rgrep/win32/main.c

   1 /*
   2  *  Regular Pattern Analyzer (RPA)
   3  *  Copyright (c) 2009-2010 Martin Stoilov
   4  *
   5  *  This program is free software: you can redistribute it and/or modify
   6  *  it under the terms of the GNU General Public License as published by
   7  *  the Free Software Foundation, either version 3 of the License, or
   8  *  (at your option) any later version.
   9  *
  10  *  This program is distributed in the hope that it will be useful,
  11  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  *  GNU General Public License for more details.
  14  *
  15  *  You should have received a copy of the GNU General Public License
  16  *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
  17  *
  18  *  Martin Stoilov <martin@rpasearch.com>
  19  */
  20
  21
  22 #include <wchar.h>
  23 #include <windows.h>
  24 #include "rlib/rmem.h"
  25 #include "rlib/rarray.h"
  26 #include "rpa/rpadbex.h"
  27 #include "rpagrep.h"
  28 #include "rpagrepdep.h"
  29
  30 rpa_buffer_t * rpa_buffer_from_wchar(const wchar_t *wstr);
  31
  32
  33 int usage(int argc, const wchar_t *argv[])
  34 {
  35             fwprintf(stderr, L"RPA Grep with RPA Engine: %s \n", rpa_dbex_version());
  36                 fwprintf(stderr, L"Copyright (C) 2010 Martin Stoilov\n\n");
  37
  38                 fwprintf(stderr, L"Usage: \n %s [OPTIONS] <filename>\n", argv[0]);
  39                 fwprintf(stderr, L" OPTIONS:\n");
  40                 fwprintf(stderr, L"\t-e patterns              BNF Expression.\n");
  41                 fwprintf(stderr, L"\t-f patternfile           Read the BNF rules from a file, the last pattern will be executed.\n");
  42                 fwprintf(stderr, L"\t-i                       Ignore case.\n");
  43                 fwprintf(stderr, L"\t-m                       Match.\n");
  44                 fwprintf(stderr, L"\t-p                       Parse.\n");
  45                 fwprintf(stderr, L"\t-l                       Line mode.\n");
  46                 fwprintf(stderr, L"\t-16                      Force UTF16 encoding.\n");
  47                 fwprintf(stderr, L"\t-b                       Force byte encoding.\n");
  48                 fwprintf(stderr, L"\t-d                       Dump a production in a tree format.\n");
  49                 fwprintf(stderr, L"\t-t                       Display time elapsed.\n");
  50                 fwprintf(stderr, L"\t-L, --list-rules         List all patterns.\n");
  51                 fwprintf(stderr, L"\t-v                       Display version information.\n");
  52                 fwprintf(stderr, L"\t-h, --help               Display this help.\n");
  53                 fwprintf(stderr, L"\t    --debug-compile      Display debug compilation information.\n");
  54                 fwprintf(stderr, L"\t    --dump-info          Display rules info.\n");
  55                 fwprintf(stderr, L"\t    --dump-code rule     Display compiled code for rule.\n");
  56                 fwprintf(stderr, L"\t    --dump-alias         Display alias info.\n");
  57                 fwprintf(stderr, L"\t    --dump-records       Display rules parsing records.\n");
  58                 fwprintf(stderr, L"\t    --no-optimizations   Disable optimizations.\n");
  59                 fwprintf(stderr, L"\t    --exec-debug         Execute in debug mode.\n");
  60                 fwprintf(stderr, L"\t    --dissable-cache     Dissable execution cache.\n");
  61
  62                 return 0;
  63 }
  64
  65
  66 int wmain(int argc, const wchar_t* argv[])
  67 {
  68         unsigned long sckb;
  69         int ret, scanned = 0, i;
  70         rpa_grep_t *pGrep = NULL;
  71         DWORD eticks, bticks = GetTickCount();
  72         rarray_t *buffers;
  73
  74         buffers = r_array_create(sizeof(rpa_buffer_t *));
  75         pGrep = rpa_grep_create();
  76
  77         if (argc <= 1) {
  78                 usage(argc, argv);
  79                 goto end;
  80         }
  81
  82         for (i = 1; i < argc; i++) {
  83                 if (wcscmp(argv[i], L"-t") == 0) {
  84                         pGrep->showtime = 1;
  85                 }
  86         }
  87
  88         for (i = 1; i < argc; i++) {
  89                 if (wcscmp(argv[i], L"--help") == 0 || wcscmp(argv[i], L"-help") == 0 || wcscmp(argv[i], L"/?") == 0 || wcscmp(argv[i], L"-h") == 0) {
  90                         usage(argc, argv);
  91                         goto end;
  92                 }
  93         }
  94         for (i = 1; i < argc; i++) {
  95                 if (wcscmp(argv[i], L"-c") == 0) {
  96                         if (++i < argc) {
  97                                 rpa_buffer_t *pattern = rpa_buffer_from_wchar(argv[i]);
  98                                 if (!pattern) {
  99                                         goto error;
 100                                 }
 101                                 rpa_grep_setup_callback(pGrep, pattern);
 102                                 rpa_buffer_destroy(pattern);
 103                         }
 104                 }
 105         }
 106
 107
 108         for (i = 1; i < argc; i++) {
 109                 if (wcscmp(argv[i], L"-f") == 0) {
 110                         if (++i < argc) {
 111                                 rpa_buffer_t *pattern = rpa_buffer_map_file(argv[i]);
 112                                 if (pattern) {
 113                                         ret = rpa_grep_load_pattern(pGrep, pattern);
 114                                         r_array_add(buffers, &pattern);
 115                                 } else {
 116                                         ret = -1;
 117                                 }
 118                                 if (ret < 0)
 119                                         goto error;
 120                         }
 121                 }
 122         }
 123
 124         for (i = 1; i < argc; i++) {
 125                 if (wcscmp(argv[i], L"-e") == 0) {
 126                         if (++i < argc) {
 127                                 rpa_buffer_t *pattern = rpa_buffer_from_wchar(argv[i]);
 128                                 if (!pattern) {
 129                                         goto error;
 130                                 }
 131                                 ret = rpa_grep_load_string_pattern(pGrep, pattern);
 132                                 rpa_buffer_destroy(pattern);
 133                                 if (ret < 0)
 134                                         goto error;
 135
 136                         }
 137                 }
 138         }
 139
 140
 141         for (i = 1; i < argc; i++) {
 142                 if (wcscmp(argv[i], L"--dump-code") == 0) {
 143                         if (rpa_dbex_compile(pGrep->hDbex) == 0) {
 144                                 if (++i < argc) {
 145                                         rpa_dbex_dumpcode(pGrep->hDbex, rpa_dbex_lookup_s(pGrep->hDbex, argv[i]));
 146                                 }
 147                         }
 148                         goto end;
 149                 }
 150         }
 151
 152
 153         for (i = 1; i < argc; i++) {
 154                 if (wcscmp(argv[i], L"--dump-info") == 0) {
 155                         rpa_grep_dump_pattern_info(pGrep);
 156                         goto end;
 157                 }
 158         }
 159
 160         for (i = 1; i < argc; i++) {
 161                 if (wcscmp(argv[i], L"--debug-compile") == 0) {
 162                         rpa_grep_debug_compile(pGrep);
 163                         goto end;
 164                 }
 165         }
 166
 167
 168         for (i = 1; i < argc; i++) {
 169                 if (wcscmp(argv[i], L"--dump-alias") == 0) {
 170                         rpa_grep_dump_alias_info(pGrep);
 171                         goto end;
 172                 }
 173         }
 174
 175         for (i = 1; i < argc; i++) {
 176                 if (wcscmp(argv[i], L"--dump-records") == 0) {
 177                         rpa_grep_dump_pattern_records(pGrep);
 178                         goto end;
 179                 }
 180         }
 181
 182         for (i = 1; i < argc; i++) {
 183                 if (wcscmp(argv[i], L"--exec-debug") == 0) {
 184                         pGrep->execdebug = 1;
 185                 }
 186         }
 187
 188         for (i = 1; i < argc; i++) {
 189                 if (wcscmp(argv[i], L"--dissable-cache") == 0) {
 190                         pGrep->disablecache = 1;
 191                 }
 192         }
 193
 194
 195         if (rpa_dbex_compile(pGrep->hDbex) < 0) {
 196                 rpa_errinfo_t errinfo;
 197                 rpa_dbex_lasterrorinfo(pGrep->hDbex, &errinfo);
 198                 if (errinfo.code == RPA_E_UNRESOLVEDSYMBOL) {
 199                         fprintf(stdout, "ERROR: Unresolved Symbol: %s\n", errinfo.name);
 200                 } else {
 201                         fprintf(stdout, "ERROR %ld: Compilation failed.\n", errinfo.code);
 202                 }
 203                 goto end;
 204         }
 205
 206
 207         for (i = 1; i < argc; i++) {
 208                 if (wcscmp(argv[i], L"-L") == 0) {
 209                         rpa_grep_list_patterns(pGrep);
 210                         goto end;
 211                 } else if (wcscmp(argv[i], L"-d") == 0) {
 212                         if (++i < argc) {
 213                                 rpa_buffer_t *pattern = rpa_buffer_from_wchar(argv[i]);
 214                                 if (!pattern) {
 215                                         goto error;
 216                                 }
 217                                 rpa_grep_dump_pattern_tree(pGrep, pattern);
 218                                 rpa_buffer_destroy(pattern);
 219                                 goto end;
 220                         }
 221                 } else if (wcscmp(argv[i], L"-i") == 0) {
 222                         pGrep->icase = 1;
 223                 } else if (wcscmp(argv[i], L"-l") == 0) {
 224                         pGrep->greptype = RPA_GREPTYPE_SCANLINES;
 225                 } else if (wcscmp(argv[i], L"-m") == 0) {
 226                         pGrep->greptype = RPA_GREPTYPE_MATCH;
 227                 } else if (wcscmp(argv[i], L"-p") == 0) {
 228                         pGrep->greptype = RPA_GREPTYPE_PARSE;
 229                 } else if (wcscmp(argv[i], L"-a") == 0) {
 230                         pGrep->greptype = RPA_GREPTYPE_PARSEAST;
 231                 } else if (wcscmp(argv[i], L"-16") == 0) {
 232                         pGrep->forceEncoding = RPA_GREP_FORCE_UTF16;
 233                 } else if (wcscmp(argv[i], L"-b") == 0) {
 234                         pGrep->forceEncoding = RPA_GREP_FORCE_BYTE;
 235                 }
 236
 237         }
 238
 239
 240         for (i = 1; i < argc; i++) {
 241                 if (wcscmp(argv[i], L"-s") == 0) {
 242                         if (++i < argc) {
 243                                 rpa_buffer_t *buf = rpa_buffer_from_wchar(argv[i]);
 244                                 rpa_grep_scan_buffer(pGrep, buf);
 245                                 rpa_buffer_destroy(buf);
 246                                 ++scanned;
 247                         }
 248                 }
 249         }
 250
 251         /* scan files */
 252         for (i = 1; i < argc; i++) {
 253                 if (argv[i][0] != L'-') {
 254                         ++scanned;
 255                         rpa_grep_scan_path(pGrep, argv[i]);
 256                 } else if (argv[i][1] == L'e' || argv[i][1] == L'f' || argv[i][1] == L'c' || argv[i][1] == L'C'){
 257                         ++i;
 258                 }
 259
 260         }
 261
 262         if (!scanned) {
 263                 rpa_buffer_t *buf = rpa_buffer_loadfile(stdin);
 264                 if (buf) {
 265                         rpa_grep_scan_buffer(pGrep, buf);
 266                         rpa_buffer_destroy(buf);
 267                 }
 268         }
 269
 270
 271         /* scan files */
 272         for (i = 1; i < argc; i++) {
 273                 if (argv[i][0] != '-') {
 274                         ++scanned;
 275                         rpa_grep_scan_path(pGrep, argv[i]);
 276                 } else if (argv[i][1] == L'e' || argv[i][1] == L'f' || argv[i][1] == L'c'){
 277                         ++i;
 278                 }
 279
 280         }
 281
 282         if (!scanned) {
 283                 rpa_buffer_t *buf = rpa_buffer_loadfile(stdin);
 284                 if (buf) {
 285                         rpa_grep_scan_buffer(pGrep, buf);
 286                         rpa_buffer_destroy(buf);
 287                 }
 288         }
 289
 290 end:
 291         for (i = 0; i < r_array_length(buffers); i++) {
 292                 rpa_buffer_destroy(r_array_index(buffers, i, rpa_buffer_t*));
 293         }
 294         r_object_destroy((robject_t*)buffers);
 295         rpa_grep_close(pGrep);
 296
 297
 298         rpa_grep_close(pGrep);
 299         sckb = (unsigned long)(pGrep->scsize/1024);
 300
 301         if (pGrep->showtime) {
 302                 unsigned long milsec;
 303                 unsigned long minutes;
 304                 float sec;
 305                 milsec = pGrep->scanmilisec;
 306                 if (milsec == 0)
 307                         milsec = 1;
 308                 minutes = milsec/60000;
 309                 sec = (milsec%60000)/1000.0;
 310                 fwprintf(stdout, L"\ntime: %0ldm%1.3fs, %ld KB (%ld KB/sec), stack: %ld KB, memory: %ld KB (leaked %ld Bytes), cachehit: %ld \n",
 311                                 minutes, sec, sckb, 1000*sckb/milsec, pGrep->usedstack / 1000, (long)r_debug_get_maxmem()/1000, (long)r_debug_get_allocmem(),
 312                                 pGrep->cachehit);
 313         }
 314
 315         rpa_grep_destroy(pGrep);
 316         return 0;
 317
 318 error:
 319         rpa_grep_destroy(pGrep);
 320         return 1;
 321 }