static tiff_document *
tiff_open_document_with_stream(fz_context *ctx, fz_stream *file)
{
tiff_document *doc;
doc = fz_new_document(ctx, tiff_document);
doc->super.drop_document = (fz_document_drop_fn *)tiff_drop_document;
doc->super.count_pages = (fz_document_count_pages_fn *)tiff_count_pages;
doc->super.load_page = (fz_document_load_page_fn *)tiff_load_page;
doc->super.lookup_metadata = (fz_document_lookup_metadata_fn *)tiff_lookup_metadata;
fz_try(ctx)
{
size_t len;
unsigned char *data;
doc->buffer = fz_read_all(ctx, file, 1024);
len = fz_buffer_storage(ctx, doc->buffer, &data);
doc->page_count = fz_load_tiff_subimage_count(ctx, data, len);
}
fz_catch(ctx)
{
fz_drop_document(ctx, &doc->super);
fz_rethrow(ctx);
}
return doc;
}
static void
pdfportfolio_extract(int argc, char **argv)
{
int entry;
const char *filename;
fz_buffer *buf;
unsigned char *data;
int len;
FILE *file;
if (fz_optind > argc-2)
usage();
entry = fz_atoi(argv[fz_optind++]);
filename = argv[fz_optind++];
buf = pdf_portfolio_entry(ctx, doc, entry);
len = fz_buffer_storage(ctx, buf, &data);
file = fopen(filename, "wb");
if (file == NULL)
{
fprintf(stderr, "Failed to open '%s' for writing\n", filename);
exit(1);
}
fwrite(data, 1, len, file);
fclose(file);
fz_drop_buffer(ctx, buf);
}
static int stream_deflate(pdf_document *xref,fz_context *ctx,int pageref,int pagegen,
int *length)
{
fz_buffer *strbuf;
int n;
unsigned char *p;
#ifdef HAVE_Z_LIB
char tempfile[512];
gzFile gz;
FILE *f;
int nw;
#endif
strbuf=pdf_load_stream(xref,pageref,pagegen);
n=fz_buffer_storage(ctx,strbuf,&p);
#ifdef HAVE_Z_LIB
wfile_abstmpnam(tempfile);
gz=gzopen(tempfile,"sab7");
gzwrite(gz,p,n);
gzclose(gz);
nw=wfile_size(tempfile);
fz_resize_buffer(ctx,strbuf,nw+1);
fz_buffer_storage(ctx,strbuf,&p);
f=fopen(tempfile,"rb");
if (f==NULL || fread(p,1,nw,f)<nw)
aprintf(ANSI_RED "** wmupdf: Error writing compressed stream to PDF file! **\n" ANSI_NORMAL);
if (f!=NULL)
fclose(f);
remove(tempfile);
p[nw]='\n';
strbuf->len=nw+1;
pdf_update_stream(xref,pageref,strbuf);
fz_drop_buffer(ctx,strbuf);
/*
printf(" After drop, xref->table[%d].stm_buf=%p, refs=%d\n",pageref,xref->table[pageref].stm_buf,
xref->table[pageref].stm_buf->refs);
*/
(*length)=nw;
return(1);
#else
fz_drop_buffer(ctx,strbuf);
(*length)=n;
return(0);
#endif
}
static int add_to_srcpage_stream(pdf_document *xref,fz_context *ctx,int pageref,int pagegen,
pdf_obj *srcdict)
{
fz_buffer *srcbuf;
fz_buffer *dstbuf;
int dstlen;
// printf("@add_to_srcpage_stream()...pageref=%d\n",pageref);
srcbuf=pdf_load_stream(xref,pdf_to_num(srcdict),pdf_to_gen(srcdict));
if (srcbuf==NULL)
{
dstbuf=pdf_load_stream(xref,pageref,pagegen);
if (dstbuf==NULL)
return(0);
dstlen=fz_buffer_storage(ctx,dstbuf,NULL);
fz_drop_buffer(ctx,dstbuf);
return(dstlen);
}
if (!pdf_is_stream(xref,pageref,pagegen))
dstbuf=fz_new_buffer(ctx,16);
else
{
dstbuf=pdf_load_stream(xref,pageref,pagegen);
if (dstbuf==NULL)
dstbuf=fz_new_buffer(ctx,16);
}
/* Concatenate srcbuf to dstbuf: (Will srcbuf->data be allowed?) */
dstlen=fz_buffer_storage(ctx,dstbuf,NULL);
/*
printf(" dstlen before = %d\n",dstlen);
printf(" srclen = %d\n",fz_buffer_storage(ctx,srcbuf,NULL));
printf(" srcptr = %p\n",srcbuf->data);
*/
fz_write_buffer(ctx,dstbuf,srcbuf->data,fz_buffer_storage(ctx,srcbuf,NULL));
dstlen=fz_buffer_storage(ctx,dstbuf,NULL);
// printf(" dstlen after = %d\n",dstlen);
fz_drop_buffer(ctx,srcbuf);
pdf_update_stream(xref,pageref,dstbuf);
fz_drop_buffer(ctx,dstbuf);
return(dstlen);
}
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 tiff_page *
tiff_load_page(fz_context *ctx, tiff_document *doc, int number)
{
fz_pixmap *pixmap = NULL;
fz_image *image = NULL;
tiff_page *page = NULL;
if (number < 0 || number >= doc->page_count)
return NULL;
fz_var(pixmap);
fz_var(image);
fz_var(page);
fz_try(ctx)
{
size_t len;
unsigned char *data;
len = fz_buffer_storage(ctx, doc->buffer, &data);
pixmap = fz_load_tiff_subimage(ctx, data, len, number);
image = fz_new_image_from_pixmap(ctx, pixmap, NULL);
page = fz_new_page(ctx, sizeof *page);
page->super.bound_page = (fz_page_bound_page_fn *)tiff_bound_page;
page->super.run_page_contents = (fz_page_run_page_contents_fn *)tiff_run_page;
page->super.drop_page = (fz_page_drop_page_fn *)tiff_drop_page;
page->image = fz_keep_image(ctx, image);
}
fz_always(ctx)
{
fz_drop_image(ctx, image);
fz_drop_pixmap(ctx, pixmap);
}
fz_catch(ctx)
{
fz_free(ctx, page);
fz_rethrow(ctx);
}
return page;
}
static void
epub_parse_ncx(fz_context *ctx, epub_document *doc, const char *path)
{
fz_archive *zip = doc->zip;
fz_buffer *buf;
fz_xml *ncx;
char base_uri[2048];
unsigned char *data;
size_t len;
fz_dirname(base_uri, path, sizeof base_uri);
buf = fz_read_archive_entry(ctx, zip, path);
fz_write_buffer_byte(ctx, buf, 0);
len = fz_buffer_storage(ctx, buf, &data);
ncx = fz_parse_xml(ctx, data, len, 0);
fz_drop_buffer(ctx, buf);
doc->outline = epub_parse_ncx_imp(ctx, doc, fz_xml_find_down(ncx, "navMap"), base_uri);
fz_drop_xml(ctx, ncx);
}
static fz_document *
svg_open_document_with_buffer(fz_context *ctx, fz_buffer *buf)
{
svg_document *doc;
fz_xml *root;
size_t len;
unsigned char *data;
len = fz_buffer_storage(ctx, buf, &data);
root = fz_parse_xml(ctx, data, len, 0);
doc = fz_new_document(ctx, svg_document);
doc->super.drop_document = svg_drop_document;
doc->super.count_pages = svg_count_pages;
doc->super.load_page = svg_load_page;
doc->root = root;
doc->idmap = NULL;
svg_build_id_map(ctx, doc, root);
return (fz_document*)doc;
}
pdf_font_desc *
pdf_load_type3_font(fz_context *ctx, pdf_document *doc, pdf_obj *rdb, pdf_obj *dict)
{
char buf[256];
char *estrings[256];
pdf_font_desc *fontdesc = NULL;
pdf_obj *encoding;
pdf_obj *widths;
pdf_obj *charprocs;
pdf_obj *obj;
int first, last;
int i, k, n;
fz_rect bbox;
fz_matrix matrix;
fz_font *font;
fz_var(fontdesc);
/* Make a new type3 font entry in the document */
if (doc->num_type3_fonts == doc->max_type3_fonts)
{
int new_max = doc->max_type3_fonts * 2;
if (new_max == 0)
new_max = 4;
doc->type3_fonts = fz_resize_array(ctx, doc->type3_fonts, new_max, sizeof(*doc->type3_fonts));
doc->max_type3_fonts = new_max;
}
fz_try(ctx)
{
obj = pdf_dict_get(ctx, dict, PDF_NAME_Name);
if (pdf_is_name(ctx, obj))
fz_strlcpy(buf, pdf_to_name(ctx, obj), sizeof buf);
else
fz_strlcpy(buf, "Unnamed-T3", sizeof buf);
fontdesc = pdf_new_font_desc(ctx);
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontMatrix);
pdf_to_matrix(ctx, obj, &matrix);
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontBBox);
fz_transform_rect(pdf_to_rect(ctx, obj, &bbox), &matrix);
font = fz_new_type3_font(ctx, buf, &matrix);
fontdesc->font = font;
fontdesc->size += sizeof(fz_font) + 256 * (sizeof(fz_buffer*) + sizeof(float));
fz_set_font_bbox(ctx, font, bbox.x0, bbox.y0, bbox.x1, bbox.y1);
/* Encoding */
for (i = 0; i < 256; i++)
estrings[i] = NULL;
encoding = pdf_dict_get(ctx, dict, PDF_NAME_Encoding);
if (!encoding)
{
fz_throw(ctx, FZ_ERROR_GENERIC, "syntaxerror: Type3 font missing Encoding");
}
if (pdf_is_name(ctx, encoding))
pdf_load_encoding(estrings, pdf_to_name(ctx, encoding));
if (pdf_is_dict(ctx, encoding))
{
pdf_obj *base, *diff, *item;
base = pdf_dict_get(ctx, encoding, PDF_NAME_BaseEncoding);
if (pdf_is_name(ctx, base))
pdf_load_encoding(estrings, pdf_to_name(ctx, base));
diff = pdf_dict_get(ctx, encoding, PDF_NAME_Differences);
if (pdf_is_array(ctx, diff))
{
n = pdf_array_len(ctx, diff);
k = 0;
for (i = 0; i < n; i++)
{
item = pdf_array_get(ctx, diff, i);
if (pdf_is_int(ctx, item))
k = pdf_to_int(ctx, item);
if (pdf_is_name(ctx, item) && k >= 0 && k < nelem(estrings))
estrings[k++] = pdf_to_name(ctx, item);
}
}
}
fontdesc->encoding = pdf_new_identity_cmap(ctx, 0, 1);
fontdesc->size += pdf_cmap_size(ctx, fontdesc->encoding);
pdf_load_to_unicode(ctx, doc, fontdesc, estrings, NULL, pdf_dict_get(ctx, dict, PDF_NAME_ToUnicode));
/* Widths */
pdf_set_default_hmtx(ctx, fontdesc, 0);
first = pdf_to_int(ctx, pdf_dict_get(ctx, dict, PDF_NAME_FirstChar));
last = pdf_to_int(ctx, pdf_dict_get(ctx, dict, PDF_NAME_LastChar));
if (first < 0 || last > 255 || first > last)
first = last = 0;
widths = pdf_dict_get(ctx, dict, PDF_NAME_Widths);
if (!widths)
{
fz_throw(ctx, FZ_ERROR_GENERIC, "syntaxerror: Type3 font missing Widths");
}
for (i = first; i <= last; i++)
{
float w = pdf_to_real(ctx, pdf_array_get(ctx, widths, i - first));
w = font->t3matrix.a * w * 1000;
font->t3widths[i] = w * 0.001f;
pdf_add_hmtx(ctx, fontdesc, i, i, w);
}
pdf_end_hmtx(ctx, fontdesc);
/* Resources -- inherit page resources if the font doesn't have its own */
font->t3freeres = pdf_t3_free_resources;
font->t3resources = pdf_dict_get(ctx, dict, PDF_NAME_Resources);
if (!font->t3resources)
font->t3resources = rdb;
if (font->t3resources)
pdf_keep_obj(ctx, font->t3resources);
if (!font->t3resources)
fz_warn(ctx, "no resource dictionary for type 3 font!");
font->t3doc = doc;
font->t3run = pdf_run_glyph_func;
/* CharProcs */
charprocs = pdf_dict_get(ctx, dict, PDF_NAME_CharProcs);
if (!charprocs)
{
fz_throw(ctx, FZ_ERROR_GENERIC, "syntaxerror: Type3 font missing CharProcs");
}
for (i = 0; i < 256; i++)
{
if (estrings[i])
{
obj = pdf_dict_gets(ctx, charprocs, estrings[i]);
if (pdf_is_stream(ctx, obj))
{
font->t3procs[i] = pdf_load_stream(ctx, obj);
fz_trim_buffer(ctx, font->t3procs[i]);
fontdesc->size += fz_buffer_storage(ctx, font->t3procs[i], NULL);
fontdesc->size += 0; // TODO: display list size calculation
}
}
}
}
fz_catch(ctx)
{
pdf_drop_font(ctx, fontdesc);
fz_rethrow(ctx);
}
doc->type3_fonts[doc->num_type3_fonts++] = fz_keep_font(ctx, font);
return fontdesc;
}
/* 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;
}
void
pdf_set_free_text_details(fz_context *ctx, pdf_annot *annot, fz_point *pos, char *text, char *font_name, float font_size, float color[3])
{
pdf_document *doc = annot->page->doc;
char nbuf[32];
pdf_obj *dr;
pdf_obj *form_fonts;
pdf_obj *font = NULL;
pdf_obj *ref;
pdf_font_desc *font_desc = NULL;
fz_matrix page_ctm, inv_page_ctm;
pdf_da_info da_info;
fz_buffer *fzbuf = NULL;
fz_point page_pos;
pdf_page_transform(ctx, annot->page, NULL, &page_ctm);
fz_invert_matrix(&inv_page_ctm, &page_ctm);
dr = pdf_dict_get(ctx, annot->page->obj, PDF_NAME_Resources);
if (!dr)
{
dr = pdf_new_dict(ctx, doc, 1);
pdf_dict_put_drop(ctx, annot->page->obj, PDF_NAME_Resources, dr);
}
/* Ensure the resource dictionary includes a font dict */
form_fonts = pdf_dict_get(ctx, dr, PDF_NAME_Font);
if (!form_fonts)
{
form_fonts = pdf_new_dict(ctx, doc, 1);
pdf_dict_put_drop(ctx, dr, PDF_NAME_Font, form_fonts);
/* form_fonts is still valid if execution continues past the above call */
}
fz_var(fzbuf);
fz_var(font);
fz_try(ctx)
{
unsigned char *da_str;
size_t da_len;
fz_rect bounds;
find_free_font_name(ctx, form_fonts, nbuf, sizeof(nbuf));
font = pdf_new_dict(ctx, doc, 5);
ref = pdf_add_object(ctx, doc, font);
pdf_dict_puts_drop(ctx, form_fonts, nbuf, ref);
pdf_dict_put_drop(ctx, font, PDF_NAME_Type, PDF_NAME_Font);
pdf_dict_put_drop(ctx, font, PDF_NAME_Subtype, PDF_NAME_Type1);
pdf_dict_put_drop(ctx, font, PDF_NAME_BaseFont, pdf_new_name(ctx, doc, font_name));
pdf_dict_put_drop(ctx, font, PDF_NAME_Encoding, PDF_NAME_WinAnsiEncoding);
memcpy(da_info.col, color, sizeof(float)*3);
da_info.col_size = 3;
da_info.font_name = nbuf;
da_info.font_size = font_size;
fzbuf = fz_new_buffer(ctx, 0);
pdf_fzbuf_print_da(ctx, fzbuf, &da_info);
da_len = fz_buffer_storage(ctx, fzbuf, &da_str);
pdf_dict_put_drop(ctx, annot->obj, PDF_NAME_DA, pdf_new_string(ctx, doc, (char *)da_str, da_len));
/* FIXME: should convert to WinAnsiEncoding */
pdf_dict_put_drop(ctx, annot->obj, PDF_NAME_Contents, pdf_new_string(ctx, doc, text, strlen(text)));
font_desc = pdf_load_font(ctx, doc, NULL, font, 0);
pdf_measure_text(ctx, font_desc, (unsigned char *)text, strlen(text), &bounds);
page_pos = *pos;
fz_transform_point(&page_pos, &inv_page_ctm);
bounds.x0 *= font_size;
bounds.x1 *= font_size;
bounds.y0 *= font_size;
bounds.y1 *= font_size;
bounds.x0 += page_pos.x;
bounds.x1 += page_pos.x;
bounds.y0 += page_pos.y;
bounds.y1 += page_pos.y;
pdf_dict_put_drop(ctx, annot->obj, PDF_NAME_Rect, pdf_new_rect(ctx, doc, &bounds));
}
fz_always(ctx)
{
pdf_drop_obj(ctx, font);
fz_drop_buffer(ctx, fzbuf);
pdf_drop_font(ctx, font_desc);
}
fz_catch(ctx)
{
fz_rethrow(ctx);
}
}
/* 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 void
epub_parse_header(fz_context *ctx, epub_document *doc)
{
fz_archive *zip = doc->zip;
fz_buffer *buf;
fz_xml *container_xml, *content_opf;
fz_xml *container, *rootfiles, *rootfile;
fz_xml *package, *manifest, *spine, *itemref, *metadata;
char base_uri[2048];
const char *full_path;
const char *version;
char ncx[2048], s[2048];
epub_chapter **tailp;
size_t len;
unsigned char *data;
if (fz_has_archive_entry(ctx, zip, "META-INF/rights.xml"))
fz_throw(ctx, FZ_ERROR_GENERIC, "EPUB is locked by DRM");
if (fz_has_archive_entry(ctx, zip, "META-INF/encryption.xml"))
fz_throw(ctx, FZ_ERROR_GENERIC, "EPUB is locked by DRM");
/* parse META-INF/container.xml to find OPF */
buf = fz_read_archive_entry(ctx, zip, "META-INF/container.xml");
fz_write_buffer_byte(ctx, buf, 0);
len = fz_buffer_storage(ctx, buf, &data);
container_xml = fz_parse_xml(ctx, data, len, 0);
fz_drop_buffer(ctx, buf);
container = fz_xml_find(container_xml, "container");
rootfiles = fz_xml_find_down(container, "rootfiles");
rootfile = fz_xml_find_down(rootfiles, "rootfile");
full_path = fz_xml_att(rootfile, "full-path");
if (!full_path)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find root file in EPUB");
fz_dirname(base_uri, full_path, sizeof base_uri);
/* parse OPF to find NCX and spine */
buf = fz_read_archive_entry(ctx, zip, full_path);
fz_write_buffer_byte(ctx, buf, 0);
len = fz_buffer_storage(ctx, buf, &data);
content_opf = fz_parse_xml(ctx, data, len, 0);
fz_drop_buffer(ctx, buf);
package = fz_xml_find(content_opf, "package");
version = fz_xml_att(package, "version");
if (!version || strcmp(version, "2.0"))
fz_warn(ctx, "unknown epub version: %s", version ? version : "<none>");
metadata = fz_xml_find_down(package, "metadata");
if (metadata)
{
doc->dc_title = find_metadata(ctx, metadata, "title");
doc->dc_creator = find_metadata(ctx, metadata, "creator");
}
manifest = fz_xml_find_down(package, "manifest");
spine = fz_xml_find_down(package, "spine");
if (path_from_idref(ncx, manifest, base_uri, fz_xml_att(spine, "toc"), sizeof ncx))
{
epub_parse_ncx(ctx, doc, ncx);
}
doc->spine = NULL;
tailp = &doc->spine;
itemref = fz_xml_find_down(spine, "itemref");
while (itemref)
{
if (path_from_idref(s, manifest, base_uri, fz_xml_att(itemref, "idref"), sizeof s))
{
*tailp = epub_parse_chapter(ctx, doc, s);
tailp = &(*tailp)->next;
}
itemref = fz_xml_find_next(itemref, "itemref");
}
fz_drop_xml(ctx, container_xml);
fz_drop_xml(ctx, content_opf);
}
int pdf_add_portfolio_entry(fz_context *ctx, pdf_document *doc,
const char *name, int name_len,
const char *desc, int desc_len,
const char *filename, int filename_len,
const char *unifile, int unifile_len, fz_buffer *buf)
{
int entry, len;
pdf_obj *ef, *f, *params, *s;
pdf_obj *key;
pdf_obj *val = NULL;
fz_var(val);
if (!doc)
fz_throw(ctx, FZ_ERROR_GENERIC, "Bad pdf_add_portfolio_entry call");
if (doc->portfolio == NULL)
load_portfolio(ctx, doc);
key = pdf_new_string(ctx, doc, name, name_len);
fz_try(ctx)
{
val = pdf_new_dict(ctx, doc, 6);
pdf_dict_put_drop(ctx, val, PDF_NAME_CI, pdf_new_dict(ctx, doc, 4));
pdf_dict_put_drop(ctx, val, PDF_NAME_EF, (ef = pdf_new_dict(ctx, doc, 4)));
pdf_dict_put_drop(ctx, val, PDF_NAME_F, pdf_new_string(ctx, doc, filename, filename_len));
pdf_dict_put_drop(ctx, val, PDF_NAME_UF, pdf_new_string(ctx, doc, unifile, unifile_len));
pdf_dict_put_drop(ctx, val, PDF_NAME_Desc, pdf_new_string(ctx, doc, desc, desc_len));
pdf_dict_put_drop(ctx, val, PDF_NAME_Type, PDF_NAME_Filespec);
pdf_dict_put_drop(ctx, ef, PDF_NAME_F, (f = pdf_add_stream(ctx, doc, buf, NULL, 0)));
len = fz_buffer_storage(ctx, buf, NULL);
pdf_dict_put_drop(ctx, f, PDF_NAME_DL, pdf_new_int(ctx, doc, len));
pdf_dict_put_drop(ctx, f, PDF_NAME_Length, pdf_new_int(ctx, doc, len));
pdf_dict_put_drop(ctx, f, PDF_NAME_Params, (params = pdf_new_dict(ctx, doc, 4)));
pdf_dict_put_drop(ctx, params, PDF_NAME_Size, pdf_new_int(ctx, doc, len));
s = pdf_dict_getl(ctx, pdf_trailer(ctx, doc), PDF_NAME_Root, PDF_NAME_Collection, NULL);
if (s == NULL)
{
s = pdf_new_dict(ctx, doc, 4);
pdf_dict_putl_drop(ctx, pdf_trailer(ctx, doc), s, PDF_NAME_Root, PDF_NAME_Collection, NULL);
}
s = pdf_dict_getl(ctx, pdf_trailer(ctx, doc), PDF_NAME_Root, PDF_NAME_Names, PDF_NAME_EmbeddedFiles, NULL);
if (s == NULL)
{
s = pdf_new_dict(ctx, doc, 4);
pdf_dict_putl_drop(ctx, pdf_trailer(ctx, doc), s, PDF_NAME_Root, PDF_NAME_Names, PDF_NAME_EmbeddedFiles, NULL);
}
entry = pdf_name_tree_insert(ctx, doc, s, key, val);
}
fz_always(ctx)
{
pdf_drop_obj(ctx, key);
pdf_drop_obj(ctx, val);
}
fz_catch(ctx)
fz_rethrow(ctx);
return entry;
}
void pdf_set_free_text_details(pdf_document *doc, pdf_annot *annot, fz_point *pos, char *text, char *font_name, float font_size, float color[3])
{
fz_context *ctx = doc->ctx;
char nbuf[32];
pdf_obj *dr;
pdf_obj *form_fonts;
pdf_obj *font = NULL;
pdf_obj *ref;
pdf_font_desc *font_desc = NULL;
pdf_da_info da_info;
fz_buffer *fzbuf = NULL;
fz_matrix ctm;
fz_point page_pos;
fz_invert_matrix(&ctm, &annot->page->ctm);
dr = pdf_dict_gets(annot->page->me, "Resources");
if (!dr)
{
dr = pdf_new_dict(doc, 1);
pdf_dict_putp_drop(annot->page->me, "Resources", dr);
}
/* Ensure the resource dictionary includes a font dict */
form_fonts = pdf_dict_gets(dr, "Font");
if (!form_fonts)
{
form_fonts = pdf_new_dict(doc, 1);
pdf_dict_puts_drop(dr, "Font", form_fonts);
/* form_fonts is still valid if execution continues past the above call */
}
fz_var(fzbuf);
fz_var(font);
fz_try(ctx)
{
unsigned char *da_str;
int da_len;
fz_rect bounds;
find_free_font_name(form_fonts, nbuf, sizeof(nbuf));
font = pdf_new_dict(doc, 5);
ref = pdf_new_ref(doc, font);
pdf_dict_puts_drop(form_fonts, nbuf, ref);
pdf_dict_puts_drop(font, "Type", pdf_new_name(doc, "Font"));
pdf_dict_puts_drop(font, "Subtype", pdf_new_name(doc, "Type1"));
pdf_dict_puts_drop(font, "BaseFont", pdf_new_name(doc, font_name));
pdf_dict_puts_drop(font, "Encoding", pdf_new_name(doc, "WinAnsiEncoding"));
memcpy(da_info.col, color, sizeof(float)*3);
da_info.col_size = 3;
da_info.font_name = nbuf;
da_info.font_size = font_size;
fzbuf = fz_new_buffer(ctx, 0);
pdf_fzbuf_print_da(ctx, fzbuf, &da_info);
da_len = fz_buffer_storage(ctx, fzbuf, &da_str);
pdf_dict_puts_drop(annot->obj, "DA", pdf_new_string(doc, (char *)da_str, da_len));
/* FIXME: should convert to WinAnsiEncoding */
pdf_dict_puts_drop(annot->obj, "Contents", pdf_new_string(doc, text, strlen(text)));
font_desc = pdf_load_font(doc, NULL, font, 0);
pdf_measure_text(ctx, font_desc, (unsigned char *)text, strlen(text), &bounds);
page_pos = *pos;
fz_transform_point(&page_pos, &ctm);
bounds.x0 *= font_size;
bounds.x1 *= font_size;
bounds.y0 *= font_size;
bounds.y1 *= font_size;
bounds.x0 += page_pos.x;
bounds.x1 += page_pos.x;
bounds.y0 += page_pos.y;
bounds.y1 += page_pos.y;
pdf_dict_puts_drop(annot->obj, "Rect", pdf_new_rect(doc, &bounds));
update_rect(ctx, annot);
}
fz_always(ctx)
{
pdf_drop_obj(font);
fz_drop_buffer(ctx, fzbuf);
pdf_drop_font(ctx, font_desc);
}
fz_catch(ctx)
{
fz_rethrow(ctx);
}
}
static void savefont(pdf_obj *dict, int num)
{
char namebuf[1024];
char *subtype;
fz_buffer *buf;
pdf_obj *stream = NULL;
pdf_obj *obj;
char *ext = "";
FILE *f;
char *fontname = "font";
int n, len;
unsigned char *data;
obj = pdf_dict_gets(dict, "FontName");
if (obj)
fontname = pdf_to_name(obj);
obj = pdf_dict_gets(dict, "FontFile");
if (obj)
{
stream = obj;
ext = "pfa";
}
obj = pdf_dict_gets(dict, "FontFile2");
if (obj)
{
stream = obj;
ext = "ttf";
}
obj = pdf_dict_gets(dict, "FontFile3");
if (obj)
{
stream = obj;
obj = pdf_dict_gets(obj, "Subtype");
if (obj && !pdf_is_name(obj))
fz_throw(ctx, FZ_ERROR_GENERIC, "invalid font descriptor subtype");
subtype = pdf_to_name(obj);
if (!strcmp(subtype, "Type1C"))
ext = "cff";
else if (!strcmp(subtype, "CIDFontType0C"))
ext = "cid";
else if (!strcmp(subtype, "OpenType"))
ext = "otf";
else
fz_throw(ctx, FZ_ERROR_GENERIC, "unhandled font type '%s'", subtype);
}
if (!stream)
{
fz_warn(ctx, "unhandled font type");
return;
}
buf = pdf_load_stream(doc, pdf_to_num(stream), pdf_to_gen(stream));
snprintf(namebuf, sizeof(namebuf), "%s-%04d.%s", fontname, num, ext);
printf("extracting font %s\n", namebuf);
f = fopen(namebuf, "wb");
if (!f)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot create font file");
len = fz_buffer_storage(ctx, buf, &data);
n = fwrite(data, 1, len, f);
if (n < len)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot write font file");
if (fclose(f) < 0)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot close font file");
fz_drop_buffer(ctx, buf);
}
static void savefont(pdf_obj *dict, int num)
{
char namebuf[1024];
fz_buffer *buf;
pdf_obj *stream = NULL;
pdf_obj *obj;
char *ext = "";
fz_output *out;
char *fontname = "font";
size_t len;
unsigned char *data;
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontName);
if (obj)
fontname = pdf_to_name(ctx, obj);
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontFile);
if (obj)
{
stream = obj;
ext = "pfa";
}
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontFile2);
if (obj)
{
stream = obj;
ext = "ttf";
}
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontFile3);
if (obj)
{
stream = obj;
obj = pdf_dict_get(ctx, obj, PDF_NAME_Subtype);
if (obj && !pdf_is_name(ctx, obj))
fz_throw(ctx, FZ_ERROR_GENERIC, "invalid font descriptor subtype");
if (pdf_name_eq(ctx, obj, PDF_NAME_Type1C))
ext = "cff";
else if (pdf_name_eq(ctx, obj, PDF_NAME_CIDFontType0C))
ext = "cid";
else if (pdf_name_eq(ctx, obj, PDF_NAME_OpenType))
ext = "otf";
else
fz_throw(ctx, FZ_ERROR_GENERIC, "unhandled font type '%s'", pdf_to_name(ctx, obj));
}
if (!stream)
{
fz_warn(ctx, "unhandled font type");
return;
}
buf = pdf_load_stream(ctx, doc, pdf_to_num(ctx, stream));
len = fz_buffer_storage(ctx, buf, &data);
fz_try(ctx)
{
snprintf(namebuf, sizeof(namebuf), "%s-%04d.%s", fontname, num, ext);
printf("extracting font %s\n", namebuf);
out = fz_new_output_with_path(ctx, namebuf, 0);
fz_try(ctx)
fz_write(ctx, out, data, len);
fz_always(ctx)
fz_drop_output(ctx, out);
fz_catch(ctx)
fz_rethrow(ctx);
}
fz_always(ctx)
fz_drop_buffer(ctx, buf);
fz_catch(ctx)
fz_rethrow(ctx);
}
static fz_image *
pdf_load_jpx(fz_context *ctx, pdf_document *doc, pdf_obj *dict, int forcemask)
{
fz_buffer *buf = NULL;
fz_colorspace *colorspace = NULL;
fz_pixmap *pix = NULL;
pdf_obj *obj;
fz_image *mask = NULL;
fz_image *img = NULL;
fz_var(pix);
fz_var(buf);
fz_var(colorspace);
fz_var(mask);
buf = pdf_load_stream(ctx, dict);
/* FIXME: We can't handle decode arrays for indexed images currently */
fz_try(ctx)
{
unsigned char *data;
size_t len;
obj = pdf_dict_get(ctx, dict, PDF_NAME(ColorSpace));
if (obj)
colorspace = pdf_load_colorspace(ctx, obj);
len = fz_buffer_storage(ctx, buf, &data);
pix = fz_load_jpx(ctx, data, len, colorspace);
obj = pdf_dict_geta(ctx, dict, PDF_NAME(SMask), PDF_NAME(Mask));
if (pdf_is_dict(ctx, obj))
{
if (forcemask)
fz_warn(ctx, "Ignoring recursive JPX soft mask");
else
mask = pdf_load_image_imp(ctx, doc, NULL, obj, NULL, 1);
}
obj = pdf_dict_geta(ctx, dict, PDF_NAME(Decode), PDF_NAME(D));
if (obj && !fz_colorspace_is_indexed(ctx, colorspace))
{
float decode[FZ_MAX_COLORS * 2];
int i;
for (i = 0; i < pix->n * 2; i++)
decode[i] = pdf_array_get_real(ctx, obj, i);
fz_decode_tile(ctx, pix, decode);
}
img = fz_new_image_from_pixmap(ctx, pix, mask);
}
fz_always(ctx)
{
fz_drop_image(ctx, mask);
fz_drop_pixmap(ctx, pix);
fz_drop_colorspace(ctx, colorspace);
fz_drop_buffer(ctx, buf);
}
fz_catch(ctx)
{
fz_rethrow(ctx);
}
return img;
}
