| 1 | /* |
| 2 | * PDF merge tool: Tool for merging pdf content. |
| 3 | * |
| 4 | * Simple test bed to work with merging pages from multiple PDFs into a single PDF. |
| 5 | */ |
| 6 | |
| 7 | #include "mupdf/fitz.h" |
| 8 | #include "mupdf/pdf.h" |
| 9 | |
| 10 | #include <stdlib.h> |
| 11 | #include <stdio.h> |
| 12 | |
| 13 | static void usage(void) |
| 14 | { |
| 15 | fprintf(stderr, |
| 16 | "usage: mutool merge [-o output.pdf] [-O options] input.pdf [pages] [input2.pdf] [pages2] ...\n" |
| 17 | "\t-o -\tname of PDF file to create\n" |
| 18 | "\t-O -\tcomma separated list of output options\n" |
| 19 | "\tinput.pdf\tname of input file from which to copy pages\n" |
| 20 | "\tpages\tcomma separated list of page numbers and ranges\n\n" |
| 21 | ); |
| 22 | fputs(fz_pdf_write_options_usage, stderr); |
| 23 | exit(1); |
| 24 | } |
| 25 | |
| 26 | static fz_context *ctx = NULL; |
| 27 | static pdf_document *doc_des = NULL; |
| 28 | static pdf_document *doc_src = NULL; |
| 29 | |
| 30 | static void page_merge(int page_from, int page_to, pdf_graft_map *graft_map) |
| 31 | { |
| 32 | pdf_obj *page_ref; |
| 33 | pdf_obj *page_dict = NULL; |
| 34 | pdf_obj *obj; |
| 35 | pdf_obj *ref = NULL; |
| 36 | int i; |
| 37 | |
| 38 | /* Copy as few key/value pairs as we can. Do not include items that reference other pages. */ |
| 39 | static pdf_obj * const copy_list[] = { |
| 40 | PDF_NAME(Contents), |
| 41 | PDF_NAME(Resources), |
| 42 | PDF_NAME(MediaBox), |
| 43 | PDF_NAME(CropBox), |
| 44 | PDF_NAME(BleedBox), |
| 45 | PDF_NAME(TrimBox), |
| 46 | PDF_NAME(ArtBox), |
| 47 | PDF_NAME(Rotate), |
| 48 | PDF_NAME(UserUnit) |
| 49 | }; |
| 50 | |
| 51 | fz_var(ref); |
| 52 | fz_var(page_dict); |
| 53 | |
| 54 | fz_try(ctx) |
| 55 | { |
| 56 | page_ref = pdf_lookup_page_obj(ctx, doc_src, page_from - 1); |
| 57 | pdf_flatten_inheritable_page_items(ctx, page_ref); |
| 58 | |
| 59 | /* Make a new page object dictionary to hold the items we copy from the source page. */ |
| 60 | page_dict = pdf_new_dict(ctx, doc_des, 4); |
| 61 | |
| 62 | pdf_dict_put(ctx, page_dict, PDF_NAME(Type), PDF_NAME(Page)); |
| 63 | |
| 64 | for (i = 0; i < nelem(copy_list); i++) |
| 65 | { |
| 66 | obj = pdf_dict_get(ctx, page_ref, copy_list[i]); |
| 67 | if (obj != NULL) |
| 68 | pdf_dict_put_drop(ctx, page_dict, copy_list[i], pdf_graft_mapped_object(ctx, graft_map, obj)); |
| 69 | } |
| 70 | |
| 71 | /* Add the page object to the destination document. */ |
| 72 | ref = pdf_add_object(ctx, doc_des, page_dict); |
| 73 | |
| 74 | /* Insert it into the page tree. */ |
| 75 | pdf_insert_page(ctx, doc_des, page_to - 1, ref); |
| 76 | } |
| 77 | fz_always(ctx) |
| 78 | { |
| 79 | pdf_drop_obj(ctx, page_dict); |
| 80 | pdf_drop_obj(ctx, ref); |
| 81 | } |
| 82 | fz_catch(ctx) |
| 83 | { |
| 84 | fz_rethrow(ctx); |
| 85 | } |
| 86 | } |
| 87 | |
| 88 | static void merge_range(const char *range) |
| 89 | { |
| 90 | int start, end, i, count; |
| 91 | pdf_graft_map *graft_map; |
| 92 | |
| 93 | count = pdf_count_pages(ctx, doc_src); |
| 94 | graft_map = pdf_new_graft_map(ctx, doc_des); |
| 95 | |
| 96 | fz_try(ctx) |
| 97 | { |
| 98 | while ((range = fz_parse_page_range(ctx, range, &start, &end, count))) |
| 99 | { |
| 100 | if (start < end) |
| 101 | for (i = start; i <= end; ++i) |
| 102 | page_merge(i, -1, graft_map); |
| 103 | else |
| 104 | for (i = start; i >= end; --i) |
| 105 | page_merge(i, -1, graft_map); |
| 106 | } |
| 107 | } |
| 108 | fz_always(ctx) |
| 109 | { |
| 110 | pdf_drop_graft_map(ctx, graft_map); |
| 111 | } |
| 112 | fz_catch(ctx) |
| 113 | { |
| 114 | fz_rethrow(ctx); |
| 115 | } |
| 116 | } |
| 117 | |
| 118 | int pdfmerge_main(int argc, char **argv) |
| 119 | { |
| 120 | pdf_write_options opts = pdf_default_write_options; |
| 121 | char *output = "out.pdf" ; |
| 122 | char *flags = "" ; |
| 123 | char *input; |
| 124 | int c; |
| 125 | |
| 126 | while ((c = fz_getopt(argc, argv, "o:O:" )) != -1) |
| 127 | { |
| 128 | switch (c) |
| 129 | { |
| 130 | case 'o': output = fz_optarg; break; |
| 131 | case 'O': flags = fz_optarg; break; |
| 132 | default: usage(); break; |
| 133 | } |
| 134 | } |
| 135 | |
| 136 | if (fz_optind == argc) |
| 137 | usage(); |
| 138 | |
| 139 | ctx = fz_new_context(NULL, NULL, FZ_STORE_UNLIMITED); |
| 140 | if (!ctx) |
| 141 | { |
| 142 | fprintf(stderr, "error: Cannot initialize MuPDF context.\n" ); |
| 143 | exit(1); |
| 144 | } |
| 145 | |
| 146 | pdf_parse_write_options(ctx, &opts, flags); |
| 147 | |
| 148 | fz_try(ctx) |
| 149 | { |
| 150 | doc_des = pdf_create_document(ctx); |
| 151 | } |
| 152 | fz_catch(ctx) |
| 153 | { |
| 154 | fprintf(stderr, "error: Cannot create destination document.\n" ); |
| 155 | fz_flush_warnings(ctx); |
| 156 | fz_drop_context(ctx); |
| 157 | exit(1); |
| 158 | } |
| 159 | |
| 160 | /* Step through the source files */ |
| 161 | while (fz_optind < argc) |
| 162 | { |
| 163 | input = argv[fz_optind++]; |
| 164 | doc_src = pdf_open_document(ctx, input); |
| 165 | |
| 166 | fz_try(ctx) |
| 167 | { |
| 168 | if (fz_optind == argc || !fz_is_page_range(ctx, argv[fz_optind])) |
| 169 | merge_range("1-N" ); |
| 170 | else |
| 171 | merge_range(argv[fz_optind++]); |
| 172 | } |
| 173 | fz_always(ctx) |
| 174 | pdf_drop_document(ctx, doc_src); |
| 175 | fz_catch(ctx) |
| 176 | fprintf(stderr, "error: Cannot merge document '%s'.\n" , input); |
| 177 | } |
| 178 | |
| 179 | if (fz_optind == argc) |
| 180 | { |
| 181 | fz_try(ctx) |
| 182 | pdf_save_document(ctx, doc_des, output, &opts); |
| 183 | fz_catch(ctx) |
| 184 | fprintf(stderr, "error: Cannot save output file: '%s'.\n" , output); |
| 185 | } |
| 186 | |
| 187 | pdf_drop_document(ctx, doc_des); |
| 188 | fz_flush_warnings(ctx); |
| 189 | fz_drop_context(ctx); |
| 190 | return 0; |
| 191 | } |
| 192 | |