static void renumberobjs(void)
{
pdf_xrefentry *oldxref;
int newlen;
int num;
/* Apply renumber map to indirect references in all objects in xref */
renumberobj(xref->trailer);
for (num = 0; num < xref->len; num++)
{
fz_obj *obj = xref->table[num].obj;
if (fz_isindirect(obj))
{
obj = fz_newindirect(renumbermap[fz_tonum(obj)], 0, xref);
pdf_updateobject(xref, num, 0, obj);
fz_dropobj(obj);
}
else
{
renumberobj(obj);
}
}
/* Create new table for the reordered, compacted xref */
oldxref = xref->table;
xref->table = fz_calloc(xref->len, sizeof(pdf_xrefentry));
xref->table[0] = oldxref[0];
/* Move used objects into the new compacted xref */
newlen = 0;
for (num = 1; num < xref->len; num++)
{
if (uselist[num])
{
if (newlen < renumbermap[num])
newlen = renumbermap[num];
xref->table[renumbermap[num]] = oldxref[num];
}
else
{
if (oldxref[num].obj)
fz_dropobj(oldxref[num].obj);
}
}
fz_free(oldxref);
/* Update the used objects count in compacted xref */
xref->len = newlen + 1;
/* Update list of used objects to fit with compacted xref */
for (num = 1; num < xref->len; num++)
uselist[num] = 1;
}
static void renumberobj(fz_obj *obj)
{
int i;
if (fz_isdict(obj))
{
for (i = 0; i < fz_dictlen(obj); i++)
{
fz_obj *key = fz_dictgetkey(obj, i);
fz_obj *val = fz_dictgetval(obj, i);
if (fz_isindirect(val))
{
val = fz_newindirect(renumbermap[fz_tonum(val)], 0, xref);
fz_dictput(obj, key, val);
fz_dropobj(val);
}
else
{
renumberobj(val);
}
}
}
else if (fz_isarray(obj))
{
for (i = 0; i < fz_arraylen(obj); i++)
{
fz_obj *val = fz_arrayget(obj, i);
if (fz_isindirect(val))
{
val = fz_newindirect(renumbermap[fz_tonum(val)], 0, xref);
fz_arrayput(obj, i, val);
fz_dropobj(val);
}
else
{
renumberobj(val);
}
}
}
}
static fz_error parseobj(fz_obj **obj, pdf_xref *xref, char **sp, struct vap *v)
{
fz_error error;
char buf[32];
int num, gen, len;
char *tmp;
char *s = *sp;
if (*s == '\0')
return fz_throw("end of data");
skipwhite(&s);
error = fz_okay;
if (v != nil && *s == '%')
{
s ++;
switch (*s)
{
case 'o': *obj = fz_keepobj(va_arg(v->ap, fz_obj*)); break;
case 'b': error = fz_newbool(obj, va_arg(v->ap, int)); break;
case 'i': error = fz_newint(obj, va_arg(v->ap, int)); break;
case 'f': error = fz_newreal(obj, (float)va_arg(v->ap, double)); break;
case 'n': error = fz_newname(obj, va_arg(v->ap, char*)); break;
case 'r':
num = va_arg(v->ap, int);
gen = va_arg(v->ap, int);
error = fz_newindirect(obj, num, gen, xref);
break;
case 's':
tmp = va_arg(v->ap, char*);
error = fz_newstring(obj, tmp, strlen(tmp));
break;
case '#':
tmp = va_arg(v->ap, char*);
len = va_arg(v->ap, int);
error = fz_newstring(obj, tmp, len);
break;
default:
error = fz_throw("unknown format specifier in packobj: '%c'", *s);
break;
}
if (error)
error = fz_rethrow(error, "cannot create object for %% format");
s ++;
}
static void retainpages(int argc, char **argv)
{
fz_error error;
fz_obj *oldroot, *root, *pages, *kids, *countobj, *parent;
/* Load the old page tree */
error = pdf_loadpagetree(xref);
if (error)
die(fz_rethrow(error, "cannot load page tree"));
/* Keep only pages/type entry to avoid references to unretained pages */
oldroot = fz_dictgets(xref->trailer, "Root");
pages = fz_dictgets(oldroot, "Pages");
root = fz_newdict(2);
fz_dictputs(root, "Type", fz_dictgets(oldroot, "Type"));
fz_dictputs(root, "Pages", fz_dictgets(oldroot, "Pages"));
pdf_updateobject(xref, fz_tonum(oldroot), fz_togen(oldroot), root);
fz_dropobj(root);
/* Create a new kids array with only the pages we want to keep */
parent = fz_newindirect(fz_tonum(pages), fz_togen(pages), xref);
kids = fz_newarray(1);
/* Retain pages specified */
while (argc - fz_optind)
{
int page, spage, epage;
char *spec, *dash;
char *pagelist = argv[fz_optind];
spec = fz_strsep(&pagelist, ",");
while (spec)
{
dash = strchr(spec, '-');
if (dash == spec)
spage = epage = pdf_getpagecount(xref);
else
spage = epage = atoi(spec);
if (dash)
{
if (strlen(dash) > 1)
epage = atoi(dash + 1);
else
epage = pdf_getpagecount(xref);
}
if (spage > epage)
page = spage, spage = epage, epage = page;
if (spage < 1)
spage = 1;
if (epage > pdf_getpagecount(xref))
epage = pdf_getpagecount(xref);
for (page = spage; page <= epage; page++)
{
fz_obj *pageobj = pdf_getpageobject(xref, page);
fz_obj *pageref = pdf_getpageref(xref, page);
fz_dictputs(pageobj, "Parent", parent);
/* Store page object in new kids array */
fz_arraypush(kids, pageref);
}
spec = fz_strsep(&pagelist, ",");
}
fz_optind++;
}
fz_dropobj(parent);
/* Update page count and kids array */
countobj = fz_newint(fz_arraylen(kids));
fz_dictputs(pages, "Count", countobj);
fz_dropobj(countobj);
fz_dictputs(pages, "Kids", kids);
fz_dropobj(kids);
}
int
copyPdfFile(
soPdfFile* inFile,
soPdfFile* outFile
)
{
fz_error *error;
int pageTreeNum, pageTreeGen;
assert(inFile != NULL);
assert(outFile != NULL);
//
// Process every page in the source file
//
{
printf("\nProcessing input page : ");
for (int pageNo = 0; pageNo < pdf_getpagecount(inFile->pageTree); pageNo++)
{
displayPageNumber(pageNo + 1, !pageNo);
// Get the page object from the source
fz_obj *pageRef = inFile->pageTree->pref[pageNo];
fz_obj *pageObj = pdf_getpageobject(inFile->pageTree, pageNo);
//
// Process the page. Each page can be split into up-to 3 pages
//
fz_rect bbRect[3];
error = processPage(inFile, pageNo, bbRect, 3);
if (error)
return soPdfError(error);
for (int ctr = 0; ctr < 3; ctr++)
{
// Check if this was a blank page
if (fz_isemptyrect(bbRect[ctr]))
break;
//
// copy the source page dictionary entry. The way this is done is basically
// by making a copy of the page dict object in the source file, and adding
// the copy in the source file. Then the copied page dict object is
// referenced and added to the destination file.
//
// This convoluted procedure is done because the copy is done by pdf_transplant
// function that accepts a source and destination. Whatever is referenced by
// destination object is deep copied
//
// allocate an object id and generation id in source file
//
// There is a bug in mupdf where the object allocation returns
// 0 oid and 0 gid when the input pdf file has iref stream
// so to work around the issue, we wrap the pdf_allocojbect
// in a for loop 10 times to get the number
//
int sNum, sGen, tries;
for (tries = 0; tries < 10; tries++)
{
error = pdf_allocobject(inFile->xref, &sNum, &sGen);
if (error)
return soPdfError(error);
// If sNum is non zero then the allocation was successful
if (sNum != 0)
break;
pdf_updateobject(inFile->xref, sNum, sGen, pageObj);
}
// If we didn't succeed even after 10 tries then this file
// is not going to work.
if (tries >= 10)
return soPdfError(fz_throw("cannot allocate object because of mupdf bug"));
// make a deep copy of the original page dict
fz_obj *pageObj2;
error = fz_deepcopydict(&pageObj2, pageObj);
if (error)
return soPdfError(error);
// update the source file with the duplicate page object
pdf_updateobject(inFile->xref, sNum, sGen, pageObj2);
fz_dropobj(pageObj2);
// create an indirect reference to the page object
fz_obj *pageRef2;
error = fz_newindirect(&pageRef2, sNum, sGen);
if (error)
return soPdfError(error);
// delete the parent dictionary entry
// Do we need to delete any other dictionary entry
// like annot, tabs, metadata, etc
fz_dictdels(pageObj2, "Parent");
// Set the media box
setPageMediaBox(inFile->xref, pageObj2, bbRect[ctr]);
// Set the rotation based on input
switch(p_mode)
{
// no rotation if fit height
case FitHeight:
case Fit2xHeight:
break;
// rotate -90 deg if fit width
case Fit2xWidth:
case FitWidth:
setPageRotate(pageObj2, p_reverseLandscape ? 90 : -90);
break;
case SmartFitHeight:
case SmartFitWidth:
default:
return soPdfError(fz_throw("Mode(%d) not yet implemented.", p_mode));
break;
}
// push the indirect reference to the destination list for copy by pdf_transplant
error = fz_arraypush(outFile->editobjs, pageRef2);
if (error)
return soPdfError(error);
}
}
}
// flush the objects into destination from source
{
fz_obj *results;
int outPages;
printf("\nCopying output page : ");
error = pdf_transplant(outFile->xref, inFile->xref, &results, outFile->editobjs);
if (error)
return soPdfError(error);
outPages = fz_arraylen(results);
for (int ctr = 0; ctr < outPages; ctr++)
{
displayPageNumber(ctr + 1, !ctr);
error = fz_arraypush(outFile->pagelist, fz_arrayget(results,
p_reverseLandscape ? outPages - 1 - ctr : ctr));
if (error)
return soPdfError(error);
}
fz_dropobj(results);
}
// flush page tree
// Create page tree and add back-links
{
fz_obj *pageTreeObj;
fz_obj *pageTreeRef;
// allocate a new object in out file for pageTree object
error = pdf_allocobject(outFile->xref, &pageTreeNum, &pageTreeGen);
if (error)
return soPdfError(error);
// Create a page tree object
error = fz_packobj(&pageTreeObj, "<</Type/Pages/Count %i/Kids %o>>",
fz_arraylen(outFile->pagelist), outFile->pagelist);
if (error)
return soPdfError(error);
// Update the xref entry with the pageTree object
pdf_updateobject(outFile->xref, pageTreeNum, pageTreeGen, pageTreeObj);
fz_dropobj(pageTreeObj);
// Create a reference to the pageTree object
error = fz_newindirect(&pageTreeRef, pageTreeNum, pageTreeGen);
if (error)
return soPdfError(error);
//
// For every page in the output file, update the parent entry
//
for (int ctr = 0; ctr < fz_arraylen(outFile->pagelist); ctr++)
{
fz_obj *pageObj;
int num = fz_tonum(fz_arrayget(outFile->pagelist, ctr));
int gen = fz_togen(fz_arrayget(outFile->pagelist, ctr));
// Get the page object from xreft
error = pdf_loadobject(&pageObj, outFile->xref, num, gen);
if (error)
return soPdfError(error);
// Update the parent entry in the page dictionary
error = fz_dictputs(pageObj, "Parent", pageTreeRef);
if (error)
return soPdfError(error);
// Update the entry with the updated page object
pdf_updateobject(outFile->xref, num, gen, pageObj);
fz_dropobj(pageObj);
}
}
// Create catalog and root entries
{
fz_obj *catObj, *infoObj;
int rootNum, rootGen;
int infoNum, infoGen;
//
// Copy the info catalog to the destination
// alloc an object id and gen id in destination file
error = pdf_allocobject(outFile->xref, &infoNum, &infoGen);
if (error)
return soPdfError(error);
// make a deep copy of the original page dict
error = fz_deepcopydict(&infoObj, inFile->xref->info);
if (error)
return soPdfError(error);
// update the dest file with object
pdf_updateobject(outFile->xref, infoNum, infoGen, infoObj);
outFile->xref->info = infoObj;
fz_dropobj(infoObj);
//
// root/catalog object creation
error = pdf_allocobject(outFile->xref, &rootNum, &rootGen);
if (error)
return soPdfError(error);
error = fz_packobj(&catObj, "<</Type/Catalog /Pages %r>>", pageTreeNum, pageTreeGen);
if (error)
return soPdfError(error);
pdf_updateobject(outFile->xref, rootNum, rootGen, catObj);
fz_dropobj(catObj);
// Create trailer
error = fz_packobj(&outFile->xref->trailer, "<</Root %r /Info %r>>",
rootNum, rootGen, infoNum, infoGen);
if (error)
return soPdfError(error);
}
// Update the info in the target file and save the xref
printf("\nSaving.\n");
error = setPageInfo(inFile, outFile);
if (error)
return soPdfError(error);
error = pdf_savexref(outFile->xref, outFile->fileName, NULL);
if (error)
return soPdfError(error);
if (g_errorCount != 0)
{
printf("\nFollowing issues encounted were ignored.\n\n");
for (int ctr = g_errorCount - 1; ctr >= 0; ctr--)
soPdfError(g_errorList[ctr]);
}
printf("\nSaved.\n");
return 0;
}
fz_error
pdf_parseindobj(fz_obj **op, pdf_xref *xref,
fz_stream *file, char *buf, int cap,
int *onum, int *ogen, int *ostmofs)
{
fz_error error = fz_okay;
fz_obj *obj = nil;
int num = 0, gen = 0, stmofs;
pdf_token_e tok;
int len;
int a, b;
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
if (tok != PDF_TINT)
return fz_throw("cannot parse indirect object (%d %d R)", num, gen);
num = atoi(buf);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
if (tok != PDF_TINT)
return fz_throw("cannot parse indirect object (%d %d R)", num, gen);
gen = atoi(buf);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
if (tok != PDF_TOBJ)
return fz_throw("cannot parse indirect object (%d %d R)", num, gen);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
switch (tok)
{
case PDF_TOARRAY:
error = pdf_parsearray(&obj, xref, file, buf, cap);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
break;
case PDF_TODICT:
error = pdf_parsedict(&obj, xref, file, buf, cap);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
break;
case PDF_TNAME: obj = fz_newname(buf); break;
case PDF_TREAL: obj = fz_newreal(atof(buf)); break;
case PDF_TSTRING: obj = fz_newstring(buf, len); break;
case PDF_TTRUE: obj = fz_newbool(1); break;
case PDF_TFALSE: obj = fz_newbool(0); break;
case PDF_TNULL: obj = fz_newnull(); break;
case PDF_TINT:
a = atoi(buf);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
if (tok == PDF_TSTREAM || tok == PDF_TENDOBJ)
{
obj = fz_newint(a);
goto skip;
}
if (tok == PDF_TINT)
{
b = atoi(buf);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
if (tok == PDF_TR)
{
obj = fz_newindirect(a, b, xref);
break;
}
}
return fz_throw("cannot parse indirect object (%d %d R)", num, gen);
case PDF_TENDOBJ:
obj = fz_newnull();
goto skip;
default:
return fz_throw("cannot parse indirect object (%d %d R)", num, gen);
}
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
{
fz_dropobj(obj);
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
}
skip:
if (tok == PDF_TSTREAM)
{
int c = fz_readbyte(file);
while (c == ' ')
c = fz_readbyte(file);
if (c == '\r')
{
c = fz_peekbyte(file);
if (c != '\n')
fz_warn("line feed missing after stream begin marker (%d %d R)", num, gen);
else
fz_readbyte(file);
}
stmofs = fz_tell(file);
}
else if (tok == PDF_TENDOBJ)
{
stmofs = 0;
}
else
{
fz_warn("expected endobj or stream keyword (%d %d R)", num, gen);
stmofs = 0;
}
if (onum) *onum = num;
if (ogen) *ogen = gen;
if (ostmofs) *ostmofs = stmofs;
*op = obj;
return fz_okay;
}
fz_error
pdf_parsedict(fz_obj **op, pdf_xref *xref, fz_stream *file, char *buf, int cap)
{
fz_error error = fz_okay;
fz_obj *dict = nil;
fz_obj *key = nil;
fz_obj *val = nil;
pdf_token_e tok;
int len;
int a, b;
dict = fz_newdict(8);
while (1)
{
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
{
fz_dropobj(dict);
return fz_rethrow(error, "cannot parse dict");
}
skip:
if (tok == PDF_TCDICT)
{
*op = dict;
return fz_okay;
}
/* for BI .. ID .. EI in content streams */
if (tok == PDF_TKEYWORD && !strcmp(buf, "ID"))
{
*op = dict;
return fz_okay;
}
if (tok != PDF_TNAME)
{
fz_dropobj(dict);
return fz_throw("invalid key in dict");;
}
key = fz_newname(buf);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
{
fz_dropobj(dict);
return fz_rethrow(error, "cannot parse dict");
}
switch (tok)
{
case PDF_TOARRAY:
error = pdf_parsearray(&val, xref, file, buf, cap);
if (error)
{
fz_dropobj(key);
fz_dropobj(dict);
return fz_rethrow(error, "cannot parse dict");
}
break;
case PDF_TODICT:
error = pdf_parsedict(&val, xref, file, buf, cap);
if (error)
{
fz_dropobj(key);
fz_dropobj(dict);
return fz_rethrow(error, "cannot parse dict");
}
break;
case PDF_TNAME: val = fz_newname(buf); break;
case PDF_TREAL: val = fz_newreal(atof(buf)); break;
case PDF_TSTRING: val = fz_newstring(buf, len); break;
case PDF_TTRUE: val = fz_newbool(1); break;
case PDF_TFALSE: val = fz_newbool(0); break;
case PDF_TNULL: val = fz_newnull(); break;
case PDF_TINT:
/* 64-bit to allow for numbers > INT_MAX and overflow */
a = (int) strtoll(buf, 0, 10);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
{
fz_dropobj(key);
fz_dropobj(dict);
return fz_rethrow(error, "cannot parse dict");
}
if (tok == PDF_TCDICT || tok == PDF_TNAME ||
(tok == PDF_TKEYWORD && !strcmp(buf, "ID")))
{
val = fz_newint(a);
fz_dictput(dict, key, val);
fz_dropobj(val);
fz_dropobj(key);
goto skip;
}
if (tok == PDF_TINT)
{
b = atoi(buf);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
{
fz_dropobj(key);
fz_dropobj(dict);
return fz_rethrow(error, "cannot parse dict");
}
if (tok == PDF_TR)
{
val = fz_newindirect(a, b, xref);
break;
}
}
fz_dropobj(key);
fz_dropobj(dict);
return fz_throw("invalid indirect reference in dict");
default:
return fz_throw("unknown token in dict");
}
fz_dictput(dict, key, val);
fz_dropobj(val);
fz_dropobj(key);
}
}
fz_error
pdf_parsearray(fz_obj **op, pdf_xref *xref, fz_stream *file, char *buf, int cap)
{
fz_error error = fz_okay;
fz_obj *ary = nil;
fz_obj *obj = nil;
int a = 0, b = 0, n = 0;
pdf_token_e tok;
int len;
ary = fz_newarray(4);
while (1)
{
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
{
fz_dropobj(ary);
return fz_rethrow(error, "cannot parse array");
}
if (tok != PDF_TINT && tok != PDF_TR)
{
if (n > 0)
{
obj = fz_newint(a);
fz_arraypush(ary, obj);
fz_dropobj(obj);
}
if (n > 1)
{
obj = fz_newint(b);
fz_arraypush(ary, obj);
fz_dropobj(obj);
}
n = 0;
}
if (tok == PDF_TINT && n == 2)
{
obj = fz_newint(a);
fz_arraypush(ary, obj);
fz_dropobj(obj);
a = b;
n --;
}
switch (tok)
{
case PDF_TCARRAY:
*op = ary;
return fz_okay;
case PDF_TINT:
if (n == 0)
a = atoi(buf);
if (n == 1)
b = atoi(buf);
n ++;
break;
case PDF_TR:
if (n != 2)
{
fz_dropobj(ary);
return fz_throw("cannot parse indirect reference in array");
}
obj = fz_newindirect(a, b, xref);
fz_arraypush(ary, obj);
fz_dropobj(obj);
n = 0;
break;
case PDF_TOARRAY:
error = pdf_parsearray(&obj, xref, file, buf, cap);
if (error)
{
fz_dropobj(ary);
return fz_rethrow(error, "cannot parse array");
}
fz_arraypush(ary, obj);
fz_dropobj(obj);
break;
case PDF_TODICT:
error = pdf_parsedict(&obj, xref, file, buf, cap);
if (error)
{
fz_dropobj(ary);
return fz_rethrow(error, "cannot parse array");
}
fz_arraypush(ary, obj);
fz_dropobj(obj);
break;
case PDF_TNAME:
obj = fz_newname(buf);
fz_arraypush(ary, obj);
fz_dropobj(obj);
break;
case PDF_TREAL:
obj = fz_newreal(atof(buf));
fz_arraypush(ary, obj);
fz_dropobj(obj);
break;
case PDF_TSTRING:
obj = fz_newstring(buf, len);
fz_arraypush(ary, obj);
fz_dropobj(obj);
break;
case PDF_TTRUE:
obj = fz_newbool(1);
fz_arraypush(ary, obj);
fz_dropobj(obj);
break;
case PDF_TFALSE:
obj = fz_newbool(0);
fz_arraypush(ary, obj);
fz_dropobj(obj);
break;
case PDF_TNULL:
obj = fz_newnull();
fz_arraypush(ary, obj);
fz_dropobj(obj);
break;
default:
fz_dropobj(ary);
return fz_throw("cannot parse token in array");
}
}
}
static void saveimage(fz_obj *obj, int num, int gen)
{
pdf_image *img = nil;
fz_obj *ref;
fz_error error;
fz_pixmap *pix;
char name[1024];
FILE *f;
int bpc;
int w;
int h;
int n;
int x;
int y;
error = fz_newindirect(&ref, num, gen, xref);
if (error)
die(error);
error = pdf_newstore(&xref->store);
if (error)
die(error);
error = pdf_loadimage(&img, xref, ref);
if (error)
die(error);
n = img->super.n;
w = img->super.w;
h = img->super.h;
bpc = img->bpc;
error = fz_newpixmap(&pix, 0, 0, w, h, n + 1);
if (error)
die(error);
error = img->super.loadtile(&img->super, pix);
if (error)
die(error);
if (bpc == 1 && n == 0)
{
fz_pixmap *temp;
error = fz_newpixmap(&temp, pix->x, pix->y, pix->w, pix->h, pdf_devicergb->n + 1);
if (error)
die(error);
for (y = 0; y < pix->h; y++)
for (x = 0; x < pix->w; x++)
{
int pixel = y * pix->w + x;
temp->samples[pixel * temp->n + 0] = 255;
temp->samples[pixel * temp->n + 1] = pix->samples[pixel];
temp->samples[pixel * temp->n + 2] = pix->samples[pixel];
temp->samples[pixel * temp->n + 3] = pix->samples[pixel];
}
fz_droppixmap(pix);
pix = temp;
}
if (img->super.cs && strcmp(img->super.cs->name, "DeviceRGB"))
{
fz_pixmap *temp;
error = fz_newpixmap(&temp, pix->x, pix->y, pix->w, pix->h, pdf_devicergb->n + 1);
if (error)
die(error);
fz_convertpixmap(img->super.cs, pix, pdf_devicergb, temp);
fz_droppixmap(pix);
pix = temp;
}
sprintf(name, "img-%04d.pnm", num);
f = fopen(name, "wb");
if (f == NULL)
die(fz_throw("Error creating image file"));
fprintf(f, "P6\n%d %d\n%d\n", w, h, 255);
for (y = 0; y < pix->h; y++)
for (x = 0; x < pix->w; x++)
{
fz_sample *sample = &pix->samples[(y * pix->w + x) * (pdf_devicergb->n + 1)];
unsigned char r = sample[1];
unsigned char g = sample[2];
unsigned char b = sample[3];
fprintf(f, "%c%c%c", r, g, b);
}
if (fclose(f) < 0)
die(fz_throw("Error closing image file"));
fz_droppixmap(pix);
pdf_dropstore(xref->store);
xref->store = nil;
fz_dropimage(&img->super);
fz_dropobj(ref);
}
void
editflushcatalog(void)
{
fz_error *error;
int rootnum, rootgen;
int listnum, listgen;
fz_obj *listref;
fz_obj *obj;
int i;
/* Create page tree and add back-links */
error = pdf_allocobject(editxref, &listnum, &listgen);
if (error)
die(error);
error = fz_packobj(&obj, "<</Type/Pages/Count %i/Kids %o>>",
fz_arraylen(editpagelist),
editpagelist);
if (error)
die(error);
pdf_updateobject(editxref, listnum, listgen, obj);
fz_dropobj(obj);
error = fz_newindirect(&listref, listnum, listgen);
if (error)
die(error);
for (i = 0; i < fz_arraylen(editpagelist); i++)
{
int num = fz_tonum(fz_arrayget(editpagelist, i));
int gen = fz_togen(fz_arrayget(editpagelist, i));
error = pdf_loadobject(&obj, editxref, num, gen);
if (error)
die(error);
error = fz_dictputs(obj, "Parent", listref);
if (error)
die(error);
pdf_updateobject(editxref, num, gen, obj);
fz_dropobj(obj);
}
/* Create catalog */
error = pdf_allocobject(editxref, &rootnum, &rootgen);
if (error)
die(error);
error = fz_packobj(&obj, "<</Type/Catalog/Pages %r>>", listnum, listgen);
if (error)
die(error);
pdf_updateobject(editxref, rootnum, rootgen, obj);
fz_dropobj(obj);
/* Create trailer */
error = fz_packobj(&editxref->trailer, "<</Root %r>>", rootnum, rootgen);
if (error)
die(error);
}
fz_error *
pdf_parseindobj(fz_obj **op, fz_stream *file, char *buf, int cap,
int *ooid, int *ogid, int *ostmofs)
{
fz_error *error = nil;
fz_obj *obj = nil;
int oid = 0, gid = 0, stmofs;
int tok, len;
int a, b;
tok = pdf_lex(file, buf, cap, &len);
if (tok != PDF_TINT)
goto cleanup;
oid = atoi(buf);
tok = pdf_lex(file, buf, cap, &len);
if (tok != PDF_TINT)
goto cleanup;
gid = atoi(buf);
tok = pdf_lex(file, buf, cap, &len);
if (tok != PDF_TOBJ)
goto cleanup;
tok = pdf_lex(file, buf, cap, &len);
switch (tok)
{
case PDF_TOARRAY: error = pdf_parsearray(&obj, file, buf, cap); break;
case PDF_TODICT: error = pdf_parsedict(&obj, file, buf, cap); break;
case PDF_TNAME: error = fz_newname(&obj, buf); break;
case PDF_TREAL: error = fz_newreal(&obj, atof(buf)); break;
case PDF_TSTRING: error = fz_newstring(&obj, buf, len); break;
case PDF_TTRUE: error = fz_newbool(&obj, 1); break;
case PDF_TFALSE: error = fz_newbool(&obj, 0); break;
case PDF_TNULL: error = fz_newnull(&obj); break;
case PDF_TINT:
a = atoi(buf);
tok = pdf_lex(file, buf, cap, &len);
if (tok == PDF_TSTREAM || tok == PDF_TENDOBJ)
{
error = fz_newint(&obj, a);
if (error) goto cleanup;
goto skip;
}
if (tok == PDF_TINT)
{
b = atoi(buf);
tok = pdf_lex(file, buf, cap, &len);
if (tok == PDF_TR)
{
error = fz_newindirect(&obj, a, b);
break;
}
}
goto cleanup;
default:
goto cleanup;
}
if (error) goto cleanup;
tok = pdf_lex(file, buf, cap, &len);
skip:
if (tok == PDF_TSTREAM)
{
int c = fz_readbyte(file);
if (c == '\r')
{
c = fz_peekbyte(file);
if (c != '\n')
fz_warn("syntaxerror: DOS format line ending after stream keyword (%d %d)\n", oid, gid);
else
c = fz_readbyte(file);
}
stmofs = fz_tell(file);
}
else if (tok == PDF_TENDOBJ)
stmofs = 0;
else
goto cleanup;
if (ooid) *ooid = oid;
if (ogid) *ogid = gid;
if (ostmofs) *ostmofs = stmofs;
*op = obj;
return nil;
cleanup:
if (obj) fz_dropobj(obj);
if (error) return error;
return fz_throw("syntaxerror: corrupt indirect object (%d %d)", oid, gid);
}
fz_error *
pdf_parsedict(fz_obj **op, fz_stream *file, char *buf, int cap)
{
fz_error *error = nil;
fz_obj *dict = nil;
fz_obj *key = nil;
fz_obj *val = nil;
int tok, len;
int a, b;
error = fz_newdict(op, 8);
if (error) return error;
dict = *op;
while (1)
{
tok = pdf_lex(file, buf, cap, &len);
skip:
if (tok == PDF_TCDICT)
return nil;
/* for BI .. ID .. EI in content streams */
if (tok == PDF_TKEYWORD && !strcmp(buf, "ID"))
return nil;
if (tok != PDF_TNAME)
goto cleanup;
error = fz_newname(&key, buf);
if (error) goto cleanup;
tok = pdf_lex(file, buf, cap, &len);
switch (tok)
{
case PDF_TOARRAY: error = pdf_parsearray(&val, file, buf, cap); break;
case PDF_TODICT: error = pdf_parsedict(&val, file, buf, cap); break;
case PDF_TNAME: error = fz_newname(&val, buf); break;
case PDF_TREAL: error = fz_newreal(&val, atof(buf)); break;
case PDF_TSTRING: error = fz_newstring(&val, buf, len); break;
case PDF_TTRUE: error = fz_newbool(&val, 1); break;
case PDF_TFALSE: error = fz_newbool(&val, 0); break;
case PDF_TNULL: error = fz_newnull(&val); break;
case PDF_TINT:
a = atoi(buf);
tok = pdf_lex(file, buf, cap, &len);
if (tok == PDF_TCDICT || tok == PDF_TNAME ||
(tok == PDF_TKEYWORD && !strcmp(buf, "ID")))
{
error = fz_newint(&val, a);
if (error) goto cleanup;
error = fz_dictput(dict, key, val);
if (error) goto cleanup;
fz_dropobj(val);
fz_dropobj(key);
key = val = nil;
goto skip;
}
if (tok == PDF_TINT)
{
b = atoi(buf);
tok = pdf_lex(file, buf, cap, &len);
if (tok == PDF_TR)
{
error = fz_newindirect(&val, a, b);
break;
}
}
goto cleanup;
default:
goto cleanup;
}
if (error) goto cleanup;
error = fz_dictput(dict, key, val);
if (error) goto cleanup;
fz_dropobj(val);
fz_dropobj(key);
key = val = nil;
}
cleanup:
if (key) fz_dropobj(key);
if (val) fz_dropobj(val);
if (dict) fz_dropobj(dict);
if (error) return error;
return fz_throw("syntaxerror: corrupt dictionary");
}
fz_error *
pdf_parsearray(fz_obj **op, fz_stream *file, char *buf, int cap)
{
fz_error *error = nil;
fz_obj *ary = nil;
fz_obj *obj = nil;
int a = 0, b = 0, n = 0;
int tok, len;
error = fz_newarray(op, 4);
if (error) return error;
ary = *op;
while (1)
{
tok = pdf_lex(file, buf, cap, &len);
if (tok != PDF_TINT && tok != PDF_TR)
{
if (n > 0)
{
error = fz_newint(&obj, a);
if (error) goto cleanup;
error = fz_arraypush(ary, obj);
if (error) goto cleanup;
fz_dropobj(obj);
obj = nil;
}
if (n > 1)
{
error = fz_newint(&obj, b);
if (error) goto cleanup;
error = fz_arraypush(ary, obj);
if (error) goto cleanup;
fz_dropobj(obj);
obj = nil;
}
n = 0;
}
if (tok == PDF_TINT && n == 2)
{
error = fz_newint(&obj, a);
if (error) goto cleanup;
error = fz_arraypush(ary, obj);
if (error) goto cleanup;
fz_dropobj(obj);
obj = nil;
a = b;
n --;
}
switch (tok)
{
case PDF_TCARRAY:
return nil;
case PDF_TINT:
if (n == 0)
a = atoi(buf);
if (n == 1)
b = atoi(buf);
n ++;
break;
case PDF_TR:
if (n != 2)
goto cleanup;
error = fz_newindirect(&obj, a, b);
if (error) goto cleanup;
n = 0;
break;
case PDF_TOARRAY: error = pdf_parsearray(&obj, file, buf, cap); break;
case PDF_TODICT: error = pdf_parsedict(&obj, file, buf, cap); break;
case PDF_TNAME: error = fz_newname(&obj, buf); break;
case PDF_TREAL: error = fz_newreal(&obj, atof(buf)); break;
case PDF_TSTRING: error = fz_newstring(&obj, buf, len); break;
case PDF_TTRUE: error = fz_newbool(&obj, 1); break;
case PDF_TFALSE: error = fz_newbool(&obj, 0); break;
case PDF_TNULL: error = fz_newnull(&obj); break;
default: goto cleanup;
}
if (error) goto cleanup;
if (obj)
{
error = fz_arraypush(ary, obj);
if (error) goto cleanup;
fz_dropobj(obj);
}
obj = nil;
}
cleanup:
if (obj) fz_dropobj(obj);
if (ary) fz_dropobj(ary);
if (error) return error;
return fz_throw("syntaxerror: corrupt array");
}
fz_error
pdf_repairxref(pdf_xref *xref, char *buf, int bufsize)
{
fz_error error;
fz_obj *dict, *obj;
fz_obj *length;
fz_obj *encrypt = nil;
fz_obj *id = nil;
fz_obj *root = nil;
fz_obj *info = nil;
struct entry *list = nil;
int listlen;
int listcap;
int maxnum = 0;
int num = 0;
int gen = 0;
int tmpofs, numofs = 0, genofs = 0;
int stmlen, stmofs = 0;
int tok;
int next;
int i, n;
pdf_logxref("repairxref %p\n", xref);
fz_seek(xref->file, 0, 0);
listlen = 0;
listcap = 1024;
list = fz_calloc(listcap, sizeof(struct entry));
/* look for '%PDF' version marker within first kilobyte of file */
n = fz_read(xref->file, (unsigned char *)buf, MAX(bufsize, 1024));
if (n < 0)
{
error = fz_rethrow(n, "cannot read from file");
goto cleanup;
}
fz_seek(xref->file, 0, 0);
for (i = 0; i < n - 4; i++)
{
if (memcmp(buf + i, "%PDF", 4) == 0)
{
fz_seek(xref->file, i, 0);
break;
}
}
while (1)
{
tmpofs = fz_tell(xref->file);
if (tmpofs < 0)
{
error = fz_throw("cannot tell in file");
goto cleanup;
}
error = pdf_lex(&tok, xref->file, buf, bufsize, &n);
if (error)
{
fz_catch(error, "ignoring the rest of the file");
break;
}
if (tok == PDF_TINT)
{
numofs = genofs;
num = gen;
genofs = tmpofs;
gen = atoi(buf);
}
if (tok == PDF_TOBJ)
{
error = fz_repairobj(xref->file, buf, bufsize, &stmofs, &stmlen, &encrypt, &id);
if (error)
{
error = fz_rethrow(error, "cannot parse object (%d %d R)", num, gen);
goto cleanup;
}
pdf_logxref("found object: (%d %d R)\n", num, gen);
if (listlen + 1 == listcap)
{
listcap = (listcap * 3) / 2;
list = fz_realloc(list, listcap, sizeof(struct entry));
}
list[listlen].num = num;
list[listlen].gen = gen;
list[listlen].ofs = numofs;
list[listlen].stmofs = stmofs;
list[listlen].stmlen = stmlen;
listlen ++;
if (num > maxnum)
maxnum = num;
}
/* trailer dictionary */
if (tok == PDF_TODICT)
{
error = pdf_parsedict(&dict, xref, xref->file, buf, bufsize);
if (error)
{
error = fz_rethrow(error, "cannot parse object");
goto cleanup;
}
obj = fz_dictgets(dict, "Encrypt");
if (obj)
{
if (encrypt)
fz_dropobj(encrypt);
encrypt = fz_keepobj(obj);
}
obj = fz_dictgets(dict, "ID");
if (obj)
{
if (id)
fz_dropobj(id);
id = fz_keepobj(obj);
}
obj = fz_dictgets(dict, "Root");
if (obj)
{
if (root)
fz_dropobj(root);
root = fz_keepobj(obj);
}
obj = fz_dictgets(dict, "Info");
if (obj)
{
if (info)
fz_dropobj(info);
info = fz_keepobj(obj);
}
fz_dropobj(dict);
}
if (tok == PDF_TERROR)
fz_readbyte(xref->file);
if (tok == PDF_TEOF)
break;
}
/* make xref reasonable */
pdf_resizexref(xref, maxnum + 1);
for (i = 0; i < listlen; i++)
{
xref->table[list[i].num].type = 'n';
xref->table[list[i].num].ofs = list[i].ofs;
xref->table[list[i].num].gen = list[i].gen;
xref->table[list[i].num].stmofs = list[i].stmofs;
/* corrected stream length */
if (list[i].stmlen >= 0)
{
pdf_logxref("correct stream length %d %d = %d\n",
list[i].num, list[i].gen, list[i].stmlen);
error = pdf_loadobject(&dict, xref, list[i].num, list[i].gen);
if (error)
{
error = fz_rethrow(error, "cannot load stream object (%d %d R)", list[i].num, list[i].gen);
goto cleanup;
}
length = fz_newint(list[i].stmlen);
fz_dictputs(dict, "Length", length);
fz_dropobj(length);
fz_dropobj(dict);
}
}
xref->table[0].type = 'f';
xref->table[0].ofs = 0;
xref->table[0].gen = 65535;
xref->table[0].stmofs = 0;
xref->table[0].obj = nil;
next = 0;
for (i = xref->len - 1; i >= 0; i--)
{
if (xref->table[i].type == 'f')
{
xref->table[i].ofs = next;
if (xref->table[i].gen < 65535)
xref->table[i].gen ++;
next = i;
}
}
/* create a repaired trailer, Root will be added later */
xref->trailer = fz_newdict(5);
obj = fz_newint(maxnum + 1);
fz_dictputs(xref->trailer, "Size", obj);
fz_dropobj(obj);
if (root)
{
fz_dictputs(xref->trailer, "Root", root);
fz_dropobj(root);
}
if (info)
{
fz_dictputs(xref->trailer, "Info", info);
fz_dropobj(info);
}
if (encrypt)
{
if (fz_isindirect(encrypt))
{
/* create new reference with non-nil xref pointer */
obj = fz_newindirect(fz_tonum(encrypt), fz_togen(encrypt), xref);
fz_dropobj(encrypt);
encrypt = obj;
}
fz_dictputs(xref->trailer, "Encrypt", encrypt);
fz_dropobj(encrypt);
}
if (id)
{
if (fz_isindirect(id))
{
/* create new reference with non-nil xref pointer */
obj = fz_newindirect(fz_tonum(id), fz_togen(id), xref);
fz_dropobj(id);
id = obj;
}
fz_dictputs(xref->trailer, "ID", id);
fz_dropobj(id);
}
fz_free(list);
return fz_okay;
cleanup:
if (encrypt) fz_dropobj(encrypt);
if (id) fz_dropobj(id);
if (root) fz_dropobj(root);
if (info) fz_dropobj(info);
fz_free(list);
return error; /* already rethrown */
}