Lines
0 %
Functions
Branches
/* -*- Mode: c; c-basic-offset: 4; indent-tabs-mode: t; tab-width: 8; -*- */
/* cairo - a vector graphics library with display and print output
*
* Copyright © 2006 Red Hat, Inc
* This library is free software; you can redistribute it and/or
* modify it either under the terms of the GNU Lesser General Public
* License version 2.1 as published by the Free Software Foundation
* (the "LGPL") or, at your option, under the terms of the Mozilla
* Public License Version 1.1 (the "MPL"). If you do not alter this
* notice, a recipient may use your version of this file under either
* the MPL or the LGPL.
* You should have received a copy of the LGPL along with this library
* in the file COPYING-LGPL-2.1; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Suite 500, Boston, MA 02110-1335, USA
* You should have received a copy of the MPL along with this library
* in the file COPYING-MPL-1.1
* The contents of this file are subject to the Mozilla Public License
* Version 1.1 (the "License"); you may not use this file except in
* compliance with the License. You may obtain a copy of the License at
* http://www.mozilla.org/MPL/
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY
* OF ANY KIND, either express or implied. See the LGPL or the MPL for
* the specific language governing rights and limitations.
* The Original Code is the cairo graphics library.
* The Initial Developer of the Original Code is Red Hat, Inc.
* Contributor(s):
* Kristian Høgsberg <krh@redhat.com>
*/
/*
* Useful links:
* http://partners.adobe.com/public/developer/en/font/T1_SPEC.PDF
#define _DEFAULT_SOURCE /* for snprintf(), strdup() */
#include "cairoint.h"
#include "cairo-array-private.h"
#include "cairo-error-private.h"
#if CAIRO_HAS_FONT_SUBSET
#include "cairo-type1-private.h"
#include "cairo-scaled-font-subsets-private.h"
#include "cairo-output-stream-private.h"
#include <ctype.h>
#define TYPE1_STACKSIZE 24 /* Defined in Type 1 Font Format */
typedef struct {
int subset_index;
double width;
const char *encrypted_charstring;
int encrypted_charstring_length;
} glyph_data_t;
typedef struct _cairo_type1_font_subset {
cairo_scaled_font_subset_t *scaled_font_subset;
struct {
unsigned int font_id;
char *base_font;
unsigned int num_glyphs; /* Num /CharStrings in font */
double x_min, y_min, x_max, y_max;
double ascent, descent;
double units_per_em;
const char *data;
unsigned long header_size;
unsigned long data_size;
unsigned long trailer_size;
} base;
/* Num glyphs in subset. May be greater than
* scaled_font_subset->num_glyphs due to glyphs required by the
* SEAC operator. */
int num_glyphs;
/* The glyphs and glyph_names arrays are indexed by the order of
* the Charstrings in the font. This is not necessarily the same
* order as the glyph index. The index_to_glyph_name() font backend
* function is used to map the glyph index to the glyph order in
* the Charstrings. */
cairo_array_t glyphs_array;
glyph_data_t *glyphs; /* pointer to first element of above array */
cairo_array_t glyph_names_array;
char **glyph_names; /* pointer to first element of above array */
int num_subrs; /* Num /Subrs routines in the font */
cairo_bool_t subset_subrs;
const char *subr_string;
int subr_length;
const char *np;
int np_length;
cairo_bool_t used;
} *subrs; /* array with num_subrs elements */
/* Maps scaled_font_subset index to glyphs_array.
* Array size = scaled_font_subset->num_glyphs. */
int *scaled_subset_index_to_glyphs;
/* Keeps track of the glyphs that will be emitted in the subset.
* Allocated size = base.num_glyphs. Number of entries = num_glyphs.
* Array values are glyph_array indexes.
int *type1_subset_index_to_glyphs;
cairo_output_stream_t *output;
cairo_array_t contents;
const char *rd, *nd, *np;
int lenIV;
char *type1_data;
unsigned int type1_length;
char *type1_end;
char *header_segment;
unsigned int header_segment_size;
char *eexec_segment;
unsigned int eexec_segment_size;
cairo_bool_t eexec_segment_is_ascii;
char *cleartext;
char *cleartext_end;
unsigned int header_size;
unsigned short eexec_key;
cairo_bool_t hex_encode;
int hex_column;
double stack[TYPE1_STACKSIZE];
int sp;
} build_stack;
int stack[TYPE1_STACKSIZE];
} ps_stack;
} cairo_type1_font_subset_t;
static cairo_status_t
_cairo_type1_font_subset_init (cairo_type1_font_subset_t *font,
cairo_scaled_font_subset_t *scaled_font_subset,
cairo_bool_t hex_encode)
{
memset (font, 0, sizeof (*font));
font->scaled_font_subset = scaled_font_subset;
_cairo_array_init (&font->glyphs_array, sizeof (glyph_data_t));
_cairo_array_init (&font->glyph_names_array, sizeof (char *));
font->scaled_subset_index_to_glyphs = _cairo_calloc_ab (scaled_font_subset->num_glyphs,
sizeof font->scaled_subset_index_to_glyphs[0]);
if (unlikely (font->scaled_subset_index_to_glyphs == NULL))
return _cairo_error (CAIRO_STATUS_NO_MEMORY);
font->type1_subset_index_to_glyphs = NULL;
font->base.num_glyphs = 0;
font->num_subrs = 0;
font->subset_subrs = TRUE;
font->subrs = NULL;
font->hex_encode = hex_encode;
font->num_glyphs = 0;
_cairo_array_init (&font->contents, sizeof (char));
return CAIRO_STATUS_SUCCESS;
}
static void
cairo_type1_font_subset_use_glyph (cairo_type1_font_subset_t *font, int glyph)
if (font->glyphs[glyph].subset_index >= 0)
return;
font->glyphs[glyph].subset_index = font->num_glyphs;
font->type1_subset_index_to_glyphs[font->num_glyphs] = glyph;
font->num_glyphs++;
static cairo_bool_t
is_ps_delimiter(int c)
static const char delimiters[] = "()[]{}<>/% \t\r\n";
return strchr (delimiters, c) != NULL;
static const char *
find_token (const char *buffer, const char *end, const char *token)
int i, length;
/* FIXME: find substring really must be find_token */
if (buffer == NULL)
return NULL;
length = strlen (token);
for (i = 0; buffer + i < end - length + 1; i++)
if (memcmp (buffer + i, token, length) == 0)
if ((i == 0 || token[0] == '/' || is_ps_delimiter(buffer[i - 1])) &&
(buffer + i == end - length || is_ps_delimiter(buffer[i + length])))
return buffer + i;
cairo_type1_font_subset_find_segments (cairo_type1_font_subset_t *font)
unsigned char *p;
const char *eexec_token;
unsigned int size, i;
p = (unsigned char *) font->type1_data;
font->type1_end = font->type1_data + font->type1_length;
if (font->type1_length >= 2 && p[0] == 0x80 && p[1] == 0x01) {
if (font->type1_end < (char *)(p + 6))
return CAIRO_INT_STATUS_UNSUPPORTED;
font->header_segment_size =
p[2] | (p[3] << 8) | (p[4] << 16) | ((unsigned int) p[5] << 24);
font->header_segment = (char *) p + 6;
p += 6 + font->header_segment_size;
font->eexec_segment_size =
font->eexec_segment = (char *) p + 6;
font->eexec_segment_is_ascii = (p[1] == 1);
p += 6 + font->eexec_segment_size;
while (font->type1_end >= (char *)(p + 6) && p[1] != 0x03) {
size = p[2] | (p[3] << 8) | (p[4] << 16) | ((unsigned int) p[5] << 24);
if (font->type1_end < (char *)(p + 6 + size))
p += 6 + size;
font->type1_end = (char *) p;
} else {
eexec_token = find_token ((char *) p, font->type1_end, "eexec");
if (eexec_token == NULL)
font->header_segment_size = eexec_token - (char *) p + strlen ("eexec\n");
font->header_segment = (char *) p;
font->eexec_segment_size = font->type1_length - font->header_segment_size;
font->eexec_segment = (char *) p + font->header_segment_size;
font->eexec_segment_is_ascii = TRUE;
for (i = 0; i < 4; i++) {
if (!_cairo_isxdigit (font->eexec_segment[i]))
font->eexec_segment_is_ascii = FALSE;
/* Search for the definition of key and erase it by overwriting with spaces.
* This function is looks for definitions of the form:
* /key1 1234 def
* /key2 [12 34 56] def
* ie a key defined as an integer or array of integers.
cairo_type1_font_erase_dict_key (cairo_type1_font_subset_t *font,
const char *key)
const char *start, *p, *segment_end;
segment_end = font->header_segment + font->header_segment_size;
start = font->header_segment;
do {
start = find_token (start, segment_end, key);
if (start) {
p = start + strlen(key);
/* skip integers or array of integers */
while (p < segment_end &&
(_cairo_isspace(*p) ||
_cairo_isdigit(*p) ||
*p == '[' ||
*p == ']'))
p++;
if (p + 3 < segment_end && memcmp(p, "def", 3) == 0) {
/* erase definition of the key */
memset((char *) start, ' ', p + 3 - start);
start += strlen(key);
} while (start);
cairo_type1_font_subset_get_matrix (cairo_type1_font_subset_t *font,
const char *name,
double *a,
double *b,
double *c,
double *d)
const char *start, *end, *segment_end;
int ret, s_max, i, j;
char *s;
const char *decimal_point;
int decimal_point_len;
decimal_point = _cairo_get_locale_decimal_point ();
decimal_point_len = strlen (decimal_point);
assert (decimal_point_len != 0);
start = find_token (font->header_segment, segment_end, name);
if (start == NULL)
end = find_token (start, segment_end, "def");
if (end == NULL)
s_max = end - start + 5*decimal_point_len + 1;
s = _cairo_malloc (s_max);
if (unlikely (s == NULL))
i = 0;
j = 0;
while (i < end - start && j < s_max - decimal_point_len) {
if (start[i] == '.') {
strncpy(s + j, decimal_point, decimal_point_len + 1);
i++;
j += decimal_point_len;
s[j++] = start[i++];
s[j] = 0;
start = strpbrk (s, "{[");
if (!start) {
free (s);
start++;
ret = 0;
if (*start)
ret = sscanf(start, "%lf %lf %lf %lf", a, b, c, d);
if (ret != 4)
cairo_type1_font_subset_get_bbox (cairo_type1_font_subset_t *font)
cairo_status_t status;
double xx, yx, xy, yy;
status = cairo_type1_font_subset_get_matrix (font, "/FontBBox",
&x_min,
&y_min,
&x_max,
&y_max);
if (unlikely (status))
return status;
status = cairo_type1_font_subset_get_matrix (font, "/FontMatrix",
&xx, &yx, &xy, &yy);
if (yy == 0.0)
/* FreeType uses 1/yy to get units per EM */
font->base.units_per_em = 1.0/yy;
/* If the FontMatrix is not a uniform scale the metrics we extract
* from the font won't match what FreeType returns */
if (xx != yy || yx != 0.0 || xy != 0.0)
font->base.x_min = x_min / font->base.units_per_em;
font->base.y_min = y_min / font->base.units_per_em;
font->base.x_max = x_max / font->base.units_per_em;
font->base.y_max = y_max / font->base.units_per_em;
font->base.ascent = font->base.y_max;
font->base.descent = font->base.y_min;
cairo_type1_font_subset_get_fontname (cairo_type1_font_subset_t *font)
int i;
start = find_token (font->header_segment, segment_end, "/FontName");
start += strlen ("/FontName");
while (end > start && _cairo_isspace(end[-1]))
end--;
s = _cairo_malloc (end - start + 1);
strncpy (s, start, end - start);
s[end - start] = 0;
start = strchr (s, '/');
if (!start++ || !start) {
/* If font name is prefixed with a subset tag, strip it off. */
if (strlen(start) > 7 && start[6] == '+') {
for (i = 0; i < 6; i++) {
if (start[i] < 'A' || start[i] > 'Z')
break;
if (i == 6)
start += 7;
font->base.base_font = strdup (start);
if (unlikely (font->base.base_font == NULL))
status = _cairo_escape_ps_name (&font->base.base_font);
cairo_type1_font_subset_write_header (cairo_type1_font_subset_t *font,
const char *name)
unsigned int i;
int glyph;
/* FIXME:
* This function assumes that /FontName always appears
* before /Encoding. This appears to always be the case with Type1
* fonts.
* The more recently added code for removing the UniqueID and XUID
* keys can not make any assumptions about the position of the
* keys in the dictionary so it is implemented by overwriting the
* key definition with spaces before we start copying the font to
* the output.
* This code should be rewritten to not make any assumptions about
* the order of dictionary keys. This will allow UniqueID to be
* stripped out instead of leaving a bunch of spaces in the
* output.
cairo_type1_font_erase_dict_key (font, "/UniqueID");
cairo_type1_font_erase_dict_key (font, "/XUID");
/* Type 1 fonts created by Fontforge have some PostScript code at
* the start of the font that skips the font if the printer has a
* cached copy of the font with the same unique id. This breaks
* our subsetted font so we disable it by searching for the
* PostScript operator "known" when used to check for the
* "/UniqueID" dictionary key. We append " pop false " after it to
* pop the result of this check off the stack and replace it with
* "false" to make the PostScript code think "/UniqueID" does not
* exist.
end = font->header_segment;
start = find_token (font->header_segment, segment_end, "/UniqueID");
start += 9;
while (start < segment_end && _cairo_isspace (*start))
if (start + 5 < segment_end && memcmp(start, "known", 5) == 0) {
_cairo_output_stream_write (font->output, font->header_segment,
start + 5 - font->header_segment);
_cairo_output_stream_printf (font->output, " pop false ");
end = start + 5;
start = find_token (end, segment_end, "/FontName");
_cairo_output_stream_write (font->output, end,
start - end);
_cairo_output_stream_printf (font->output, "/FontName /%s def", name);
end += 3;
start = find_token (end, segment_end, "/Encoding");
_cairo_output_stream_write (font->output, end, start - end);
_cairo_output_stream_printf (font->output,
"/Encoding 256 array\n"
"0 1 255 {1 index exch /.notdef put} for\n");
if (font->scaled_font_subset->is_latin) {
for (i = 1; i < 256; i++) {
int subset_glyph = font->scaled_font_subset->latin_to_subset_glyph_index[i];
if (subset_glyph > 0) {
"dup %d /%s put\n",
i,
_cairo_winansi_to_glyphname (i));
for (i = 1; i < font->scaled_font_subset->num_glyphs; i++) {
glyph = font->scaled_subset_index_to_glyphs[i];
font->glyph_names[glyph]);
_cairo_output_stream_printf (font->output, "readonly def");
/* There are some buggy fonts that contain more than one /Encoding */
if (find_token (end, segment_end, "/Encoding"))
_cairo_output_stream_write (font->output, end, segment_end - end);
return font->output->status;
static int
hex_to_int (int ch)
if (ch <= '9')
return ch - '0';
else if (ch <= 'F')
return ch - 'A' + 10;
else
return ch - 'a' + 10;
cairo_type1_font_subset_write_encrypted (cairo_type1_font_subset_t *font,
const char *data, unsigned int length)
const unsigned char *in, *end;
int c, p;
static const char hex_digits[16] = "0123456789abcdef";
char digits[3];
in = (const unsigned char *) data;
end = (const unsigned char *) data + length;
while (in < end) {
p = *in++;
c = p ^ (font->eexec_key >> 8);
font->eexec_key = (c + font->eexec_key) * CAIRO_TYPE1_ENCRYPT_C1 + CAIRO_TYPE1_ENCRYPT_C2;
if (font->hex_encode) {
digits[0] = hex_digits[c >> 4];
digits[1] = hex_digits[c & 0x0f];
digits[2] = '\n';
font->hex_column += 2;
if (font->hex_column == 78) {
_cairo_output_stream_write (font->output, digits, 3);
font->hex_column = 0;
_cairo_output_stream_write (font->output, digits, 2);
digits[0] = c;
_cairo_output_stream_write (font->output, digits, 1);
cairo_type1_font_subset_decrypt_eexec_segment (cairo_type1_font_subset_t *font)
unsigned short r = CAIRO_TYPE1_PRIVATE_DICT_KEY;
unsigned char *in, *end;
char *out;
in = (unsigned char *) font->eexec_segment;
end = (unsigned char *) in + font->eexec_segment_size;
font->cleartext = _cairo_malloc (font->eexec_segment_size + 1);
if (unlikely (font->cleartext == NULL))
out = font->cleartext;
if (font->eexec_segment_is_ascii) {
c = *in++;
if (_cairo_isspace (c))
continue;
c = (hex_to_int (c) << 4) | hex_to_int (*in++);
p = c ^ (r >> 8);
r = (c + r) * CAIRO_TYPE1_ENCRYPT_C1 + CAIRO_TYPE1_ENCRYPT_C2;
*out++ = p;
font->cleartext_end = out;
/* Overwrite random bytes with spaces.
* The first 4 bytes of the cleartext are the random bytes
* required by the encryption algorithm. When encrypting the
* cleartext, the first ciphertext byte must not be a white space
* character and the first 4 bytes must not be an ASCII Hex
* character. Some fonts do not check that their randomly chosen
* bytes results in ciphertext that complies with this
* restriction. This may cause problems for some PDF consumers. By
* replacing the random bytes with spaces, the first four bytes of
* ciphertext will always be 0xf9, 0x83, 0xef, 0x00 which complies
* with this restriction. Using spaces also means we don't have to
* skip over the random bytes when parsing the cleartext.
for (i = 0; i < 4 && i < font->eexec_segment_size; i++)
font->cleartext[i] = ' ';
/* Ensure strtol() can not scan past the end of the cleartext */
font->cleartext[font->eexec_segment_size] = 0;
skip_token (const char *p, const char *end)
while (p < end && _cairo_isspace(*p))
while (p < end && !_cairo_isspace(*p))
if (p == end)
return p;
cairo_type1_font_subset_decrypt_charstring (const unsigned char *in, int size, unsigned char *out)
unsigned short r = CAIRO_TYPE1_CHARSTRING_KEY;
int c, p, i;
for (i = 0; i < size; i++) {
static const unsigned char *
cairo_type1_font_subset_decode_integer (const unsigned char *p, int *integer)
if (*p <= 246) {
*integer = *p++ - 139;
} else if (*p <= 250) {
*integer = (p[0] - 247) * 256 + p[1] + 108;
p += 2;
} else if (*p <= 254) {
*integer = -(p[0] - 251) * 256 - p[1] - 108;
*integer = ((uint32_t)p[1] << 24) | (p[2] << 16) | (p[3] << 8) | p[4];
p += 5;
use_standard_encoding_glyph (cairo_type1_font_subset_t *font, int index)
const char *glyph_name;
if (index < 0 || index > 255)
glyph_name = _cairo_ps_standard_encoding_to_glyphname (index);
if (glyph_name == NULL)
for (i = 0; i < font->base.num_glyphs; i++) {
if (font->glyph_names[i] && strcmp (font->glyph_names[i], glyph_name) == 0) {
cairo_type1_font_subset_use_glyph (font, i);
#define TYPE1_CHARSTRING_COMMAND_HSTEM 0x01
#define TYPE1_CHARSTRING_COMMAND_VSTEM 0x03
#define TYPE1_CHARSTRING_COMMAND_VMOVETO 0x04
#define TYPE1_CHARSTRING_COMMAND_RLINETO 0x05
#define TYPE1_CHARSTRING_COMMAND_HLINETO 0x06
#define TYPE1_CHARSTRING_COMMAND_VLINETO 0x07
#define TYPE1_CHARSTRING_COMMAND_RRCURVETO 0x08
#define TYPE1_CHARSTRING_COMMAND_CLOSEPATH 0x09
#define TYPE1_CHARSTRING_COMMAND_CALLSUBR 0x0a
#define TYPE1_CHARSTRING_COMMAND_RETURN 0x0b
#define TYPE1_CHARSTRING_COMMAND_ESCAPE 0x0c
#define TYPE1_CHARSTRING_COMMAND_HSBW 0x0d
#define TYPE1_CHARSTRING_COMMAND_ENDCHAR 0x0e
#define TYPE1_CHARSTRING_COMMAND_RMOVETO 0x15
#define TYPE1_CHARSTRING_COMMAND_HMOVETO 0x16
#define TYPE1_CHARSTRING_COMMAND_VHCURVETO 0x1e
#define TYPE1_CHARSTRING_COMMAND_HVCURVETO 0x1f
#define TYPE1_CHARSTRING_COMMAND_DOTSECTION 0x0c00
#define TYPE1_CHARSTRING_COMMAND_VSTEM3 0x0c01
#define TYPE1_CHARSTRING_COMMAND_HSTEM3 0x0c02
#define TYPE1_CHARSTRING_COMMAND_SEAC 0x0c06
#define TYPE1_CHARSTRING_COMMAND_SBW 0x0c07
#define TYPE1_CHARSTRING_COMMAND_DIV 0x0c0c
#define TYPE1_CHARSTRING_COMMAND_CALLOTHERSUBR 0x0c10
#define TYPE1_CHARSTRING_COMMAND_POP 0x0c11
#define TYPE1_CHARSTRING_COMMAND_SETCURRENTPOINT 0x0c21
/* Parse the charstring, including recursing into subroutines. Find
* the glyph width, subroutines called, and glyphs required by the
cairo_type1_font_subset_parse_charstring (cairo_type1_font_subset_t *font,
int glyph,
const char *encrypted_charstring,
int encrypted_charstring_length)
unsigned char *charstring;
const unsigned char *end;
const unsigned char *p;
int command;
charstring = _cairo_malloc (encrypted_charstring_length);
if (unlikely (charstring == NULL))
cairo_type1_font_subset_decrypt_charstring ((const unsigned char *)
encrypted_charstring,
encrypted_charstring_length,
charstring);
end = charstring + encrypted_charstring_length;
p = charstring + font->lenIV;
status = CAIRO_STATUS_SUCCESS;
while (p < end) {
if (*p < 32) {
command = *p++;
switch (command) {
case TYPE1_CHARSTRING_COMMAND_HSTEM:
case TYPE1_CHARSTRING_COMMAND_VSTEM:
case TYPE1_CHARSTRING_COMMAND_VMOVETO:
case TYPE1_CHARSTRING_COMMAND_RLINETO:
case TYPE1_CHARSTRING_COMMAND_HLINETO:
case TYPE1_CHARSTRING_COMMAND_VLINETO:
case TYPE1_CHARSTRING_COMMAND_RRCURVETO:
case TYPE1_CHARSTRING_COMMAND_CLOSEPATH:
case TYPE1_CHARSTRING_COMMAND_RMOVETO:
case TYPE1_CHARSTRING_COMMAND_HMOVETO:
case TYPE1_CHARSTRING_COMMAND_VHCURVETO:
case TYPE1_CHARSTRING_COMMAND_HVCURVETO:
case TYPE1_CHARSTRING_COMMAND_RETURN:
case TYPE1_CHARSTRING_COMMAND_ENDCHAR:
default:
/* stack clearing operator */
font->build_stack.sp = 0;
case TYPE1_CHARSTRING_COMMAND_CALLSUBR:
if (font->subset_subrs && font->build_stack.sp > 0) {
double int_val;
if (modf(font->build_stack.stack[--font->build_stack.sp], &int_val) == 0.0) {
int subr_num = int_val;
if (subr_num >= 0 && subr_num < font->num_subrs) {
font->subrs[subr_num].used = TRUE;
status = cairo_type1_font_subset_parse_charstring (
font,
glyph,
font->subrs[subr_num].subr_string,
font->subrs[subr_num].subr_length);
font->subset_subrs = FALSE;
case TYPE1_CHARSTRING_COMMAND_HSBW:
if (font->build_stack.sp < 2) {
status = CAIRO_INT_STATUS_UNSUPPORTED;
goto cleanup;
font->glyphs[glyph].width = font->build_stack.stack[1]/font->base.units_per_em;
case TYPE1_CHARSTRING_COMMAND_ESCAPE:
command = command << 8 | *p++;
case TYPE1_CHARSTRING_COMMAND_DOTSECTION:
case TYPE1_CHARSTRING_COMMAND_VSTEM3:
case TYPE1_CHARSTRING_COMMAND_HSTEM3:
case TYPE1_CHARSTRING_COMMAND_SETCURRENTPOINT:
case TYPE1_CHARSTRING_COMMAND_SEAC:
/* The seac command takes five integer arguments. The
* last two are glyph indices into the PS standard
* encoding give the names of the glyphs that this
* glyph is composed from. All we need to do is to
* make sure those glyphs are present in the subset
* under their standard names. */
if (font->build_stack.sp < 5) {
status = use_standard_encoding_glyph (font, font->build_stack.stack[3]);
status = use_standard_encoding_glyph (font, font->build_stack.stack[4]);
case TYPE1_CHARSTRING_COMMAND_SBW:
if (font->build_stack.sp < 4) {
font->glyphs[glyph].width = font->build_stack.stack[2]/font->base.units_per_em;
case TYPE1_CHARSTRING_COMMAND_DIV:
double num1 = font->build_stack.stack[font->build_stack.sp - 2];
double num2 = font->build_stack.stack[font->build_stack.sp - 1];
font->build_stack.sp--;
if (num2 == 0.0) {
font->build_stack.stack[font->build_stack.sp - 1] = num1/num2;
case TYPE1_CHARSTRING_COMMAND_CALLOTHERSUBR:
if (font->build_stack.sp < 1) {
font->ps_stack.sp = 0;
while (font->build_stack.sp)
font->ps_stack.stack[font->ps_stack.sp++] = font->build_stack.stack[--font->build_stack.sp];
case TYPE1_CHARSTRING_COMMAND_POP:
if (font->ps_stack.sp < 1) {
/* T1 spec states that if the interpreter does not
* support executing the callothersub, the results
* must be taken from the callothersub arguments. */
font->build_stack.stack[font->build_stack.sp++] = font->ps_stack.stack[--font->ps_stack.sp];
/* integer argument */
if (font->build_stack.sp < TYPE1_STACKSIZE) {
int val;
p = cairo_type1_font_subset_decode_integer (p, &val);
font->build_stack.stack[font->build_stack.sp++] = val;
cleanup:
free (charstring);
cairo_type1_font_subset_build_subr_list (cairo_type1_font_subset_t *font,
int subr_number,
const char *encrypted_charstring, int encrypted_charstring_length,
const char *np, int np_length)
font->subrs[subr_number].subr_string = encrypted_charstring;
font->subrs[subr_number].subr_length = encrypted_charstring_length;
font->subrs[subr_number].np = np;
font->subrs[subr_number].np_length = np_length;
write_used_subrs (cairo_type1_font_subset_t *font,
const char *subr_string, int subr_string_length,
char buffer[256];
int length;
if (!font->subrs[subr_number].used)
length = snprintf (buffer, sizeof buffer,
"dup %d %d %s ",
subr_number, subr_string_length, font->rd);
status = cairo_type1_font_subset_write_encrypted (font, buffer, length);
status = cairo_type1_font_subset_write_encrypted (font,
subr_string,
subr_string_length);
if (np) {
status = cairo_type1_font_subset_write_encrypted (font, np, np_length);
length = snprintf (buffer, sizeof buffer, "%s\n", font->np);
typedef cairo_status_t (*subr_func_t) (cairo_type1_font_subset_t *font,
const char *np, int np_length);
cairo_type1_font_for_each_subr (cairo_type1_font_subset_t *font,
const char *array_start,
const char *cleartext_end,
subr_func_t func,
const char **array_end)
const char *p, *subr_string;
char *end;
int subr_num, subr_length;
/* We're looking at "dup" at the start of the first subroutine. The subroutines
* definitions are on the form:
* dup 5 23 RD <23 binary bytes> NP
* or alternatively using -| and |- instead of RD and ND.
* The first number is the subroutine number.
p = array_start;
while (p + 3 < cleartext_end && strncmp (p, "dup", 3) == 0) {
p = skip_token (p, cleartext_end);
/* get subr number */
subr_num = strtol (p, &end, 10);
if (subr_num < 0 || subr_num >= font->num_subrs)
/* get subr length */
p = end;
subr_length = strtol (p, &end, 10);
/* Skip past -| or RD to binary data. There is exactly one space
* between the -| or RD token and the encrypted data, thus '+ 1'. */
subr_string = skip_token (end, cleartext_end) + 1;
np = NULL;
np_length = 0;
/* Skip binary data and | or NP token. */
p = skip_token (subr_string + subr_length, cleartext_end);
if (p == NULL)
while (p < cleartext_end && _cairo_isspace(*p))
/* Some fonts have "noaccess put" instead of "NP" */
if (p + 3 < cleartext_end && strncmp (p, "put", 3) == 0) {
np = subr_string + subr_length;
np_length = p - np;
status = func (font, subr_num,
subr_string, subr_length, np, np_length);
*array_end = (char *) p;
cairo_type1_font_subset_build_glyph_list (cairo_type1_font_subset_t *font,
int glyph_number,
const char *name, int name_length,
const char *encrypted_charstring, int encrypted_charstring_length)
glyph_data_t glyph;
s = _cairo_malloc (name_length + 1);
strncpy (s, name, name_length);
s[name_length] = 0;
status = _cairo_array_append (&font->glyph_names_array, &s);
glyph.subset_index = -1;
glyph.width = 0;
glyph.encrypted_charstring = encrypted_charstring;
glyph.encrypted_charstring_length = encrypted_charstring_length;
status = _cairo_array_append (&font->glyphs_array, &glyph);
write_used_glyphs (cairo_type1_font_subset_t *font,
const char *charstring, int charstring_length)
unsigned int subset_id;
int ch;
const char *wa_name;
if (font->glyphs[glyph_number].subset_index < 0)
/* When using the WinAnsi encoding in PDF, the /Encoding array
* is ignored and instead glyphs are keyed by glyph names. To
* ensure correct rendering we replace the glyph name in the
* font with the standard name.
**/
subset_id = font->glyphs[glyph_number].subset_index;
/* Any additional glyph included for use by the seac operator
* will either have subset_id >= font->scaled_font_subset->num_glyphs
* or will not map to a winansi name (wa_name = NULL). In this
* case the original name is used.
if (subset_id > 0 && subset_id < font->scaled_font_subset->num_glyphs) {
ch = font->scaled_font_subset->to_latin_char[subset_id];
wa_name = _cairo_winansi_to_glyphname (ch);
if (wa_name) {
name = wa_name;
name_length = strlen(name);
"/%.*s %d %s ",
name_length, name, charstring_length, font->rd);
charstring,
charstring_length);
length = snprintf (buffer, sizeof buffer, "%s\n", font->nd);
typedef cairo_status_t (*glyph_func_t) (cairo_type1_font_subset_t *font,
const char *charstring, int charstring_length);
cairo_type1_font_subset_for_each_glyph (cairo_type1_font_subset_t *font,
const char *dict_start,
const char *dict_end,
glyph_func_t func,
const char **dict_out)
int charstring_length, name_length;
const char *p, *charstring, *name;
int glyph_count;
/* We're looking at '/' in the name of the first glyph. The glyph
* /name 23 RD <23 binary bytes> ND
* We parse the glyph name and see if it is in the subset. If it
* is, we call the specified callback with the glyph name and
* glyph data, otherwise we just skip it. We need to parse
* through a glyph definition; we can't just find the next '/',
* since the binary data could contain a '/'.
p = dict_start;
glyph_count = 0;
while (*p == '/') {
name = p + 1;
p = skip_token (p, dict_end);
name_length = p - name;
charstring_length = strtol (p, &end, 10);
charstring = skip_token (end, dict_end) + 1;
/* Skip binary data and |- or ND token. */
p = skip_token (charstring + charstring_length, dict_end);
while (p < dict_end && _cairo_isspace(*p))
/* In case any of the skip_token() calls above reached EOF, p will
* be equal to dict_end. */
if (p == dict_end)
status = func (font, glyph_count++,
name, name_length,
charstring, charstring_length);
*dict_out = p;
cairo_type1_font_subset_write_private_dict (cairo_type1_font_subset_t *font,
const char *p, *subrs, *charstrings, *array_start, *array_end, *dict_start, *dict_end;
const char *lenIV_start, *lenIV_end, *closefile_token;
char buffer[32], *lenIV_str, *subr_count_end, *glyph_count_end;
int ret, lenIV, length;
const cairo_scaled_font_backend_t *backend;
int glyph, j;
/* The private dict holds hint information, common subroutines and
* the actual glyph definitions (charstrings).
* What we do here is scan directly to the /Subrs token, which
* marks the beginning of the subroutines. We read in all the
* subroutines, then move on to the /CharString token, which marks
* the beginning of the glyph definitions, and read in the charstrings.
* The charstrings are parsed to extract glyph widths, work out
* which subroutines are called, and to see if any extra glyphs
* need to be included due to the use of the seac glyph combining
* operator.
* Finally, the private dict is copied to the subset font minus the
* subroutines and charstrings not required.
/* Determine lenIV, the number of random characters at the start of
each encrypted charstring. The default is 4, but this can be
overridden in the private dict. */
font->lenIV = 4;
if ((lenIV_start = find_token (font->cleartext, font->cleartext_end, "/lenIV")) != NULL) {
lenIV_start += 6;
lenIV_end = find_token (lenIV_start, font->cleartext_end, "def");
if (lenIV_end == NULL)
lenIV_str = _cairo_malloc (lenIV_end - lenIV_start + 1);
if (unlikely (lenIV_str == NULL))
strncpy (lenIV_str, lenIV_start, lenIV_end - lenIV_start);
lenIV_str[lenIV_end - lenIV_start] = 0;
ret = sscanf(lenIV_str, "%d", &lenIV);
free(lenIV_str);
if (unlikely (ret <= 0))
/* Apparently some fonts signal unencrypted charstrings with a negative lenIV,
though this is not part of the Type 1 Font Format specification. See, e.g.
http://lists.gnu.org/archive/html/freetype-devel/2000-06/msg00064.html. */
if (unlikely (lenIV < 0))
font->lenIV = lenIV;
/* Find start of Subrs */
subrs = find_token (font->cleartext, font->cleartext_end, "/Subrs");
if (subrs == NULL) {
p = font->cleartext;
array_start = NULL;
goto skip_subrs;
/* Scan past /Subrs and get the array size. */
p = subrs + strlen ("/Subrs");
font->num_subrs = strtol (p, &subr_count_end, 10);
if (subr_count_end == p)
if (font->num_subrs <= 0)
font->subrs = _cairo_calloc_ab (font->num_subrs, sizeof (font->subrs[0]));
if (unlikely (font->subrs == NULL))
/* look for "dup" which marks the beginning of the first subr */
array_start = find_token (subr_count_end, font->cleartext_end, "dup");
if (array_start == NULL)
/* Read in the subroutines */
status = cairo_type1_font_for_each_subr (font,
array_start,
font->cleartext_end,
cairo_type1_font_subset_build_subr_list,
&array_end);
if (unlikely(status))
p = array_end;
skip_subrs:
/* Find start of CharStrings */
charstrings = find_token (p, font->cleartext_end, "/CharStrings");
if (charstrings == NULL)
/* Scan past /CharStrings and the integer following it. */
p = charstrings + strlen ("/CharStrings");
strtol (p, &glyph_count_end, 10);
if (p == glyph_count_end)
/* Look for a '/' which marks the beginning of the first glyph
* definition. */
for (p = glyph_count_end; p < font->cleartext_end; p++)
if (*p == '/')
if (p == font->cleartext_end)
dict_start = p;
/* Now that we have the private dictionary broken down in
* sections, do the first pass through the glyph definitions to
* build a list of glyph names and charstrings. */
status = cairo_type1_font_subset_for_each_glyph (font,
dict_start,
cairo_type1_font_subset_build_glyph_list,
&dict_end);
font->glyphs = _cairo_array_index (&font->glyphs_array, 0);
font->glyph_names = _cairo_array_index (&font->glyph_names_array, 0);
font->base.num_glyphs = _cairo_array_num_elements (&font->glyphs_array);
font->type1_subset_index_to_glyphs = _cairo_calloc_ab (font->base.num_glyphs,
sizeof font->type1_subset_index_to_glyphs[0]);
if (unlikely (font->type1_subset_index_to_glyphs == NULL))
backend = font->scaled_font_subset->scaled_font->backend;
if (!backend->index_to_glyph_name)
/* Find the glyph number corresponding to each glyph in the subset
* and mark it as in use */
for (i = 0; i < font->scaled_font_subset->num_glyphs; i++) {
unsigned long index;
status = backend->index_to_glyph_name (font->scaled_font_subset->scaled_font,
font->glyph_names,
font->base.num_glyphs,
font->scaled_font_subset->glyphs[i],
&index);
cairo_type1_font_subset_use_glyph (font, index);
font->scaled_subset_index_to_glyphs[i] = index;
/* Go through the charstring of each glyph in use, get the glyph
* width and figure out which extra glyphs may be required by the
* seac operator (which may cause font->num_glyphs to increase
* while this loop is executing). Also subset the Subrs. */
for (j = 0; j < font->num_glyphs; j++) {
glyph = font->type1_subset_index_to_glyphs[j];
status = cairo_type1_font_subset_parse_charstring (font,
font->glyphs[glyph].encrypted_charstring,
font->glyphs[glyph].encrypted_charstring_length);
/* Always include the first five subroutines in case the Flex/hint mechanism is
* being used. */
for (j = 0; j < MIN (font->num_subrs, 5); j++) {
font->subrs[j].used = TRUE;
closefile_token = find_token (dict_end, font->cleartext_end, "closefile");
if (closefile_token == NULL)
/* We're ready to start outputting. First write the header,
* i.e. the public part of the font dict.*/
status = cairo_type1_font_subset_write_header (font, name);
font->base.header_size = _cairo_output_stream_get_position (font->output);
/* Start outputting the private dict */
if (font->subset_subrs) {
/* First output everything up to the start of the Subrs array. */
status = cairo_type1_font_subset_write_encrypted (font, font->cleartext,
array_start - font->cleartext);
/* Write out the subr definitions for each of the glyphs in
* the subset. */
write_used_subrs,
&p);
/* If subr subsetting, output everything from end of subrs to
* start of /CharStrings token. If not subr subsetting, output
* everything start of private dict to start of /CharStrings
* token. */
status = cairo_type1_font_subset_write_encrypted (font, p, charstrings - p);
/* Write out new charstring count */
"/CharStrings %d", font->num_glyphs);
/* Write out text between the charstring count and the first
* charstring definition */
status = cairo_type1_font_subset_write_encrypted (font, glyph_count_end,
dict_start - glyph_count_end);
/* Write out the charstring definitions for each of the glyphs in
write_used_glyphs,
/* Output what's left between the end of the glyph definitions and
* the end of the private dict to the output. */
status = cairo_type1_font_subset_write_encrypted (font, p,
closefile_token - p + strlen ("closefile") + 1);
if (font->hex_encode)
_cairo_output_stream_write (font->output, "\n", 1);
cairo_type1_font_subset_write_trailer(cairo_type1_font_subset_t *font)
const char *cleartomark_token;
static const char zeros[65] =
"0000000000000000000000000000000000000000000000000000000000000000\n";
for (i = 0; i < 8; i++)
_cairo_output_stream_write (font->output, zeros, sizeof zeros);
cleartomark_token = find_token (font->type1_data, font->type1_end, "cleartomark");
if (cleartomark_token) {
/* Some fonts have conditional save/restore around the entire
* font dict, so we need to retain whatever postscript code
* that may come after 'cleartomark'. */
_cairo_output_stream_write (font->output, cleartomark_token,
font->type1_end - cleartomark_token);
if (*(font->type1_end - 1) != '\n')
_cairo_output_stream_printf (font->output, "\n");
} else if (!font->eexec_segment_is_ascii) {
/* Fonts embedded in PDF may omit the fixed-content portion
* that includes the 'cleartomark' operator. Type 1 in PDF is
* always binary. */
_cairo_output_stream_printf (font->output, "cleartomark\n");
/* some fonts do not have a newline at the end of the last line */
type1_font_write (void *closure, const unsigned char *data, unsigned int length)
cairo_type1_font_subset_t *font = closure;
return _cairo_array_append_multiple (&font->contents, data, length);
cairo_type1_font_subset_write (cairo_type1_font_subset_t *font,
status = cairo_type1_font_subset_find_segments (font);
status = cairo_type1_font_subset_decrypt_eexec_segment (font);
/* Determine which glyph definition delimiters to use. */
if (find_token (font->cleartext, font->cleartext_end, "/-|") != NULL) {
font->rd = "-|";
font->nd = "|-";
font->np = "|";
} else if (find_token (font->cleartext, font->cleartext_end, "/RD") != NULL) {
font->rd = "RD";
font->nd = "ND";
font->np = "NP";
/* Don't know *what* kind of font this is... */
font->eexec_key = CAIRO_TYPE1_PRIVATE_DICT_KEY;
status = cairo_type1_font_subset_get_bbox (font);
status = cairo_type1_font_subset_get_fontname (font);
status = cairo_type1_font_subset_write_private_dict (font, name);
font->base.data_size = _cairo_output_stream_get_position (font->output) -
font->base.header_size;
status = cairo_type1_font_subset_write_trailer (font);
font->base.trailer_size =
_cairo_output_stream_get_position (font->output) -
font->base.header_size - font->base.data_size;
check_fontdata_is_type1 (const unsigned char *data, long length)
/* Test for Type 1 Binary (PFB) */
if (length > 2 && data[0] == 0x80 && data[1] == 0x01)
return TRUE;
/* Test for Type 1 1 ASCII (PFA) */
if (length > 2 && data[0] == '%' && data[1] == '!')
return FALSE;
cairo_type1_font_subset_generate (void *abstract_font,
cairo_type1_font_subset_t *font = abstract_font;
cairo_scaled_font_t *scaled_font;
unsigned long data_length;
scaled_font = font->scaled_font_subset->scaled_font;
if (!scaled_font->backend->load_type1_data)
status = scaled_font->backend->load_type1_data (scaled_font, 0, NULL, &data_length);
if (status)
font->type1_length = data_length;
font->type1_data = _cairo_malloc (font->type1_length);
if (unlikely (font->type1_data == NULL))
status = scaled_font->backend->load_type1_data (scaled_font, 0,
(unsigned char *) font->type1_data,
&data_length);
if (!check_fontdata_is_type1 ((unsigned char *)font->type1_data, data_length))
status = _cairo_array_grow_by (&font->contents, 4096);
font->output = _cairo_output_stream_create (type1_font_write, NULL, font);
if (unlikely ((status = font->output->status)))
status = cairo_type1_font_subset_write (font, name);
font->base.data = _cairo_array_index (&font->contents, 0);
_cairo_type1_font_subset_fini (cairo_type1_font_subset_t *font)
cairo_status_t status = CAIRO_STATUS_SUCCESS;
/* If the subset generation failed, some of the pointers below may
* be NULL depending on at which point the error occurred. */
_cairo_array_fini (&font->contents);
free (font->type1_data);
for (i = 0; i < _cairo_array_num_elements (&font->glyph_names_array); i++) {
char **s;
s = _cairo_array_index (&font->glyph_names_array, i);
free (*s);
_cairo_array_fini (&font->glyph_names_array);
_cairo_array_fini (&font->glyphs_array);
free (font->subrs);
if (font->output != NULL)
status = _cairo_output_stream_destroy (font->output);
free (font->base.base_font);
free (font->scaled_subset_index_to_glyphs);
free (font->type1_subset_index_to_glyphs);
free (font->cleartext);
cairo_status_t
_cairo_type1_subset_init (cairo_type1_subset_t *type1_subset,
cairo_type1_font_subset_t font;
cairo_bool_t is_synthetic;
unsigned long length;
char buf[30];
/* We need to use a fallback font if this font differs from the type1 outlines. */
if (scaled_font_subset->scaled_font->backend->is_synthetic) {
status = scaled_font_subset->scaled_font->backend->is_synthetic (scaled_font_subset->scaled_font, &is_synthetic);
if (is_synthetic)
status = _cairo_type1_font_subset_init (&font, scaled_font_subset, hex_encode);
status = cairo_type1_font_subset_generate (&font, name);
goto fail1;
if (font.base.base_font) {
type1_subset->base_font = strdup (font.base.base_font);
snprintf(buf, sizeof (buf), "CairoFont-%u-%u",
scaled_font_subset->font_id, scaled_font_subset->subset_id);
type1_subset->base_font = strdup (buf);
if (unlikely (type1_subset->base_font == NULL))
type1_subset->widths = _cairo_calloc_ab (scaled_font_subset->num_glyphs, sizeof (double));
if (unlikely (type1_subset->widths == NULL))
goto fail2;
for (i = 0; i < font.scaled_font_subset->num_glyphs; i++) {
glyph = font.scaled_subset_index_to_glyphs[i];
type1_subset->widths[i] = font.glyphs[glyph].width;
type1_subset->x_min = font.base.x_min;
type1_subset->y_min = font.base.y_min;
type1_subset->x_max = font.base.x_max;
type1_subset->y_max = font.base.y_max;
type1_subset->ascent = font.base.ascent;
type1_subset->descent = font.base.descent;
length = font.base.header_size +
font.base.data_size +
font.base.trailer_size;
type1_subset->data = _cairo_malloc (length);
if (unlikely (type1_subset->data == NULL))
goto fail3;
memcpy (type1_subset->data,
_cairo_array_index (&font.contents, 0), length);
type1_subset->header_length = font.base.header_size;
type1_subset->data_length = font.base.data_size;
type1_subset->trailer_length = font.base.trailer_size;
return _cairo_type1_font_subset_fini (&font);
fail3:
free (type1_subset->widths);
fail2:
free (type1_subset->base_font);
fail1:
_cairo_type1_font_subset_fini (&font);
void
_cairo_type1_subset_fini (cairo_type1_subset_t *subset)
free (subset->base_font);
free (subset->widths);
free (subset->data);
cairo_bool_t
_cairo_type1_scaled_font_is_type1 (cairo_scaled_font_t *scaled_font)
unsigned char buf[64];
status = scaled_font->backend->load_type1_data (scaled_font, 0, NULL, &length);
/* We only need a few bytes to test for Type 1 */
if (length > sizeof (buf))
length = sizeof (buf);
status = scaled_font->backend->load_type1_data (scaled_font, 0, buf, &length);
return check_fontdata_is_type1 (buf, length);
#endif /* CAIRO_HAS_FONT_SUBSET */