static fz_error
pdf_loadpagecontents(fz_buffer **bufp, pdf_xref *xref, fz_obj *obj)
{
fz_error error;
if (fz_isarray(obj))
{
error = pdf_loadpagecontentsarray(bufp, xref, obj);
if (error)
return fz_rethrow(error, "cannot load content stream array (%d 0 R)", fz_tonum(obj));
}
else if (pdf_isstream(xref, fz_tonum(obj), fz_togen(obj)))
{
error = pdf_loadstream(bufp, xref, fz_tonum(obj), fz_togen(obj));
if (error)
return fz_rethrow(error, "cannot load content stream (%d 0 R)", fz_tonum(obj));
}
else
{
fz_warn("page contents missing, leaving page blank");
*bufp = fz_newbuffer(0);
}
return fz_okay;
}
static void sweepref(fz_obj *obj)
{
int num = fz_tonum(obj);
int gen = fz_togen(obj);
if (num < 0 || num >= xref->len)
return;
if (uselist[num])
return;
uselist[num] = 1;
/* Bake in /Length in stream objects */
if (pdf_isstream(xref, num, gen))
{
fz_obj *len = fz_dictgets(obj, "Length");
if (fz_isindirect(len))
{
uselist[fz_tonum(len)] = 0;
len = fz_resolveindirect(len);
fz_dictputs(obj, "Length", len);
}
}
sweepobj(fz_resolveindirect(obj));
}
static fz_error
loadcharproc(fz_tree **treep, pdf_xref *xref, fz_obj *rdb, fz_obj *stmref)
{
fz_error error;
pdf_csi *csi;
fz_stream *stm;
error = pdf_newcsi(&csi, 1);
if (error)
return fz_rethrow(error, "cannot create interpreter");
error = pdf_openstream(&stm, xref, fz_tonum(stmref), fz_togen(stmref));
if (error)
{
pdf_dropcsi(csi);
return fz_rethrow(error, "cannot open glyph content stream");
}
error = pdf_runcsi(csi, xref, rdb, stm);
if (error)
{
fz_dropstream(stm);
pdf_dropcsi(csi);
return fz_rethrow(error, "cannot interpret glyph content stream (%d %d R)", fz_tonum(stmref), fz_togen(stmref));
}
*treep = csi->tree;
csi->tree = nil;
fz_dropstream(stm);
pdf_dropcsi(csi);
return fz_okay;
}
void
pdf_storeitem(pdf_store *store, pdf_itemkind kind, fz_obj *key, void *val)
{
pdf_item *item;
item = fz_malloc(sizeof(pdf_item));
item->kind = kind;
item->key = fz_keepobj(key);
item->val = val;
item->age = 0;
item->next = nil;
if (fz_isindirect(key))
{
struct refkey refkey;
pdf_logrsrc("store item %s (%d %d R) ptr=%p\n", kindstr(kind), fz_tonum(key), fz_togen(key), val);
refkey.kind = kind;
refkey.oid = fz_tonum(key);
refkey.gen = fz_togen(key);
fz_hashinsert(store->hash, &refkey, item);
}
else
{
pdf_logrsrc("store item %s: ... = %p\n", kindstr(kind), val);
item->next = store->root;
store->root = item;
}
keepitem(kind, val);
}
fz_error
pdf_loadembeddedfont(pdf_fontdesc *fontdesc, pdf_xref *xref, fz_obj *stmref)
{
fz_error error;
fz_buffer *buf;
pdf_logfont("load embedded font\n");
error = pdf_loadstream(&buf, xref, fz_tonum(stmref), fz_togen(stmref));
if (error)
return fz_rethrow(error, "cannot load font stream (%d %d R)", fz_tonum(stmref), fz_togen(stmref));
error = fz_newfontfrombuffer(&fontdesc->font, buf->data, buf->len, 0);
if (error)
{
fz_dropbuffer(buf);
return fz_rethrow(error, "cannot load embedded font (%d %d R)", fz_tonum(stmref), fz_togen(stmref));
}
fontdesc->buffer = buf->data; /* save the buffer so we can free it later */
fz_free(buf); /* only free the fz_buffer struct, not the contained data */
fontdesc->isembedded = 1;
return fz_okay;
}
static fz_error
gatherpsobjs(int page, fz_obj *pageobj, fz_obj *dict)
{
int i;
for (i = 0; i < fz_dictlen(dict); i++)
{
fz_obj *ref;
fz_obj *xobjdict;
fz_obj *type;
fz_obj *subtype;
int k;
xobjdict = ref = fz_dictgetval(dict, i);
if (!fz_isdict(xobjdict))
return fz_throw("not a xobject dict (%d %d R)", fz_tonum(ref), fz_togen(ref));
type = fz_dictgets(xobjdict, "Subtype");
if (!fz_isname(type))
return fz_throw("not a xobject type (%d %d R)", fz_tonum(ref), fz_togen(ref));
if (strcmp(fz_toname(type), "Form"))
continue;
subtype = fz_dictgets(xobjdict, "Subtype2");
if (subtype && !fz_isname(subtype))
return fz_throw("not a xobject subtype (%d %d R)", fz_tonum(ref), fz_togen(ref));
if (strcmp(fz_toname(type), "PS") &&
(strcmp(fz_toname(type), "Form") || strcmp(fz_toname(subtype), "PS")))
continue;
for (k = 0; k < psobjs; k++)
if (fz_tonum(psobj[k]->ref) == fz_tonum(ref) &&
fz_togen(psobj[k]->ref) == fz_togen(ref))
break;
if (k < psobjs)
continue;
psobjs++;
psobj = fz_realloc(psobj, psobjs * sizeof (struct info *));
if (!psobj)
return fz_throw("out of memory");
psobj[psobjs - 1] = fz_malloc(sizeof (struct info));
if (!psobj[psobjs - 1])
return fz_throw("out of memory");
psobj[psobjs - 1]->page = page;
psobj[psobjs - 1]->pageobj = pageobj;
psobj[psobjs - 1]->ref = ref;
}
return fz_okay;
}
int
pdf_findpageobject(pdf_xref *xref, fz_obj *page)
{
int num = fz_tonum(page);
int gen = fz_togen(page);
int i;
for (i = 0; i < xref->pagelen; i++)
if (num == fz_tonum(xref->pagerefs[i]) && gen == fz_togen(xref->pagerefs[i]))
return i + 1;
return 0;
}
int
epdf_index_item_page_get(const Epdf_Document *doc, const Epdf_Index_Item *item)
{
fz_obj *dest;
int p;
int n;
int g;
if (!item || !item->link)
return -1;
if (PDF_LGOTO != item->link->kind)
return -1;
dest = item->link->dest;
p = 0;
if (fz_isint(dest))
{
p = fz_toint(dest);
return p;
}
if (fz_isdict(dest))
{
/* The destination is linked from a Go-To action's D array */
fz_obj *D;
D = fz_dictgets(dest, "D");
if (D && fz_isarray(D))
dest = fz_arrayget(D, 0);
}
n = fz_tonum(dest);
g = fz_togen(dest);
for (p = 1; p <= epdf_document_page_count_get(doc); p++)
{
fz_obj *page;
int np;
int gp;
page = pdf_getpageobject(doc->xref, p);
if (!page)
continue;
np = fz_tonum(page);
gp = fz_togen(page);
if (n == np && g == gp)
return p-1;
}
return 0;
}
fz_error *
pdf_storeitem(pdf_store *store, pdf_itemkind kind, fz_obj *key, void *val)
{
fz_error *error;
switch (kind)
{
case PDF_KCOLORSPACE: fz_keepcolorspace(val); break;
case PDF_KFUNCTION: pdf_keepfunction(val); break;
case PDF_KXOBJECT: pdf_keepxobject(val); break;
case PDF_KIMAGE: fz_keepimage(val); break;
case PDF_KPATTERN: pdf_keeppattern(val); break;
case PDF_KSHADE: fz_keepshade(val); break;
case PDF_KCMAP: pdf_keepcmap(val); break;
case PDF_KFONT: fz_keepfont(val); break;
}
if (fz_isindirect(key))
{
struct refkey item;
pdf_logrsrc("store item %d: %d %d R = %p\n", kind, fz_tonum(key), fz_togen(key), val);
item.kind = kind;
item.oid = fz_tonum(key);
item.gen = fz_togen(key);
error = fz_hashinsert(store->hash, &item, val);
if (error)
return error;
}
else
{
pdf_item *item;
item = fz_malloc(sizeof(pdf_item));
if (!item)
return fz_outofmem;
pdf_logrsrc("store item %d: ... = %p\n", kind, val);
item->kind = kind;
item->key = fz_keepobj(key);
item->val = val;
item->next = store->root;
store->root = item;
}
return nil;
}
static fz_error
gathershadings(int page, fz_obj *pageobj, fz_obj *dict)
{
int i;
for (i = 0; i < fz_dictlen(dict); i++)
{
fz_obj *ref;
fz_obj *shade;
fz_obj *type;
int k;
shade = ref = fz_dictgetval(dict, i);
if (!fz_isdict(shade))
return fz_throw("not a shading dict (%d %d R)", fz_tonum(ref), fz_togen(ref));
type = fz_dictgets(shade, "ShadingType");
if (!fz_isint(type) || fz_toint(type) < 1 || fz_toint(type) > 7)
{
fz_warn("not a shading type (%d %d R)", fz_tonum(ref), fz_togen(ref));
type = nil;
}
for (k = 0; k < shadings; k++)
if (fz_tonum(shading[k]->ref) == fz_tonum(ref) &&
fz_togen(shading[k]->ref) == fz_togen(ref))
break;
if (k < shadings)
continue;
shadings++;
shading = fz_realloc(shading, shadings * sizeof (struct info *));
if (!shading)
return fz_throw("out of memory");
shading[shadings - 1] = fz_malloc(sizeof (struct info));
if (!shading[shadings - 1])
return fz_throw("out of memory");
shading[shadings - 1]->page = page;
shading[shadings - 1]->pageobj = pageobj;
shading[shadings - 1]->ref = ref;
shading[shadings - 1]->u.shading.type = type;
}
return fz_okay;
}
fz_error *
pdf_loadembeddedfont(pdf_font *font, pdf_xref *xref, fz_obj *stmref)
{
fz_error *error;
int fterr;
FT_Face face;
fz_buffer *buf;
error = initfontlibs();
if (error)
return fz_rethrow(error, "cannot init font libraries");
pdf_logfont("load embedded font\n");
error = pdf_loadstream(&buf, xref, fz_tonum(stmref), fz_togen(stmref));
if (error)
return fz_rethrow(error, "cannot load font stream");
fterr = FT_New_Memory_Face(ftlib, buf->rp, buf->wp - buf->rp, 0, &face);
if (fterr)
{
fz_dropbuffer(buf);
return fz_throw("freetype: cannot load embedded font: %s", ft_errstr(fterr));
}
font->ftface = face;
font->fontdata = buf;
return fz_okay;
}
static void evictitem(pdf_item *item)
{
pdf_logrsrc("evicting item %s (%d %d R) at age %d\n", kindstr(item->kind), fz_tonum(item->key), fz_togen(item->key), item->age);
fz_dropobj(item->key);
dropitem(item->kind, item->val);
fz_free(item);
}
void *
pdf_finditem(pdf_store *store, pdf_itemkind kind, fz_obj *key)
{
pdf_item *item;
struct refkey refkey;
if (key == nil)
return nil;
if (fz_isindirect(key))
{
refkey.kind = kind;
refkey.oid = fz_tonum(key);
refkey.gen = fz_togen(key);
item = fz_hashfind(store->hash, &refkey);
if (item)
{
item->age = 0;
return item->val;
}
}
else
{
for (item = store->root; item; item = item->next)
if (item->kind == kind && !fz_objcmp(item->key, key))
{
item->age = 0;
return item->val;
}
}
return nil;
}
fz_error *
pdf_loadembeddedfont(pdf_font *font, pdf_xref *xref, fz_obj *stmref)
{
fz_error *error;
int fterr;
FT_Face face;
fz_buffer *buf;
error = initfontlibs();
if (error)
return error;
pdf_logfont("load embedded font\n");
error = pdf_loadstream(&buf, xref, fz_tonum(stmref), fz_togen(stmref));
if (error)
return error;
fterr = FT_New_Memory_Face(ftlib, buf->rp, buf->wp - buf->rp, 0, &face);
if (fterr) {
fz_free(buf);
return fz_throw("freetype could not load embedded font: %s", pdf_fterrorstring(fterr));
}
font->ftface = face;
font->fontdata = buf;
return nil;
}
static fz_error
loadtype0(pdf_fontdesc **fontdescp, pdf_xref *xref, fz_obj *dict)
{
fz_error error;
fz_obj *dfonts;
fz_obj *dfont;
fz_obj *subtype;
fz_obj *encoding;
fz_obj *tounicode;
dfonts = fz_dictgets(dict, "DescendantFonts");
if (!dfonts)
return fz_throw("cid font is missing descendant fonts");
dfont = fz_arrayget(dfonts, 0);
subtype = fz_dictgets(dfont, "Subtype");
encoding = fz_dictgets(dict, "Encoding");
tounicode = fz_dictgets(dict, "ToUnicode");
if (fz_isname(subtype) && !strcmp(fz_toname(subtype), "CIDFontType0"))
error = loadcidfont(fontdescp, xref, dfont, encoding, tounicode);
else if (fz_isname(subtype) && !strcmp(fz_toname(subtype), "CIDFontType2"))
error = loadcidfont(fontdescp, xref, dfont, encoding, tounicode);
else
error = fz_throw("syntaxerror: unknown cid font type");
if (error)
return fz_rethrow(error, "cannot load descendant font (%d %d R)", fz_tonum(dfont), fz_togen(dfont));
return fz_okay;
}
void
pdf_removeitem(pdf_store *store, pdf_itemkind kind, fz_obj *key)
{
pdf_item *item, *prev;
struct refkey refkey;
if (key == nil)
return;
if (fz_isindirect(key))
{
refkey.kind = kind;
refkey.oid = fz_tonum(key);
refkey.gen = fz_togen(key);
item = fz_hashfind(store->hash, &refkey);
if (!item)
return;
fz_hashremove(store->hash, &refkey);
pdf_logrsrc("remove item %s (%d %d R) ptr=%p\n", kindstr(kind), fz_tonum(key), fz_togen(key), item->val);
dropitem(kind, item->val);
fz_dropobj(item->key);
fz_free(item);
}
else
{
prev = nil;
for (item = store->root; item; item = item->next)
{
if (item->kind == kind && !fz_objcmp(item->key, key))
{
if (!prev)
store->root = item->next;
else
prev->next = item->next;
dropitem(kind, item->val);
fz_dropobj(item->key);
fz_free(item);
break;
}
prev = item;
}
}
}
fz_error
pdf_loadfont(pdf_fontdesc **fontdescp, pdf_xref *xref, fz_obj *rdb, fz_obj *dict)
{
fz_error error;
char *subtype;
fz_obj *dfonts;
fz_obj *charprocs;
if ((*fontdescp = pdf_finditem(xref->store, pdf_dropfont, dict)))
{
pdf_keepfont(*fontdescp);
return fz_okay;
}
subtype = fz_toname(fz_dictgets(dict, "Subtype"));
dfonts = fz_dictgets(dict, "DescendantFonts");
charprocs = fz_dictgets(dict, "CharProcs");
if (subtype && !strcmp(subtype, "Type0"))
error = loadtype0(fontdescp, xref, dict);
else if (subtype && !strcmp(subtype, "Type1"))
error = loadsimplefont(fontdescp, xref, dict);
else if (subtype && !strcmp(subtype, "MMType1"))
error = loadsimplefont(fontdescp, xref, dict);
else if (subtype && !strcmp(subtype, "TrueType"))
error = loadsimplefont(fontdescp, xref, dict);
else if (subtype && !strcmp(subtype, "Type3"))
error = pdf_loadtype3font(fontdescp, xref, rdb, dict);
else if (charprocs)
{
fz_warn("unknown font format, guessing type3.");
error = pdf_loadtype3font(fontdescp, xref, rdb, dict);
}
else if (dfonts)
{
fz_warn("unknown font format, guessing type0.");
error = loadtype0(fontdescp, xref, dict);
}
else
{
fz_warn("unknown font format, guessing type1 or truetype.");
error = loadsimplefont(fontdescp, xref, dict);
}
if (error)
return fz_rethrow(error, "cannot load font (%d %d R)", fz_tonum(dict), fz_togen(dict));
/* Save the widths to stretch non-CJK substitute fonts */
if ((*fontdescp)->font->ftsubstitute && !(*fontdescp)->tottfcmap)
pdf_makewidthtable(*fontdescp);
/* SumatraPDF: this font renders wrong without hinting */
if (strstr((*fontdescp)->font->name, "MingLiU"))
(*fontdescp)->font->fthint = 1;
pdf_storeitem(xref->store, pdf_keepfont, pdf_dropfont, dict, *fontdescp);
return fz_okay;
}
static void
printglobalinfo(void)
{
printf("\nPDF-%d.%d\n", xref->version / 10, xref->version % 10);
if (info->u.info.obj)
{
printf("Info object (%d %d R):\n", fz_tonum(info->ref), fz_togen(info->ref));
fz_debugobj(fz_resolveindirect(info->u.info.obj));
}
if (cryptinfo->u.crypt.obj)
{
printf("\nEncryption object (%d %d R):\n", fz_tonum(cryptinfo->ref), fz_togen(cryptinfo->ref));
fz_debugobj(cryptinfo->u.crypt.obj);
}
printf("\nPages: %d\n\n", pagecount);
}
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);
}
}
}
}
void pdfapp_gotopage(pdfapp_t *app, fz_obj *obj)
{
int oid = fz_tonum(obj);
int i;
for (i = 0; i < pdf_getpagecount(app->pages); i++)
{
if (fz_tonum(app->pages->pref[i]) == oid)
{
if (app->histlen + 1 == 256)
{
memmove(app->hist, app->hist + 1, sizeof(int) * 255);
app->histlen --;
}
app->hist[app->histlen++] = app->pageno;
app->pageno = i + 1;
pdfapp_showpage(app, 1, 1);
return;
}
}
}
fz_error
pdf_loadimage(fz_pixmap **pixp, pdf_xref *xref, fz_obj *dict)
{
fz_error error;
if ((*pixp = pdf_finditem(xref->store, fz_droppixmap, dict)))
{
fz_keeppixmap(*pixp);
return fz_okay;
}
pdf_logimage("load image (%d 0 R) {\n", fz_tonum(dict));
error = pdf_loadimageimp(pixp, xref, nil, dict, nil, 0);
if (error)
return fz_rethrow(error, "cannot load image (%d 0 R)", fz_tonum(dict));
pdf_storeitem(xref->store, fz_keeppixmap, fz_droppixmap, dict, *pixp);
pdf_logimage("}\n");
return fz_okay;
}
static fz_error
gatherdimensions(int page, fz_obj *pageobj)
{
fz_obj *ref;
fz_rect bbox;
fz_obj *obj;
int j;
obj = ref = fz_dictgets(pageobj, "MediaBox");
if (!fz_isarray(obj))
return fz_throw("cannot find page bounds (%d %d R)", fz_tonum(ref), fz_togen(ref));
bbox = pdf_torect(obj);
for (j = 0; j < dims; j++)
if (!memcmp(dim[j]->u.dim.bbox, &bbox, sizeof (fz_rect)))
break;
if (j < dims)
return fz_okay;
dims++;
dim = fz_realloc(dim, dims * sizeof (struct info *));
if (!dim)
return fz_throw("out of memory");
dim[dims - 1] = fz_malloc(sizeof (struct info));
if (!dim[dims - 1])
return fz_throw("out of memory");
dim[dims - 1]->u.dim.bbox = fz_malloc(sizeof (fz_rect));
if (!dim[dims - 1]->u.dim.bbox)
return fz_throw("out of memory");
dim[dims - 1]->page = page;
dim[dims - 1]->pageobj = pageobj;
dim[dims - 1]->ref = nil;
memcpy(dim[dims - 1]->u.dim.bbox, &bbox, sizeof (fz_rect));
return fz_okay;
}
void
editcopy(int pagenum)
{
fz_error *error;
fz_obj *obj;
fz_obj *ref;
fz_obj *num;
printf("copy %s page %d\n", srcname, pagenum);
ref = srcpages->pref[pagenum - 1];
obj = pdf_getpageobject(srcpages, pagenum - 1);
fz_dictdels(obj, "Parent");
/*
fz_dictdels(obj, "B");
fz_dictdels(obj, "PieceInfo");
fz_dictdels(obj, "Metadata");
fz_dictdels(obj, "Annots");
fz_dictdels(obj, "Tabs");
*/
pdf_updateobject(src, fz_tonum(ref), fz_togen(ref), obj);
error = fz_arraypush(editobjects, ref);
if (error)
die(error);
error = fz_newint(&num, editmode);
if (error)
die(error);
error = fz_arraypush(editmodelist, num);
if (error)
die(error);
fz_dropobj(num);
}
static fz_error
pdf_loadpagecontentsarray(fz_buffer **bigbufp, pdf_xref *xref, fz_obj *list)
{
fz_error error;
fz_buffer *big;
fz_buffer *one;
int i;
pdf_logpage("multiple content streams: %d\n", fz_arraylen(list));
/* TODO: openstream, read, close into big buffer at once */
big = fz_newbuffer(32 * 1024);
for (i = 0; i < fz_arraylen(list); i++)
{
fz_obj *stm = fz_arrayget(list, i);
error = pdf_loadstream(&one, xref, fz_tonum(stm), fz_togen(stm));
if (error)
{
fz_dropbuffer(big);
return fz_rethrow(error, "cannot load content stream part %d/%d (%d %d R)", i + 1, fz_arraylen(list), fz_tonum(stm), fz_togen(stm));
}
if (big->len + one->len + 1 > big->cap)
fz_resizebuffer(big, big->len + one->len + 1);
memcpy(big->data + big->len, one->data, one->len);
big->data[big->len + one->len] = ' ';
big->len += one->len + 1;
fz_dropbuffer(one);
}
*bigbufp = big;
return fz_okay;
}
fz_error
pdf_loadpattern(pdf_pattern **patp, pdf_xref *xref, fz_obj *dict)
{
fz_error error;
pdf_pattern *pat;
fz_obj *obj;
if ((*patp = pdf_finditem(xref->store, pdf_droppattern, dict)))
{
pdf_keeppattern(*patp);
return fz_okay;
}
pdf_logrsrc("load pattern (%d %d R) {\n", fz_tonum(dict), fz_togen(dict));
pat = fz_malloc(sizeof(pdf_pattern));
pat->refs = 1;
pat->resources = nil;
pat->contents = nil;
/* Store pattern now, to avoid possible recursion if objects refer back to this one */
pdf_storeitem(xref->store, pdf_keeppattern, pdf_droppattern, dict, pat);
pat->ismask = fz_toint(fz_dictgets(dict, "PaintType")) == 2;
pat->xstep = fz_toreal(fz_dictgets(dict, "XStep"));
pat->ystep = fz_toreal(fz_dictgets(dict, "YStep"));
pdf_logrsrc("mask %d\n", pat->ismask);
pdf_logrsrc("xstep %g\n", pat->xstep);
pdf_logrsrc("ystep %g\n", pat->ystep);
obj = fz_dictgets(dict, "BBox");
pat->bbox = pdf_torect(obj);
pdf_logrsrc("bbox [%g %g %g %g]\n",
pat->bbox.x0, pat->bbox.y0,
pat->bbox.x1, pat->bbox.y1);
obj = fz_dictgets(dict, "Matrix");
if (obj)
pat->matrix = pdf_tomatrix(obj);
else
pat->matrix = fz_identity;
pdf_logrsrc("matrix [%g %g %g %g %g %g]\n",
pat->matrix.a, pat->matrix.b,
pat->matrix.c, pat->matrix.d,
pat->matrix.e, pat->matrix.f);
pat->resources = fz_dictgets(dict, "Resources");
if (pat->resources)
fz_keepobj(pat->resources);
error = pdf_loadstream(&pat->contents, xref, fz_tonum(dict), fz_togen(dict));
if (error)
{
pdf_removeitem(xref->store, pdf_droppattern, dict);
pdf_droppattern(pat);
return fz_rethrow(error, "cannot load pattern stream (%d %d R)", fz_tonum(dict), fz_togen(dict));
}
pdf_logrsrc("}\n");
*patp = pat;
return fz_okay;
}
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);
}
static fz_error *
loadshadedict(fz_shade **shadep, pdf_xref *xref, fz_obj *dict, fz_obj *ref, fz_matrix matrix)
{
fz_error *error;
fz_shade *shade;
fz_obj *obj;
int type;
int i;
pdf_logshade("load shade dict %d %d {\n", fz_tonum(ref), fz_togen(ref));
shade = fz_malloc(sizeof(fz_shade));
if (!shade)
return fz_outofmem;
shade->refs = 1;
shade->usebackground = 0;
shade->usefunction = 0;
shade->matrix = matrix;
shade->bbox = fz_infiniterect;
shade->meshlen = 0;
shade->meshcap = 0;
shade->mesh = nil;
obj = fz_dictgets(dict, "ShadingType");
type = fz_toint(obj);
pdf_logshade("type %d\n", type);
/* TODO: flatten indexed... */
obj = fz_dictgets(dict, "ColorSpace");
if (obj)
{
shade->cs = pdf_finditem(xref->store, PDF_KCOLORSPACE, obj);
if (shade->cs)
fz_keepcolorspace(shade->cs);
else
{
error = pdf_resolve(&obj, xref);
if (error)
return error;
error = pdf_loadcolorspace(&shade->cs, xref, obj);
if (error)
return error;
fz_dropobj(obj);
}
}
pdf_logshade("colorspace %s\n", shade->cs->name);
obj = fz_dictgets(dict, "Background");
if (obj)
{
pdf_logshade("background\n");
shade->usebackground = 1;
for (i = 0; i < shade->cs->n; i++)
shade->background[i] = fz_toreal(fz_arrayget(obj, i));
}
obj = fz_dictgets(dict, "BBox");
if (fz_isarray(obj))
{
shade->bbox = pdf_torect(obj);
pdf_logshade("bbox [%g %g %g %g]\n",
shade->bbox.x0, shade->bbox.y0,
shade->bbox.x1, shade->bbox.y1);
}
switch(type)
{
case 1:
error = pdf_loadtype1shade(shade, xref, dict, ref);
if (error) goto cleanup;
break;
case 2:
error = pdf_loadtype2shade(shade, xref, dict, ref);
if (error) goto cleanup;
break;
case 3:
error = pdf_loadtype3shade(shade, xref, dict, ref);
if (error) goto cleanup;
break;
case 4:
error = pdf_loadtype4shade(shade, xref, dict, ref);
if (error) goto cleanup;
break;
case 5:
error = pdf_loadtype5shade(shade, xref, dict, ref);
if (error) goto cleanup;
break;
case 6:
error = pdf_loadtype6shade(shade, xref, dict, ref);
if (error) goto cleanup;
break;
case 7:
error = pdf_loadtype7shade(shade, xref, dict, ref);
if (error) goto cleanup;
break;
default:
fz_warn("syntaxerror: unknown shading type: %d", type);
break;
};
pdf_logshade("}\n");
*shadep = shade;
return nil;
cleanup:
fz_dropshade(shade);
return error;
}
fz_error *
pdf_loadshade(fz_shade **shadep, pdf_xref *xref, fz_obj *dict, fz_obj *ref)
{
fz_error *error;
fz_matrix mat;
fz_obj *obj;
fz_obj *shd;
if ((*shadep = pdf_finditem(xref->store, PDF_KSHADE, ref)))
return nil;
/*
* Type 2 pattern dictionary
*/
if (fz_dictgets(dict, "PatternType"))
{
pdf_logshade("load shade pattern %d %d {\n", fz_tonum(ref), fz_togen(ref));
obj = fz_dictgets(dict, "Matrix");
if (obj)
{
mat = pdf_tomatrix(obj);
pdf_logshade("matrix [%g %g %g %g %g %g]\n",
mat.a, mat.b, mat.c, mat.d, mat.e, mat.f);
}
else
{
mat = fz_identity();
}
obj = fz_dictgets(dict, "ExtGState");
if (obj)
{
pdf_logshade("extgstate ...\n");
}
obj = fz_dictgets(dict, "Shading");
if (!obj)
return fz_throw("syntaxerror: missing shading dictionary");
shd = obj;
error = pdf_resolve(&shd, xref);
if (error)
return error;
error = loadshadedict(shadep, xref, shd, obj, mat);
fz_dropobj(shd);
if (error)
return error;
pdf_logshade("}\n");
}
/*
* Naked shading dictionary
*/
else
{
error = loadshadedict(shadep, xref, dict, ref, fz_identity());
if (error)
return error;
}
error = pdf_storeitem(xref->store, PDF_KSHADE, ref, *shadep);
if (error)
{
fz_dropshade(*shadep);
return error;
}
return nil;
}
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;
}