	$type = ord($d[self::TYPE_POS]);
278
279		// sectorID of first sector or short sector, if this entry refers to a stream (the case with workbook)
280		// sectorID of first sector of the short-stream container stream, if this entry is root entry
281	29	$startBlock = self::getInt4d($d, self::START_BLOCK_POS);
282
283	29	$size = self::getInt4d($d, self::SIZE_POS);
284
285	29	$name = str_replace("\x00", '', substr($d, 0, $nameSize));
286
287	29	$this->props[] = [
288	29	'name' => $name,
289	29	'type' => $type,
290	29	'startBlock' => $startBlock,
291	29	'size' => $size,
292		];
293
294		// tmp helper to simplify checks
295	29	$upName = strtoupper($name);
296
297		// Workbook directory entry (BIFF5 uses Book, BIFF8 uses Workbook)
298	29	if (($upName === 'WORKBOOK') \|\| ($upName === 'BOOK')) {
299	29	$this->wrkbook = count($this->props) - 1;
300	29	} elseif ($upName === 'ROOT ENTRY' \|\| $upName === 'R') {
301		// Root entry
302	29	$this->rootentry = count($this->props) - 1;
303		}
304
305		// Summary information
306	29	if ($name == chr(5) . 'SummaryInformation') {
307	27	$this->summaryInformation = count($this->props) - 1;
308		}
309
310		// Additional Document Summary information
311	29	if ($name == chr(5) . 'DocumentSummaryInformation') {
312	27	$this->documentSummaryInformation = count($this->props) - 1;
313		}
314
315	29	$offset += self::PROPERTY_STORAGE_BLOCK_SIZE;
316		}
317	29	}
318
319		/**
320		* Read 4 bytes of data at specified position.
321		*
322		* @param string $data
323		* @param int $pos
324		*
325		* @return int
326		*/
327	29	private static function getInt4d($data, $pos)
328		{
329	29	if ($pos < 0) {
330		// Invalid position
331		throw new ReaderException('Parameter pos=' . $pos . ' is invalid.');
332		}
333
334	29	$len = strlen($data);
335	29	if ($len < $pos + 4) {
336		$data .= str_repeat("\0", $pos + 4 - $len);
337		}
338
339		// FIX: represent numbers correctly on 64-bit system
340		// http://sourceforge.net/tracker/index.php?func=detail&aid=1487372&group_id=99160&atid=623334
341		// Changed by Andreas Rehm 2006 to ensure correct result of the <<24 block on 32 and 64bit systems
342	29	$_or_24 = ord($data[$pos + 3]);
343	29	if ($_or_24 >= 128) {
344		// negative number
345	29	$_ord_24 = -abs((256 - $_or_24) << 24);
346		} else {
347	29	$_ord_24 = ($_or_24 & 127) << 24;
348		}
349
350	29	return ord($data[$pos]) \| (ord($data[$pos + 1]) << 8) \| (ord($data[$pos + 2]) << 16) \| $_ord_24;
351		}
352		}
353

PHPOffice / PhpSpreadsheet

Push — develop ( ae9dd1...08b445 )

OLERead::getStream() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like