modules/officeart.c

   1 // This file is part of Deark.
   2 // Copyright (C) 2019 Jason Summers
   3 // See the file COPYING for terms of use.
   4
   5 // Microsoft Office Art / Office Drawing / "Escher" / "Blip"
   6 // Refer to Microsoft's "[MS-ODRAW]" document.
   7 // Found in some PowerPoint and Publisher files.
   8
   9 #include <deark-config.h>
  10 #include <deark-private.h>
  11 #include <deark-fmtutil.h>
  12 DE_DECLARE_MODULE(de_module_officeart);
  13
  14 struct officeart_rectype {
  15         u16 rectype;
  16         u16 flags;
  17         const char *name;
  18         void *reserved;
  19 };
  20
  21 static const struct officeart_rectype officeart_rectype_arr[] = {
  22         { 0xf000, 0, "DggContainer", NULL },
  23         { 0xf001, 0, "BStoreContainer", NULL },
  24         { 0xf006, 0, "FDGGBlock", NULL },
  25         { 0xf007, 0, "FBSE", NULL },
  26         { 0xf00b, 0, "FOPT", NULL },
  27         { 0xf01a, 0, "BlipEMF", NULL },
  28         { 0xf01b, 0, "BlipWMF", NULL },
  29         { 0xf01c, 0, "BlipPICT", NULL },
  30         { 0xf01d, 0, "BlipJPEG", NULL },
  31         { 0xf01e, 0, "BlipPNG", NULL },
  32         { 0xf01f, 0, "BlipDIB", NULL },
  33         { 0xf029, 0, "BlipTIFF", NULL },
  34         { 0xf02a, 0, "BlipJPEG", NULL },
  35         { 0xf11a, 0, "ColorMRUContainer", NULL },
  36         { 0xf11e, 0, "SplitMenuColorContainer", NULL },
  37         { 0xf122, 0, "TertiaryFOPT", NULL }
  38 };
  39
  40 static const char *get_officeart_rectype_name(unsigned int t)
  41 {
  42         size_t k;
  43
  44         for(k=0; k<DE_ARRAYCOUNT(officeart_rectype_arr); k++) {
  45                 if((unsigned int)officeart_rectype_arr[k].rectype == t) {
  46                         return officeart_rectype_arr[k].name;
  47                 }
  48         }
  49         return "?";
  50 }
  51
  52 struct officeartctx {
  53 #define OACTX_STACKSIZE 10
  54         i64 container_end_stack[OACTX_STACKSIZE];
  55         size_t container_end_stackptr;
  56
  57         // Passed to do_OfficeArtStream_record():
  58         i64 record_pos;
  59
  60         // Returned from do_OfficeArtStream_record():
  61         i64 record_bytes_consumed;
  62         int is_container;
  63         i64 container_endpos; // valid if (is_container)
  64 };
  65
  66 static int do_OfficeArtStream_record(deark *c, struct officeartctx *oactx,
  67         dbuf *inf)
  68 {
  69         unsigned int rectype;
  70         unsigned int recinstance;
  71         unsigned int recver;
  72         unsigned int n;
  73         i64 reclen;
  74         i64 extra_bytes = 0;
  75         dbuf *outf = NULL;
  76         const char *ext = "bin";
  77         int has_metafileHeader = 0;
  78         int has_zlib_cmpr = 0;
  79         int is_dib = 0;
  80         int is_pict = 0;
  81         int retval = 0;
  82         int is_blip = 0;
  83         int saved_indent_level;
  84         i64 pos1 = oactx->record_pos;
  85         i64 pos = pos1;
  86
  87         oactx->record_bytes_consumed = 0;
  88         oactx->is_container = 0;
  89         oactx->container_endpos = 0;
  90
  91         de_dbg_indent_save(c, &saved_indent_level);
  92
  93         n = (unsigned int)dbuf_getu16le_p(inf, &pos);
  94         recver = n&0x0f;
  95         if(recver==0x0f) oactx->is_container = 1;
  96         recinstance = n>>4;
  97
  98         rectype = (unsigned int)dbuf_getu16le_p(inf, &pos);
  99         if((rectype&0xf000)!=0xf000) {
 100                 // Assume this is the end of data, not necessarily an error.
 101                 goto done;
 102         }
 103
 104         reclen = dbuf_getu32le_p(inf, &pos);
 105
 106         de_dbg(c, "record at [%"I64_FMT"], ver=0x%x, inst=0x%03x, type=0x%04x (%s), dlen=%"I64_FMT,
 107                 pos1, recver, recinstance,
 108                 rectype, get_officeart_rectype_name(rectype), reclen);
 109         de_dbg_indent(c, 1);
 110
 111         if(pos + reclen > inf->len) goto done;
 112         if(oactx->is_container) {
 113                 // A container is described as *being* its header record. It does have
 114                 // a recLen, but it should be safe to ignore it if all we care about is
 115                 // reading the records at a low level.
 116                 oactx->record_bytes_consumed = 8;
 117                 oactx->container_endpos = oactx->record_pos + 8 + reclen;
 118         }
 119         else {
 120                 oactx->record_bytes_consumed = (pos-pos1) + reclen;
 121         }
 122         retval = 1;
 123
 124         if(rectype>=0xf018 && rectype<=0xf117) is_blip = 1;
 125         if(!is_blip) goto done;
 126
 127         if(rectype==0xf01a) {
 128                 ext = "emf";
 129                 if(recinstance==0x3d4) extra_bytes=50;
 130                 else if(recinstance==0x3d5) extra_bytes=66;
 131                 if(extra_bytes) has_metafileHeader=1;
 132         }
 133         else if(rectype==0xf01b) {
 134                 ext = "wmf";
 135                 if(recinstance==0x216) extra_bytes=50;
 136                 else if(recinstance==0x217) extra_bytes=66;
 137                 if(extra_bytes) has_metafileHeader=1;
 138         }
 139         else if(rectype==0xf01c) {
 140                 ext = "pict";
 141                 if(recinstance==0x542) extra_bytes=50;
 142                 else if(recinstance==0x543) extra_bytes=66;
 143                 if(extra_bytes) has_metafileHeader=1;
 144                 is_pict = 1;
 145         }
 146         else if(rectype==0xf01d) {
 147                 ext = "jpg";
 148                 if(recinstance==0x46a || recinstance==0x6e2) extra_bytes = 17;
 149                 else if(recinstance==0x46b || recinstance==0x6e3) extra_bytes = 33;
 150         }
 151         else if(rectype==0xf01e) {
 152                 ext = "png";
 153                 if(recinstance==0x6e0) extra_bytes = 17;
 154                 else if(recinstance==0x6e1) extra_bytes = 33;
 155         }
 156         else if(rectype==0xf01f) {
 157                 ext = "dib";
 158                 if(recinstance==0x7a8) extra_bytes = 17;
 159                 else if(recinstance==0x7a9) extra_bytes = 33;
 160                 if(extra_bytes) is_dib=1;
 161         }
 162         else if(rectype==0xf029) {
 163                 ext = "tif";
 164                 if(recinstance==0x6e4) extra_bytes = 17;
 165                 else if(recinstance==0x6e5) extra_bytes = 33;
 166         }
 167
 168         if(extra_bytes==0) {
 169                 de_warn(c, "Unsupported OfficeArtBlip format (recInstance=0x%03x, recType=0x%04x)",
 170                         recinstance, rectype);
 171                 goto done;
 172         }
 173
 174         if(has_metafileHeader) {
 175                 // metafileHeader starts at pos+extra_bytes-34
 176                 u8 cmpr = dbuf_getbyte(inf, pos+extra_bytes-2);
 177                 // 0=DEFLATE, 0xfe=NONE
 178                 de_dbg(c, "compression type: %u", (unsigned int)cmpr);
 179                 has_zlib_cmpr = (cmpr==0);
 180         }
 181
 182         pos += extra_bytes;
 183
 184         if(is_dib) {
 185                 de_run_module_by_id_on_slice2(c, "dib", "X", inf, pos, reclen-extra_bytes);
 186                 goto done;
 187         }
 188
 189         outf = dbuf_create_output_file(c, ext, NULL, DE_CREATEFLAG_IS_AUX);
 190         if(is_pict) {
 191                 dbuf_write_zeroes(outf, 512);
 192         }
 193
 194         if(has_zlib_cmpr) {
 195                 i64 cmprlen;
 196
 197                 cmprlen = reclen-extra_bytes;
 198                 fmtutil_decompress_deflate(inf, pos, cmprlen, outf, 0, NULL, DE_DEFLATEFLAG_ISZLIB);
 199                 de_dbg(c, "decompressed %"I64_FMT" to %"I64_FMT" bytes", cmprlen, outf->len);
 200         }
 201         else {
 202                 dbuf_copy(inf, pos, reclen-extra_bytes, outf);
 203         }
 204
 205 done:
 206         de_dbg_indent_restore(c, saved_indent_level);
 207         dbuf_close(outf);
 208         return retval;
 209 }
 210
 211 static void de_run_officeart(deark *c, de_module_params *mparams)
 212 {
 213         struct officeartctx * oactx = NULL;
 214         dbuf *inf = c->infile;
 215         int saved_indent_level;
 216
 217         de_dbg_indent_save(c, &saved_indent_level);
 218         oactx = de_malloc(c, sizeof(struct officeartctx));
 219
 220         oactx->record_pos = 0;
 221         while(1) {
 222                 i64 ret;
 223
 224                 if(oactx->record_pos >= inf->len-8) break;
 225
 226                 // Have we reached the end of any containers?
 227                 while(oactx->container_end_stackptr>0 &&
 228                         oactx->record_pos>=oactx->container_end_stack[oactx->container_end_stackptr-1])
 229                 {
 230                         oactx->container_end_stackptr--;
 231                         de_dbg_indent(c, -1);
 232                 }
 233
 234                 ret = do_OfficeArtStream_record(c, oactx, inf);
 235                 if(!ret || oactx->record_bytes_consumed<=0) break;
 236
 237                 oactx->record_pos += oactx->record_bytes_consumed;
 238
 239                 // Is a new container open?
 240                 if(oactx->is_container && oactx->container_end_stackptr<OACTX_STACKSIZE) {
 241                         oactx->container_end_stack[oactx->container_end_stackptr++] = oactx->container_endpos;
 242                         de_dbg_indent(c, 1);
 243                 }
 244         }
 245
 246         de_free(c, oactx);
 247         de_dbg_indent_restore(c, saved_indent_level);
 248 }
 249
 250 void de_module_officeart(deark *c, struct deark_module_info *mi)
 251 {
 252         mi->id = "officeart";
 253         mi->desc = "Office Art data";
 254         mi->run_fn = de_run_officeart;
 255         mi->flags |= DE_MODFLAG_HIDDEN;
 256 }