void pdf_clean_file(fz_context *ctx, char *infile, char *outfile, char *password, fz_write_options *opts, char *argv[], int argc)
{
globals glo = { 0 };
glo.ctx = ctx;
fz_try(ctx)
{
glo.doc = pdf_open_document_no_run(ctx, infile);
if (pdf_needs_password(ctx, glo.doc))
if (!pdf_authenticate_password(ctx, glo.doc, password))
fz_throw(glo.ctx, FZ_ERROR_GENERIC, "cannot authenticate password: %s", infile);
/* Only retain the specified subset of the pages */
if (argc)
retainpages(ctx, &glo, argc, argv);
pdf_write_document(ctx, glo.doc, outfile, opts);
}
fz_always(ctx)
{
pdf_close_document(ctx, glo.doc);
}
fz_catch(ctx)
{
if (opts && opts->errors)
*opts->errors = *opts->errors+1;
}
}
int pdfsign_main(int argc, char **argv)
{
fz_context *ctx;
pdf_document *doc;
char *password = "";
int i, n, c;
pdf_page *page = NULL;
while ((c = fz_getopt(argc, argv, "p:")) != -1)
{
switch (c)
{
case 'p': password = fz_optarg; break;
default: usage(); break;
}
}
if (argc - fz_optind < 1)
usage();
filename = argv[fz_optind++];
ctx = fz_new_context(NULL, NULL, FZ_STORE_UNLIMITED);
if (!ctx)
{
fprintf(stderr, "cannot initialize context\n");
exit(1);
}
fz_var(page);
doc = pdf_open_document(ctx, filename);
fz_try(ctx)
{
if (pdf_needs_password(ctx, doc))
if (!pdf_authenticate_password(ctx, doc, password))
fz_warn(ctx, "cannot authenticate password: %s", filename);
n = pdf_count_pages(ctx, doc);
for (i = 0; i < n; ++i)
{
page = pdf_load_page(ctx, doc, i);
verify_page(ctx, doc, i, page);
fz_drop_page(ctx, (fz_page*)page);
page = NULL;
}
}
fz_always(ctx)
pdf_drop_document(ctx, doc);
fz_catch(ctx)
{
fz_drop_page(ctx, (fz_page*)page);
fprintf(stderr, "error verify signatures: %s\n", fz_caught_message(ctx));
}
fz_flush_warnings(ctx);
fz_drop_context(ctx);
return 0;
}
int pdfextract_main(int argc, char **argv)
{
char *infile;
char *password = "";
int c, o;
while ((c = fz_getopt(argc, argv, "p:r")) != -1)
{
switch (c)
{
case 'p':
password = fz_optarg;
break;
case 'r':
dorgb++;
break;
default:
usage();
break;
}
}
if (fz_optind == argc)
usage();
infile = argv[fz_optind++];
ctx = fz_new_context(NULL, NULL, FZ_STORE_UNLIMITED);
if (!ctx)
{
fprintf(stderr, "cannot initialise context\n");
exit(1);
}
doc = pdf_open_document(ctx, infile);
if (pdf_needs_password(ctx, doc))
if (!pdf_authenticate_password(ctx, doc, password))
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot authenticate password: %s", infile);
if (fz_optind == argc)
{
int len = pdf_count_objects(ctx, doc);
for (o = 1; o < len; o++)
showobject(o);
}
else
{
while (fz_optind < argc)
{
showobject(atoi(argv[fz_optind]));
fz_optind++;
}
}
pdf_close_document(ctx, doc);
fz_flush_warnings(ctx);
fz_drop_context(ctx);
return 0;
}
int pdfposter_main(int argc, char **argv)
{
char *infile;
char *outfile = "out.pdf";
char *password = "";
int c;
fz_write_options opts = { 0 };
pdf_document *doc;
fz_context *ctx;
opts.do_incremental = 0;
opts.do_garbage = 0;
opts.do_expand = 0;
opts.do_ascii = 0;
opts.do_linear = 0;
while ((c = fz_getopt(argc, argv, "x:y:")) != -1)
{
switch (c)
{
case 'p': password = fz_optarg; break;
case 'x': x_factor = atoi(fz_optarg); break;
case 'y': y_factor = atoi(fz_optarg); break;
default: usage(); break;
}
}
if (argc - fz_optind < 1)
usage();
infile = argv[fz_optind++];
if (argc - fz_optind > 0 &&
(strstr(argv[fz_optind], ".pdf") || strstr(argv[fz_optind], ".PDF")))
{
outfile = argv[fz_optind++];
}
ctx = fz_new_context(NULL, NULL, FZ_STORE_UNLIMITED);
if (!ctx)
{
fprintf(stderr, "cannot initialise context\n");
exit(1);
}
doc = pdf_open_document(ctx, infile);
if (pdf_needs_password(ctx, doc))
if (!pdf_authenticate_password(ctx, doc, password))
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot authenticate password: %s", infile);
decimatepages(ctx, doc);
pdf_write_document(ctx, doc, outfile, &opts);
pdf_close_document(ctx, doc);
fz_drop_context(ctx);
return 0;
}
int pdfposter_main(int argc, char **argv)
{
char *infile;
char *outfile = "out.pdf";
char *password = "";
int c;
fz_write_options opts;
pdf_document *xref;
fz_context *ctx;
opts.do_garbage = 0;
opts.do_expand = 0;
opts.do_ascii = 0;
while ((c = fz_getopt(argc, argv, "x:y:")) != -1)
{
switch (c)
{
case 'p': password = fz_optarg; break;
case 'x': x_factor = atoi(fz_optarg); break;
case 'y': y_factor = atoi(fz_optarg); break;
default: usage(); break;
}
}
if (argc - fz_optind < 1)
usage();
infile = argv[fz_optind++];
if (argc - fz_optind > 0 &&
(strstr(argv[fz_optind], ".pdf") || strstr(argv[fz_optind], ".PDF")))
{
outfile = argv[fz_optind++];
}
ctx = fz_new_context(NULL, NULL, FZ_STORE_UNLIMITED);
if (!ctx)
{
fprintf(stderr, "cannot initialise context\n");
exit(1);
}
xref = pdf_open_document_no_run(ctx, infile);
if (pdf_needs_password(xref))
if (!pdf_authenticate_password(xref, password))
fz_throw(ctx, "cannot authenticate password: %s", infile);
/* Only retain the specified subset of the pages */
decimatepages(xref);
pdf_write_document(xref, outfile, &opts);
pdf_close_document(xref);
fz_free_context(ctx);
return 0;
}
/*
** Get /Info strings, e.g. wmupdf_info_field("myfile.pdf","Author",buf,255);
** Info labels:
** Title Document title
** Author Name of the person who created the doc.
** Subject Subject of the doc
** Keywords Keywords associated with the document.
** Creator If doc was converted to PDF from another format, the name
** of the product that created the original document.
** Producer If doc was converted to PDF from another format, the name
** of the product that converted it to PDF.
** CreationDate Date/Time document was created.
** ModDate Date/Time of most recent mod.
*/
int wmupdf_info_field(char *infile,char *label,char *buf,int maxlen)
{
pdf_document *xref;
fz_context *ctx;
pdf_obj *info,*obj;
char *password="";
xref=NULL;
buf[0]='\0';
ctx = fz_new_context(NULL,NULL,FZ_STORE_UNLIMITED);
if (!ctx)
return(-1);
fz_try(ctx)
{
xref=pdf_open_document_no_run(ctx,infile);
if (!xref)
{
fz_free_context(ctx);
return(-2);
}
if (pdf_needs_password(xref) && !pdf_authenticate_password(xref,password))
{
pdf_close_document(xref);
fz_free_context(ctx);
return(-3);
}
if (xref->trailer!=NULL
&& (info=pdf_dict_gets(xref->trailer,"Info"))!=NULL
&& (obj=pdf_dict_gets(info,label))!=NULL
&& pdf_is_string(obj))
{
strncpy(buf,pdf_to_str_buf(obj),maxlen-1);
buf[maxlen-1]='\0';
}
}
fz_always(ctx)
{
pdf_close_document(xref);
}
fz_catch(ctx)
{
}
fz_free_context(ctx);
return(0);
}
static int
pdfpages_pages(fz_context *ctx, fz_output *out, char *filename, char *password, char *argv[], int argc)
{
enum { NO_FILE_OPENED, NO_INFO_GATHERED, INFO_SHOWN } state;
int argidx = 0;
pdf_document *doc = NULL;
int ret = 0;
state = NO_FILE_OPENED;
while (argidx < argc)
{
if (state == NO_FILE_OPENED || !fz_is_page_range(ctx, argv[argidx]))
{
if (state == NO_INFO_GATHERED)
{
showpages(ctx, doc, out, "1-N");
}
pdf_drop_document(ctx, doc);
filename = argv[argidx];
fz_printf(ctx, out, "%s:\n", filename);
doc = pdf_open_document(ctx, filename);
if (pdf_needs_password(ctx, doc))
if (!pdf_authenticate_password(ctx, doc, password))
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot authenticate password: %s", filename);
state = NO_INFO_GATHERED;
}
else
{
ret |= showpages(ctx, doc, out, argv[argidx]);
state = INFO_SHOWN;
}
argidx++;
}
if (state == NO_INFO_GATHERED)
showpages(ctx, doc, out, "1-N");
pdf_drop_document(ctx, doc);
return ret;
}
pdf_document *openxref(fz_context *ctx, char *filename)
{
pdf_document *xref = pdf_open_document(ctx, filename);
fz_try(ctx)
{
if (pdf_needs_password(xref))
{
logbench("Warning: password protected document\n");
fz_throw(ctx, "document requires password");
}
pdf_count_pages(xref);
}
fz_catch(ctx)
{
pdf_close_document(xref);
fz_rethrow(ctx);
}
return xref;
}
static int _pdf_doc_load(struct _pdf_doc *self, mume_stream_t *stm)
{
fz_error error;
fz_stream *fzstm;
fz_rect *mbox;
fz_obj *page_obj, *box_obj;
int i, c;
_pdf_doc_clear(self);
fzstm = fz_new_stream(stm, _pdf_stream_read, _pdf_stream_close);
mume_stream_reference(stm);
fzstm->seek = _pdf_stream_seek;
error = pdf_open_xref_with_stream(&self->xref, fzstm, NULL);
fz_close(fzstm);
if (error) {
mume_error(("Read xref failed\n", error));
return 0;
}
assert(!pdf_needs_password(self->xref));
/* Load meta information. */
error = pdf_load_page_tree(self->xref);
if (error) {
mume_error(("Cannot load page tree\n"));
return 0;
}
c = pdf_count_pages(self->xref);
self->glyph_cache = fz_new_glyph_cache();
self->pages = calloc_abort(c, sizeof(pdf_page*));
self->disps = calloc_abort(c, sizeof(fz_display_list*));
self->media_boxes = malloc_abort(c * sizeof(fz_rect));
self->page_rotates = malloc_abort(c * sizeof(int));
/* Extract each pages' media box and rotation. */
for (i = 0; i < c; ++i) {
mbox = self->media_boxes + i;
page_obj = self->xref->page_objs[i];
if (!page_obj) {
*mbox = fz_empty_rect;
continue;
}
box_obj = fz_dict_gets(page_obj, "MediaBox");
*mbox = pdf_to_rect(box_obj);
if (fz_is_empty_rect(*mbox)) {
fz_warn("Cannot find page bounds, guessing page bounds.");
mbox->x1 = 612;
mbox->y1 = 792;
}
box_obj = fz_dict_gets(page_obj, "CropBox");
if (fz_is_array(box_obj))
*mbox = fz_intersect_rect(*mbox, pdf_to_rect(box_obj));
self->page_rotates[i] = fz_to_int(
fz_dict_gets(page_obj, "Rotate"));
if (self->page_rotates[i] % 90)
self->page_rotates[i] = 0;
}
return 1;
}
int pdfinfo_main(int argc, char **argv)
{
enum { NO_FILE_OPENED, NO_INFO_GATHERED, INFO_SHOWN } state;
char *filename = "";
char *password = "";
int show = ALL;
int c;
while ((c = fz_getopt(argc, argv, "mfispxd:")) != -1)
{
switch (c)
{
case 'm': if (show == ALL) show = DIMENSIONS; else show |= DIMENSIONS; break;
case 'f': if (show == ALL) show = FONTS; else show |= FONTS; break;
case 'i': if (show == ALL) show = IMAGES; else show |= IMAGES; break;
case 's': if (show == ALL) show = SHADINGS; else show |= SHADINGS; break;
case 'p': if (show == ALL) show = PATTERNS; else show |= PATTERNS; break;
case 'x': if (show == ALL) show = XOBJS; else show |= XOBJS; break;
case 'd': password = fz_optarg; break;
default:
infousage();
break;
}
}
if (fz_optind == argc)
infousage();
ctx = fz_new_context(NULL, NULL, FZ_STORE_UNLIMITED);
if (!ctx)
{
fprintf(stderr, "cannot initialise context\n");
exit(1);
}
state = NO_FILE_OPENED;
while (fz_optind < argc)
{
if (state == NO_FILE_OPENED || !arg_is_page_range(argv[fz_optind]))
{
if (state == NO_INFO_GATHERED)
{
showinfo(filename, show, "1-");
}
closexref();
filename = argv[fz_optind];
printf("%s:\n", filename);
doc = pdf_open_document_no_run(ctx, filename);
if (pdf_needs_password(doc))
if (!pdf_authenticate_password(doc, password))
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot authenticate password: %s", filename);
pagecount = pdf_count_pages(doc);
showglobalinfo();
state = NO_INFO_GATHERED;
}
else
{
showinfo(filename, show, argv[fz_optind]);
state = INFO_SHOWN;
}
fz_optind++;
}
if (state == NO_INFO_GATHERED)
showinfo(filename, show, "1-");
closexref();
fz_free_context(ctx);
return 0;
}
int pdfportfolio_main(int argc, char **argv)
{
char *password = "";
char *outfile = NULL;
char *outopts = "compress";
char *infile;
int exit_code = 0;
int do_save = 0;
int has_old_file = 0;
int c;
while ((c = fz_getopt(argc, argv, "p:o:O:")) != -1)
{
switch (c)
{
case 'p': password = fz_optarg; break;
case 'o': outfile = fz_optarg; break;
case 'O': outopts = fz_optarg; break;
default: usage(); break;
}
}
if (fz_optind == argc)
usage();
infile = argv[fz_optind++];
if (!outfile)
outfile = infile;
ctx = fz_new_context(NULL, NULL, FZ_STORE_UNLIMITED);
if (!ctx)
{
fprintf(stderr, "cannot initialise context\n");
exit(1);
}
if (fz_file_exists(ctx, infile))
{
doc = pdf_open_document(ctx, infile);
if (pdf_needs_password(ctx, doc))
if (!pdf_authenticate_password(ctx, doc, password))
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot authenticate password: %s", infile);
has_old_file = 1;
}
else
{
doc = pdf_create_document(ctx);
/* add a blank page */
{
const char *template = "BT /Tm 16 Tf 50 434 TD (This is a portfolio document.) Tj ET\n";
const unsigned char *data;
int size;
fz_font *font;
pdf_obj *font_obj, *page_obj;
pdf_obj *resources;
fz_buffer *contents;
fz_rect mediabox = { 0, 0, 400, 500 };
data = fz_lookup_base14_font(ctx, "Times-Roman", &size);
font = fz_new_font_from_memory(ctx, "Times-Roman", data, size, 0, 0);
font_obj = pdf_add_simple_font(ctx, doc, font, PDF_SIMPLE_ENCODING_LATIN);
fz_drop_font(ctx, font);
resources = pdf_add_new_dict(ctx, doc, 1);
pdf_dict_putp_drop(ctx, resources, "Font/Tm", font_obj);
contents = fz_new_buffer_from_shared_data(ctx, (const unsigned char *)template, strlen(template));
void pdfapp_open(pdfapp_t *app, char *filename)
{
#if 0
fz_error error;
fz_obj *obj;
char *password = "";
#else
fz_error error;
fz_stream *file;
char *password = "";
fz_obj *obj;
fz_obj *info;
int fd;
fd = open(filename, O_BINARY | O_RDONLY, 0666);
if (fd < 0)
fprintf(stderr, "error, file %s does not exist\n", filename);
#endif
/*
* Open PDF and load xref table
*/
app->filename = filename;
#if 0
app->xref = pdf_newxref();
error = pdf_loadxref(app->xref, filename);
if (error)
{
fz_catch(error, "trying to air");
error = pdf_repairxref(app->xref, filename);
if (error)
pdfapp_error(app, error);
}
error = pdf_decryptxref(app->xref);
if (error)
pdfapp_error(app, error);
#else
file = fz_open_fd(fd);
error = pdf_open_xref_with_stream(&app->xref, file, NULL);
if (error)
pdfapp_error(app, fz_rethrow(error, "cannot open document '%s'", filename));
fz_close(file);
#endif
/*
* Handle encrypted PDF files
*/
if (pdf_needs_password(app->xref))
{
int okay = pdf_authenticate_password(app->xref, password);
while (!okay)
{
//password = winpassword(app, filename);
if (!password)
exit(1);
okay = pdf_authenticate_password(app->xref, password);
if (!okay)
pdfapp_warn(app, "Invalid password.");
}
}
/*
* Load meta information
* TODO: move this into mupdf library
*/
#if 0
obj = fz_dictgets(app->xref->trailer, "Root");
app->xref->root = fz_resolveindirect(obj);
if (!app->xref->root)
pdfapp_error(app, fz_throw("syntaxerror: missing Root object"));
fz_keepobj(app->xref->root);
obj = fz_dictgets(app->xref->trailer, "Info");
app->xref->info = fz_resolveindirect(obj);
if (!app->xref->info)
pdfapp_warn(app, "Could not load PDF meta information.");
if (app->xref->info)
fz_keepobj(app->xref->info);
/*app->outline = pdf_loadoutline(app->xref);*/
app->doctitle = filename;
if (strrchr(app->doctitle, '\\'))
app->doctitle = strrchr(app->doctitle, '\\') + 1;
if (strrchr(app->doctitle, '/'))
app->doctitle = strrchr(app->doctitle, '/') + 1;
if (app->xref->info)
{
obj = fz_dictgets(app->xref->info, "Title");
if (obj)
{
app->doctitle = pdf_toutf8(obj);
}
}
#else
app->outline = pdf_load_outline(app->xref);
app->doctitle = filename;
if (strrchr(app->doctitle, '\\'))
app->doctitle = strrchr(app->doctitle, '\\') + 1;
if (strrchr(app->doctitle, '/'))
app->doctitle = strrchr(app->doctitle, '/') + 1;
info = fz_dict_gets(app->xref->trailer, "Info");
if (info)
{
obj = fz_dict_gets(info, "Title");
if (obj)
app->doctitle = pdf_to_utf8(obj);
}
#endif
/*
* Start at first page
*/
#if 0
app->pagecount = pdf_getpagecount(app->xref);
#else
error = pdf_load_page_tree(app->xref);
if (error)
pdfapp_error(app, fz_rethrow(error, "cannot load page tree"));
app->pagecount = pdf_count_pages(app->xref);
#endif
app->rotate = 0;
}
static void pdfapp_open_pdf(pdfapp_t *app, char *filename, int fd)
{
fz_error error;
fz_stream *file;
char *password = "";
fz_obj *obj;
fz_obj *info;
/*
* Open PDF and load xref table
*/
file = fz_open_fd(fd);
error = pdf_open_xref_with_stream(&app->xref, file, NULL);
if (error)
pdfapp_error(app, fz_rethrow(error, "cannot open document '%s'", filename));
fz_close(file);
/*
* Handle encrypted PDF files
*/
if (pdf_needs_password(app->xref))
{
int okay = pdf_authenticate_password(app->xref, password);
while (!okay)
{
password = winpassword(app, filename);
if (!password)
exit(1);
okay = pdf_authenticate_password(app->xref, password);
if (!okay)
pdfapp_warn(app, "Invalid password.");
}
}
/*
* Load meta information
*/
app->outline = pdf_load_outline(app->xref);
app->doctitle = filename;
if (strrchr(app->doctitle, '\\'))
app->doctitle = strrchr(app->doctitle, '\\') + 1;
if (strrchr(app->doctitle, '/'))
app->doctitle = strrchr(app->doctitle, '/') + 1;
info = fz_dict_gets(app->xref->trailer, "Info");
if (info)
{
obj = fz_dict_gets(info, "Title");
if (obj)
app->doctitle = pdf_to_utf8(obj);
}
/*
* Start at first page
*/
error = pdf_load_page_tree(app->xref);
if (error)
pdfapp_error(app, fz_rethrow(error, "cannot load page tree"));
app->pagecount = pdf_count_pages(app->xref);
}
fz_error
pdf_open_xref_with_stream(pdf_xref **xrefp, fz_stream *file, char *password)
{
pdf_xref *xref;
fz_error error;
fz_obj *encrypt, *id;
fz_obj *dict, *obj;
int i, repaired = 0;
/* install pdf specific callback */
fz_resolve_indirect = pdf_resolve_indirect;
xref = fz_malloc(sizeof(pdf_xref));
memset(xref, 0, sizeof(pdf_xref));
xref->file = fz_keep_stream(file);
error = pdf_load_xref(xref, xref->scratch, sizeof xref->scratch);
if (error)
{
fz_catch(error, "trying to repair");
if (xref->table)
{
fz_free(xref->table);
xref->table = NULL;
xref->len = 0;
}
if (xref->trailer)
{
fz_drop_obj(xref->trailer);
xref->trailer = NULL;
}
error = pdf_repair_xref(xref, xref->scratch, sizeof xref->scratch);
if (error)
{
pdf_free_xref(xref);
return fz_rethrow(error, "cannot repair document");
}
repaired = 1;
}
encrypt = fz_dict_gets(xref->trailer, "Encrypt");
id = fz_dict_gets(xref->trailer, "ID");
if (fz_is_dict(encrypt))
{
error = pdf_new_crypt(&xref->crypt, encrypt, id);
if (error)
{
pdf_free_xref(xref);
return fz_rethrow(error, "cannot decrypt document");
}
}
if (pdf_needs_password(xref))
{
/* Only care if we have a password */
if (password)
{
int okay = pdf_authenticate_password(xref, password);
if (!okay)
{
pdf_free_xref(xref);
return fz_throw("invalid password");
}
}
}
if (repaired)
{
int hasroot, hasinfo;
error = pdf_repair_obj_stms(xref);
if (error)
{
pdf_free_xref(xref);
return fz_rethrow(error, "cannot repair document");
}
hasroot = fz_dict_gets(xref->trailer, "Root") != NULL;
hasinfo = fz_dict_gets(xref->trailer, "Info") != NULL;
for (i = 1; i < xref->len; i++)
{
if (xref->table[i].type == 0 || xref->table[i].type == 'f')
continue;
error = pdf_load_object(&dict, xref, i, 0);
if (error)
{
fz_catch(error, "ignoring broken object (%d 0 R)", i);
continue;
}
if (!hasroot)
{
obj = fz_dict_gets(dict, "Type");
if (fz_is_name(obj) && !strcmp(fz_to_name(obj), "Catalog"))
{
obj = fz_new_indirect(i, 0, xref);
fz_dict_puts(xref->trailer, "Root", obj);
fz_drop_obj(obj);
}
}
if (!hasinfo)
{
if (fz_dict_gets(dict, "Creator") || fz_dict_gets(dict, "Producer"))
{
obj = fz_new_indirect(i, 0, xref);
fz_dict_puts(xref->trailer, "Info", obj);
fz_drop_obj(obj);
}
}
fz_drop_obj(dict);
}
}
error = pdf_read_ocg(xref);
if (error)
{
pdf_free_xref(xref);
return fz_rethrow(error, "Broken Optional Content");
}
*xrefp = xref;
return fz_okay;
}
int main(int argc, char **argv)
#endif
{
char *infile;
char *outfile = "out.pdf";
char *password = "";
int c, num;
int subset;
while ((c = fz_getopt(argc, argv, "adfgip:")) != -1)
{
switch (c)
{
case 'p': password = fz_optarg; break;
case 'g': dogarbage ++; break;
case 'd': doexpand ^= expand_all; break;
case 'f': doexpand ^= expand_fonts; break;
case 'i': doexpand ^= expand_images; break;
case 'a': doascii ++; break;
default: usage(); break;
}
}
if (argc - fz_optind < 1)
usage();
infile = argv[fz_optind++];
if (argc - fz_optind > 0 &&
(strstr(argv[fz_optind], ".pdf") || strstr(argv[fz_optind], ".PDF")))
{
outfile = argv[fz_optind++];
}
subset = 0;
if (argc - fz_optind > 0)
subset = 1;
ctx = fz_new_context(NULL, NULL, FZ_STORE_UNLIMITED);
if (!ctx)
{
fprintf(stderr, "cannot initialise context\n");
exit(1);
}
xref = pdf_open_document(ctx, infile);
if (pdf_needs_password(xref))
if (!pdf_authenticate_password(xref, password))
fz_throw(ctx, "cannot authenticate password: %s\n", infile);
out = fopen(outfile, "wb");
if (!out)
fz_throw(ctx, "cannot open output file '%s'", outfile);
fprintf(out, "%%PDF-%d.%d\n", xref->version / 10, xref->version % 10);
fprintf(out, "%%\316\274\341\277\246\n\n");
uselist = fz_malloc_array(ctx, xref->len + 1, sizeof(char));
ofslist = fz_malloc_array(ctx, xref->len + 1, sizeof(int));
genlist = fz_malloc_array(ctx, xref->len + 1, sizeof(int));
renumbermap = fz_malloc_array(ctx, xref->len + 1, sizeof(int));
for (num = 0; num < xref->len; num++)
{
uselist[num] = 0;
ofslist[num] = 0;
genlist[num] = 0;
renumbermap[num] = num;
}
/* Make sure any objects hidden in compressed streams have been loaded */
preloadobjstms();
/* Only retain the specified subset of the pages */
if (subset)
retainpages(argc, argv);
/* Sweep & mark objects from the trailer */
if (dogarbage >= 1)
sweepobj(xref->trailer);
/* Coalesce and renumber duplicate objects */
if (dogarbage >= 3)
removeduplicateobjs();
/* Compact xref by renumbering and removing unused objects */
if (dogarbage >= 2)
compactxref();
/* Make renumbering affect all indirect references and update xref */
/* Do not renumber objects if encryption is in use, as the object
* numbers are baked into the streams/strings, and we can't currently
* cope with moving them. See bug 692627. */
if (dogarbage >= 2 && !xref->crypt)
renumberobjs();
writepdf();
if (fclose(out))
fz_throw(ctx, "cannot close output file '%s'", outfile);
fz_free(xref->ctx, uselist);
fz_free(xref->ctx, ofslist);
fz_free(xref->ctx, genlist);
fz_free(xref->ctx, renumbermap);
pdf_close_document(xref);
fz_free_context(ctx);
return 0;
}
int pdfclean_main(int argc, char **argv)
{
char *infile;
char *outfile = "out.pdf";
char *password = "";
int c;
int subset;
fz_write_options opts;
int write_failed = 0;
opts.do_garbage = 0;
opts.do_expand = 0;
opts.do_ascii = 0;
opts.do_linear = 0;
while ((c = fz_getopt(argc, argv, "adfgilp:")) != -1)
{
switch (c)
{
case 'p': password = fz_optarg; break;
case 'g': opts.do_garbage ++; break;
case 'd': opts.do_expand ^= fz_expand_all; break;
case 'f': opts.do_expand ^= fz_expand_fonts; break;
case 'i': opts.do_expand ^= fz_expand_images; break;
case 'l': opts.do_linear ++; break;
case 'a': opts.do_ascii ++; break;
default: usage(); break;
}
}
if (argc - fz_optind < 1)
usage();
infile = argv[fz_optind++];
if (argc - fz_optind > 0 &&
(strstr(argv[fz_optind], ".pdf") || strstr(argv[fz_optind], ".PDF")))
{
outfile = argv[fz_optind++];
}
subset = 0;
if (argc - fz_optind > 0)
subset = 1;
ctx = fz_new_context(NULL, NULL, FZ_STORE_UNLIMITED);
if (!ctx)
{
fprintf(stderr, "cannot initialise context\n");
exit(1);
}
fz_try(ctx)
{
xref = pdf_open_document_no_run(ctx, infile);
if (pdf_needs_password(xref))
if (!pdf_authenticate_password(xref, password))
fz_throw(ctx, "cannot authenticate password: %s", infile);
/* Only retain the specified subset of the pages */
if (subset)
retainpages(argc, argv);
pdf_write_document(xref, outfile, &opts);
}
fz_always(ctx)
{
pdf_close_document(xref);
}
fz_catch(ctx)
{
write_failed = 1;
}
fz_free_context(ctx);
return write_failed ? 1 : 0;
}
/*
** Reconstruct PDF file per the information in pageinfo.
** use_forms==0: Old-style reconstruction where the pages are not turned into XObject Forms.
** use_forms==1: New-style where pages are turned into XObject forms.
*/
int wmupdf_remake_pdf(char *infile,char *outfile,WPDFPAGEINFO *pageinfo,int use_forms,FILE *out)
{
pdf_document *xref;
fz_context *ctx;
fz_write_options fzopts;
char *password="";
int status;
int write_failed;
fzopts.do_garbage=1; /* 2 and 3 don't work for this. */
fzopts.do_expand=0;
fzopts.do_ascii=0;
fzopts.do_linear=0;
write_failed=0;
wpdfpageinfo_sort(pageinfo);
xref=NULL;
/* New context */
ctx = fz_new_context(NULL,NULL,FZ_STORE_UNLIMITED);
if (!ctx)
{
nprintf(out,"wmupdf_remake_pdf: Cannot initialize context.\n");
return(-1);
}
fz_try(ctx)
{
xref=pdf_open_document_no_run(ctx,infile);
if (!xref)
{
fz_free_context(ctx);
nprintf(out,"wmupdf_remake_pdf: Cannot open PDF file %s.\n",infile);
return(-2);
}
if (pdf_needs_password(xref) && !pdf_authenticate_password(xref,password))
{
pdf_close_document(xref);
fz_free_context(ctx);
nprintf(out,"wmupdf_remake_pdf: Cannot authenticate PDF file %s.\n",infile);
return(-3);
}
status=wmupdf_pdfdoc_newpages(xref,ctx,pageinfo,use_forms,out);
if (status<0)
{
pdf_close_document(xref);
fz_free_context(ctx);
nprintf(out,"wmupdf_remake_pdf: Error re-paginating PDF file %s.\n",infile);
return(status);
}
info_update(ctx,xref,pageinfo->producer);
/* Write output */
pdf_write_document(xref,outfile,&fzopts);
}
fz_always(ctx)
{
pdf_close_document(xref);
}
fz_catch(ctx)
{
write_failed=1;
}
fz_free_context(ctx);
if (write_failed)
{
nprintf(out,"wmupdf_remake_pdf: Error writing output PDF file %s.\n",outfile);
return(-10);
}
return(0);
}