modules/ar.c

   1 // This file is part of Deark.
   2 // Copyright (C) 2016 Jason Summers
   3 // See the file COPYING for terms of use.
   4
   5 #include <deark-config.h>
   6 #include <deark-private.h>
   7 DE_DECLARE_MODULE(de_module_ar);
   8
   9 typedef struct localctx_struct {
  10         i64 extended_name_table_pos; // 0=none
  11         i64 extended_name_table_size;
  12 } lctx;
  13
  14 static int do_ar_item(deark *c, lctx *d, i64 pos1, i64 *p_item_len)
  15 {
  16         char name_orig[17];
  17         size_t name_orig_len;
  18         de_ucstring *rawname_ucstring = NULL;
  19         de_ucstring *filename_ucstring = NULL;
  20         char timestamp_buf[64];
  21         i64 mod_time;
  22         i64 file_mode;
  23         i64 file_offset;
  24         i64 file_size = 0;
  25         i64 name_offset;
  26         de_finfo *fi = NULL;
  27         i64 k;
  28         int retval = 0;
  29         int ret;
  30         i64 foundpos;
  31         i64 ext_name_len;
  32
  33         de_dbg(c, "archive member at %d", (int)pos1);
  34         de_dbg_indent(c, 1);
  35
  36         fi = de_finfo_create(c);
  37
  38         de_read((u8*)name_orig, pos1, 16);
  39         // Strip trailing spaces
  40         name_orig[16] = '\0';
  41         for(k=15; k>=0; k--) {
  42                 if(name_orig[k]!=' ') break;
  43                 name_orig[k]='\0';
  44         }
  45         name_orig_len = de_strlen(name_orig);
  46
  47         rawname_ucstring = ucstring_create(c);
  48         ucstring_append_bytes(rawname_ucstring, (const u8*)name_orig, name_orig_len, 0, DE_ENCODING_UTF8);
  49
  50         de_dbg(c, "member raw name: \"%s\"", ucstring_getpsz(rawname_ucstring));
  51
  52         (void)dbuf_read_ascii_number(c->infile, pos1+16, 12, 10, &mod_time);
  53         de_unix_time_to_timestamp(mod_time, &fi->timestamp[DE_TIMESTAMPIDX_MODIFY], 0x1);
  54         de_timestamp_to_string(&fi->timestamp[DE_TIMESTAMPIDX_MODIFY], timestamp_buf, sizeof(timestamp_buf), 0);
  55         de_dbg(c, "mod time: %" I64_FMT " (%s)", mod_time, timestamp_buf);
  56
  57         (void)dbuf_read_ascii_number(c->infile, pos1+40, 8, 8, &file_mode);
  58         de_dbg(c, "file mode: octal(%06o)", (int)file_mode);
  59         if((file_mode & 0111)!=0) {
  60                 fi->mode_flags |= DE_MODEFLAG_EXE;
  61         }
  62         else {
  63                 fi->mode_flags |= DE_MODEFLAG_NONEXE;
  64         }
  65
  66         file_offset = pos1 + 60;
  67         (void)dbuf_read_ascii_number(c->infile, pos1+48, 10, 10, &file_size);
  68         de_dbg(c, "member data at %d, size: %d",
  69                 (int)file_offset, (int)file_size);
  70
  71         if(name_orig_len<1) {
  72                 de_warn(c, "Missing filename");
  73                 retval = 1;
  74                 goto done;
  75         }
  76         else if(!de_strcmp(name_orig, "/")) {
  77                 de_dbg(c, "symbol table (ignoring)");
  78                 retval = 1;
  79                 goto done;
  80         }
  81         else if(!de_strcmp(name_orig, "//")) {
  82                 de_dbg(c, "extended name table");
  83                 d->extended_name_table_pos = file_offset;
  84                 d->extended_name_table_size = file_size;
  85                 retval = 1;
  86                 goto done;
  87         }
  88         else if(name_orig[0]=='/' && name_orig[1]>='0' && name_orig[1]<='9') {
  89                 if(d->extended_name_table_pos==0) {
  90                         de_err(c, "Missing extended name table");
  91                         goto done;
  92                 }
  93
  94                 (void)dbuf_read_ascii_number(c->infile, pos1+1, 15, 10, &name_offset);
  95                 if(name_offset >= d->extended_name_table_size) {
  96                         goto done;
  97                 }
  98
  99                 ret = dbuf_search_byte(c->infile, '\x0a', d->extended_name_table_pos+name_offset,
 100                         d->extended_name_table_size-name_offset, &foundpos);
 101                 if(!ret) goto done;
 102                 ext_name_len = foundpos - (d->extended_name_table_pos+name_offset);
 103
 104                 // TODO: Consolidate the filename extraction code.
 105                 if(ext_name_len>0 && de_getbyte(d->extended_name_table_pos+name_offset+ext_name_len-1)=='/') {
 106                         // Strip trailing slash.
 107                         ext_name_len--;
 108                 }
 109
 110                 filename_ucstring = ucstring_create(c);
 111                 dbuf_read_to_ucstring(c->infile, d->extended_name_table_pos+name_offset,
 112                         ext_name_len, filename_ucstring, 0, DE_ENCODING_UTF8);
 113
 114                 de_dbg(c, "extended filename: \"%s\"", ucstring_getpsz(filename_ucstring));
 115
 116                 de_finfo_set_name_from_ucstring(c, fi, filename_ucstring, 0);
 117                 fi->original_filename_flag = 1;
 118         }
 119         else if(name_orig[0]=='/') {
 120                 de_warn(c, "Unsupported extension: \"%s\"", ucstring_getpsz(rawname_ucstring));
 121                 retval = 1;
 122                 goto done;
 123         }
 124         else {
 125                 i64 adjusted_len;
 126
 127                 filename_ucstring = ucstring_create(c);
 128
 129                 adjusted_len = name_orig_len;
 130                 if(name_orig[name_orig_len-1]=='/') {
 131                         // Filenames are often terminated with a '/', to allow for
 132                         // trailing spaces. Strip off the '/'.
 133                         adjusted_len--;
 134                 }
 135                 ucstring_append_bytes(filename_ucstring, (u8*)name_orig, adjusted_len,
 136                         0, DE_ENCODING_UTF8);
 137
 138                 de_dbg(c, "filename: \"%s\"", ucstring_getpsz(filename_ucstring));
 139                 de_finfo_set_name_from_ucstring(c, fi, filename_ucstring, 0);
 140                 fi->original_filename_flag = 1;
 141         }
 142
 143         dbuf_create_file_from_slice(c->infile, file_offset, file_size, NULL, fi, 0);
 144
 145         retval = 1;
 146 done:
 147         *p_item_len = 60 + file_size;
 148         if(*p_item_len % 2) (*p_item_len)++; // padding byte
 149         de_dbg_indent(c, -1);
 150         de_finfo_destroy(c, fi);
 151         ucstring_destroy(rawname_ucstring);
 152         ucstring_destroy(filename_ucstring);
 153         return retval;
 154 }
 155
 156 static void de_run_ar(deark *c, de_module_params *mparams)
 157 {
 158         lctx *d = NULL;
 159         i64 pos;
 160         i64 item_len;
 161         int ret;
 162
 163         d = de_malloc(c, sizeof(lctx));
 164
 165         pos = 8;
 166         while(1) {
 167                 if(pos >= c->infile->len) break;
 168                 ret = do_ar_item(c, d, pos, &item_len);
 169                 if(!ret || item_len<1) break;
 170                 pos += item_len;
 171         }
 172
 173         de_free(c, d);
 174 }
 175
 176 static int de_identify_ar(deark *c)
 177 {
 178         if(!dbuf_memcmp(c->infile, 0, "!<arch>\x0a", 8))
 179                 return 100;
 180         return 0;
 181 }
 182
 183 void de_module_ar(deark *c, struct deark_module_info *mi)
 184 {
 185         mi->id = "ar";
 186         mi->desc = "ar archive";
 187         mi->run_fn = de_run_ar;
 188         mi->identify_fn = de_identify_ar;
 189 }