TiffSplitter.phpclass

<?php
/**************************************************************************************************************

    NAME
        TiffSplitter.phpclass

    DESCRIPTION
        A class for splitting multipage TIFF files.

    AUTHOR
        Christian Vigh, 04/2017.

    REFERENCES
	- https://www.itu.int/itudoc/itu-t/com16/tiff-fx/docs/tiff6.pdf :
		Specifications of the TIFF file format version 6.
	- http://www.awaresystems.be/imaging/tiff.html :
		Where to get a list of available TIFF IFD tags.

    HISTORY
    [Version : 1.0]	[Date : 2017-04-08]     [Author : CV]
        Initial version.

    [Version : 1.0.1]	[Date : 2017-04-14]     [Author : CV]
	. Moved most of the code from TiffSplitter to TiffImage.

    [Version : 1.0.2]	[Date : 2017-04-16]     [Author : CV]
	. Completely rewrote the AsString() method to use the new TiffIFD and TiffIFDEntry classes.

    [Version : 1.0.3]	[Date : 2017-05-03]     [Author : CV]
	. Corrected an offset inconsistency in the output when the STRIP_BYTE_OFFSETS (and therefore the 
	  STRIP_BYTE_COUNT) tag contains only one entry, directly contained in the IFD entry.

 **************************************************************************************************************/

require_once ( dirname ( __FILE__ ) . '/TiffBase.phpclass' ) ;


/*==============================================================================================================

    class TiffSplitter -
        A class for splitting a mutipage TIFF file (or string) into individual TIFF files.

  ==============================================================================================================*/
class	TiffSplitter		extends		TiffImage
				implements	\ArrayAccess, \Countable, \IteratorAggregate
   {

	/*--------------------------------------------------------------------------------------------------------------
	
	    Constructor -
		Should not be called directly. Call the static Load() or LoadFromString() methods instead.

	 *-------------------------------------------------------------------------------------------------------------*/
	public function  __construct ( )
	   {
		parent::__construct ( ) ;
	    }


	/*--------------------------------------------------------------------------------------------------------------

	    GetPageInstance -
		Returns an instance of a (splitted) TIFF page.
	    	
	 *-------------------------------------------------------------------------------------------------------------*/
	protected function  GetPageInstance ( $parent, $ifd_data ) 
	   {
		return ( new  TiffSplitterPage ( $parent, $ifd_data ) ) ;
	    }

     }


/*==============================================================================================================

    class TiffSplitterPage -
        Allows to extract TIFF data from a page of a multipage TIFF file.

  ==============================================================================================================*/
class  TiffSplitterPage			extends  TiffPage 
   {
	/*--------------------------------------------------------------------------------------------------------------

	    Constructor - 
		Instantiates a TIFF page and retrieves the page number, if any.

	 *-------------------------------------------------------------------------------------------------------------*/
	public function  __construct ( $parent, $ifd_data )
	   {
		parent::__construct ( $parent, $ifd_data ) ;
	    }


	/*--------------------------------------------------------------------------------------------------------------
	
	    NAME
	        AsString - Converts the page to a TIFF or PDF file.
	
	    PROTOTYPE
	        $data		=  $page -> AsString ( $format = TiffImage::OUTPUT_FORMAT_TIFF ) ;
	
	    DESCRIPTION
	        Returns the contents of the TIFF page either as a TIFF or PDF file.
	
	    PARAMETERS
	        $format (integer) -
	                Either TiffImage::OUTPUT_FORMAT_TIFF or TiffImage::OUTPUT_FORMAT_PDF.
	
	    RETURN VALUE
	        A string value that can be directly used to create a TIFF or PDF file containing this page.

	    NOTES
		The PDF file format is not yet implemented.
	
	 *-------------------------------------------------------------------------------------------------------------*/
	public function  __tostring ( )
	   { return ( $this -> AsString ( ) ) ; }


	public function  AsString ( $format = TiffImage::OUTPUT_FORMAT_TIFF )
	   {
		$tiff			=  $this -> Parent -> TiffData ;
		$ifd			=  $this -> IFD ;
		$ifd_count		=  $ifd -> GetRealCount ( ) ;

		// Build the header  (either the WORD16 0x4949 or 0x4D4D, followed by the WORD16 42)
		$tiff_data		=  $tiff -> ToWord16 ( $this -> Parent -> Endianness, TiffImage::TIFF_IDENTIFIER ) ;
		$offset			=  4 ;

		// Put the address of the first (and unique) IFD of this file. Naturally, it will come right after this entry
		$tiff_data		.=  $tiff -> ToWord32 ( 8 ) ;
		$offset			+=  4 ;

		// Now, put the number of IFD entries
		$tiff_data		.=  $tiff -> ToWord16 ( $ifd_count ) ;
		$offset			+=  2 ;

		// Compute the offset AFTER the IFD, including the pointer to the next IFD, which will be zero in our case
		$post_offset		 =  $offset + ( $ifd_count * TiffImage::IFD_ENTRY_SIZE ) + 4 ;

		// Data after the IFD entries
		$post_data		 =  '' ;

		// Loop through each IFD entry
		foreach  ( $ifd  as  $ifd_entry )
		   {
			// Don't try to process unsupported tags
			if  ( ! $ifd_entry -> Supported )
				continue ;

			// Get the Tag, Type and Count values converted to their original endianness
			$tag	=  $tiff -> ToWord16 ( $ifd_entry -> Tag ) ;
			$type	=  $tiff -> ToWord16 ( $ifd_entry -> Type ) ;
			$count	=  $tiff -> ToWord32 ( $ifd_entry -> Count ) ;


			// Normal processing
			if  ( $ifd_entry -> Tag  !==  TiffImage::TAG_STRIP_OFFSETS )
			   {
				// IFD entry is not an offset to some data : simply reuse the original Value
				if  ( ! $ifd_entry -> IsOffset )
					$value		=  $ifd_entry -> Value ;
				// Otherwise, retrieve the data it points to
				else
				   {
					// The original IFD value will be replaced with the current output offset
					$value		 =  $post_offset ;

					// Get real data
					$data		 =  $ifd_entry -> UnderlyingValue ;

					$post_data	.=  $data ;
					$post_offset	+=  strlen ( $data ) ;
				    }
			    }
			// Special processing for StripOffsets
			else
			   {
				// The original IFD value will be replaced with the current output offset
				$value		 =  $post_offset ;

				// Get real data
				$data			=  $ifd_entry -> UnderlyingValue ;

				// Get the StripByteCounts value
				$byte_counts_entry	=  $this -> IFD -> GetTagById ( TiffImage::TAG_STRIP_BYTE_COUNTS ) ;

				// When the Count field is > 1, the Value field is necessarily an offset
				if  ( $byte_counts_entry -> IsOffset )
				   {
					// Get the StripByteCounts entries
					$strip_byte_counts	=  $tiff -> Unpack32 ( $byte_counts_entry -> UnderlyingValue ) ;

					// Then strip offset data
					$strip_offsets		=  $tiff -> Unpack32 ( $data ) ;
					$strip_count		=  count ( $strip_offsets ) ;
				    }
				// Otherwise, the Value field gives the strip byte counts
				else
				   {
					$strip_byte_counts	=  array ( 1 => $byte_counts_entry -> Value ) ;
					$strip_offsets		=  array ( 1 => $ifd_entry -> Value ) ;
					$strip_count		=  1 ;
				    }

				// Rebuild the new offset table
				$strip_data		=  '' ;
				$strip_data_offset	=  $post_offset + ( $strip_count * 4 ) ;
				$new_offsets		=  array ( ) ;

				for  ( $i = 1 ; $i  <=  $strip_count ; $i ++ )
				   {
					$strip_data		.=  $tiff -> GetRawBytes ( $strip_offsets [$i], $strip_byte_counts [$i] ) ;
					$new_offsets []		 =  $strip_data_offset ;
					$strip_data_offset	+=  $strip_byte_counts [$i] ;
				    }
					
				// Catenate the new offset table + strip contents to the data to be added after the IFD structure
				// but take into account the fact that there is only one byte count and strip data
				if  ( $byte_counts_entry -> IsOffset )
					$extra_data	 =  $tiff -> ToWord32 ( $new_offsets ) . $strip_data ;
				else
					$extra_data	 =  $strip_data ;

				$post_data	.=  $extra_data ;
				$post_offset    +=  strlen ( $extra_data ) ;
			    }

			// Add this IFD entry tot he output data
			$entry_value	 =  $tiff -> ToWord32 ( $value ) ;
			$tiff_data	.=  $tag . $type . $count . $entry_value ;
		    }
		// Add the pointer to the next IFD, which is zero since there will be only one IFD here, then the data we extracted 
		// in the above loop
		$tiff_data	.=  "\x00\x00\x00\x00" . $post_data ;

		// All done, return
		return ( $tiff_data ) ;
	    }


	/*--------------------------------------------------------------------------------------------------------------
	
	    NAME
	        SaveTo - Saves the current page to the specified output file.
	
	    PROTOTYPE
	        $page -> SaveTo ( $filename, $format = TiffImage::OUTPUT_FORMAT_TIFF ) ;
	
	    DESCRIPTION
	        Saves the current page to the specified output file, in the specified format.
	
	    PARAMETERS
		$filename (string) -
			Output filename.

	        $format (integer) -
	                Either TiffImage::OUTPUT_FORMAT_TIFF or TiffImage::OUTPUT_FORMAT_PDF.
	
	    NOTES
		The PDF file format is not yet implemented.
	
	 *-------------------------------------------------------------------------------------------------------------*/
	public function  SaveTo ( $filename, $format = TiffImage::OUTPUT_FORMAT_TIFF )
	   {
		$data	=  $this -> AsString ( $format ) ;

		switch  ( $format )
		   {
			//case	TiffImage::OUTPUT_FORMAT_PDF :
			//	$expected_extensions	=  array ( 'pdf' ) ;

			case	TiffImage::OUTPUT_FORMAT_TIFF :
			default :
				$expected_extensions	=  array ( 'tiff', 'tif' ) ;
		    }
		
		$extension	=  strtolower ( pathinfo ( $filename, PATHINFO_EXTENSION ) ) ;

		if  ( ! in_array ( $extension, $expected_extensions ) )
			$filename	.=  '.' . $expected_extensions [0] ;

		file_put_contents ( $filename, $data ) ;
	    }
    }