fz_document *
fz_open_document_with_stream(fz_context *ctx, char *magic, fz_stream *stream)
{
char *ext = strrchr(magic, '.');
if (ext)
{
if (!fz_strcasecmp(ext, ".xps") || !fz_strcasecmp(ext, ".rels"))
return (fz_document*) xps_open_document_with_stream(ctx, stream);
if (!fz_strcasecmp(ext, ".cbz") || !fz_strcasecmp(ext, ".zip"))
return (fz_document*) cbz_open_document_with_stream(ctx, stream);
if (!fz_strcasecmp(ext, ".pdf"))
return (fz_document*) pdf_open_document_with_stream(ctx, stream);
}
if (!strcmp(magic, "cbz") || !strcmp(magic, "application/x-cbz"))
return (fz_document*) cbz_open_document_with_stream(ctx, stream);
if (!strcmp(magic, "xps") || !strcmp(magic, "application/vnd.ms-xpsdocument"))
return (fz_document*) xps_open_document_with_stream(ctx, stream);
if (!strcmp(magic, "pdf") || !strcmp(magic, "application/pdf"))
return (fz_document*) pdf_open_document_with_stream(ctx, stream);
/* last guess: pdf */
return (fz_document*) pdf_open_document_with_stream(ctx, stream);
}
static int
tiff_recognize(fz_context *doc, const char *magic)
{
char *ext = strrchr(magic, '.');
if (ext)
{
if (!fz_strcasecmp(ext, ".tiff") || !fz_strcasecmp(ext, ".tif"))
return 100;
#ifdef _TINSPIRE
if (!fz_strcasecmp(ext, ".tns"))
{
while (--ext >= magic)
{
if (*ext == '.')
break;
}
if (ext >= magic && (!fz_strcasecmp(ext, ".tiff.tns") || !fz_strcasecmp(ext, ".tif.tns")))
return 100;
}
#endif
}
if (!strcmp(magic, "tif") || !strcmp(magic, "image/tiff") ||
!strcmp(magic, "tiff") || !strcmp(magic, "image/x-tiff"))
return 100;
return 0;
}
static int
cbz_recognize(fz_context *doc, const char *magic)
{
char *ext = strrchr(magic, '.');
if (ext)
{
if (!fz_strcasecmp(ext, ".cbz") || !fz_strcasecmp(ext, ".zip"))
return 100;
}
if (!strcmp(magic, "cbz") || !strcmp(magic, "application/x-cbz"))
return 100;
return 0;
}
static int
htdoc_recognize(fz_context *doc, const char *magic)
{
char *ext = strrchr(magic, '.');
if (ext)
{
if (!fz_strcasecmp(ext, ".xml") || !fz_strcasecmp(ext, ".xhtml") || !fz_strcasecmp(ext, ".html"))
return 100;
}
if (!strcmp(magic, "application/html+xml") || !strcmp(magic, "application/xml") || !strcmp(magic, "text/xml"))
return 100;
return 0;
}
static void
cbz_create_page_list(fz_context *ctx, cbz_document *doc)
{
fz_archive *arch = doc->arch;
int i, k, count;
count = fz_count_archive_entries(ctx, arch);
doc->page_count = 0;
doc->page = fz_malloc_array(ctx, count, sizeof *doc->page);
for (i = 0; i < count; i++)
{
const char *name = fz_list_archive_entry(ctx, arch, i);
const char *ext = name ? strrchr(name, '.') : NULL;
for (k = 0; cbz_ext_list[k]; k++)
{
if (ext && !fz_strcasecmp(ext, cbz_ext_list[k]))
{
doc->page[doc->page_count++] = name;
break;
}
}
}
qsort((char **)doc->page, doc->page_count, sizeof *doc->page, cbz_compare_page_names);
}
static int
tiff_recognize(fz_context *doc, const char *magic)
{
char *ext = strrchr(magic, '.');
if (ext)
{
if (!fz_strcasecmp(ext, ".tiff") || !fz_strcasecmp(ext, ".tif"))
return 100;
}
if (!strcmp(magic, "tif") || !strcmp(magic, "image/tiff") ||
!strcmp(magic, "tiff") || !strcmp(magic, "image/x-tiff"))
return 100;
return 0;
}
fz_document *
fz_open_document(fz_context *ctx, char *filename)
{
char *ext = strrchr(filename, '.');
if (ext)
{
if (!fz_strcasecmp(ext, ".xps") || !fz_strcasecmp(ext, ".rels"))
return (fz_document*) xps_open_document(ctx, filename);
if (!fz_strcasecmp(ext, ".cbz") || !fz_strcasecmp(ext, ".zip"))
return (fz_document*) cbz_open_document(ctx, filename);
if (!fz_strcasecmp(ext, ".pdf"))
return (fz_document*) pdf_open_document(ctx, filename);
}
/* last guess: pdf */
return (fz_document*) pdf_open_document(ctx, filename);
}
static int
svg_recognize(fz_context *ctx, const char *magic)
{
char *ext = strrchr(magic, '.');
if (ext && !fz_strcasecmp(ext, ".svg"))
return 100;
if (!strcmp(magic, "svg") || !strcmp(magic, "image/svg+xml"))
return 100;
return 0;
}
fz_document *
fz_open_document(fz_context *ctx, char *filename)
{
char *ext = strrchr(filename, '.');
if (ext && (!fz_strcasecmp(ext, ".xps") || !fz_strcasecmp(ext, ".rels")))
return (fz_document*) xps_open_document(ctx, filename);
if (ext && !fz_strcasecmp(ext, ".cbz"))
return (fz_document*) cbz_open_document(ctx, filename);
#if 0
/* We used to only open pdf files if they ended in .pdf. For now,
* until we move to detecting filetypes by their content, we disable
* this code, and assume that any file that hasn't matched an
* extension already, is a PDF. */
if (ext && !fz_strcasecmp(ext, ".pdf"))
return (fz_document*) pdf_open_document(ctx, filename);
fz_throw(ctx, "unknown document type: '%s'", filename);
return NULL;
#else
return (fz_document*) pdf_open_document(ctx, filename);
#endif
}
static int
epub_recognize(fz_context *doc, const char *magic)
{
char *ext = strrchr(magic, '.');
if (ext)
if (!fz_strcasecmp(ext, ".epub"))
return 100;
if (strstr(magic, "META-INF/container.xml") || strstr(magic, "META-INF\\container.xml"))
return 200;
if (!strcmp(magic, "application/epub+zip"))
return 100;
return 0;
}
fz_document_writer *
fz_new_document_writer(fz_context *ctx, const char *path, const char *format, const char *options)
{
if (!format)
{
format = strrchr(path, '.');
if (!format)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot detect document format");
format += 1; /* skip the '.' */
}
if (!fz_strcasecmp(format, "cbz"))
return fz_new_cbz_writer(ctx, path, options);
if (!fz_strcasecmp(format, "png"))
return fz_new_png_writer(ctx, path, options);
#if FZ_ENABLE_PDF
if (!fz_strcasecmp(format, "pdf"))
return fz_new_pdf_writer(ctx, path, options);
#endif
fz_throw(ctx, FZ_ERROR_GENERIC, "unknown output document format: %s", format);
}
static fz_css_rule *
html_load_css(fz_context *ctx, fz_archive *zip, const char *base_uri, fz_css_rule *css, fz_xml *root)
{
fz_xml *node;
fz_buffer *buf;
char path[2048];
for (node = root; node; node = fz_xml_next(node))
{
const char *tag = fz_xml_tag(node);
if (tag && !strcmp(tag, "link"))
{
char *rel = fz_xml_att(node, "rel");
if (rel && !fz_strcasecmp(rel, "stylesheet"))
{
char *type = fz_xml_att(node, "type");
if ((type && !strcmp(type, "text/css")) || !type)
{
char *href = fz_xml_att(node, "href");
if (href)
{
fz_strlcpy(path, base_uri, sizeof path);
fz_strlcat(path, "/", sizeof path);
fz_strlcat(path, href, sizeof path);
fz_cleanname(path);
buf = fz_read_archive_entry(ctx, zip, path);
fz_write_buffer_byte(ctx, buf, 0);
css = fz_parse_css(ctx, css, (char*)buf->data, path);
fz_drop_buffer(ctx, buf);
}
}
}
}
if (tag && !strcmp(tag, "style"))
{
char *s = concat_text(ctx, node);
css = fz_parse_css(ctx, css, s, "<style>");
fz_free(ctx, s);
}
if (fz_xml_down(node))
css = html_load_css(ctx, zip, base_uri, css, fz_xml_down(node));
}
return css;
}
static fz_css_rule *
html_load_css(fz_context *ctx, fz_archive *zip, const char *base_uri, fz_css_rule *css, fz_xml *root)
{
fz_xml *html, *head, *node;
fz_buffer *buf;
char path[2048];
fz_var(buf);
html = fz_xml_find(root, "html");
head = fz_xml_find_down(html, "head");
for (node = fz_xml_down(head); node; node = fz_xml_next(node))
{
if (fz_xml_is_tag(node, "link"))
{
char *rel = fz_xml_att(node, "rel");
if (rel && !fz_strcasecmp(rel, "stylesheet"))
{
char *type = fz_xml_att(node, "type");
if ((type && !strcmp(type, "text/css")) || !type)
{
char *href = fz_xml_att(node, "href");
if (href)
{
fz_strlcpy(path, base_uri, sizeof path);
fz_strlcat(path, "/", sizeof path);
fz_strlcat(path, href, sizeof path);
fz_urldecode(path);
fz_cleanname(path);
buf = NULL;
fz_try(ctx)
{
buf = fz_read_archive_entry(ctx, zip, path);
fz_write_buffer_byte(ctx, buf, 0);
css = fz_parse_css(ctx, css, (char*)buf->data, path);
}
fz_always(ctx)
fz_drop_buffer(ctx, buf);
fz_catch(ctx)
fz_warn(ctx, "ignoring stylesheet %s", path);
}
}
}
/*
Open an output stream that writes to a
given path.
filename: The filename to write to (specified in UTF-8).
append: non-zero if we should append to the file, rather than
overwriting it.
*/
fz_output *
fz_new_output_with_path(fz_context *ctx, const char *filename, int append)
{
FILE *file;
fz_output *out;
if (!strcmp(filename, "/dev/null") || !fz_strcasecmp(filename, "nul:"))
return fz_new_output(ctx, 0, NULL, null_write, NULL, NULL);
#ifdef _WIN32
/* Ensure we create a brand new file. We don't want to clobber our old file. */
if (!append)
{
if (fz_remove_utf8(filename) < 0)
if (errno != ENOENT)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot remove file '%s': %s", filename, strerror(errno));
}
file = fz_fopen_utf8(filename, append ? "rb+" : "wb+");
#else
/* Ensure we create a brand new file. We don't want to clobber our old file. */
if (!append)
{
if (remove(filename) < 0)
if (errno != ENOENT)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot remove file '%s': %s", filename, strerror(errno));
}
file = fopen(filename, append ? "rb+" : "wb+");
#endif
if (!file)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot open file '%s': %s", filename, strerror(errno));
setvbuf(file, NULL, _IONBF, 0); /* we do our own buffering */
out = fz_new_output(ctx, 8192, file, file_write, NULL, file_drop);
out->seek = file_seek;
out->tell = file_tell;
out->as_stream = file_as_stream;
return out;
}
fz_output *
fz_new_output_with_path(fz_context *ctx, const char *filename, int append)
{
fz_output *out = NULL;
FILE *file;
if (!strcmp(filename, "/dev/null") || !fz_strcasecmp(filename, "nul:"))
return NULL;
file = fz_fopen(filename, append ? "ab" : "wb");
if (!file)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot open file '%s': %s", filename, strerror(errno));
fz_try(ctx)
{
out = fz_new_output_with_file_ptr(ctx, file, 1);
}
fz_catch(ctx)
{
fclose(file);
fz_rethrow(ctx);
}
return out;
}
static int
img_recognize(fz_context *doc, const char *magic)
{
char *ext = strrchr(magic, '.');
if (ext)
{
if (!fz_strcasecmp(ext, ".png") || !fz_strcasecmp(ext, ".jpg") ||
!fz_strcasecmp(ext, ".jpeg") || !fz_strcasecmp(ext, ".jfif") ||
!fz_strcasecmp(ext, ".jfif-tbnl") || !fz_strcasecmp(ext, ".jpe") ||
!fz_strcasecmp(ext, ".gif") || !fz_strcasecmp(ext, ".bmp"))
return 100;
#ifdef _TINSPIRE
if (!fz_strcasecmp(ext, ".tns"))
{
while (--ext >= magic)
{
if (*ext == '.')
break;
}
if (ext >= magic &&
(!fz_strcasecmp(ext, ".png.tns") || !fz_strcasecmp(ext, ".jpg.tns") ||
!fz_strcasecmp(ext, ".jpeg.tns") || !fz_strcasecmp(ext, ".jfif.tns") ||
!fz_strcasecmp(ext, ".jfif-tbnl.tns") || !fz_strcasecmp(ext, ".jpe.tns") ||
!fz_strcasecmp(ext, ".gif.tns")))
return 100;
}
#endif
}
if (!strcmp(magic, "png") || !strcmp(magic, "image/png") ||
!strcmp(magic, "jpg") || !strcmp(magic, "image/jpeg") ||
!strcmp(magic, "jpeg") || !strcmp(magic, "image/pjpeg") ||
!strcmp(magic, "jpe") || !strcmp(magic, "jfif") ||
!strcmp(magic, "gif") || !strcmp(magic, "image/gif") ||
!strcmp(magic, "bmp") || !strcmp(magic, "image/bmp"))
return 100;
return 0;
}
