/* Convert Unicode/PdfDocEncoding string into ucs-2 */
unsigned short *
pdf_to_ucs2(fz_context *ctx, pdf_obj *src)
{
unsigned char *srcptr = (unsigned char *) pdf_to_str_buf(src);
unsigned short *dstptr, *dst;
int srclen = pdf_to_str_len(src);
int i;
if (srclen >= 2 && srcptr[0] == 254 && srcptr[1] == 255)
{
dstptr = dst = fz_malloc_array(ctx, (srclen - 2) / 2 + 1, sizeof(short));
for (i = 2; i + 1 < srclen; i += 2)
*dstptr++ = srcptr[i] << 8 | srcptr[i+1];
}
else if (srclen >= 2 && srcptr[0] == 255 && srcptr[1] == 254)
{
dstptr = dst = fz_malloc_array(ctx, (srclen - 2) / 2 + 1, sizeof(short));
for (i = 2; i + 1 < srclen; i += 2)
*dstptr++ = srcptr[i] | srcptr[i+1] << 8;
}
else
{
dstptr = dst = fz_malloc_array(ctx, srclen + 1, sizeof(short));
for (i = 0; i < srclen; i++)
*dstptr++ = pdf_doc_encoding[srcptr[i]];
}
*dstptr = '\0';
return dst;
}
/* (buffer must be at least (fz_to_str_len(src) + 1) * 2 bytes in size) */
void
pdf_to_ucs2_buf(unsigned short *buffer, pdf_obj *src)
{
unsigned char *srcptr = (unsigned char *) pdf_to_str_buf(src);
unsigned short *dstptr = buffer;
int srclen = pdf_to_str_len(src);
int i;
if (srclen >= 2 && srcptr[0] == 254 && srcptr[1] == 255)
{
for (i = 2; i + 1 < srclen; i += 2)
*dstptr++ = srcptr[i] << 8 | srcptr[i+1];
}
else if (srclen >= 2 && srcptr[0] == 255 && srcptr[1] == 254)
{
for (i = 2; i + 1 < srclen; i += 2)
*dstptr++ = srcptr[i] | srcptr[i+1] << 8;
}
else
{
for (i = 0; i < srclen; i++)
*dstptr++ = pdf_doc_encoding[srcptr[i]];
}
*dstptr = '\0';
}
/* Convert Unicode/PdfDocEncoding string into utf-8 */
char *
pdf_to_utf8(fz_context *ctx, pdf_obj *src)
{
unsigned char *srcptr = (unsigned char *) pdf_to_str_buf(src);
char *dstptr, *dst;
int srclen = pdf_to_str_len(src);
int dstlen = 0;
int ucs;
int i;
if (srclen >= 2 && srcptr[0] == 254 && srcptr[1] == 255)
{
for (i = 2; i + 1 < srclen; i += 2)
{
ucs = srcptr[i] << 8 | srcptr[i+1];
dstlen += fz_runelen(ucs);
}
dstptr = dst = fz_malloc(ctx, dstlen + 1);
for (i = 2; i + 1 < srclen; i += 2)
{
ucs = srcptr[i] << 8 | srcptr[i+1];
dstptr += fz_runetochar(dstptr, ucs);
}
}
else if (srclen >= 2 && srcptr[0] == 255 && srcptr[1] == 254)
{
for (i = 2; i + 1 < srclen; i += 2)
{
ucs = srcptr[i] | srcptr[i+1] << 8;
dstlen += fz_runelen(ucs);
}
dstptr = dst = fz_malloc(ctx, dstlen + 1);
for (i = 2; i + 1 < srclen; i += 2)
{
ucs = srcptr[i] | srcptr[i+1] << 8;
dstptr += fz_runetochar(dstptr, ucs);
}
}
else
{
for (i = 0; i < srclen; i++)
dstlen += fz_runelen(pdf_doc_encoding[srcptr[i]]);
dstptr = dst = fz_malloc(ctx, dstlen + 1);
for (i = 0; i < srclen; i++)
{
ucs = pdf_doc_encoding[srcptr[i]];
dstptr += fz_runetochar(dstptr, ucs);
}
}
*dstptr = '\0';
return dst;
}
static fz_colorspace *
load_indexed(pdf_document *doc, pdf_obj *array)
{
fz_context *ctx = doc->ctx;
pdf_obj *baseobj = pdf_array_get(array, 1);
pdf_obj *highobj = pdf_array_get(array, 2);
pdf_obj *lookupobj = pdf_array_get(array, 3);
fz_colorspace *base = NULL;
fz_colorspace *cs;
int i, n, high;
unsigned char *lookup = NULL;
fz_var(base);
fz_var(lookup);
fz_try(ctx)
{
base = pdf_load_colorspace(doc, baseobj);
high = pdf_to_int(highobj);
high = fz_clampi(high, 0, 255);
n = base->n * (high + 1);
lookup = fz_malloc_array(ctx, 1, n);
if (pdf_is_string(lookupobj) && pdf_to_str_len(lookupobj) >= n)
{
unsigned char *buf = (unsigned char *) pdf_to_str_buf(lookupobj);
for (i = 0; i < n; i++)
lookup[i] = buf[i];
}
else if (pdf_is_indirect(lookupobj))
{
fz_stream *file = NULL;
fz_var(file);
fz_try(ctx)
{
file = pdf_open_stream(doc, pdf_to_num(lookupobj), pdf_to_gen(lookupobj));
i = fz_read(file, lookup, n);
if (i < n)
memset(lookup+i, 0, n-i);
}
fz_always(ctx)
{
fz_close(file);
}
fz_catch(ctx)
{
fz_rethrow_message(ctx, "cannot open colorspace lookup table (%d 0 R)", pdf_to_num(lookupobj));
}
}
else
{
fz_rethrow_message(ctx, "cannot parse colorspace lookup table");
}
cs = fz_new_indexed_colorspace(ctx, base, high, lookup);
}
static void
safe_print_pdf_obj(fz_context *ctx, pdf_obj *obj, const char *dflt)
{
if (obj == NULL)
fprintf(stderr, "%s", dflt);
else if (pdf_is_string(ctx, obj))
safe_print_pdf_string(ctx, (unsigned char *)pdf_to_str_buf(ctx, obj), pdf_to_str_len(ctx, obj));
else
pdf_print_obj(ctx, fz_stderr(ctx), obj, 1);
}
static wchar_t *get_prop_str(fz_context* ctx, pdf_obj *obj, const char *name)
{
pdf_obj* p_obj = pdf_dict_getp(ctx, obj, name);
int len = pdf_to_str_len(ctx, p_obj) + 1;
wchar_t *buf = AllocArray<wchar_t>(len);
pdf_to_ucs2_buf(ctx, (unsigned short *)buf, p_obj);
return p_obj ? pdf_clean_string(buf) : nullptr;
}
static char *get_string_or_stream(pdf_document *doc, pdf_obj *obj)
{
fz_context *ctx = doc->ctx;
int len = 0;
char *buf = NULL;
fz_buffer *strmbuf = NULL;
char *text = NULL;
fz_var(strmbuf);
fz_var(text);
fz_try(ctx)
{
if (pdf_is_string(obj))
{
len = pdf_to_str_len(obj);
buf = pdf_to_str_buf(obj);
}
else if (pdf_is_stream(doc, pdf_to_num(obj), pdf_to_gen(obj)))
{
strmbuf = pdf_load_stream(doc, pdf_to_num(obj), pdf_to_gen(obj));
len = fz_buffer_storage(ctx, strmbuf, (unsigned char **)&buf);
}
if (buf)
{
text = fz_malloc(ctx, len+1);
memcpy(text, buf, len);
text[len] = 0;
}
}
fz_always(ctx)
{
fz_drop_buffer(ctx, strmbuf);
}
fz_catch(ctx)
{
fz_free(ctx, text);
fz_rethrow(ctx);
}
return text;
}
static pdf_font_desc *
load_cid_font(pdf_document *xref, pdf_obj *dict, pdf_obj *encoding, pdf_obj *to_unicode)
{
pdf_obj *widths;
pdf_obj *descriptor;
pdf_font_desc *fontdesc = NULL;
FT_Face face;
int kind;
char collection[256];
char *basefont;
int i, k, fterr;
pdf_obj *obj;
int dw;
fz_context *ctx = xref->ctx;
fz_var(fontdesc);
fz_try(ctx)
{
/* Get font name and CID collection */
basefont = pdf_to_name(pdf_dict_gets(dict, "BaseFont"));
{
pdf_obj *cidinfo;
char tmpstr[64];
int tmplen;
cidinfo = pdf_dict_gets(dict, "CIDSystemInfo");
if (!cidinfo)
fz_throw(ctx, "cid font is missing info");
obj = pdf_dict_gets(cidinfo, "Registry");
tmplen = fz_mini(sizeof tmpstr - 1, pdf_to_str_len(obj));
memcpy(tmpstr, pdf_to_str_buf(obj), tmplen);
tmpstr[tmplen] = '\0';
fz_strlcpy(collection, tmpstr, sizeof collection);
fz_strlcat(collection, "-", sizeof collection);
obj = pdf_dict_gets(cidinfo, "Ordering");
tmplen = fz_mini(sizeof tmpstr - 1, pdf_to_str_len(obj));
memcpy(tmpstr, pdf_to_str_buf(obj), tmplen);
tmpstr[tmplen] = '\0';
fz_strlcat(collection, tmpstr, sizeof collection);
}
/* Load font file */
fontdesc = pdf_new_font_desc(ctx);
descriptor = pdf_dict_gets(dict, "FontDescriptor");
if (!descriptor)
fz_throw(ctx, "syntaxerror: missing font descriptor");
pdf_load_font_descriptor(fontdesc, xref, descriptor, collection, basefont);
face = fontdesc->font->ft_face;
kind = ft_kind(face);
/* Encoding */
if (pdf_is_name(encoding))
{
if (!strcmp(pdf_to_name(encoding), "Identity-H"))
fontdesc->encoding = pdf_new_identity_cmap(ctx, 0, 2);
else if (!strcmp(pdf_to_name(encoding), "Identity-V"))
fontdesc->encoding = pdf_new_identity_cmap(ctx, 1, 2);
else
fontdesc->encoding = pdf_load_system_cmap(ctx, pdf_to_name(encoding));
}
else if (pdf_is_indirect(encoding))
{
fontdesc->encoding = pdf_load_embedded_cmap(xref, encoding);
}
else
{
fz_throw(ctx, "syntaxerror: font missing encoding");
}
fontdesc->size += pdf_cmap_size(ctx, fontdesc->encoding);
pdf_set_font_wmode(ctx, fontdesc, pdf_cmap_wmode(ctx, fontdesc->encoding));
if (kind == TRUETYPE)
{
pdf_obj *cidtogidmap;
cidtogidmap = pdf_dict_gets(dict, "CIDToGIDMap");
if (pdf_is_indirect(cidtogidmap))
{
fz_buffer *buf;
buf = pdf_load_stream(xref, pdf_to_num(cidtogidmap), pdf_to_gen(cidtogidmap));
fontdesc->cid_to_gid_len = (buf->len) / 2;
fontdesc->cid_to_gid = fz_malloc_array(ctx, fontdesc->cid_to_gid_len, sizeof(unsigned short));
fontdesc->size += fontdesc->cid_to_gid_len * sizeof(unsigned short);
for (i = 0; i < fontdesc->cid_to_gid_len; i++)
fontdesc->cid_to_gid[i] = (buf->data[i * 2] << 8) + buf->data[i * 2 + 1];
fz_drop_buffer(ctx, buf);
}
/* if truetype font is external, cidtogidmap should not be identity */
/* so we map from cid to unicode and then map that through the (3 1) */
/* unicode cmap to get a glyph id */
else if (fontdesc->font->ft_substitute)
{
fterr = FT_Select_Charmap(face, ft_encoding_unicode);
if (fterr)
{
fz_throw(ctx, "fonterror: no unicode cmap when emulating CID font: %s", ft_error_string(fterr));
}
if (!strcmp(collection, "Adobe-CNS1"))
fontdesc->to_ttf_cmap = pdf_load_system_cmap(ctx, "Adobe-CNS1-UCS2");
else if (!strcmp(collection, "Adobe-GB1"))
fontdesc->to_ttf_cmap = pdf_load_system_cmap(ctx, "Adobe-GB1-UCS2");
else if (!strcmp(collection, "Adobe-Japan1"))
fontdesc->to_ttf_cmap = pdf_load_system_cmap(ctx, "Adobe-Japan1-UCS2");
else if (!strcmp(collection, "Adobe-Japan2"))
fontdesc->to_ttf_cmap = pdf_load_system_cmap(ctx, "Adobe-Japan2-UCS2");
else if (!strcmp(collection, "Adobe-Korea1"))
fontdesc->to_ttf_cmap = pdf_load_system_cmap(ctx, "Adobe-Korea1-UCS2");
}
}
pdf_load_to_unicode(xref, fontdesc, NULL, collection, to_unicode);
/* Horizontal */
dw = 1000;
obj = pdf_dict_gets(dict, "DW");
if (obj)
dw = pdf_to_int(obj);
pdf_set_default_hmtx(ctx, fontdesc, dw);
widths = pdf_dict_gets(dict, "W");
if (widths)
{
int c0, c1, w, n, m;
n = pdf_array_len(widths);
for (i = 0; i < n; )
{
c0 = pdf_to_int(pdf_array_get(widths, i));
obj = pdf_array_get(widths, i + 1);
if (pdf_is_array(obj))
{
m = pdf_array_len(obj);
for (k = 0; k < m; k++)
{
w = pdf_to_int(pdf_array_get(obj, k));
pdf_add_hmtx(ctx, fontdesc, c0 + k, c0 + k, w);
}
i += 2;
}
else
{
c1 = pdf_to_int(obj);
w = pdf_to_int(pdf_array_get(widths, i + 2));
pdf_add_hmtx(ctx, fontdesc, c0, c1, w);
i += 3;
}
}
}
pdf_end_hmtx(ctx, fontdesc);
/* Vertical */
if (pdf_cmap_wmode(ctx, fontdesc->encoding) == 1)
{
int dw2y = 880;
int dw2w = -1000;
obj = pdf_dict_gets(dict, "DW2");
if (obj)
{
dw2y = pdf_to_int(pdf_array_get(obj, 0));
dw2w = pdf_to_int(pdf_array_get(obj, 1));
}
pdf_set_default_vmtx(ctx, fontdesc, dw2y, dw2w);
widths = pdf_dict_gets(dict, "W2");
if (widths)
{
int c0, c1, w, x, y, n;
n = pdf_array_len(widths);
for (i = 0; i < n; )
{
c0 = pdf_to_int(pdf_array_get(widths, i));
obj = pdf_array_get(widths, i + 1);
if (pdf_is_array(obj))
{
int m = pdf_array_len(obj);
for (k = 0; k * 3 < m; k ++)
{
w = pdf_to_int(pdf_array_get(obj, k * 3 + 0));
x = pdf_to_int(pdf_array_get(obj, k * 3 + 1));
y = pdf_to_int(pdf_array_get(obj, k * 3 + 2));
pdf_add_vmtx(ctx, fontdesc, c0 + k, c0 + k, x, y, w);
}
i += 2;
}
else
{
c1 = pdf_to_int(obj);
w = pdf_to_int(pdf_array_get(widths, i + 2));
x = pdf_to_int(pdf_array_get(widths, i + 3));
y = pdf_to_int(pdf_array_get(widths, i + 4));
pdf_add_vmtx(ctx, fontdesc, c0, c1, x, y, w);
i += 5;
}
}
}
pdf_end_vmtx(ctx, fontdesc);
}
}
fz_catch(ctx)
{
pdf_drop_font(ctx, fontdesc);
fz_throw(ctx, "cannot load cid font (%d %d R)", pdf_to_num(dict), pdf_to_gen(dict));
}
return fontdesc;
}
static fz_colorspace *
load_indexed(pdf_document *xref, pdf_obj *array)
{
struct indexed *idx = NULL;
fz_context *ctx = xref->ctx;
pdf_obj *baseobj = pdf_array_get(array, 1);
pdf_obj *highobj = pdf_array_get(array, 2);
pdf_obj *lookup = pdf_array_get(array, 3);
fz_colorspace *base = NULL;
fz_colorspace *cs = NULL;
int i, n;
fz_var(idx);
fz_var(base);
fz_var(cs);
fz_try(ctx)
{
base = pdf_load_colorspace(xref, baseobj);
/* "cannot load base colorspace (%d %d R)", pdf_to_num(baseobj), pdf_to_gen(baseobj) */
idx = fz_malloc_struct(ctx, struct indexed);
idx->lookup = NULL;
idx->base = base;
idx->high = pdf_to_int(highobj);
idx->high = CLAMP(idx->high, 0, 255);
n = base->n * (idx->high + 1);
idx->lookup = fz_malloc_array(ctx, 1, n);
cs = fz_new_colorspace(ctx, "Indexed", 1);
cs->to_rgb = indexed_to_rgb;
cs->free_data = free_indexed;
cs->data = idx;
cs->size += sizeof(*idx) + n + (base ? base->size : 0);
if (pdf_is_string(lookup) && pdf_to_str_len(lookup) == n)
{
unsigned char *buf = (unsigned char *) pdf_to_str_buf(lookup);
for (i = 0; i < n; i++)
idx->lookup[i] = buf[i];
}
else if (pdf_is_indirect(lookup))
{
fz_stream *file = NULL;
fz_try(ctx)
{
file = pdf_open_stream(xref, pdf_to_num(lookup), pdf_to_gen(lookup));
}
fz_catch(ctx)
{
fz_throw(ctx, "cannot open colorspace lookup table (%d 0 R)", pdf_to_num(lookup));
}
i = fz_read(file, idx->lookup, n);
if (i < 0)
{
fz_close(file);
fz_throw(ctx, "cannot read colorspace lookup table (%d 0 R)", pdf_to_num(lookup));
}
fz_close(file);
}
else
{
fz_throw(ctx, "cannot parse colorspace lookup table");
}
}
/* Convert Unicode/PdfDocEncoding string into utf-8 */
char *
pdf_to_utf8(pdf_document *doc, pdf_obj *src)
{
fz_context *ctx = doc->ctx;
fz_buffer *strmbuf = NULL;
unsigned char *srcptr;
char *dstptr, *dst;
int srclen;
int dstlen = 0;
int ucs;
int i;
fz_var(strmbuf);
fz_try(ctx)
{
if (pdf_is_string(src))
{
srcptr = (unsigned char *) pdf_to_str_buf(src);
srclen = pdf_to_str_len(src);
}
else if (pdf_is_stream(doc, pdf_to_num(src), pdf_to_gen(src)))
{
strmbuf = pdf_load_stream(doc, pdf_to_num(src), pdf_to_gen(src));
srclen = fz_buffer_storage(ctx, strmbuf, (unsigned char **)&srcptr);
}
else
{
srclen = 0;
}
if (srclen >= 2 && srcptr[0] == 254 && srcptr[1] == 255)
{
for (i = 2; i + 1 < srclen; i += 2)
{
ucs = srcptr[i] << 8 | srcptr[i+1];
dstlen += fz_runelen(ucs);
}
dstptr = dst = fz_malloc(ctx, dstlen + 1);
for (i = 2; i + 1 < srclen; i += 2)
{
ucs = srcptr[i] << 8 | srcptr[i+1];
dstptr += fz_runetochar(dstptr, ucs);
}
}
else if (srclen >= 2 && srcptr[0] == 255 && srcptr[1] == 254)
{
for (i = 2; i + 1 < srclen; i += 2)
{
ucs = srcptr[i] | srcptr[i+1] << 8;
dstlen += fz_runelen(ucs);
}
dstptr = dst = fz_malloc(ctx, dstlen + 1);
for (i = 2; i + 1 < srclen; i += 2)
{
ucs = srcptr[i] | srcptr[i+1] << 8;
dstptr += fz_runetochar(dstptr, ucs);
}
}
else
{
for (i = 0; i < srclen; i++)
dstlen += fz_runelen(pdf_doc_encoding[srcptr[i]]);
dstptr = dst = fz_malloc(ctx, dstlen + 1);
for (i = 0; i < srclen; i++)
{
ucs = pdf_replace_undefined(pdf_doc_encoding[srcptr[i]]);
dstptr += fz_runetochar(dstptr, ucs);
}
}
}
fz_always(ctx)
{
fz_drop_buffer(ctx, strmbuf);
}
fz_catch(ctx)
{
fz_rethrow(ctx);
}
*dstptr = '\0';
return dst;
}
/* Convert Unicode/PdfDocEncoding string into utf-8 */
char *
pdf_to_utf8(fz_context *ctx, pdf_document *doc, pdf_obj *src)
{
fz_buffer *stmbuf = NULL;
unsigned char *srcptr;
char *dstptr, *dst;
size_t srclen;
size_t dstlen = 0;
int ucs;
size_t i;
fz_var(stmbuf);
fz_try(ctx)
{
if (pdf_is_string(ctx, src))
{
srcptr = (unsigned char *) pdf_to_str_buf(ctx, src);
srclen = pdf_to_str_len(ctx, src);
}
else if (pdf_is_stream(ctx, src))
{
stmbuf = pdf_load_stream(ctx, doc, pdf_to_num(ctx, src), pdf_to_gen(ctx, src));
srclen = fz_buffer_storage(ctx, stmbuf, (unsigned char **)&srcptr);
}
else
{
srclen = 0;
}
/* UTF-16BE */
if (srclen >= 2 && srcptr[0] == 254 && srcptr[1] == 255)
{
i = 2;
while (i + 2 <= srclen)
{
/* skip language escape codes */
if (i + 6 <= srclen &&
srcptr[i+0] == 0 && srcptr[i+1] == 27 &&
srcptr[i+4] == 0 && srcptr[i+5] == 27)
{
i += 6;
}
else if (i + 8 <= srclen &&
srcptr[i+0] == 0 && srcptr[i+1] == 27 &&
srcptr[i+6] == 0 && srcptr[i+7] == 27)
{
i += 8;
}
else
{
i += rune_from_utf16be(&ucs, srcptr + i, srcptr + srclen);
dstlen += fz_runelen(ucs);
}
}
dstptr = dst = fz_malloc(ctx, dstlen + 1);
i = 2;
while (i + 2 <= srclen)
{
/* skip language escape codes */
if (i + 6 <= srclen &&
srcptr[i+0] == 0 && srcptr[i+1] == 27 &&
srcptr[i+4] == 0 && srcptr[i+5] == 27)
{
i += 6;
}
else if (i + 8 <= srclen &&
srcptr[i+0] == 0 && srcptr[i+1] == 27 &&
srcptr[i+6] == 0 && srcptr[i+7] == 27)
{
i += 8;
}
else
{
i += rune_from_utf16be(&ucs, srcptr + i, srcptr + srclen);
dstptr += fz_runetochar(dstptr, ucs);
}
}
}
/* PDFDocEncoding */
else
{
for (i = 0; i < srclen; i++)
dstlen += fz_runelen(pdf_doc_encoding[srcptr[i]]);
dstptr = dst = fz_malloc(ctx, dstlen + 1);
for (i = 0; i < srclen; i++)
{
ucs = pdf_doc_encoding[srcptr[i]];
dstptr += fz_runetochar(dstptr, ucs);
}
}
}
fz_always(ctx)
{
fz_drop_buffer(ctx, stmbuf);
}
fz_catch(ctx)
{
fz_rethrow(ctx);
}
*dstptr = '\0';
return dst;
}
static pdf_font_desc *
load_cid_font(pdf_document *doc, pdf_obj *dict, pdf_obj *encoding, pdf_obj *to_unicode)
{
pdf_obj *widths;
pdf_obj *descriptor;
pdf_font_desc *fontdesc = NULL;
FT_Face face;
int kind;
char collection[256];
char *basefont;
int i, k, fterr;
pdf_obj *obj;
int dw;
fz_context *ctx = doc->ctx;
fz_var(fontdesc);
fz_try(ctx)
{
/* Get font name and CID collection */
basefont = pdf_to_name(pdf_dict_gets(dict, "BaseFont"));
{
pdf_obj *cidinfo;
char tmpstr[64];
int tmplen;
cidinfo = pdf_dict_gets(dict, "CIDSystemInfo");
if (!cidinfo)
fz_throw(ctx, FZ_ERROR_GENERIC, "cid font is missing info");
obj = pdf_dict_gets(cidinfo, "Registry");
tmplen = fz_mini(sizeof tmpstr - 1, pdf_to_str_len(obj));
memcpy(tmpstr, pdf_to_str_buf(obj), tmplen);
tmpstr[tmplen] = '\0';
fz_strlcpy(collection, tmpstr, sizeof collection);
fz_strlcat(collection, "-", sizeof collection);
obj = pdf_dict_gets(cidinfo, "Ordering");
tmplen = fz_mini(sizeof tmpstr - 1, pdf_to_str_len(obj));
memcpy(tmpstr, pdf_to_str_buf(obj), tmplen);
tmpstr[tmplen] = '\0';
fz_strlcat(collection, tmpstr, sizeof collection);
}
/* Load font file */
fontdesc = pdf_new_font_desc(ctx);
descriptor = pdf_dict_gets(dict, "FontDescriptor");
if (!descriptor)
fz_throw(ctx, FZ_ERROR_GENERIC, "syntaxerror: missing font descriptor");
pdf_load_font_descriptor(fontdesc, doc, descriptor, collection, basefont, 1, 1);
face = fontdesc->font->ft_face;
kind = ft_kind(face);
/* Encoding */
if (pdf_is_name(encoding))
{
if (!strcmp(pdf_to_name(encoding), "Identity-H"))
fontdesc->encoding = pdf_new_identity_cmap(ctx, 0, 2);
else if (!strcmp(pdf_to_name(encoding), "Identity-V"))
fontdesc->encoding = pdf_new_identity_cmap(ctx, 1, 2);
else
fontdesc->encoding = pdf_load_system_cmap(ctx, pdf_to_name(encoding));
}
else if (pdf_is_indirect(encoding))
{
fontdesc->encoding = pdf_load_embedded_cmap(doc, encoding);
}
else
{
fz_throw(ctx, FZ_ERROR_GENERIC, "syntaxerror: font missing encoding");
}
fontdesc->size += pdf_cmap_size(ctx, fontdesc->encoding);
pdf_set_font_wmode(ctx, fontdesc, pdf_cmap_wmode(ctx, fontdesc->encoding));
if (kind == TRUETYPE ||
/* cf. http://code.google.com/p/sumatrapdf/issues/detail?id=1565 */
!strcmp(pdf_to_name(pdf_dict_gets(dict, "Subtype")), "CIDFontType2") ||
/* cf. http://code.google.com/p/sumatrapdf/issues/detail?id=1997 */
pdf_is_indirect(pdf_dict_gets(dict, "CIDToGIDMap")))
{
pdf_obj *cidtogidmap;
cidtogidmap = pdf_dict_gets(dict, "CIDToGIDMap");
if (pdf_is_indirect(cidtogidmap))
{
fz_buffer *buf;
buf = pdf_load_stream(doc, pdf_to_num(cidtogidmap), pdf_to_gen(cidtogidmap));
fontdesc->cid_to_gid_len = (buf->len) / 2;
fontdesc->cid_to_gid = fz_malloc_array(ctx, fontdesc->cid_to_gid_len, sizeof(unsigned short));
fontdesc->size += fontdesc->cid_to_gid_len * sizeof(unsigned short);
for (i = 0; i < fontdesc->cid_to_gid_len; i++)
fontdesc->cid_to_gid[i] = (buf->data[i * 2] << 8) + buf->data[i * 2 + 1];
fz_drop_buffer(ctx, buf);
}
/* if truetype font is external, cidtogidmap should not be identity */
/* so we map from cid to unicode and then map that through the (3 1) */
/* unicode cmap to get a glyph id */
else if (fontdesc->font->ft_substitute)
{
fterr = FT_Select_Charmap(face, ft_encoding_unicode);
if (fterr)
{
fz_throw(ctx, FZ_ERROR_GENERIC, "fonterror: no unicode cmap when emulating CID font: %s", ft_error_string(fterr));
}
if (!strcmp(collection, "Adobe-CNS1"))
fontdesc->to_ttf_cmap = pdf_load_system_cmap(ctx, "Adobe-CNS1-UCS2");
else if (!strcmp(collection, "Adobe-GB1"))
fontdesc->to_ttf_cmap = pdf_load_system_cmap(ctx, "Adobe-GB1-UCS2");
else if (!strcmp(collection, "Adobe-Japan1"))
fontdesc->to_ttf_cmap = pdf_load_system_cmap(ctx, "Adobe-Japan1-UCS2");
else if (!strcmp(collection, "Adobe-Japan2"))
fontdesc->to_ttf_cmap = pdf_load_system_cmap(ctx, "Adobe-Japan2-UCS2");
else if (!strcmp(collection, "Adobe-Korea1"))
fontdesc->to_ttf_cmap = pdf_load_system_cmap(ctx, "Adobe-Korea1-UCS2");
/* cf. http://code.google.com/p/sumatrapdf/issues/detail?id=2318 */
else if (!strcmp(collection, "Adobe-Identity") && fontdesc->font->ft_file)
fontdesc->font->ft_substitute = 0;
}
}
/* cf. http://code.google.com/p/sumatrapdf/issues/detail?id=1961 */
fz_try(ctx)
{
pdf_load_to_unicode(doc, fontdesc, NULL, collection, to_unicode);
}
fz_catch(ctx)
{
fz_warn(ctx, "cannot load ToUnicode CMap");
}
/* If we have an identity encoding, we're supposed to use the glyph ids directly.
* If we only have a substitute font, that won't work.
* Make a last ditch attempt by using
* the ToUnicode table if it exists to map via the substitute font's cmap. */
if (strstr(fontdesc->encoding->cmap_name, "Identity-") && fontdesc->font->ft_substitute)
{
fz_warn(ctx, "non-embedded font using identity encoding: %s", basefont);
if (fontdesc->to_unicode && !fontdesc->to_ttf_cmap)
fontdesc->to_ttf_cmap = pdf_keep_cmap(ctx, fontdesc->to_unicode);
}
/* Horizontal */
dw = 1000;
obj = pdf_dict_gets(dict, "DW");
if (obj)
dw = pdf_to_int(obj);
pdf_set_default_hmtx(ctx, fontdesc, dw);
widths = pdf_dict_gets(dict, "W");
if (widths)
{
int c0, c1, w, n, m;
n = pdf_array_len(widths);
for (i = 0; i < n; )
{
c0 = pdf_to_int(pdf_array_get(widths, i));
obj = pdf_array_get(widths, i + 1);
if (pdf_is_array(obj))
{
m = pdf_array_len(obj);
for (k = 0; k < m; k++)
{
w = pdf_to_int(pdf_array_get(obj, k));
pdf_add_hmtx(ctx, fontdesc, c0 + k, c0 + k, w);
}
i += 2;
}
else
{
c1 = pdf_to_int(obj);
w = pdf_to_int(pdf_array_get(widths, i + 2));
pdf_add_hmtx(ctx, fontdesc, c0, c1, w);
i += 3;
}
}
}
pdf_end_hmtx(ctx, fontdesc);
/* Vertical */
if (pdf_cmap_wmode(ctx, fontdesc->encoding) == 1)
{
int dw2y = 880;
int dw2w = -1000;
obj = pdf_dict_gets(dict, "DW2");
if (obj)
{
dw2y = pdf_to_int(pdf_array_get(obj, 0));
dw2w = pdf_to_int(pdf_array_get(obj, 1));
}
pdf_set_default_vmtx(ctx, fontdesc, dw2y, dw2w);
widths = pdf_dict_gets(dict, "W2");
if (widths)
{
int c0, c1, w, x, y, n;
n = pdf_array_len(widths);
for (i = 0; i < n; )
{
c0 = pdf_to_int(pdf_array_get(widths, i));
obj = pdf_array_get(widths, i + 1);
if (pdf_is_array(obj))
{
int m = pdf_array_len(obj);
for (k = 0; k * 3 < m; k ++)
{
w = pdf_to_int(pdf_array_get(obj, k * 3 + 0));
x = pdf_to_int(pdf_array_get(obj, k * 3 + 1));
y = pdf_to_int(pdf_array_get(obj, k * 3 + 2));
pdf_add_vmtx(ctx, fontdesc, c0 + k, c0 + k, x, y, w);
}
i += 2;
}
else
{
c1 = pdf_to_int(obj);
w = pdf_to_int(pdf_array_get(widths, i + 2));
x = pdf_to_int(pdf_array_get(widths, i + 3));
y = pdf_to_int(pdf_array_get(widths, i + 4));
pdf_add_vmtx(ctx, fontdesc, c0, c1, x, y, w);
i += 5;
}
}
}
pdf_end_vmtx(ctx, fontdesc);
}
}
fz_catch(ctx)
{
pdf_drop_font(ctx, fontdesc);
fz_rethrow_message(ctx, "cannot load cid font (%d %d R)", pdf_to_num(dict), pdf_to_gen(dict));
}
return fontdesc;
}
