Added separate image/font extraction tool.

author: Sebastian Rasmussen <sebras@hotmail.com> 2009-05-23 02:16:50 +0200
committer: Sebastian Rasmussen <sebras@hotmail.com> 2009-05-23 02:16:50 +0200
commit: da6f61c67c5c591f805a6a933ff7ce03c85563e7 (patch)
tree: 9786c3f40dbf571f93650676b5f489bf0e9902ed
parent: ac244e57a9c64451b75447620509b038e239e0e8 (diff)
download: mupdf-da6f61c67c5c591f805a6a933ff7ce03c85563e7.tar.xz
2 files changed, 335 insertions, 0 deletions
diff --git a/apps/Jamfile b/apps/Jamfile
index 1c70f554..f2f36e5d 100644
--- a/apps/Jamfile
+++ b/apps/Jamfile
@@ -16,6 +16,9 @@ LinkLibraries pdfdraw : $(FITZLIBS) ;
 Main pdfinfo : pdfinfo.c ;
 LinkLibraries pdfinfo : $(FITZLIBS) ;
 
+Main pdfextract : pdfextract.c ;
+LinkLibraries pdfextract : $(FITZLIBS) ;
+
 SubDir TOP apps common ;
 Library libpdfapp : pdfapp.c ;
 
diff --git a/apps/pdfextract.c b/apps/pdfextract.c
new file mode 100644
index 00000000..0e3d71f4
--- /dev/null
+++ b/apps/pdfextract.c
@@ -0,0 +1,332 @@
+/*
+ * pdfextract -- the ultimate way to extract images and fonts from pdfs
+ */
+
+#include "fitz.h"
+#include "mupdf.h"
+
+pdf_xref *xref = NULL;
+
+void die(fz_error eo)
+{
+    fz_catch(eo, "aborting");
+    exit(1);
+}
+
+void openxref(char *filename, char *password)
+{
+    fz_error error;
+    fz_obj *obj;
+
+    error = pdf_newxref(&xref);
+    if (error)
+	die(error);
+
+    error = pdf_loadxref(xref, filename);
+    if (error)
+    {
+	fz_catch(error, "trying to repair");
+	error = pdf_repairxref(xref, filename);
+	if (error)
+	    die(error);
+    }
+
+    error = pdf_decryptxref(xref);
+    if (error)
+	die(error);
+
+    if (xref->crypt)
+    {
+	int okay = pdf_setpassword(xref->crypt, password);
+	if (!okay)
+	    die(fz_throw("invalid password"));
+    }
+
+    /* TODO: move into mupdf lib, see pdfapp_open in pdfapp.c */
+    obj = fz_dictgets(xref->trailer, "Root");
+    if (!obj)
+	die(error);
+
+    error = pdf_loadindirect(&xref->root, xref, obj);
+    if (error)
+	die(error);
+
+    obj = fz_dictgets(xref->trailer, "Info");
+    if (obj)
+    {
+	error = pdf_loadindirect(&xref->info, xref, obj);
+	if (error)
+	    die(error);
+    }
+}
+
+void closexref()
+{
+    pdf_closexref(xref);
+    xref = nil;
+}
+
+int showcolumn;
+
+void showusage(void)
+{
+    fprintf(stderr, "usage: pdfextract [-d password] <file> [object numbers]\n");
+    fprintf(stderr, "  -d  \tdecrypt password\n");
+    exit(1);
+}
+
+int isimage(fz_obj *obj)
+{
+    fz_obj *type = fz_dictgets(obj, "Subtype");
+    return fz_isname(type) && !strcmp(fz_toname(type), "Image");
+}
+
+int isfontdesc(fz_obj *obj)
+{
+    fz_obj *type = fz_dictgets(obj, "Type");
+    return fz_isname(type) && !strcmp(fz_toname(type), "FontDescriptor");
+}
+
+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);
+    if (error)
+        die(error);
+
+    error = pdf_newstore(&xref->store);
+    if (error)
+        die(error);
+
+    error = pdf_loadimage(&img, xref, obj, 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 savefont(fz_obj *dict, int num, int gen)
+{
+    fz_error error;
+    char name[1024];
+    char *subtype;
+    fz_buffer *buf;
+    fz_obj *stream;
+    fz_obj *obj;
+    char *ext;
+    FILE *f;
+    unsigned char *p;
+    char *fontname;
+
+    obj = fz_dictgets(dict, "FontName");
+    if (obj)
+        fontname = fz_toname(obj);
+
+    obj = fz_dictgets(dict, "FontFile");
+    if (obj)
+    {
+        stream = obj;
+        ext = "pfa";
+    }
+
+    obj = fz_dictgets(dict, "FontFile2");
+    if (obj)
+    {
+        stream = obj;
+        ext = "ttf";
+    }
+
+    obj = fz_dictgets(dict, "FontFile3");
+    if (obj)
+    {
+
+        stream = obj;
+
+        error = pdf_resolve(&obj, xref);
+        if (error)
+            die(error);
+
+        obj = fz_dictgets(obj, "Subtype");
+        if (obj && !fz_isname(obj))
+            die(fz_throw("Invalid font descriptor subtype"));
+
+        subtype = fz_toname(obj);
+        if (!strcmp(subtype, "Type1C"))
+            ext = "cff";
+        else if (!strcmp(subtype, "CIDFontType0C"))
+            ext = "cid";
+        else
+            die(fz_throw("Unhandled font type '%s'", subtype));
+    }
+
+    if (!stream)
+    {
+        fz_warn("Unhandled font type");
+        return;
+    }
+
+    error = fz_newbuffer(&buf, 0);
+    if (error)
+        die(error);
+
+    error = pdf_loadstream(&buf, xref, fz_tonum(stream), fz_togen(stream));
+    if (error)
+        die(error);
+
+    sprintf(name, "%s-%04d.%s", fontname, num, ext);
+
+    f = fopen(name, "wb");
+    if (f == NULL)
+        die(fz_throw("Error creating image file"));
+
+    for (p = buf->rp; p < buf->wp; p ++)
+        fprintf(f, "%c", *p);
+
+    if (fclose(f) < 0)
+        die(fz_throw("Error closing image file"));
+
+    fz_dropbuffer(buf);
+}
+
+void showobject(int num, int gen)
+{
+    fz_error error;
+    fz_obj *obj;
+
+    if (!xref)
+	die(fz_throw("no file specified"));
+
+    error = pdf_loadobject(&obj, xref, num, gen);
+    if (error)
+	die(error);
+
+    if (isimage(obj))
+        saveimage(obj, num, gen);
+    else if (isfontdesc(obj))
+        savefont(obj, num, gen);
+
+    fz_dropobj(obj);
+}
+
+int main(int argc, char **argv)
+{
+    char *password = "";
+    int c, o;
+
+    while ((c = getopt(argc, argv, "d:")) != -1)
+    {
+	switch (c)
+	{
+	    case 'd': password = optarg; break;
+	    default:
+		      showusage();
+		      break;
+	}
+    }
+
+    if (optind == argc)
+	showusage();
+
+    openxref(argv[optind++], password);
+
+    if (optind == argc)
+        for (o = 0; o < xref->len; o++)
+            showobject(o, 0);
+    else
+        while (optind < argc)
+        {
+            showobject(atoi(argv[optind]), 0);
+            optind++;
+        }
+
+    closexref();
+}
+
author	Sebastian Rasmussen <sebras@hotmail.com>	2009-05-23 02:16:50 +0200
committer	Sebastian Rasmussen <sebras@hotmail.com>	2009-05-23 02:16:50 +0200
commit	da6f61c67c5c591f805a6a933ff7ce03c85563e7 (patch)
tree	9786c3f40dbf571f93650676b5f489bf0e9902ed
parent	ac244e57a9c64451b75447620509b038e239e0e8 (diff)
download	mupdf-da6f61c67c5c591f805a6a933ff7ce03c85563e7.tar.xz