I have modified ofxPdf a bit so that it reads pdfs with images…
here is the core of the code:
void ofxPDF::loadPDF(string path, int pagenumber)
{path = ofToDataPath(path); fz_context *ctx; fz_document* doc; fz_device *dev; fz_bbox box; fz_rect bounds; fz_matrix ctm; fz_pixmap *image; fz_colorspace *colorspace; int i,j,rotation = 0; int pagecount = 0; fz_page *page; ctx = fz_new_context(NULL, NULL, FZ_STORE_UNLIMITED); fz_try(ctx){ doc = fz_open_document(ctx, (char*)path.c_str()); }fz_catch(ctx){ fz_free_context(ctx); return false; } colorspace = fz_device_rgb; pagecount = fz_count_pages(doc); for(i=0;i<pagecount;i++){ cout<<i<<endl; page = fz_load_page(doc,i); if(i == 0){ bounds= fz_bound_page(doc,page); box.x0 =bounds.x0; box.x1 = bounds.x1; box.y0 = bounds.y0; box.y1 = bounds.y1; width = bounds.x1 - bounds.x0; height = bounds.y1 - bounds.y0; } // colorspace= fz_pixmap_colorspace(ctx, NULL); image = fz_new_pixmap_with_bbox(ctx,colorspace,box); dev = fz_new_draw_device(ctx,image); fz_clear_pixmap_with_value(ctx, image, 255); fz_try(ctx){ colorspace= fz_device_rgb; cout<<colorspace<<endl; fz_run_page(doc,page,dev,fz_identity,NULL); ofPixels pixs; pixs.setFromExternalPixels(fz_pixmap_samples(ctx, image) , fz_pixmap_width(ctx,image), fz_pixmap_height(ctx,image), OF_PIXELS_RGBA); img_texture.setFromPixels(pixs); fz_free_device(dev); dev = fz_new_ofpath_device(ctx, paths); fz_run_page(doc,page,dev,fz_identity,NULL); // fz_write_png(ctx, image, (char*)path.c_str(), false); }fz_catch(ctx){ fz_drop_pixmap(ctx,image); fz_free_device(dev); fz_free_page(doc, page); continue; } fz_drop_pixmap(ctx,image); fz_free_device(dev); fz_free_page(doc, page); } fz_close_document(doc); fz_free_context(ctx); }
currently it only works with rgbalpha pdfs,
the reason why it doesn’t work with other types is because I can’t tell from the type of pixels what
colorspace is using, if is grayscale rgb cmyk or brg
…
the newer version of muPdf has a function that returns the fz_colorspace from the ctx before you allocate the image,
not sure if this version of mupdf has a way to tell the type of the pdf…
Any ideas??
is there a smart way to tell from an unsigned char * what type of pixels it has?
maybe divide by with and height? and see if it leaves 1 2 or 3 or 4 etc?
dev_ofpath.cpp (5.9 KB)
ofxPDF.h (606 Bytes)
ofxPDF.cpp (4.8 KB)
