char *
fz_copy_selection(fz_context *ctx, fz_stext_page *page, fz_rect rect)
{
fz_buffer *buffer;
fz_rect hitbox;
int c, i, block_num, seen = 0;
unsigned char *s;
float x0 = rect.x0;
float x1 = rect.x1;
float y0 = rect.y0;
float y1 = rect.y1;
buffer = fz_new_buffer(ctx, 1024);
for (block_num = 0; block_num < page->len; block_num++)
{
fz_stext_block *block;
fz_stext_line *line;
fz_stext_span *span;
if (page->blocks[block_num].type != FZ_PAGE_BLOCK_TEXT)
continue;
block = page->blocks[block_num].u.text;
for (line = block->lines; line < block->lines + block->len; line++)
{
for (span = line->first_span; span; span = span->next)
{
if (seen)
{
fz_write_buffer_byte(ctx, buffer, '\n');
}
seen = 0;
for (i = 0; i < span->len; i++)
{
fz_stext_char_bbox(ctx, &hitbox, span, i);
c = span->text[i].c;
if (c < 32)
c = 0xFFFD;
if (hitbox.x1 >= x0 && hitbox.x0 <= x1 && hitbox.y1 >= y0 && hitbox.y0 <= y1)
{
fz_write_buffer_rune(ctx, buffer, c);
seen = 1;
}
}
seen = (seen && span == line->last_span);
}
}
}
fz_write_buffer_byte(ctx, buffer, 0);
fz_buffer_extract(ctx, buffer, &s);
fz_drop_buffer(ctx, buffer);
return (char*)s;
}
char *
fz_copy_selection(fz_context *ctx, fz_text_page *page, fz_rect rect)
{
fz_buffer *buffer;
fz_rect hitbox;
fz_text_block *block;
fz_text_line *line;
int c, i, seen = 0;
char *s;
float x0 = rect.x0;
float x1 = rect.x1;
float y0 = rect.y0;
float y1 = rect.y1;
buffer = fz_new_buffer(ctx, 1024);
for (block = page->blocks; block < page->blocks + page->len; block++)
{
for (line = block->lines; line < block->lines + block->len; line++)
{
int span_num;
for (span_num = 0; span_num < line->len; span_num++)
{
fz_text_span *span = line->spans[span_num];
if (seen)
{
fz_write_buffer_byte(ctx, buffer, '\n');
}
seen = 0;
for (i = 0; i < span->len; i++)
{
fz_text_char_bbox(&hitbox, span, i);
c = span->text[i].c;
if (c < 32)
c = '?';
if (hitbox.x1 >= x0 && hitbox.x0 <= x1 && hitbox.y1 >= y0 && hitbox.y0 <= y1)
{
fz_write_buffer_rune(ctx, buffer, c);
seen = 1;
}
}
seen = (seen && span_num + 1 == line->len);
}
}
}
fz_write_buffer_byte(ctx, buffer, 0);
s = (char*)buffer->data;
fz_free(ctx, buffer);
return s;
}
static fz_document *
htdoc_open_document(fz_context *ctx, const char *filename)
{
char dirname[2048];
fz_buffer *buf;
html_document *doc;
fz_dirname(dirname, filename, sizeof dirname);
doc = fz_malloc_struct(ctx, html_document);
doc->super.close = htdoc_close_document;
doc->super.layout = htdoc_layout;
doc->super.count_pages = htdoc_count_pages;
doc->super.load_page = htdoc_load_page;
doc->zip = fz_open_directory(ctx, dirname);
doc->set = fz_new_html_font_set(ctx);
buf = fz_read_file(ctx, filename);
fz_write_buffer_byte(ctx, buf, 0);
doc->box = fz_parse_html(ctx, doc->set, doc->zip, ".", buf, fz_user_css(ctx));
fz_drop_buffer(ctx, buf);
return (fz_document*)doc;
}
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);
}
}
}
static fz_document *
svg_open_document_with_stream(fz_context *ctx, fz_stream *file)
{
fz_buffer *buf;
fz_document *doc;
buf = fz_read_all(ctx, file, 0);
fz_try(ctx)
{
fz_write_buffer_byte(ctx, buf, 0);
doc = svg_open_document_with_buffer(ctx, buf);
}
fz_always(ctx)
fz_drop_buffer(ctx, buf);
fz_catch(ctx)
fz_rethrow(ctx);
return doc;
}
void pdfapp_open_progressive(pdfapp_t *app, char *filename, int reload, int bps)
{
fz_context *ctx = app->ctx;
char *password = "";
fz_try(ctx)
{
fz_register_document_handlers(ctx);
if (app->layout_css)
{
fz_buffer *buf = fz_read_file(ctx, app->layout_css);
fz_write_buffer_byte(ctx, buf, 0);
fz_set_user_css(ctx, (char*)buf->data);
fz_drop_buffer(ctx, buf);
}
#ifdef HAVE_CURL
if (!strncmp(filename, "http://", 7))
{
app->stream = fz_stream_from_curl(ctx, filename, pdfapp_more_data, app);
while (1)
{
fz_try(ctx)
{
fz_seek(ctx, app->stream, 0, SEEK_SET);
app->doc = fz_open_document_with_stream(ctx, filename, app->stream);
}
fz_catch(ctx)
{
if (fz_caught(ctx) == FZ_ERROR_TRYLATER)
{
pdfapp_warn(app, "not enough data to open yet");
continue;
}
fz_rethrow(ctx);
}
break;
}
}
else
#endif
if (bps == 0)
static epub_chapter *
epub_parse_chapter(fz_context *ctx, epub_document *doc, const char *path)
{
fz_archive *zip = doc->zip;
fz_buffer *buf;
epub_chapter *ch;
char base_uri[2048];
fz_dirname(base_uri, path, sizeof base_uri);
buf = fz_read_archive_entry(ctx, zip, path);
fz_write_buffer_byte(ctx, buf, 0);
ch = fz_malloc_struct(ctx, epub_chapter);
ch->box = fz_parse_html(ctx, doc->set, zip, base_uri, buf, fz_user_css(ctx));
ch->next = NULL;
fz_drop_buffer(ctx, buf);
return ch;
}
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 *
htdoc_open_document_with_stream(fz_context *ctx, fz_stream *file)
{
html_document *doc;
fz_buffer *buf;
doc = fz_malloc_struct(ctx, html_document);
doc->super.close = htdoc_close_document;
doc->super.layout = htdoc_layout;
doc->super.count_pages = htdoc_count_pages;
doc->super.load_page = htdoc_load_page;
doc->zip = fz_open_directory(ctx, ".");
doc->set = fz_new_html_font_set(ctx);
buf = fz_read_all(ctx, file, 0);
fz_write_buffer_byte(ctx, buf, 0);
doc->box = fz_parse_html(ctx, doc->set, doc->zip, ".", buf, fz_user_css(ctx));
fz_drop_buffer(ctx, buf);
return (fz_document*)doc;
}
/*
* Read and interleave split parts from a ZIP file.
*/
xps_part *
xps_read_part(fz_context *ctx, xps_document *doc, char *partname)
{
fz_archive *zip = doc->zip;
fz_buffer *buf, *tmp;
char path[2048];
unsigned char *data;
int size;
int count;
char *name;
int seen_last;
name = partname;
if (name[0] == '/')
name ++;
/* All in one piece */
if (fz_has_archive_entry(ctx, zip, name))
{
buf = fz_read_archive_entry(ctx, zip, name);
}
/* Assemble all the pieces */
else
{
buf = fz_new_buffer(ctx, 512);
seen_last = 0;
for (count = 0; !seen_last; ++count)
{
sprintf(path, "%s/[%d].piece", name, count);
if (fz_has_archive_entry(ctx, zip, path))
{
tmp = fz_read_archive_entry(ctx, zip, path);
fz_append_buffer(ctx, buf, tmp);
fz_drop_buffer(ctx, tmp);
}
else
{
sprintf(path, "%s/[%d].last.piece", name, count);
if (fz_has_archive_entry(ctx, zip, path))
{
tmp = fz_read_archive_entry(ctx, zip, path);
fz_append_buffer(ctx, buf, tmp);
fz_drop_buffer(ctx, tmp);
seen_last = 1;
}
else
{
fz_drop_buffer(ctx, buf);
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find all pieces for part '%s'", partname);
}
}
}
}
fz_write_buffer_byte(ctx, buf, 0); /* zero-terminate */
/* take over the data */
data = buf->data;
/* size doesn't include the added zero-terminator */
size = buf->len - 1;
fz_free(ctx, buf);
return xps_new_part(ctx, doc, partname, data, size);
}
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 main(int argc, char **argv)
#endif
{
const GLFWvidmode *video_mode;
char filename[2048];
char *password = "";
float layout_w = DEFAULT_LAYOUT_W;
float layout_h = DEFAULT_LAYOUT_H;
float layout_em = DEFAULT_LAYOUT_EM;
char *layout_css = NULL;
int c;
while ((c = fz_getopt(argc, argv, "p:r:W:H:S:U:")) != -1)
{
switch (c)
{
default: usage(argv[0]); break;
case 'p': password = fz_optarg; break;
case 'r': currentzoom = fz_atof(fz_optarg); break;
case 'W': layout_w = fz_atof(fz_optarg); break;
case 'H': layout_h = fz_atof(fz_optarg); break;
case 'S': layout_em = fz_atof(fz_optarg); break;
case 'U': layout_css = fz_optarg; break;
}
}
if (fz_optind < argc)
{
fz_strlcpy(filename, argv[fz_optind], sizeof filename);
}
else
{
#ifdef _WIN32
win_install();
if (!win_open_file(filename, sizeof filename))
exit(0);
#else
usage(argv[0]);
#endif
}
title = strrchr(filename, '/');
if (!title)
title = strrchr(filename, '\\');
if (title)
++title;
else
title = filename;
memset(&ui, 0, sizeof ui);
search_input.p = search_input.text;
search_input.q = search_input.p;
search_input.end = search_input.p;
if (!glfwInit()) {
fprintf(stderr, "cannot initialize glfw\n");
exit(1);
}
video_mode = glfwGetVideoMode(glfwGetPrimaryMonitor());
screen_w = video_mode->width;
screen_h = video_mode->height;
glfwSetErrorCallback(on_error);
window = glfwCreateWindow(DEFAULT_WINDOW_W, DEFAULT_WINDOW_H, filename, NULL, NULL);
if (!window) {
fprintf(stderr, "cannot create glfw window\n");
exit(1);
}
glfwMakeContextCurrent(window);
ctx = fz_new_context(NULL, NULL, 0);
fz_register_document_handlers(ctx);
if (layout_css)
{
fz_buffer *buf = fz_read_file(ctx, layout_css);
fz_write_buffer_byte(ctx, buf, 0);
fz_set_user_css(ctx, (char*)buf->data);
fz_drop_buffer(ctx, buf);
}
has_ARB_texture_non_power_of_two = glfwExtensionSupported("GL_ARB_texture_non_power_of_two");
if (!has_ARB_texture_non_power_of_two)
fz_warn(ctx, "OpenGL implementation does not support non-power of two texture sizes");
glGetIntegerv(GL_MAX_TEXTURE_SIZE, &max_texture_size);
ui.fontsize = DEFAULT_UI_FONTSIZE;
ui.baseline = DEFAULT_UI_BASELINE;
ui.lineheight = DEFAULT_UI_LINEHEIGHT;
ui_init_fonts(ctx, ui.fontsize);
doc = fz_open_document(ctx, filename);
if (fz_needs_password(ctx, doc))
{
if (!fz_authenticate_password(ctx, doc, password))
{
fprintf(stderr, "Invalid password.\n");
exit(1);
}
}
outline = fz_load_outline(ctx, doc);
pdf = pdf_specifics(ctx, doc);
if (pdf)
pdf_enable_js(ctx, pdf);
fz_layout_document(ctx, doc, layout_w, layout_h, layout_em);
render_page();
update_title();
shrinkwrap();
glfwSetFramebufferSizeCallback(window, on_reshape);
glfwSetCursorPosCallback(window, on_mouse_motion);
glfwSetMouseButtonCallback(window, on_mouse_button);
glfwSetScrollCallback(window, on_scroll);
glfwSetCharModsCallback(window, on_char);
glfwSetKeyCallback(window, on_key);
glfwSetWindowRefreshCallback(window, on_display);
glfwGetFramebufferSize(window, &window_w, &window_h);
ui_needs_update = 1;
while (!glfwWindowShouldClose(window))
{
glfwWaitEvents();
if (ui_needs_update)
run_main_loop();
}
ui_finish_fonts(ctx);
fz_drop_link(ctx, links);
fz_drop_page(ctx, page);
fz_drop_outline(ctx, outline);
fz_drop_document(ctx, doc);
fz_drop_context(ctx);
glfwTerminate();
return 0;
}
int muconvert_main(int argc, char **argv)
{
int i, c;
while ((c = fz_getopt(argc, argv, "p:A:W:H:S:U:o:F:O:")) != -1)
{
switch (c)
{
default: usage(); break;
case 'p': password = fz_optarg; break;
case 'A': alphabits = atoi(fz_optarg); break;
case 'W': layout_w = atof(fz_optarg); break;
case 'H': layout_h = atof(fz_optarg); break;
case 'S': layout_em = atof(fz_optarg); break;
case 'U': layout_css = fz_optarg; break;
case 'o': output = fz_optarg; break;
case 'F': format = fz_optarg; break;
case 'O': options = fz_optarg; break;
}
}
if (fz_optind == argc || (!format && !output))
usage();
/* Create a context to hold the exception stack and various caches. */
ctx = fz_new_context(NULL, NULL, FZ_STORE_UNLIMITED);
if (!ctx)
{
fprintf(stderr, "cannot create mupdf context\n");
return EXIT_FAILURE;
}
/* Register the default file types to handle. */
fz_try(ctx)
fz_register_document_handlers(ctx);
fz_catch(ctx)
{
fprintf(stderr, "cannot register document handlers: %s\n", fz_caught_message(ctx));
fz_drop_context(ctx);
return EXIT_FAILURE;
}
fz_set_aa_level(ctx, alphabits);
if (layout_css)
{
fz_buffer *buf = fz_read_file(ctx, layout_css);
fz_write_buffer_byte(ctx, buf, 0);
fz_set_user_css(ctx, (char*)buf->data);
fz_drop_buffer(ctx, buf);
}
/* Open the output document. */
fz_try(ctx)
out = fz_new_document_writer(ctx, output, format, options);
fz_catch(ctx)
{
fprintf(stderr, "cannot create document: %s\n", fz_caught_message(ctx));
fz_drop_context(ctx);
return EXIT_FAILURE;
}
for (i = fz_optind; i < argc; ++i)
{
doc = fz_open_document(ctx, argv[i]);
if (fz_needs_password(ctx, doc))
if (!fz_authenticate_password(ctx, doc, password))
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot authenticate password: %s", argv[i]);
fz_layout_document(ctx, doc, layout_w, layout_h, layout_em);
count = fz_count_pages(ctx, doc);
if (i+1 < argc && fz_is_page_range(ctx, argv[i+1]))
runrange(argv[++i]);
else
runrange("1-N");
fz_drop_document(ctx, doc);
}
fz_drop_document_writer(ctx, out);
fz_drop_context(ctx);
return EXIT_SUCCESS;
}
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;
char base_uri[2048];
const char *full_path;
const char *version;
char ncx[2048], s[2048];
epub_chapter *head, *tail;
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);
container_xml = fz_parse_xml(ctx, buf->data, buf->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");
printf("epub: found root: %s\n", full_path);
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);
content_opf = fz_parse_xml(ctx, buf->data, buf->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>");
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))
{
/* TODO: parse NCX to create fz_outline */
printf("epub: found outline: %s\n", ncx);
}
head = tail = NULL;
itemref = fz_xml_find_down(spine, "itemref");
while (itemref)
{
if (path_from_idref(s, manifest, base_uri, fz_xml_att(itemref, "idref"), sizeof s))
{
printf("epub: found spine %s\n", s);
if (!head)
head = tail = epub_parse_chapter(ctx, doc, s);
else
tail = tail->next = epub_parse_chapter(ctx, doc, s);
}
itemref = fz_xml_find_next(itemref, "itemref");
}
doc->spine = head;
printf("epub: done.\n");
fz_drop_xml(ctx, container_xml);
fz_drop_xml(ctx, content_opf);
}
