75
out = []
for i in range(len(source)-(shingleLen-1)):
out.append (binascii.crc32(' '.join( [x for x in source[i:i+shingleLen]]
).encode('utf-8')))
return out
def compaire (source1,source2):
same = 0
for i in range(len(source1)):
if source1[i] in source2:
same = same + 1
return float(same*2)/float(len(source1) + len(source2))*100
class doc extends cfb {
public function parse() {
parent::parse();
$wdStreamID = $this->getStreamIdByName("WordDocument");
if ($wdStreamID === false) { return false; }
$wdStream = $this->getStreamById($wdStreamID);
$bytes = $this->getShort(0x000A, $wdStream);
$fComplex = ($bytes & 0x0004) == 0x0004;
$fWhichTblStm = ($bytes & 0x0200) == 0x020;
$fcClx = $this->getLong(0x01A2, $wdStream);
$lcbClx = $this->getLong(0x01A6, $wdStream);
76
$ccpText = $this->getLong(0x004C, $wdStream);
$ccpFtn = $this->getLong(0x0050, $wdStream);
$ccpHdd = $this->getLong(0x0054, $wdStream);
$ccpMcr = $this->getLong(0x0058, $wdStream);
$ccpAtn = $this->getLong(0x005C, $wdStream);
$ccpEdn = $this->getLong(0x0060, $wdStream);
$ccpTxbx = $this->getLong(0x0064, $wdStream);
$ccpHdrTxbx = $this->getLong(0x0068, $wdStream);
$lastCP = $ccpFtn + $ccpHdd + $ccpMcr + $ccpAtn + $ccpEdn + $ccpTxbx +
$ccpHdrTxbx;
$lastCP += ($lastCP != 0) + $ccpText;
$tStreamID = $this->getStreamIdByName(intval($fWhichTblStm)."Table");
if ($tStreamID === false) { return false; }
$tStream = $this->getStreamById($tStreamID);
$clx = substr($tStream, $fcClx, $lcbClx);
$lcbPieceTable = 0;
$pieceTable = "";
$pieceCount = 0;
$from = 0;
while (($i = strpos($clx, chr(0x02), $from)) !== false) {
$lcbPieceTable = $this->getLong($i + 1, $clx);
$pieceTable = substr($clx, $i + 5);
if (strlen($pieceTable) != $lcbPieceTable) {
$from = $i + 1;
77
continue;
}
break;
}
$cp = array(); $i = 0;
while (($cp[] = $this->getLong($i, $pieceTable)) != $lastCP)
$i += 4;
$pcd = str_split(substr($pieceTable, $i + 4), 8);
$text = "";
for ($i = 0; $i < count($pcd); $i++) {
$fcValue = $this->getLong(2, $pcd[$i]);
$isANSI = ($fcValue & 0x40000000) == 0x40000000;
$fc = $fcValue & 0x3FFFFFFF;
$lcb = $cp[$i + 1] - $cp[$i];
if (!$isANSI)
$lcb *= 2;
else
$fc /= 2;
$part = substr($wdStream, $fc, $lcb);
if (!$isANSI)
$part = $this->unicode_to_utf8($part);
$text .= $part;
}
return $text;
}
}
Размещено на Allbest.ru