diff options
Diffstat (limited to 'test')
| -rw-r--r-- | test/pdfclean.c | 526 | ||||
| -rw-r--r-- | test/pdfdebug.c | 152 |
2 files changed, 89 insertions, 589 deletions
diff --git a/test/pdfclean.c b/test/pdfclean.c index 00721842..7eb72cda 100644 --- a/test/pdfclean.c +++ b/test/pdfclean.c @@ -1,30 +1,6 @@ #include <fitz.h> #include <mupdf.h> -#define encrypt encrypt3am - -int (*printobj)(FILE*,fz_obj*) = fz_fprintobj; - -/* - * Rewrite PDF with cleaned up syntax, and consolidate the xref table. - * Remove encryption while we're at it :) - */ - -static FILE *out; - -static pdf_xref *xt; - -static int *ofslist; -static int *genlist; -static int *uselist; - -static int dorebuild = 0; -static int doexpand = 0; -static pdf_crypt *encrypt = nil; -static fz_obj *encryptdict = nil; -static fz_obj *id = nil; -static int dogc = 0; - void usage() { fprintf(stderr, @@ -32,469 +8,129 @@ void usage() " -r\treconstruct broken xref table\n" " -g\tgarbage collect unused objects\n" " -x\texpand compressed streams\n" - " -c\twrite compact objects\n" " -d -\tset user password for decryption\n" " -e\tencrypt outfile\n" - " -u -\tset user password for encryption\n" - " -o -\tset owner password\n" - " -p -\tset permissions\n" - " -n -\tkey length in bits: 40 <= n <= 128\n" + " -u -\tset user password for encryption\n" + " -o -\tset owner password\n" + " -p -\tset permissions\n" + " -n -\tkey length in bits: 40 <= n <= 128\n" ); exit(1); } -void garbagecollect(fz_obj *ref); - -void gc0(fz_obj *obj) -{ - int i; - - if (fz_isdict(obj)) - for (i = 0; i < fz_dictlen(obj); i++) - gc0(fz_dictgetval(obj, i)); - - if (fz_isarray(obj)) - for (i = 0; i < fz_arraylen(obj); i++) - gc0(fz_arrayget(obj, i)); - - if (fz_isindirect(obj)) - garbagecollect(obj); -} - -void garbagecollect(fz_obj *ref) -{ - fz_obj *obj; - int stmofs; - - if (uselist[fz_toobjid(ref)]) - return; - - uselist[fz_toobjid(ref)] = 1; - - pdf_loadindirect(&obj, xt, ref, &stmofs); - - if (stmofs != -1) { - fz_obj *len = fz_dictgets(obj, "Length"); - if (fz_isindirect(len)) { - pdf_loadindirect(&len, xt, len, nil); - fz_dictputs(obj, "Length", len); - } - } - - gc0(obj); -} - -void decodestream(fz_obj *obj, int ofs, int oid, int gid) -{ - fz_error *error; - unsigned char buf[512]; - fz_filter *filter; - fz_file *sf; - int n; - int len; - fz_obj *lenobj; - fz_obj *newdict; - - /* count length of decoded data */ - len = 0; - - error = pdf_buildstream(&filter, xt, obj, oid, gid); - if (error) fz_abort(error); - - n = fz_seek(xt->file, ofs); - - fz_chainfile(&sf, xt->file, filter); - - while (1) { - n = fz_read(sf, buf, sizeof buf); - if (n < 0) - fz_abort(fz_ferror(sf)); - if (n == 0) - break; - len += n; - } - - fz_unchainfile(sf); - - /* change dictionary of object... */ - fz_copydict(&newdict, obj); - - fz_newint(&lenobj, len); - fz_dictputs(newdict, "Length", lenobj); - fz_dropobj(lenobj); - - fz_dictdels(newdict, "Filter"); - fz_dictdels(newdict, "DecodeParms"); - - /* save object */ - fprintf(out, "%d %d obj\n", oid, gid); - printobj(out, newdict); - fprintf(out, "\n"); - fprintf(out, "stream\n"); - - fz_dropobj(newdict); - - /* now decode stream for real */ - error = pdf_buildstream(&filter, xt, obj, oid, gid); - if (error) fz_abort(error); - - fz_seek(xt->file, ofs); - - if (encrypt) { - fz_filter *cf; - pdf_cryptstm(&cf, encrypt, oid, gid); - fz_newpipeline(&filter, filter, cf); - } - - fz_chainfile(&sf, xt->file, filter); - while (1) { - n = fz_read(sf, buf, sizeof buf); - if (n < 0) - fz_abort(fz_ferror(sf)); - if (n == 0) - break; - fwrite(buf, 1, n, out); - } - fz_unchainfile(sf); - - /* the end */ - fprintf(out, "endstream\nendobj\n\n"); - - return; -} - -void savestream(fz_obj *obj, int ofs, int oid, int gid) -{ - unsigned char buf[512]; - fz_filter *filter; - fz_file *sf; - int len; - int n; - - /* save object */ - fprintf(out, "%d %d obj\n", oid, gid); - printobj(out, obj); - fprintf(out, "\n"); - fprintf(out, "stream\n"); - - /* copy stream */ - obj = fz_dictgets(obj, "Length"); - if (fz_isindirect(obj)) { - pdf_loadindirect(&obj, xt, obj, nil); - len = fz_toint(obj); - fz_dropobj(obj); - } - else { - len = fz_toint(obj); - } - - fz_newnullfilter(&filter, len); - - if (xt->crypt) { - fz_filter *cf; - pdf_cryptstm(&cf, xt->crypt, oid, gid); - fz_newpipeline(&filter, cf, filter); - } - - if (encrypt) { - fz_filter *cf; - pdf_cryptstm(&cf, encrypt, oid, gid); - fz_newpipeline(&filter, filter, cf); - } - - fz_seek(xt->file, ofs); - fz_chainfile(&sf, xt->file, filter); - while (1) - { - n = fz_read(sf, buf, sizeof buf); - if (n == 0) - break; - if (n < 0) - fz_abort(fz_ferror(sf)); - fwrite(buf, 1, n, out); - } - fz_unchainfile(sf); - - /* the end */ - fprintf(out, "endstream\nendobj\n\n"); -} - -void deleteobject(int oid, int gid) -{ - uselist[oid] = 0; -} - -void saveobject(int oid, int gid) +void expandstreams(pdf_xref *xref) { fz_error *error; - fz_obj *obj; - fz_obj *t; + fz_obj *stmobj; int stmofs; + fz_buffer *buf; + fz_obj *stmlen; + int i, gen; - error = pdf_loadobj(&obj, xt, oid, gid, &stmofs); - if (error) fz_abort(error); - - /* trash ObjStm and XrefStm objects */ - if (fz_isdict(obj)) { - t = fz_dictgets(obj, "Type"); - if (fz_isname(t) && strcmp(fz_toname(t), "ObjStm") == 0) { - deleteobject(oid, gid); - fz_dropobj(obj); - return; - } - if (fz_isname(t) && strcmp(fz_toname(t), "XRef") == 0) { - deleteobject(oid, gid); - fz_dropobj(obj); - return; + for (i = 0; i < xref->size; i++) + { + if (xref->table[i].type == 'n') + { + gen = xref->table[i].gen; + + error = pdf_loadobject0(&stmobj, xref, i, gen, &stmofs); + if (error) fz_abort(error); + + if (stmofs != -1) + { + error = pdf_readstream0(&buf, xref, stmobj, i, gen, stmofs); + if (error) fz_abort(error); + + fz_dictdels(stmobj, "Filter"); + fz_dictdels(stmobj, "DecodeParms"); + + error = fz_newint(&stmlen, buf->wp - buf->rp); + if (error) fz_abort(error); + error = fz_dictputs(stmobj, "Length", stmlen); + if (error) fz_abort(error); + fz_dropobj(stmlen); + + error = pdf_saveobject(xref, i, gen, stmobj); + if (error) fz_abort(error); + error = pdf_savestream(xref, i, gen, buf); + if (error) fz_abort(error); + } } } - - if (encrypt) - pdf_cryptobj(encrypt, obj, oid, gid); - - if (stmofs == -1) { - fprintf(out, "%d %d obj\n", oid, gid); - printobj(out, obj); - fprintf(out, "\nendobj\n\n"); - } - else if (doexpand) { - decodestream(obj, stmofs, oid, gid); - } - else { - savestream(obj, stmofs, oid, gid); - } - - fz_dropobj(obj); -} - -void savexref(void) -{ - fz_obj *newtrailer; - fz_obj *obj; - int startxref; - int i; - - startxref = ftell(out); - - fprintf(out, "xref\n0 %d\n", xt->size); - for (i = 0; i < xt->size; i++) { - if (uselist[i]) - fprintf(out, "%010d %05d n \n", ofslist[i], genlist[i]); - else - fprintf(out, "%010d %05d f \n", ofslist[i], genlist[i]); - } - fprintf(out, "\n"); - - fz_newdict(&newtrailer, 5); - - fz_newint(&obj, xt->size); - fz_dictputs(newtrailer, "Size", obj); - fz_dropobj(obj); - - obj = fz_dictgets(xt->trailer, "Info"); - if (obj) fz_dictputs(newtrailer, "Info", obj); - - obj = fz_dictgets(xt->trailer, "Root"); - if (obj) fz_dictputs(newtrailer, "Root", obj); - - fz_dictputs(newtrailer, "ID", id); - - if (encryptdict) { - fz_newindirect(&obj, xt->size - 1, 0); - fz_dictputs(newtrailer, "Encrypt", obj); - fz_dropobj(obj); - } - - fprintf(out, "trailer\n"); - printobj(out, newtrailer); - fprintf(out, "\n\n"); - - fprintf(out, "startxref\n%d\n%%%%EOF\n", startxref); } int main(int argc, char **argv) { fz_error *error; - fz_obj *obj; - int lastfree; - char *filename; - int i; + char *infile; + char *outfile; + pdf_xref *xref; int c; int doencrypt = 0; - char *password = ""; + int dorepair = 0; + int doexpand = 0; + int dogc = 0; + char *userpw = ""; char *ownerpw = ""; int perms = -4; /* 0xfffffffc */ int keylen = 40; + char *password = ""; - while (1) + while ((c = getopt(argc, argv, "rgxd:eu:o:p:n:")) != -1) { - c = getopt(argc, argv, "rcxgeu:o:p:n:d:"); - - if (c == -1) - break; - switch (c) { - case 'r': - dorebuild ++; - break; - case 'x': - doexpand ++; - break; - case 'g': - dogc ++; - break; - case 'c': - printobj = fz_fprintcobj; - break; - case 'd': - password = optarg; - break; - case 'e': - doencrypt ++; - break; - case 'u': - userpw = optarg; - break; - case 'o': - ownerpw = optarg; - break; - case 'p': - perms = atoi(optarg); - break; - case 'n': - keylen = atoi(optarg); - break; - default: - usage(); + case 'r': ++ dorepair; break; + case 'x': ++ doexpand; break; + case 'g': ++ dogc; break; + case 'e': ++ doencrypt; break; + case 'u': userpw = optarg; break; + case 'o': ownerpw = optarg; break; + case 'p': perms = atoi(optarg); break; + case 'n': keylen = atoi(optarg); break; + case 'd': password = optarg; break; + default: usage(); } } - if (argc - optind != 2) + if (argc - optind < 2) usage(); - filename = argv[optind]; + infile = argv[optind++]; + outfile = argv[optind++]; - if (dorebuild) - error = pdf_rebuildxref(&xt, filename); - else - error = pdf_openxref(&xt, filename); - if (error) fz_abort(error); - - if (doencrypt && keylen > 40 && xt->version < 1.4) - xt->version = 1.4; - - id = fz_dictgets(xt->trailer, "ID"); - obj = fz_dictgets(xt->trailer, "Encrypt"); - if (fz_isindirect(obj)) { - pdf_loadindirect(&obj, xt, obj, nil); - } - if (obj && id) { - pdf_newdecrypt(&xt->crypt, obj, id); - error = pdf_setpassword(xt->crypt, password); - if (error) fz_abort(error); - } - - id = fz_dictgets(xt->trailer, "ID"); - if (!id) - fz_parseobj(&id, "[ (foobar) (foobar) ]"); - - if (doencrypt) - pdf_newencrypt(&encrypt, &encryptdict, userpw, ownerpw, perms, keylen, id); - - out = fopen(argv[optind + 1], "w"); - if (!out) { - fz_abort(fz_throw("open(%s): %s", argv[optind + 1], strerror(errno))); - } - - fprintf(out, "%%PDF-%.1f\n\n", xt->version); - - ofslist = fz_malloc(sizeof(int) * (xt->size + 1)); - genlist = fz_malloc(sizeof(int) * (xt->size + 1)); - uselist = fz_malloc(sizeof(int) * (xt->size + 1)); - - lastfree = 0; - - ofslist[0] = 0; - genlist[0] = 65535; - uselist[0] = 0; - - for (i = 1; i < xt->size; i++) { - ofslist[i] = 0; - genlist[i] = 0; - uselist[i] = 1; - } + error = pdf_newxref(&xref); + if (error) + fz_abort(error); - /* garbage collect from roots in trailer */ - if (dogc) - { - for (i = 1; i < xt->size; i++) - uselist[i] = 0; - - obj = fz_dictgets(xt->trailer, "Info"); - if (fz_isindirect(obj)) - garbagecollect(obj); - - obj = fz_dictgets(xt->trailer, "Root"); - if (fz_isindirect(obj)) - garbagecollect(obj); + if (dorepair) + error = pdf_repairxref(xref, infile); + else + error = pdf_openxref(xref, infile); + if (error) + fz_abort(error); - obj = fz_dictgets(xt->trailer, "ID"); - if (fz_isindirect(obj)) - garbagecollect(obj); - } + error = pdf_decryptxref(xref); + if (error) + fz_abort(error); - /* pretty print objects */ - for (i = 0; i < xt->size; i++) + if (xref->crypt) { - if (xt->table[i].type == 0) - uselist[i] = 0; - - if (xt->table[i].type == 0) - genlist[i] = xt->table[i].gen; - if (xt->table[i].type == 1) - genlist[i] = xt->table[i].gen; - if (xt->table[i].type == 2) - genlist[i] = 0; - - if (dogc && !uselist[i]) - continue; - - if (xt->table[i].type == 1 || xt->table[i].type == 2) - { - ofslist[i] = ftell(out); - saveobject(i, genlist[i]); - } - } - - /* add encryption dictionary if we crypted */ - if (encryptdict) { - xt->size ++; - ofslist[xt->size - 1] = ftell(out); - genlist[xt->size - 1] = 0; - uselist[xt->size - 1] = 1; - fprintf(out, "%d %d obj\n", xt->size - 1, 0); - printobj(out, encryptdict); - fprintf(out, "\nendobj\n\n"); - } - - /* construct linked list of free object slots */ - lastfree = 0; - for (i = 1; i < xt->size; i++) { - if (!uselist[i]) { - genlist[i] ++; - ofslist[lastfree] = i; - lastfree = i; - } + error = pdf_setpassword(xref->crypt, password); + if (error) fz_abort(error); } - savexref(); + if (doexpand) + expandstreams(xref); - pdf_closexref(xt); +printf("saving %s...\n", outfile); + error = pdf_savepdf(xref, outfile); + if (error) + fz_abort(error); - fclose(out); + pdf_closexref(xref); return 0; } diff --git a/test/pdfdebug.c b/test/pdfdebug.c index d63a720c..ddc10dff 100644 --- a/test/pdfdebug.c +++ b/test/pdfdebug.c @@ -5,11 +5,10 @@ static char *password = ""; static int dodecode = 0; static int dorepair = 0; static int doprintxref = 0; -static int doprintpages = 0; void usage() { - fprintf(stderr, "usage: mupdf [-drxp] [-u password] file.pdf\n"); + fprintf(stderr, "usage: pdfdebug [-drx] [-u password] file.pdf [oid ...]\n"); exit(1); } @@ -127,114 +126,11 @@ void printobject(pdf_xref *xref, int oid, int gid) fz_dropobj(obj); } -/* - * Draw page - */ - -void runcsi(pdf_xref *xref, pdf_csi *csi, pdf_resources *rdb, fz_obj *stmref) -{ - fz_error *error; - - error = pdf_openstream(xref, stmref); - if (error) fz_abort(error); - - error = pdf_runcsi(csi, rdb, xref->file); - if (error) fz_abort(error); - - pdf_closestream(xref); -} - -void showpage(pdf_xref *xref, fz_obj *page) -{ - fz_error *error; - pdf_csi *csi; - pdf_resources *rdb = nil; - fz_obj *resources; - fz_obj *contents; - int i; - - fz_debugobj(page); - printf("\n"); - - resources = fz_dictgets(page, "Resources"); - if (resources) - { - error = pdf_resolve(&resources, xref); - if (error) fz_abort(error); - - error = pdf_loadresources(&rdb, xref, resources); - if (error) fz_abort(error); - - // parse resources into native res dict - fz_dropobj(resources); - } - else - fz_abort(fz_throw("syntaxerror: missing resource dictionary")); - -printf("resources:\n"); -printf(" font:\n"); -fz_debugobj(rdb->font); -printf("\n extgstate:\n"); -fz_debugobj(rdb->extgstate); -printf("\nfitz tree:\n"); - - error = pdf_newcsi(&csi); - if (error) fz_abort(error); - - contents = fz_dictgets(page, "Contents"); - if (contents) - { - if (fz_isarray(contents)) - { - for (i = 0; i < fz_arraylen(contents); i++) - { - runcsi(xref, csi, rdb, fz_arrayget(contents, i)); - } - } - else - { - // XXX resolve and check if it is an array - runcsi(xref, csi, rdb, contents); - } - } - - fz_debugtree(csi->tree); - - { - fz_pixmap *pix; - fz_renderer *gc; - fz_matrix ctm; - -#define W 612 -#define H 792 - -#define xW 1106 -#define xH 1548 - - fz_newrenderer(&gc); - fz_newpixmap(&pix, 0, 0, W, H, 1, 0); - ctm = fz_concat(fz_translate(0, -H), fz_scale(1,-1)); - - memset(pix->samples, 0x00, pix->stride * pix->h * 2); - -printf("rendering!\n"); - fz_rendernode(gc, csi->tree->root, ctm, pix); -printf("done!\n"); - fz_debugpixmap(pix); - - fz_freepixmap(pix); - fz_freerenderer(gc); - } - - pdf_freecsi(csi); -} - int main(int argc, char **argv) { fz_error *error; char *filename; pdf_xref *xref; - pdf_pagetree *pages; int c; while ((c = getopt(argc, argv, "drxopu:")) != -1) @@ -250,9 +146,6 @@ int main(int argc, char **argv) case 'x': doprintxref ++; break; - case 'p': - doprintpages ++; - break; case 'u': password = optarg; break; @@ -290,48 +183,19 @@ int main(int argc, char **argv) if (doprintxref) pdf_debugxref(xref); - if (doprintpages) + if (optind == argc) { - error = pdf_loadpagetree(&pages, xref); - if (error) fz_abort(error); - - if (optind == argc) - { - printf("pagetree\n"); - pdf_debugpagetree(pages); - printf("\n"); - } - else - { - for ( ; optind < argc; optind++) - { - int page = atoi(argv[optind]); - if (page < 1 || page > pages->count) - fprintf(stderr, "page out of bounds: %d\n", page); - printf("page %d\n", page); - showpage(xref, pages->pobj[page - 1]); - } - } + printf("trailer\n"); + fz_debugobj(xref->trailer); + printf("\n"); } - else + for ( ; optind < argc; optind++) { - if (optind == argc) - { - printf("trailer\n"); - fz_debugobj(xref->trailer); - printf("\n"); - } - - for ( ; optind < argc; optind++) - { - printobject(xref, atoi(argv[optind]), 0); - printf("\n"); - } + printobject(xref, atoi(argv[optind]), 0); + printf("\n"); } -printf("done.\n"); - pdf_closexref(xref); return 0; |