#include <stdio.h>
#include <pdflib.h>
#include <errno.h>
#include <unistd.h>
#include <string.h>

#include "debug.h"
#include "clean_spaces.h"
#include "suppress_spaces.h"
#include "document.h"

#define NO_PARENT	-25
#define SS(X,Y)		strcasecmp( X, Y ) == 0

/* Initialize defaults */
int	document_init( document *d )
{
	d -> chars_per_line = 135;
	d -> lines_per_page = 66;

	d -> width = letter_height;
	d -> height = letter_width;

	d -> left_margin = 40;
	d -> top_margin = 40;
	d -> font_size = 9;
	d -> output_filename = NULL;
	d -> font_face = NULL;

	d -> image_dpi = 72;
	d -> image_filename = NULL;

	d -> index_count = 0;
	d -> index_def = NULL;
	d -> default_reparent = DEFAULT_REPARENT;

	d -> subject = NULL;
	d -> title = NULL;
	d -> author = NULL;
	d -> extra_key = NULL;
	d -> extra_data = NULL;

	d -> strip_non_printable = 1;

	d -> input_filename = NULL;

	d -> pdf = PDF_new();

	return( 1 );
}

int	document_add_index( document *d, const indexd *i )
{
	indexd	*t;
	int	dump_index;
	int	error;
	char	errmsg[ 255 ];

	if( d -> index_count == 0 )
	{
		dmsg( 3, "Initializing new indexes" );
		d -> index_def = (indexd *)malloc( 1 );
	}

	/* Validate it looks OK */
	dump_index = 0;
	if( i -> start_char >= i -> stop_char ) 
	{
		fprintf( stderr, "Start char must be smaller than stop char\n" );
		dump_index = 1;
	}

	if( dump_index )
	{
		fprintf( stderr, "Index definition:\n"
				"Page:            %i\n"
				"Line:            %i\n"
				"Start:           %i\n"
				"Stop:            %i\n"
				"Suppress Spaces: %i\n"
				"Reparent:        %i\n"
				"Format:          '%s'\n",
			i -> page_no,
			i -> line_no,
			i -> start_char,
			i -> stop_char,
			i -> suppress_spaces,
			i -> reparent,
			i -> format );
		return( 1 );
	}

	/* Allocate space and copy in index */
	d -> index_count++;
	d -> index_def = (indexd *)realloc( d -> index_def, 
		sizeof( indexd ) * d -> index_count );
	t = &(d -> index_def[ d -> index_count - 1 ]);
	memcpy( t,
		i, sizeof( indexd ) );
	/* Null this out, we'll set and use it later */
	t -> last_index = NULL;
	t -> indexid = NO_PARENT;

	/* Make it 1 based */
	t -> start_char--;
	t -> stop_char--;

	dmsg( 2, "Added index #%d", d -> index_count );
	dmsg( 3, "page_no: %i, line_no: %i, start_char: %i, end_char: %i, suppress_chars: %i",
		t -> page_no,
		t -> line_no,
		t -> start_char,
		t -> stop_char,
		t -> suppress_spaces 
	);

	/* Compile regex */
	if( t -> index_type == IDX_REGEX )
	{
		dmsg( 3, "Compiling regex: '%s'", t -> regstring );
		error = regcomp( &t -> regcomp, t -> regstring, REG_EXTENDED );

		if( error )
		{
			regerror( error, &t -> regcomp, errmsg, 255 );
			fprintf( stderr, "Regex error: (%d) %s\n", error, errmsg );
		}
	}

	return( 1 );
}

int	document_start( document *d )
{
	if( d -> output_filename == NULL )
	{
		dmsg( 2, "Using STDOUT" );
		PDF_open_file( d -> pdf, "-" );
	}else
	{
		dmsg( 2, "Using file: '%s'", d -> output_filename );
		PDF_open_file( d -> pdf, d -> output_filename );
	}

	/* Calculate image scale */
	if( d -> image_filename != NULL )
	{
		dmsg( 3, "Using background image: '%s'", d -> image_filename );
		d -> image_id = PDF_open_image_file( d -> pdf, 
			"tiff", d -> image_filename, NULL, 0 );

		d -> image_scale_x = (float)72.0 / d -> image_dpi;
		d -> image_scale_y = (float)72.0 / d -> image_dpi;
		dmsg( 3, "Scale is: %fx%f", d -> image_scale_x, d -> image_scale_y );
	}

	if( d -> font_face == NULL )
	{
		d -> font_index = 
			PDF_findfont( d -> pdf, "Courier", "host", 0 );
	}else
	{
		d -> font_index = 
			PDF_findfont( d -> pdf, d -> font_face, "host", 0 );
	}

	d -> line_count = 0;
	d -> total_line_count = 0;
	d -> page_count = 0;

	/* Set up info */
	PDF_set_info( d -> pdf, "Subject", d -> subject ? d -> subject : "PDF Document" );
	PDF_set_info( d -> pdf, "Title", d -> title ? d -> title : d -> input_filename );
	PDF_set_info( d -> pdf, "Creator", "ipdf V." VERSION " by Steve Slaven <bpk@hoopajoo.net> - http://hoopajoo.net" );
	PDF_set_info( d -> pdf, "Author", d -> author ? d -> author : "Unknown" );
	if( d -> extra_data != NULL )
	{
		PDF_set_info( d -> pdf, d -> extra_key ? d -> extra_key : "Keywords", d -> extra_data );
	}

	/* Set the bookmarkdest to be zoom */
	if( PDF_get_majorversion() > 3 )
	{
		PDF_set_parameter( d -> pdf, "bookmarkdest", "fitwidth" );
	}

	return( 0 );
}

int	document_addline( document *d, const char *line )
{
	char	*t;
	int	found_index, c, parent_id, end_page;
	indexd	*i;
	regmatch_t	junk;

	d -> line_count++;
	d -> total_line_count++;

	dmsg( 4, "Adding line, line_count: %i total_line_count: %i",
		d -> line_count,
		d -> total_line_count );

	if( d -> line_count == 1 )
	{
		d -> page_count++;
		dmsg( 2, "Starting new page, #%i", d -> page_count );
		PDF_begin_page( d -> pdf, d -> width, d -> height );

		/* Add image */
		if( d -> image_filename != NULL )
		{
			dmsg( 4, "Setting background" );
			PDF_save( d -> pdf );
			PDF_scale( d -> pdf, d -> image_scale_x, d -> image_scale_y );
			PDF_place_image( d -> pdf, d -> image_id, 0, 0, 1 );
			PDF_restore( d -> pdf );
		}

		PDF_set_text_pos( d -> pdf, d -> left_margin,
			d -> height - d -> top_margin );
		PDF_setfont( d -> pdf, d -> font_index, d -> font_size );
	}

	/* Text out */
	dmsg( 5, "Truncating line: '%s'", line );
	t = (char *)malloc( d -> chars_per_line + 1 );
	bzero( t, d -> chars_per_line + 1 );
	strncpy( t, line, d -> chars_per_line );

	if( d -> strip_non_printable ) {
		/* Strip non-printable */
		for( c = 0; c < strlen( t ); c++ )
		{
			if( ( t[ c ] != '\t' ) && ( ! isprint( (char)t[ c ] ) ) )
			{
				t[ c ] = ' ';
			}
		}
	}

	dmsg( 5, "    Adding line: '%s'", t );
	PDF_continue_text( d -> pdf, t );
	free( t );

	/* Now try and index this line */
	found_index = 0;
	parent_id = 0;
	for( c = 0; c < d -> index_count; c++ )
	{
		i = &d -> index_def[ c ];
		if( found_index )
		{
			/* Zero this, index found higher up so we reset */
			dmsg( 4, "Index already happened, resetting me at #%i", c );
			if( i -> last_index != NULL )
			{
				free( i -> last_index );
				i -> last_index = NULL;
			}

			/* Clear it's index */
			i -> indexid = NO_PARENT;
		}else
		{
			switch( i -> index_type )
			{
				case IDX_RELATIVE:
					dmsg( 4, "Trying to index, line is %i, looking for %i",
						d -> line_count, i -> line_no );
					if( i -> line_no == d -> line_count )
					{
						found_index = document_do_index( d, i, line, parent_id, c );
					}
					break;

				case IDX_ABSOLUTE:
					/* Check page_no and line_count */
					if( 
						(
							( i -> page_no == 0 ) && /* page 0 means index on total linecount */
							( i -> line_no == d -> total_line_count )
						) || (
							( i -> line_no == d -> line_count ) && /* normal index */
							( i -> page_no == d -> page_count ) 
						) || (
							( i -> page_no == -1 ) && /* page -1 means index this line on every page */
							( i -> line_no == d -> line_count )
						)
					)
					{
						document_do_index( d, i, line, parent_id, c );
					}
					break;

				case IDX_REGEX:
					/* do a compare */
					if( regexec( &i -> regcomp, line, 0, &junk, 0 ) == 0 )
					{
						found_index = document_do_index( d, i, line, parent_id, c );
					}
					break;

			}
		}

		/* Save pid */
		parent_id = i -> indexid;
	}

	/* Form feed if needed */
	end_page = 0;
	if( strchr( line, '\f' ) )
	{
		/* Form feed */
		dmsg( 2, "Form feed on line %i", d -> line_count );
		end_page = 1;
	}else if( d -> line_count > d -> lines_per_page )
	{
		/* Advance */
		dmsg( 2, "Reached max line count, new page" );
		end_page = 1;
	}

	if( end_page )
	{
		dmsg( 3, "Ending current page" );
		PDF_end_page( d -> pdf );
		d -> line_count = 0;
	}

	return( 1 );
}

int	document_end( document *d )
{
	/* Flush any extra data */
	if( d -> line_count > 0 )
	{
		dmsg( 2, "Flushing final data page" );
		PDF_end_page( d -> pdf );
	}

	dmsg( 3, "Closing PDF" );
	PDF_close( d -> pdf );
	PDF_delete( d-> pdf );
}

int	document_do_index( document *d, indexd *i, const char *line, int parent_id, int depth )
{
	char 	*t, 
		*n = NULL;
	int	index_len, reparent;

	/* And index line, pull our substring and
		match it */
	index_len = i -> stop_char - i -> start_char;
	t = (char *)malloc( index_len + 1 );
	bzero( t, index_len + 1 );
	t = strncpy( t, &line[ i -> start_char ], index_len );

	if( i -> suppress_spaces )
	{
		dmsg( 4, "Normalizing space" );
		suppress_spaces( t );
	}

	/* Clean spaces */
	clean_spaces( t, CL_BOTH );

	dmsg( 4, "index substring: '%s'", t );
	/* Compare */
	if( ( ! i -> suppress_duplicates ) ||
		( 
			( i -> last_index == NULL ) ||
			( strcmp( i -> last_index, t ) != 0 )
		)
	)
	{
		dmsg( 3, "New index: %s", t );
		if( i -> reparent )
		{
			dmsg( 2, "Reparenting requested, depth is: %i, reparent is: %i", depth, i -> reparent );

			/* Reparent back N levels */
			reparent = depth - i -> reparent - 1;
			if( reparent < 0 )
			{
				dmsg( 2, "Re-parented index to top" );
				parent_id = 0;
			}else
			{
				dmsg( 2, "Re-parented to sub index" );
				parent_id = d -> index_def[ reparent ].indexid;
			}
		}

		/* Make sure parent is up */
		if( parent_id != NO_PARENT )
		{
			dmsg( 2, "OK, indexing: '%s'", t );

			/* Save this for future ref */
			if( i -> last_index != NULL )
			{
				free( i -> last_index );
			}
			i -> last_index = strdup( t );

			/* Slap it in the format string, if it's defined */
			/* Coming out of this, *t should have the index string */
			if( i -> format != NULL )
			{
				n = (char *)malloc( strlen( i -> format ) + strlen( t ) + 1 );
				sprintf( n, i -> format, t );
				free( t );
				t = n;
			}

			/* Get around NULL bookmark */
			if( t[ 0 ] == 0 )
			{
				free( t );
				t = strdup( "NULL" );
			}

			dmsg( 3, "Parentid for index: %i", parent_id );
			i -> indexid = PDF_add_bookmark( d -> pdf, t, parent_id, 0 );

			/* If we alloc'd n then dealloc */
			free( t );

			/* found index */
			return( 1 );
		}else
		{
			dmsg( 2, "Not ready for this index, discarding" );
			free( t );
		}
	}else
	{
		dmsg( 4, "Repeat" );
		free( t );
	}

	return( 0 );
}

int document_set_page( document *d, const char *ptype )
{
	/* page mode */
	if( SS( ptype, "letter-p" ) )
	{
		d -> width = letter_width;
		d -> height = letter_height;
	}else if( SS( ptype, "letter-l" ) ) {
		d -> width = letter_height;
		d -> height = letter_width;
	}else if( SS( ptype, "a0-p" ) ) {
		d -> width = a0_width;
		d -> height = a0_height;
	}else if( SS( ptype, "a0-l" ) ) {
		d -> width = a0_height;
		d -> height = a0_width;
	}else if( SS( ptype, "a1-p" ) ) {
		d -> width = a1_width;
		d -> height = a1_height;
	}else if( SS( ptype, "a1-l" ) ) {
		d -> width = a1_height;
		d -> height = a1_width;
	}else if( SS( ptype, "a2-p" ) ) {
		d -> width = a2_width;
		d -> height = a2_height;
	}else if( SS( ptype, "a2-l" ) ) {
		d -> width = a2_height;
		d -> height = a2_width;
	}else if( SS( ptype, "a3-p" ) ) {
		d -> width = a3_width; 
		d -> height = a3_height;
	}else if( SS( ptype, "a3-l" ) ) {
		d -> width = a3_height;
		d -> height = a3_width;
	}else if( SS( ptype, "a4-p" ) ) {
		d -> width = a4_width;
		d -> height = a4_height;
	}else if( SS( ptype, "a4-l" ) ) {
		d -> width = a4_height;
		d -> height = a4_width;
	}else if( SS( ptype, "a5-p" ) ) {
		d -> width = a5_width;
		d -> height = a5_height;
	}else if( SS( ptype, "a5-l" ) ) {
		d -> width = a5_height;
		d -> height = a5_width;
	}else if( SS( ptype, "a6-p" ) ) {
		d -> width = a6_width;
		d -> height = a6_height;
	}else if( SS( ptype, "a6-l" ) ) {
		d -> width = a6_height;
		d -> height = a6_width;
	}else if( SS( ptype, "b5-p" ) ) {
		d -> width = b5_width;
		d -> height = b5_height;
	}else if( SS( ptype, "b5-l" ) ) {
		d -> width = b5_height;
		d -> height = b5_width;
	}else if( SS( ptype, "legal-p" ) ) {
		d -> width = legal_width;
		d -> height = legal_height;
	}else if( SS( ptype, "legal-l" ) ) {
		d -> width = legal_height;
		d -> height = legal_width;
	}else if( SS( ptype, "ledger-p" ) ) {
		d -> width = ledger_width;
		d -> height = ledger_height;
	}else if( SS( ptype, "ledger-l" ) ) {
		d -> width = ledger_height;
		d -> height = ledger_width;
	}else if( SS( ptype, "11x17-p" ) ) {
		d -> width = p11x17_width;
		d -> height = p11x17_height; 
	}else if( SS( ptype, "11x17-l" ) ) {
		d -> width = p11x17_height;
		d -> height = p11x17_width;
	}else {
		fprintf( stderr, "Unknown page mode: '%s'\n", ptype );
		exit(0);
	}
}