Details | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 4680 | right-hear | 1 | /* |
| 2 | * pdfextract -- the ultimate way to extract images and fonts from pdfs |
||
| 3 | */ |
||
| 4 | |||
| 5 | #include "fitz.h" |
||
| 6 | #include "mupdf.h" |
||
| 7 | |||
| 8 | static pdf_xref *xref = NULL; |
||
| 9 | static int dorgb = 0; |
||
| 10 | |||
| 11 | void die(fz_error error) |
||
| 12 | { |
||
| 13 | fz_catch(error, "aborting"); |
||
| 14 | if (xref) |
||
| 15 | pdf_free_xref(xref); |
||
| 16 | exit(1); |
||
| 17 | } |
||
| 18 | |||
| 19 | static void usage(void) |
||
| 20 | { |
||
| 21 | fprintf(stderr, "usage: pdfextract [options] file.pdf [object numbers]\n"); |
||
| 22 | fprintf(stderr, "\t-p\tpassword\n"); |
||
| 23 | fprintf(stderr, "\t-r\tconvert images to rgb\n"); |
||
| 24 | exit(1); |
||
| 25 | } |
||
| 26 | |||
| 27 | static int isimage(fz_obj *obj) |
||
| 28 | { |
||
| 29 | fz_obj *type = fz_dict_gets(obj, "Subtype"); |
||
| 30 | return fz_is_name(type) && !strcmp(fz_to_name(type), "Image"); |
||
| 31 | } |
||
| 32 | |||
| 33 | static int isfontdesc(fz_obj *obj) |
||
| 34 | { |
||
| 35 | fz_obj *type = fz_dict_gets(obj, "Type"); |
||
| 36 | return fz_is_name(type) && !strcmp(fz_to_name(type), "FontDescriptor"); |
||
| 37 | } |
||
| 38 | |||
| 39 | static void saveimage(int num) |
||
| 40 | { |
||
| 41 | fz_error error; |
||
| 42 | fz_pixmap *img; |
||
| 43 | fz_obj *ref; |
||
| 44 | char name[1024]; |
||
| 45 | |||
| 46 | ref = fz_new_indirect(num, 0, xref); |
||
| 47 | |||
| 48 | /* TODO: detect DCTD and save as jpeg */ |
||
| 49 | |||
| 50 | error = pdf_load_image(&img, xref, ref); |
||
| 51 | if (error) |
||
| 52 | die(error); |
||
| 53 | |||
| 54 | if (dorgb && img->colorspace && img->colorspace != fz_device_rgb) |
||
| 55 | { |
||
| 56 | fz_pixmap *temp; |
||
| 57 | temp = fz_new_pixmap_with_rect(fz_device_rgb, fz_bound_pixmap(img)); |
||
| 58 | fz_convert_pixmap(img, temp); |
||
| 59 | fz_drop_pixmap(img); |
||
| 60 | img = temp; |
||
| 61 | } |
||
| 62 | |||
| 63 | if (img->n <= 4) |
||
| 64 | { |
||
| 65 | sprintf(name, "img-%04d.png", num); |
||
| 66 | printf("extracting image %s\n", name); |
||
| 67 | fz_write_png(img, name, 0); |
||
| 68 | } |
||
| 69 | else |
||
| 70 | { |
||
| 71 | sprintf(name, "img-%04d.pam", num); |
||
| 72 | printf("extracting image %s\n", name); |
||
| 73 | fz_write_pam(img, name, 0); |
||
| 74 | } |
||
| 75 | |||
| 76 | fz_drop_pixmap(img); |
||
| 77 | fz_drop_obj(ref); |
||
| 78 | } |
||
| 79 | |||
| 80 | static void savefont(fz_obj *dict, int num) |
||
| 81 | { |
||
| 82 | fz_error error; |
||
| 83 | char name[1024]; |
||
| 84 | char *subtype; |
||
| 85 | fz_buffer *buf; |
||
| 86 | fz_obj *stream = NULL; |
||
| 87 | fz_obj *obj; |
||
| 88 | char *ext = ""; |
||
| 89 | FILE *f; |
||
| 90 | char *fontname = "font"; |
||
| 91 | int n; |
||
| 92 | |||
| 93 | obj = fz_dict_gets(dict, "FontName"); |
||
| 94 | if (obj) |
||
| 95 | fontname = fz_to_name(obj); |
||
| 96 | |||
| 97 | obj = fz_dict_gets(dict, "FontFile"); |
||
| 98 | if (obj) |
||
| 99 | { |
||
| 100 | stream = obj; |
||
| 101 | ext = "pfa"; |
||
| 102 | } |
||
| 103 | |||
| 104 | obj = fz_dict_gets(dict, "FontFile2"); |
||
| 105 | if (obj) |
||
| 106 | { |
||
| 107 | stream = obj; |
||
| 108 | ext = "ttf"; |
||
| 109 | } |
||
| 110 | |||
| 111 | obj = fz_dict_gets(dict, "FontFile3"); |
||
| 112 | if (obj) |
||
| 113 | { |
||
| 114 | stream = obj; |
||
| 115 | |||
| 116 | obj = fz_dict_gets(obj, "Subtype"); |
||
| 117 | if (obj && !fz_is_name(obj)) |
||
| 118 | die(fz_throw("Invalid font descriptor subtype")); |
||
| 119 | |||
| 120 | subtype = fz_to_name(obj); |
||
| 121 | if (!strcmp(subtype, "Type1C")) |
||
| 122 | ext = "cff"; |
||
| 123 | else if (!strcmp(subtype, "CIDFontType0C")) |
||
| 124 | ext = "cid"; |
||
| 125 | else |
||
| 126 | die(fz_throw("Unhandled font type '%s'", subtype)); |
||
| 127 | } |
||
| 128 | |||
| 129 | if (!stream) |
||
| 130 | { |
||
| 131 | fz_warn("Unhandled font type"); |
||
| 132 | return; |
||
| 133 | } |
||
| 134 | |||
| 135 | buf = fz_new_buffer(0); |
||
| 136 | |||
| 137 | error = pdf_load_stream(&buf, xref, fz_to_num(stream), fz_to_gen(stream)); |
||
| 138 | if (error) |
||
| 139 | die(error); |
||
| 140 | |||
| 141 | sprintf(name, "%s-%04d.%s", fontname, num, ext); |
||
| 142 | printf("extracting font %s\n", name); |
||
| 143 | |||
| 144 | f = fopen(name, "wb"); |
||
| 145 | if (f == NULL) |
||
| 146 | die(fz_throw("Error creating font file")); |
||
| 147 | |||
| 148 | n = fwrite(buf->data, 1, buf->len, f); |
||
| 149 | if (n < buf->len) |
||
| 150 | die(fz_throw("Error writing font file")); |
||
| 151 | |||
| 152 | if (fclose(f) < 0) |
||
| 153 | die(fz_throw("Error closing font file")); |
||
| 154 | |||
| 155 | fz_drop_buffer(buf); |
||
| 156 | } |
||
| 157 | |||
| 158 | static void showobject(int num) |
||
| 159 | { |
||
| 160 | fz_error error; |
||
| 161 | fz_obj *obj; |
||
| 162 | |||
| 163 | if (!xref) |
||
| 164 | die(fz_throw("no file specified")); |
||
| 165 | |||
| 166 | error = pdf_load_object(&obj, xref, num, 0); |
||
| 167 | if (error) |
||
| 168 | die(error); |
||
| 169 | |||
| 170 | if (isimage(obj)) |
||
| 171 | saveimage(num); |
||
| 172 | else if (isfontdesc(obj)) |
||
| 173 | savefont(obj, num); |
||
| 174 | |||
| 175 | fz_drop_obj(obj); |
||
| 176 | } |
||
| 177 | |||
| 178 | int main(int argc, char **argv) |
||
| 179 | { |
||
| 180 | fz_error error; |
||
| 181 | char *infile; |
||
| 182 | char *password = ""; |
||
| 183 | int c, o; |
||
| 184 | |||
| 185 | while ((c = fz_getopt(argc, argv, "p:r")) != -1) |
||
| 186 | { |
||
| 187 | switch (c) |
||
| 188 | { |
||
| 189 | case 'p': password = fz_optarg; break; |
||
| 190 | case 'r': dorgb++; break; |
||
| 191 | default: usage(); break; |
||
| 192 | } |
||
| 193 | } |
||
| 194 | |||
| 195 | if (fz_optind == argc) |
||
| 196 | usage(); |
||
| 197 | |||
| 198 | infile = argv[fz_optind++]; |
||
| 199 | error = pdf_open_xref(&xref, infile, password); |
||
| 200 | if (error) |
||
| 201 | die(fz_rethrow(error, "cannot open input file '%s'", infile)); |
||
| 202 | |||
| 203 | if (fz_optind == argc) |
||
| 204 | { |
||
| 205 | for (o = 0; o < xref->len; o++) |
||
| 206 | showobject(o); |
||
| 207 | } |
||
| 208 | else |
||
| 209 | { |
||
| 210 | while (fz_optind < argc) |
||
| 211 | { |
||
| 212 | showobject(atoi(argv[fz_optind])); |
||
| 213 | fz_optind++; |
||
| 214 | } |
||
| 215 | } |
||
| 216 | |||
| 217 | pdf_free_xref(xref); |
||
| 218 | |||
| 219 | fz_flush_warnings(); |
||
| 220 | |||
| 221 | return 0; |
||
| 222 | } |