@ -1,4 +1,5 @@
<?php <?php
/** /**
* @file * @file
* Classes and functions for datastream validation. * Classes and functions for datastream validation.
@ -16,7 +17,7 @@
* The hexidecimal string. * The hexidecimal string.
* *
* @throws Exception * @throws Exception
* i f something horrible happens during the actual conversion. * I f something horrible happens during the actual conversion.
* *
* @return bool|int * @return bool|int
* FALSE on failure, or the integer on success. * FALSE on failure, or the integer on success.
@ -221,14 +222,14 @@ class TIFFDatastreamValidator extends DatastreamValidator {
/** /**
* Asserts that the TIFF contains an appropriate header. * Asserts that the TIFF contains an appropriate header.
*/ */
public function assertTIFF HeaderHex() { public function assertTiff HeaderHex() {
$datastream_header_hex = self::getTIFFHeaderHex(); $datastream_header_hex = self::getTIFFHeaderHex();
if ($datastream_header_hex == "49492a00") { if ($datastream_header_hex == "49492a00") {
// In this case, the ingested TIFF is designated as using the "Intel // In this case, the ingested TIFF is designated as using the "Intel
// byte-order" (i.e. little-endian) by starting with the characters "II" // byte-order" (i.e. little-endian) by starting with the characters "II"
// (repeated so that byte order does not yet need to be significant). // (repeated so that byte order does not yet need to be significant).
// The number that follows is '42' in little-endian hex, a number of // The number that follows is '42' in little-endian hex, a number of
// 'deep philosophical significance' to the TIFF format creators.' // 'deep philosophical significance' to the TIFF format creators.'.
$this->addResult(TRUE, "{$this->datastream} datastream asserts that it is a valid Intel-byte-orderded TIF/TIFF file."); $this->addResult(TRUE, "{$this->datastream} datastream asserts that it is a valid Intel-byte-orderded TIF/TIFF file.");
} }
elseif ($datastream_header_hex == "4d4d002a") { elseif ($datastream_header_hex == "4d4d002a") {
@ -248,7 +249,7 @@ class TIFFDatastreamValidator extends DatastreamValidator {
* @return string * @return string
* The ... thing I just wrote up there. * The ... thing I just wrote up there.
*/ */
protected function getTIFF HeaderHex() { protected function getTiff HeaderHex() {
return substr(bin2hex($this->datastreamContent), 0, 8); return substr(bin2hex($this->datastreamContent), 0, 8);
} }
@ -265,7 +266,7 @@ class JP2DatastreamValidator extends DatastreamValidator {
* JP2 files begin with an offset header at the second 32-bit integer, * JP2 files begin with an offset header at the second 32-bit integer,
* 0x6A502020. This header is in all .jp2s, and we check for it here. * 0x6A502020. This header is in all .jp2s, and we check for it here.
*/ */
protected function assertJP 2Header() { protected function assertJp 2Header() {
$assertion = substr(bin2hex($this->datastreamContent), 8, 8) == '6a502020'; $assertion = substr(bin2hex($this->datastreamContent), 8, 8) == '6a502020';
$pass = "Datastream {$this->datastream} contains the appropriate JP2 header."; $pass = "Datastream {$this->datastream} contains the appropriate JP2 header.";
$fail = "Datastream {$this->datastream} does not contain the appropriate JP2 header."; $fail = "Datastream {$this->datastream} does not contain the appropriate JP2 header.";
@ -279,13 +280,14 @@ class JP2DatastreamValidator extends DatastreamValidator {
* JP2 files have their codestream capped with a marker, 0xFFD9. We're just * JP2 files have their codestream capped with a marker, 0xFFD9. We're just
* checking for it here to see if the .jp2 encoder finished okay. * checking for it here to see if the .jp2 encoder finished okay.
*/ */
protected function assertJP 2Marker() { protected function assertJp 2Marker() {
$assertion = substr(bin2hex($this->datastreamContent), strlen(bin2hex($this->datastreamContent)) - 4, 4) == 'ffd9'; $assertion = substr(bin2hex($this->datastreamContent), strlen(bin2hex($this->datastreamContent)) - 4, 4) == 'ffd9';
$pass = "Datastream {$this->datastream} contains the appropriate JP2 ending marker."; $pass = "Datastream {$this->datastream} contains the appropriate JP2 ending marker.";
$fail = "Datastream {$this->datastream} does not contain the appropriate JP2 ending marker. If this is the only JP2 validator that failed, it is likely that derivative generation was interrupted."; $fail = "Datastream {$this->datastream} does not contain the appropriate JP2 ending marker. If this is the only JP2 validator that failed, it is likely that derivative generation was interrupted.";
$message = $assertion ? $pass : $fail; $message = $assertion ? $pass : $fail;
$this->addResult($assertion, $message); $this->addResult($assertion, $message);
} }
} }
/** /**
@ -296,7 +298,7 @@ class PDFDatastreamValidator extends DatastreamValidator {
/** /**
* Validates the PDF signature. * Validates the PDF signature.
*/ */
protected function assertPDF Signature() { protected function assertPdf Signature() {
$assertion = substr($this->datastreamContent, 0, 5) == '%PDF-'; $assertion = substr($this->datastreamContent, 0, 5) == '%PDF-';
$pdf_version = substr($this->datastreamContent, 5, 3); $pdf_version = substr($this->datastreamContent, 5, 3);
$pass = "{$this->datastream} datastream asserts that it is a valid PDF file using PDF version {$pdf_version}"; $pass = "{$this->datastream} datastream asserts that it is a valid PDF file using PDF version {$pdf_version}";
@ -308,7 +310,7 @@ class PDFDatastreamValidator extends DatastreamValidator {
/** /**
* Counts the number of signatures in this PDF file and asserts there are any. * Counts the number of signatures in this PDF file and asserts there are any.
*/ */
protected function assertPDF StreamCount() { protected function assertPdf StreamCount() {
$pdf_stream_count = substr_count(bin2hex($this->datastreamContent), '0a73747265616d0a'); $pdf_stream_count = substr_count(bin2hex($this->datastreamContent), '0a73747265616d0a');
$assertion = $pdf_stream_count !== 0; $assertion = $pdf_stream_count !== 0;
$pass = "{$this->datastream} datastream reports the existence of {$pdf_stream_count} PDF streams. Note that an extremely low number could still indicate corruption."; $pass = "{$this->datastream} datastream reports the existence of {$pdf_stream_count} PDF streams. Note that an extremely low number could still indicate corruption.";
@ -319,17 +321,15 @@ class PDFDatastreamValidator extends DatastreamValidator {
/** /**
* Validates the PDF closing tag. * Validates the PDF closing tag.
*
* @return bool
* TRUE if it was present; FALSE otherwise.
*/ */
protected function assertPDF ClosingTag() { protected function assertPdfClosingTag() {
$assertion = strpos(bin2hex($this->datastreamContent), '0a2525454f460a') == TRUE; $assertion = strpos(bin2hex($this->datastreamContent), '0a2525454f460a') == TRUE;
$pass = "{$this->datastream} datastream reports the existence of the closing 'EOF' tag required at the end of PDFs"; $pass = "{$this->datastream} datastream reports the existence of the closing 'EOF' tag required at the end of PDFs";
$fail = "{$this->datastream} datastream does not contain the closing 'EOF' tag. If this is the only PDF validation that failed, it is likely that derivative generation was interrupted."; $fail = "{$this->datastream} datastream does not contain the closing 'EOF' tag. If this is the only PDF validation that failed, it is likely that derivative generation was interrupted.";
$message = $assertion ? $pass : $fail; $message = $assertion ? $pass : $fail;
$this->addResult($assertion, $message); $this->addResult($assertion, $message);
} }
} }
/** /**
@ -340,6 +340,7 @@ class PDFDatastreamValidator extends DatastreamValidator {
* integer representing the number of times it should appear in the datastream. * integer representing the number of times it should appear in the datastream.
*/ */
class TextDatastreamValidator extends DatastreamValidator { class TextDatastreamValidator extends DatastreamValidator {
/** /**
* Constructor override; blow up if we don't have our two values. * Constructor override; blow up if we don't have our two values.
*/ */
@ -369,6 +370,7 @@ class TextDatastreamValidator extends DatastreamValidator {
protected function getTextStringCount() { protected function getTextStringCount() {
return substr_count($this->datastreamContent, $this->params[0]); return substr_count($this->datastreamContent, $this->params[0]);
} }
} }
/** /**
@ -401,7 +403,7 @@ class WAVDatastreamValidator extends DatastreamValidator {
/** /**
* Asserts that the datastream contains a valid WAV signature. * Asserts that the datastream contains a valid WAV signature.
*/ */
protected function assertWAV Signature() { protected function assertWav Signature() {
$signatures = str_split(substr($this->datastreamContent, 0, 24), 8); $signatures = str_split(substr($this->datastreamContent, 0, 24), 8);
$assertion = $signatures[0] == '52494646' & & $signatures[2] == '57415645'; $assertion = $signatures[0] == '52494646' & & $signatures[2] == '57415645';
$pass = "Header of the {$this->datastream} datastream contains a valid file signature."; $pass = "Header of the {$this->datastream} datastream contains a valid file signature.";
@ -413,7 +415,7 @@ class WAVDatastreamValidator extends DatastreamValidator {
/** /**
* Asserts that the chunksize in the header is correct. * Asserts that the chunksize in the header is correct.
*/ */
protected function assertWAV ChunkSize() { protected function assertWav ChunkSize() {
$assertion = islandora_hex2int(substr($this->datastreamContent, 8, 8)) === 36 + self::getDataSubChunkSize(); $assertion = islandora_hex2int(substr($this->datastreamContent, 8, 8)) === 36 + self::getDataSubChunkSize();
$pass = "{$this->datastream} datastream chunksize in WAV header is correct"; $pass = "{$this->datastream} datastream chunksize in WAV header is correct";
$fail = "{$this->datastream} datastream chunksize in WAV header does not match actual chunksize."; $fail = "{$this->datastream} datastream chunksize in WAV header does not match actual chunksize.";
@ -424,7 +426,7 @@ class WAVDatastreamValidator extends DatastreamValidator {
/** /**
* Asserts that the datastream contains a 'fmt' subchunk. * Asserts that the datastream contains a 'fmt' subchunk.
*/ */
protected function assertWAV FmtSubChunk() { protected function assertWav FmtSubChunk() {
$assertion = substr($this->datastreamContent, 24, 8) === '666d7420'; $assertion = substr($this->datastreamContent, 24, 8) === '666d7420';
$pass = "{$this->datastream} datastream contains a 'fmt' subchunk."; $pass = "{$this->datastream} datastream contains a 'fmt' subchunk.";
$fail = "{$this->datastream} datastream is missing the required 'fmt' subchunk."; $fail = "{$this->datastream} datastream is missing the required 'fmt' subchunk.";
@ -435,7 +437,7 @@ class WAVDatastreamValidator extends DatastreamValidator {
/** /**
* Asserts that the byterate reported by the WAV header is valid. * Asserts that the byterate reported by the WAV header is valid.
*/ */
protected function assertWAV ByteRate() { protected function assertWav ByteRate() {
$wav_samplerate = islandora_hex2int(substr($this->datastreamContent, 48, 8)); $wav_samplerate = islandora_hex2int(substr($this->datastreamContent, 48, 8));
$assertion = islandora_hex2int(substr($this->datastreamContent, 56, 8)) === $wav_samplerate * self::getNumChannels() * self::getBytesPerSample(); $assertion = islandora_hex2int(substr($this->datastreamContent, 56, 8)) === $wav_samplerate * self::getNumChannels() * self::getBytesPerSample();
$pass = "{$this->datastream} datastream byterate in the WAV header is correct."; $pass = "{$this->datastream} datastream byterate in the WAV header is correct.";
@ -447,7 +449,7 @@ class WAVDatastreamValidator extends DatastreamValidator {
/** /**
* Asserts that the block alignment is correct. * Asserts that the block alignment is correct.
*/ */
protected function assertWAV BlockAlignment() { protected function assertWav BlockAlignment() {
$assertion = islandora_hex2int(substr($this->datastreamContent, 64, 4)) === self::getNumChannels() * self::getBytesPerSample(); $assertion = islandora_hex2int(substr($this->datastreamContent, 64, 4)) === self::getNumChannels() * self::getBytesPerSample();
$pass = "{$this->datastream} datastream block alignment is set correctly."; $pass = "{$this->datastream} datastream block alignment is set correctly.";
$fail = "{$this->datastream} datastream block alignment is off."; $fail = "{$this->datastream} datastream block alignment is off.";
@ -460,7 +462,7 @@ class WAVDatastreamValidator extends DatastreamValidator {
* *
* Also asserts that the subchunk size is correct. * Also asserts that the subchunk size is correct.
*/ */
protected function assertWAV DataSubChunk() { protected function assertWav DataSubChunk() {
if (substr($this->datastreamContent, 72, 8) !== '64617461') { if (substr($this->datastreamContent, 72, 8) !== '64617461') {
$this->addResult(FALSE, "{$this->datastream} datastream is missing the 'data' subchunk."); $this->addResult(FALSE, "{$this->datastream} datastream is missing the 'data' subchunk.");
return; return;
@ -505,6 +507,7 @@ class WAVDatastreamValidator extends DatastreamValidator {
protected function getDataSubChunkSize() { protected function getDataSubChunkSize() {
return islandora_hex2int(substr($this->datastreamContent, 80, 8)); return islandora_hex2int(substr($this->datastreamContent, 80, 8));
} }
} }
/** /**
@ -526,7 +529,7 @@ class MP3DatastreamValidator extends DatastreamValidator {
* breaking my own rules here and using a single assert function so that I * breaking my own rules here and using a single assert function so that I
* can handle the weird logic. * can handle the weird logic.
*/ */
protected function assertValidMP 3() { protected function assertValidMp 3() {
$this->datastreamContent = bin2hex($this->datastreamContent); $this->datastreamContent = bin2hex($this->datastreamContent);
// If it's not a VBR MP3, we don't have to check much, so let's get that // If it's not a VBR MP3, we don't have to check much, so let's get that
@ -612,7 +615,7 @@ class MP4DatastreamValidator extends DatastreamValidator {
/** /**
* Asserts that the datastream is ISO-formatted video. * Asserts that the datastream is ISO-formatted video.
*/ */
protected function assertISO Video() { protected function assertIso Video() {
$mp4_ftyp = substr(strpos($this->datastreamContent, 'ftyp'), 4, 4); $mp4_ftyp = substr(strpos($this->datastreamContent, 'ftyp'), 4, 4);
$assertion = strpos($this->datastreamContent, 'ftyp') !== 0; $assertion = strpos($this->datastreamContent, 'ftyp') !== 0;
$pass = "{$this->datastream} datastream asserts that it is a valid ISO-formatted video file using ftyp {$mp4_ftyp}"; $pass = "{$this->datastream} datastream asserts that it is a valid ISO-formatted video file using ftyp {$mp4_ftyp}";
@ -620,6 +623,7 @@ class MP4DatastreamValidator extends DatastreamValidator {
$message = $assertion ? $pass : $fail; $message = $assertion ? $pass : $fail;
$this->addResult($assertion, $message); $this->addResult($assertion, $message);
} }
} }
/** /**
@ -636,7 +640,7 @@ class OGGDatastreamValidator extends DatastreamValidator {
/** /**
* Asserts that the datastream contains ogg pages. * Asserts that the datastream contains ogg pages.
*/ */
protected function assertOGG Pages() { protected function assertOgg Pages() {
$ogg_pages = substr_count($this->datastreamContent, 'OggS'); $ogg_pages = substr_count($this->datastreamContent, 'OggS');
$assertion = $ogg_pages !== 0; $assertion = $ogg_pages !== 0;
$pass = "{$this->datastream} datastream asserts that it contains {$ogg_pages} Ogg pages (even a very small file should contain several)."; $pass = "{$this->datastream} datastream asserts that it contains {$ogg_pages} Ogg pages (even a very small file should contain several).";
@ -666,6 +670,7 @@ class OGGDatastreamValidator extends DatastreamValidator {
$message = $assertion ? $pass : $fail; $message = $assertion ? $pass : $fail;
$this->addResult($assertion, $message); $this->addResult($assertion, $message);
} }
} }
/** /**
@ -683,7 +688,7 @@ class MKVDatastreamValidator extends DatastreamValidator {
/** /**
* Asserts that the datastream is an EBML-format file. * Asserts that the datastream is an EBML-format file.
*/ */
protected function assertEBML Format() { protected function assertEbml Format() {
$assertion = substr(bin2hex($this->datastreamContent), 0, 8) == '1a45dfa3'; $assertion = substr(bin2hex($this->datastreamContent), 0, 8) == '1a45dfa3';
$pass = "{$this->datastream} datastream asserts that it is an EBML-formatted file"; $pass = "{$this->datastream} datastream asserts that it is an EBML-formatted file";
$fail = "{$this->datastream} datastream is not an EBML-formatted file."; $fail = "{$this->datastream} datastream is not an EBML-formatted file.";
@ -701,4 +706,5 @@ class MKVDatastreamValidator extends DatastreamValidator {
$message = $assertion ? $pass : $fail; $message = $assertion ? $pass : $fail;
$this->addResult($assertion, $message); $this->addResult($assertion, $message);
} }
} }
Jonathan Green
7 years ago