/* Copyright (C) 1997, 2000 Aladdin Enterprises.  All rights reserved.
  
  This software is provided AS-IS with no warranty, either express or
  implied.
  
  This software is distributed under license and may not be copied,
  modified or distributed except as expressly authorized under the terms
  of the license contained in the file LICENSE in this distribution.
  
  For more information about licensing, please refer to
  http://www.ghostscript.com/licensing/. For information on
  commercial licensing, go to http://www.artifex.com/licensing/ or
  contact Artifex Software, Inc., 101 Lucas Valley Road #110,
  San Rafael, CA  94903, U.S.A., +1(415)492-9861.
*/

/* $Id: gdevpdfr.c,v 1.9 2005/04/25 12:28:49 igor Exp $ */
/* Named object pdfmark processing */
#include "memory_.h"
#include "string_.h"
#include "gx.h"
#include "gserrors.h"
#include "gsutil.h"		/* for bytes_compare */
#include "gdevpdfx.h"
#include "gdevpdfo.h"
#include "scanchar.h"
#include "strimpl.h"
#include "sstring.h"

#ifndef gs_error_syntaxerror
#  define gs_error_syntaxerror gs_error_rangecheck
#endif

/* Test whether an object name has valid syntax, {name}. */
bool
pdf_objname_is_valid(const byte *data, uint size)
{
    return (size >= 2 && data[0] == '{' &&
	    (const byte *)memchr(data, '}', size) == data + size - 1);
}

/*
 * Look up a named object.  Return e_rangecheck if the syntax is invalid.
 * If the object is missing, return e_undefined.
 */
int
pdf_find_named(gx_device_pdf * pdev, const gs_param_string * pname,
	       cos_object_t **ppco)
{
    const cos_value_t *pvalue;

    if (!pdf_objname_is_valid(pname->data, pname->size))
	return_error(gs_error_rangecheck);
    if ((pvalue = cos_dict_find(pdev->local_named_objects, pname->data,
				pname->size)) != 0 ||
	(pvalue = cos_dict_find(pdev->global_named_objects, pname->data,
				pname->size)) != 0
	) {
	*ppco = pvalue->contents.object;
	return 0;
    }
    return_error(gs_error_undefined);
}

/*
 * Create a (local) named object.  id = -1L means do not assign an id.
 * pname = 0 means just create the object, do not name it.  Note that
 * during initialization, local_named_objects == global_named_objects.
 */
int
pdf_create_named(gx_device_pdf *pdev, const gs_param_string *pname,
		 cos_type_t cotype, cos_object_t **ppco, long id)
{
    cos_object_t *pco;
    cos_value_t value;

    *ppco = pco = cos_object_alloc(pdev, "pdf_create_named");
    if (pco == 0)
	return_error(gs_error_VMerror);
    pco->id =
	(id == -1 ? 0L : id == 0 ? pdf_obj_ref(pdev) : id);
    if (pname) {
	int code = cos_dict_put(pdev->local_named_objects, pname->data,
				pname->size, cos_object_value(&value, pco));

	if (code < 0)
	    return code;
    }
    if (cotype != cos_type_generic)
	cos_become(pco, cotype);
    *ppco = pco;
    return 0;
}
int
pdf_create_named_dict(gx_device_pdf *pdev, const gs_param_string *pname,
		      cos_dict_t **ppcd, long id)
{
    cos_object_t *pco;
    int code = pdf_create_named(pdev, pname, cos_type_dict, &pco, id);

    *ppcd = (cos_dict_t *)pco;
    return code;
}

/*
 * Look up a named object as for pdf_find_named.  If the object does not
 * exist, create it (as a dictionary if it is one of the predefined names
 * {ThisPage}, {NextPage}, {PrevPage}, or {Page<#>}, otherwise as a
 * generic object) and return 1.
 */
int
pdf_refer_named(gx_device_pdf * pdev, const gs_param_string * pname_orig,
		cos_object_t **ppco)
{
    const gs_param_string *pname = pname_orig;
    int code = pdf_find_named(pdev, pname, ppco);
    char page_name_chars[6 + 10 + 2]; /* {Page<n>}, enough for an int */
    gs_param_string pnstr;
    int page_number;

    if (code != gs_error_undefined)
	return code;
    /*
     * Check for a predefined name.  Map ThisPage, PrevPage, and NextPage
     * to the appropriate Page<#> name.
     */
    if (pname->size >= 7 &&
	sscanf((const char *)pname->data, "{Page%d}", &page_number) == 1
	)
	goto cpage;
    if (pdf_key_eq(pname, "{ThisPage}"))
	page_number = pdev->next_page + 1;
    else if (pdf_key_eq(pname, "{NextPage}"))
	page_number = pdev->next_page + 2;
    else if (pdf_key_eq(pname, "{PrevPage}"))
	page_number = pdev->next_page;
    else {
	code = pdf_create_named(pdev, pname, cos_type_generic, ppco, 0L);
	return (code < 0 ? code : 1);
    }
    if (page_number <= 0)
	return code;
    sprintf(page_name_chars, "{Page%d}", page_number);
    param_string_from_string(pnstr, page_name_chars);
    pname = &pnstr;
    code = pdf_find_named(pdev, pname, ppco);
    if (code != gs_error_undefined)
	return code;
 cpage:
    if (pdf_page_id(pdev, page_number) <= 0)
	return_error(gs_error_rangecheck);
    *ppco = COS_OBJECT(pdev->pages[page_number - 1].Page);
    return 0;
}

/*
 * Look up a named object as for pdf_refer_named.  If the object already
 * exists and is not simply a forward reference, return e_rangecheck;
 * if it exists as a forward reference, set its type and return 0;
 * otherwise, create the object with the given type and return 1.
 */
int
pdf_make_named(gx_device_pdf * pdev, const gs_param_string * pname,
	       cos_type_t cotype, cos_object_t **ppco, bool assign_id)
{
    if (pname) {
	int code = pdf_refer_named(pdev, pname, ppco);
	cos_object_t *pco = *ppco;

	if (code < 0)
	    return code;
	if (cos_type(pco) != cos_type_generic)
	    return_error(gs_error_rangecheck);
	if (assign_id && pco->id == 0)
	    pco->id = pdf_obj_ref(pdev);
	cos_become(pco, cotype);
	return code;
    } else {
	int code = pdf_create_named(pdev, pname, cotype, ppco,
				    (assign_id ? 0L : -1L));

	return (code < 0 ? code : 1);
    }
}
int
pdf_make_named_dict(gx_device_pdf * pdev, const gs_param_string * pname,
		    cos_dict_t **ppcd, bool assign_id)
{
    cos_object_t *pco;
    int code = pdf_make_named(pdev, pname, cos_type_dict, &pco, assign_id);

    *ppcd = (cos_dict_t *)pco;
    return code;
}

/*
 * Look up a named object as for pdf_refer_named.  If the object does not
 * exist, return e_undefined; if the object exists but has the wrong type,
 * return e_typecheck.
 */
int
pdf_get_named(gx_device_pdf * pdev, const gs_param_string * pname,
	      cos_type_t cotype, cos_object_t **ppco)
{
    int code = pdf_refer_named(pdev, pname, ppco);

    if (code < 0)
	return code;
    if (cos_type(*ppco) != cotype)
	return_error(gs_error_typecheck);
    return code;
}

/*
 * Push the current local namespace onto the namespace stack, and reset it
 * to an empty namespace.
 */
int
pdf_push_namespace(gx_device_pdf *pdev)
{
    int code = cos_array_add_object(pdev->Namespace_stack,
				    COS_OBJECT(pdev->local_named_objects));
    cos_dict_t *pcd =
	cos_dict_alloc(pdev, "pdf_push_namespace(local_named_objects)");
    cos_array_t *pca =
	cos_array_alloc(pdev, "pdf_push_namespace(NI_stack)");

    if (code < 0 ||
	(code = cos_array_add_object(pdev->Namespace_stack,
				     COS_OBJECT(pdev->NI_stack))) < 0
	)
	return code;
    if (pcd == 0 || pca == 0)
	return_error(gs_error_VMerror);
    pdev->local_named_objects = pcd;
    pdev->NI_stack = pca;
    return 0;
}

/*
 * Pop the top local namespace from the namespace stack.  Return an error if
 * the stack is empty.
 */
int
pdf_pop_namespace(gx_device_pdf *pdev)
{
    cos_value_t nis_value, lno_value;
    int code = cos_array_unadd(pdev->Namespace_stack, &nis_value);
    
    if (code < 0 ||
	(code = cos_array_unadd(pdev->Namespace_stack, &lno_value)) < 0
	)
	return code;
    COS_FREE(pdev->local_named_objects,
	     "pdf_pop_namespace(local_named_objects)");
    pdev->local_named_objects = (cos_dict_t *)lno_value.contents.object;
    COS_FREE(pdev->NI_stack, "pdf_pop_namespace(NI_stack)");
    pdev->NI_stack = (cos_array_t *)nis_value.contents.object;
    return 0;
}

/*
 * Scan a token from a string.  <<, >>, [, and ] are treated as tokens.
 * Return 1 if a token was scanned, 0 if we reached the end of the string,
 * or an error.  On a successful return, the token extends from *ptoken up
 * to but not including *pscan.
 *
 * Note that this scanner expects a subset of PostScript syntax, not PDF
 * syntax.  In particular, it doesn't understand ASCII85 strings,
 * doesn't process the PDF #-escape syntax within names, and does only
 * minimal syntax checking.  It also recognizes one extension to PostScript
 * syntax, to allow gs_pdfwr.ps to pass names that include non-regular
 * characters: If a name is immediately preceded by two null characters,
 * the name includes everything up to a following null character.  The only
 * place that currently generates this convention is the PostScript code
 * that pre-processes the arguments for pdfmarks, in lib/gs_pdfwr.ps.
 */
int
pdf_scan_token(const byte **pscan, const byte * end, const byte **ptoken)
{
    const byte *p = *pscan;

    while (p < end && scan_char_decoder[*p] == ctype_space) {
	++p;
	if (p[-1] == 0 && p + 1 < end && *p == 0 && p[1] == '/') {
	/* Special handling for names delimited by a null character. */
	    *ptoken = ++p;
	    while (*p != 0)
		if (++p >= end)
		    return_error(gs_error_syntaxerror);	/* no terminator */
	    *pscan = p;
	    return 1;
	}
    }
    *ptoken = p;
    if (p >= end) {
	*pscan = p;
	return 0;
    }
    switch (*p) {
    case '%':
    case ')':
	return_error(gs_error_syntaxerror);
    case '(': {
	/* Skip over the string. */
	byte buf[50];		/* size is arbitrary */
	stream_cursor_read r;
	stream_cursor_write w;
	stream_PSSD_state ss;
	int status;

	s_PSSD_init((stream_state *)&ss);
	r.ptr = p;		/* skip the '(' */
	r.limit = end - 1;
	w.limit = buf + sizeof(buf) - 1;
	do {
	    /* One picky compiler complains if we initialize to buf - 1. */
	    w.ptr = buf;  w.ptr--;
	    status = (*s_PSSD_template.process)
		((stream_state *) & ss, &r, &w, true);
	}
	while (status == 1);
	*pscan = r.ptr + 1;
	return 1;
    }
    case '<':
	if (end - p < 2)
	    return_error(gs_error_syntaxerror);
	if (p[1] != '<') {
	    /*
	     * We need the cast because some compilers declare memchar as
	     * returning a char * rather than a void *.
	     */
	    p = (const byte *)memchr(p + 1, '>', end - p - 1);
	    if (p == 0)
		return_error(gs_error_syntaxerror);
	}
	goto m2;
    case '>':
	if (end - p < 2 || p[1] != '>')
	    return_error(gs_error_syntaxerror);
m2:	*pscan = p + 2;
	return 1;
    case '[': case ']': case '{': case '}':
	*pscan = p + 1;
	return 1;
    case '/':
	++p;
    default:
	break;
    }
    while (p < end && scan_char_decoder[*p] <= ctype_name)
	++p;
    *pscan = p;
    if (p == *ptoken)		/* no chars scanned, i.e., not ctype_name */
	return_error(gs_error_syntaxerror);
    return 1;
}
/*
 * Scan a possibly composite token: arrays and dictionaries are treated as
 * single tokens.
 */
int
pdf_scan_token_composite(const byte **pscan, const byte * end,
			 const byte **ptoken_orig)
{
    int level = 0;
    const byte *ignore_token;
    const byte **ptoken = ptoken_orig;
    int code;

    do {
	code = pdf_scan_token(pscan, end, ptoken);
	if (code <= 0)
	    return (code < 0 || level == 0 ? code :
		    gs_note_error(gs_error_syntaxerror));
	switch (**ptoken) {
	case '<': case '[': case '{':
	    ++level; break;
	case '>': case ']': case '}':
	    if (level == 0)
		return_error(gs_error_syntaxerror);
	    --level; break;
	}
	ptoken = &ignore_token;
    } while (level);
    return code;
}

/* Replace object names with object references in a (parameter) string. */
private const byte *
pdfmark_next_object(const byte * scan, const byte * end, const byte **pname,
		    cos_object_t **ppco, gx_device_pdf * pdev)
{
    /*
     * Starting at scan, find the next object reference, set *pname
     * to point to it in the string, store the object at *ppco,
     * and return a pointer to the first character beyond the
     * reference.  If there are no more object references, set
     * *pname = end, *ppco = 0, and return end.
     */
    int code;

    while ((code = pdf_scan_token(&scan, end, pname)) != 0) {
	gs_param_string sname;

	if (code < 0) {
	    ++scan;
	    continue;
	}
	if (**pname != '{')
	    continue;
	/* Back up over the { and rescan as a single token. */
	scan = *pname;
	code = pdf_scan_token_composite(&scan, end, pname);
	if (code < 0) {
	    ++scan;
	    continue;
	}
	sname.data = *pname;
	sname.size = scan - sname.data;
	/*
	 * Forward references are allowed.  If there is an error,
	 * simply retain the name as a literal string.
	 */
	code = pdf_refer_named(pdev, &sname, ppco);
	if (code < 0)
	    continue;
	return scan;
    }
    *ppco = 0;
    return end;
}
int
pdf_replace_names(gx_device_pdf * pdev, const gs_param_string * from,
		  gs_param_string * to)
{
    const byte *start = from->data;
    const byte *end = start + from->size;
    const byte *scan;
    uint size = 0;
    cos_object_t *pco;
    bool any = false;
    byte *sto;
    char ref[1 + 10 + 5 + 1];	/* max obj number is 10 digits */

    /* Do a first pass to compute the length of the result. */
    for (scan = start; scan < end;) {
	const byte *sname;
	const byte *next =
	    pdfmark_next_object(scan, end, &sname, &pco, pdev);

	size += sname - scan;
	if (pco) {
	    sprintf(ref, " %ld 0 R ", pco->id);
	    size += strlen(ref);
	}
	scan = next;
	any |= next != sname;
    }
    to->persistent = true;	/* ??? */
    if (!any) {
	to->data = start;
	to->size = size;
	return 0;
    }
    sto = gs_alloc_bytes(pdev->pdf_memory, size, "pdf_replace_names");
    if (sto == 0)
	return_error(gs_error_VMerror);
    to->data = sto;
    to->size = size;
    /* Do a second pass to do the actual substitutions. */
    for (scan = start; scan < end;) {
	const byte *sname;
	const byte *next =
	    pdfmark_next_object(scan, end, &sname, &pco, pdev);
	uint copy = sname - scan;
	int rlen;

	memcpy(sto, scan, copy);
	sto += copy;
	if (pco) {
	    sprintf(ref, " %ld 0 R ", pco->id);
	    rlen = strlen(ref);
	    memcpy(sto, ref, rlen);
	    sto += rlen;
	}
	scan = next;
    }
    return 0;
}