src/ssgrep.c

   1 /*
   2  * ssgrep.c: Search spreadsheets of selected strings
   3  *
   4  * Copyright (C) 2008 Jody Goldberg
   5  * Copyright (C) 2008-2009 Morten Welinder (terra@gnome.org)
   6  */
   7 #include <gnumeric-config.h>
   8 #include <gnumeric.h>
   9 #include <libgnumeric.h>
  10 #include <goffice/goffice.h>
  11 #include <command-context-stderr.h>
  12 #include <workbook-view.h>
  13 #include <workbook.h>
  14 #include <application.h>
  15 #include <gutils.h>
  16 #include <gnm-plugin.h>
  17 #include <search.h>
  18 #include <sheet.h>
  19 #include <cell.h>
  20 #include <value.h>
  21 #include <func.h>
  22 #include <parse-util.h>
  23 #include <sheet-object-cell-comment.h>
  24 #include <gnumeric-conf.h>
  25
  26 #include <gsf/gsf-input-stdio.h>
  27 #include <gsf/gsf-input-textline.h>
  28 #include <glib/gi18n.h>
  29 #include <string.h>
  30
  31 static gboolean ssgrep_locus_values = TRUE;
  32 static gboolean ssgrep_locus_expressions = TRUE;
  33 static gboolean ssgrep_locus_results = FALSE;
  34 static gboolean ssgrep_locus_comments = TRUE;
  35 static gboolean ssgrep_locus_scripts = TRUE;
  36 static gboolean ssgrep_ignore_case = FALSE;
  37 static gboolean ssgrep_match_words = FALSE;
  38 static gboolean ssgrep_quiet = FALSE;
  39 static gboolean ssgrep_count = FALSE;
  40 static gboolean ssgrep_print_filenames = (gboolean)2;
  41 static gboolean ssgrep_print_matching_filenames = FALSE;
  42 static gboolean ssgrep_print_nonmatching_filenames = FALSE;
  43 static gboolean ssgrep_print_locus = FALSE;
  44 static gboolean ssgrep_print_type = FALSE;
  45 static char *ssgrep_pattern = NULL;
  46 static gboolean ssgrep_fixed_strings = FALSE;
  47 static gboolean ssgrep_recalc = FALSE;
  48 static gboolean ssgrep_invert_match = FALSE;
  49 static gboolean ssgrep_string_table = FALSE;
  50
  51 static gboolean ssgrep_show_version = FALSE;
  52 static char *ssgrep_pattern_file = NULL;
  53
  54 static gboolean ssgrep_error = FALSE;
  55 static gboolean ssgrep_any_matches = FALSE;
  56
  57 static GOptionEntry const ssgrep_options [] = {
  58         {
  59                 "count", 'c',
  60                 0, G_OPTION_ARG_NONE, &ssgrep_count,
  61                 N_("Only print a count of matches per file"),
  62                 NULL
  63         },
  64         {
  65                 "string-table-count", 'C',
  66                 0, G_OPTION_ARG_NONE, &ssgrep_string_table,
  67                 N_("Search only via the string table, display a count of the references."),
  68                 NULL
  69         },
  70
  71         {
  72                 "pattern-file", 'f',
  73                 0, G_OPTION_ARG_STRING, &ssgrep_pattern_file,
  74                 N_("Get patterns from a file, one per line"),
  75                 N_("FILE")
  76         },
  77
  78         {
  79                 "fixed-strings", 'F',
  80                 0, G_OPTION_ARG_NONE, &ssgrep_fixed_strings,
  81                 N_("Pattern is a set of fixed strings"),
  82                 NULL
  83         },
  84
  85         {
  86                 "with-filename", 'H',
  87                 0, G_OPTION_ARG_NONE, &ssgrep_print_filenames,
  88                 N_("Print the filename for each match"),
  89                 NULL
  90         },
  91
  92         {
  93                 "without-filename", 'h',
  94                 G_OPTION_FLAG_REVERSE, G_OPTION_ARG_NONE, &ssgrep_print_filenames,
  95                 N_("Do not print the filename for each match"),
  96                 NULL
  97         },
  98
  99         {
 100                 "ignore-case", 'i',
 101                 0, G_OPTION_ARG_NONE, &ssgrep_ignore_case,
 102                 N_("Ignore differences in letter case"),
 103                 NULL
 104         },
 105
 106         {
 107                 "files-with-matches", 'l',
 108                 0, G_OPTION_ARG_NONE, &ssgrep_print_matching_filenames,
 109                 N_("Print filenames with matches"),
 110                 NULL
 111         },
 112
 113         {
 114                 "files-without-matches", 'L',
 115                 0, G_OPTION_ARG_NONE, &ssgrep_print_nonmatching_filenames,
 116                 N_("Print filenames without matches"),
 117                 NULL
 118         },
 119
 120         {
 121                 "print-locus", 'n',
 122                 0, G_OPTION_ARG_NONE, &ssgrep_print_locus,
 123                 N_("Print the location of each match"),
 124                 NULL
 125         },
 126
 127         {
 128                 "quiet", 'q',
 129                 0, G_OPTION_ARG_NONE, &ssgrep_quiet,
 130                 N_("Suppress all normal output"),
 131                 NULL
 132         },
 133
 134         {
 135                 "search-results", 'R',
 136                 0, G_OPTION_ARG_NONE, &ssgrep_locus_results,
 137                 N_("Search results of expressions too"),
 138                 NULL
 139         },
 140
 141         {
 142                 "print-type", 'T',
 143                 0, G_OPTION_ARG_NONE, &ssgrep_print_type,
 144                 N_("Print the location type of each match"),
 145                 NULL
 146         },
 147
 148         {
 149                 "invert-match", 'v',
 150                 0, G_OPTION_ARG_NONE, &ssgrep_invert_match,
 151                 N_("Search for cells that do not match"),
 152                 NULL
 153         },
 154
 155         {
 156                 "version", 'V',
 157                 0, G_OPTION_ARG_NONE, &ssgrep_show_version,
 158                 N_("Display program version"),
 159                 NULL
 160         },
 161
 162         {
 163                 "word-regexp", 'w',
 164                 0, G_OPTION_ARG_NONE, &ssgrep_match_words,
 165                 N_("Match only whole words"),
 166                 NULL
 167         },
 168
 169         {
 170                 "recalc", 0,
 171                 0, G_OPTION_ARG_NONE, &ssgrep_recalc,
 172                 N_("Recalculate all cells"),
 173                 NULL
 174         },
 175
 176         /* ---------------------------------------- */
 177
 178         { NULL }
 179 };
 180
 181 typedef struct {
 182         Workbook   *wb;
 183         GHashTable *targets;
 184         GHashTable *results;
 185         char const *lc_code;
 186 } StringTableSearch;
 187
 188 static void
 189 add_result (StringTableSearch *state, char const *clean, unsigned int n)
 190 {
 191         gpointer prev;
 192
 193         if (NULL == state->results)
 194                 state->results = g_hash_table_new (g_str_hash, g_str_equal);
 195         else if (NULL != (prev = g_hash_table_lookup (state->results, clean)))
 196                 n += GPOINTER_TO_UINT (prev);
 197         g_hash_table_replace (state->results, (gpointer) clean, GUINT_TO_POINTER (n));
 198 }
 199
 200 static void
 201 cb_check_strings (G_GNUC_UNUSED gpointer key, gpointer str, gpointer user_data)
 202 {
 203         StringTableSearch *state = user_data;
 204         char *clean = g_utf8_strdown (key, -1);
 205         char const *orig = g_hash_table_lookup (state->targets, clean);
 206         if (NULL != orig)
 207                 add_result (state, clean, go_string_get_ref_count (str));
 208         g_free (clean);
 209 }
 210
 211 static void
 212 cb_check_func (gpointer clean, gpointer orig, gpointer user_data)
 213 {
 214         StringTableSearch *state = user_data;
 215         GnmFunc *func = gnm_func_lookup (clean, state->wb);
 216         if (func && gnm_func_get_in_use (func))
 217                 add_result (state, clean, 1);
 218 }
 219
 220 static void
 221 cb_find_target_in_module (gpointer clean, gpointer orig, gpointer user_data)
 222 {
 223         StringTableSearch *state = user_data;
 224         unsigned n = 0;
 225         char const *ptr = state->lc_code;
 226
 227         while (NULL != (ptr = strstr (ptr, clean))) {
 228                 n++;
 229                 ptr++;
 230         }
 231
 232         if (n > 0)
 233                 add_result (state, clean, n);
 234 }
 235
 236 static void
 237 cb_check_module (gpointer name, gpointer code, gpointer user_data)
 238 {
 239         StringTableSearch *state = user_data;
 240         state->lc_code = g_utf8_strdown (code, -1);
 241         g_hash_table_foreach (state->targets, &cb_find_target_in_module, state);
 242         g_free ((gpointer)state->lc_code);
 243         state->lc_code = NULL;
 244 }
 245
 246 static void
 247 cb_dump_results (gpointer name, gpointer count)
 248 {
 249         g_print ("\t%s : %u\n", (char const *)name, GPOINTER_TO_UINT (count));
 250 }
 251
 252 static void
 253 search_string_table (Workbook *wb, char const *file_name, GHashTable *targets)
 254 {
 255         StringTableSearch        state;
 256         GHashTable *modules;
 257
 258         state.wb        = wb;
 259         state.targets   = targets;
 260         state.results   = NULL;
 261         go_string_foreach_base (&cb_check_strings, &state);
 262         g_hash_table_foreach (targets, &cb_check_func, &state);
 263
 264         if (NULL != (modules = g_object_get_data (G_OBJECT (wb), "VBA")))
 265                 g_hash_table_foreach (modules, &cb_check_module, &state);
 266         if (NULL != state.results) {
 267                 g_print ("%s\n", file_name);
 268                 g_hash_table_foreach (state.results, (GHFunc)&cb_dump_results, NULL);
 269                 g_hash_table_destroy (state.results);
 270         }
 271 }
 272
 273 static void
 274 ssgrep (const char *arg, char const *uri, GOIOContext *ioc, GHashTable *targets, char const *pattern)
 275 {
 276         WorkbookView *wbv;
 277         Workbook *wb;
 278         GnmSearchReplace *search;
 279         GPtrArray *cells;
 280         GPtrArray *matches;
 281         gboolean has_match;
 282
 283         wbv = workbook_view_new_from_uri (uri, NULL, ioc, NULL);
 284         if (wbv == NULL) {
 285                 ssgrep_error = TRUE;
 286                 return;
 287         }
 288         wb = wb_view_get_workbook (wbv);
 289
 290         if (ssgrep_locus_results) {
 291                 if (ssgrep_recalc)
 292                         workbook_recalc_all (wb);
 293                 gnm_app_recalc ();
 294         }
 295
 296         if (ssgrep_string_table) {
 297                 search_string_table (wb, arg, targets);
 298                 g_object_unref (wb);
 299                 return;
 300         }
 301
 302         search = (GnmSearchReplace*)
 303                 g_object_new (GNM_SEARCH_REPLACE_TYPE,
 304                               "search-text", ssgrep_pattern,
 305                               "is-regexp", TRUE,
 306                               "invert", ssgrep_invert_match,
 307                               "ignore-case", ssgrep_ignore_case,
 308                               "match-words", ssgrep_match_words,
 309                               "search-strings", ssgrep_locus_values,
 310                               "search-other-values", ssgrep_locus_values,
 311                               "search-expressions", ssgrep_locus_expressions,
 312                               "search-expression-results", ssgrep_locus_results,
 313                               "search-comments", ssgrep_locus_comments,
 314                               "search-scripts", ssgrep_locus_scripts,
 315                               "sheet", workbook_sheet_by_index (wb, 0),
 316                               "scope", GNM_SRS_WORKBOOK,
 317                               NULL);
 318
 319         cells = gnm_search_collect_cells (search);
 320         matches = gnm_search_filter_matching (search, cells);
 321         has_match = (matches->len > 0);
 322
 323         if (has_match)
 324                 ssgrep_any_matches = TRUE;
 325
 326         if (ssgrep_quiet) {
 327                 /* Nothing */
 328         } else if (ssgrep_print_nonmatching_filenames) {
 329                 if (!has_match)
 330                         g_print ("%s\n", arg);
 331         } else if (ssgrep_print_matching_filenames) {
 332                 if (has_match)
 333                         g_print ("%s\n", arg);
 334         } else if (ssgrep_count) {
 335                 if (ssgrep_print_filenames)
 336                         g_print ("%s:", arg);
 337                 g_print ("%u\n", matches->len);
 338         } else {
 339                 unsigned ui;
 340                 for (ui = 0; ui < matches->len; ui++) {
 341                         const GnmSearchFilterResult *item = g_ptr_array_index (matches, ui);
 342                         char *txt = NULL;
 343                         const char *locus_type = "";
 344
 345                         switch (item->locus) {
 346                         case GNM_SRL_CONTENTS: {
 347                                 GnmCell const *cell =
 348                                         sheet_cell_get (item->ep.sheet,
 349                                                         item->ep.eval.col,
 350                                                         item->ep.eval.row);
 351                                 txt = gnm_cell_get_entered_text (cell);
 352                                 locus_type = _("cell");
 353                                 break;
 354                         }
 355
 356                         case GNM_SRL_VALUE: {
 357                                 GnmCell const *cell =
 358                                         sheet_cell_get (item->ep.sheet,
 359                                                         item->ep.eval.col,
 360                                                         item->ep.eval.row);
 361                                 if (cell && cell->value)
 362                                         txt = value_get_as_string (cell->value);
 363                                 locus_type = _("result");
 364                                 break;
 365                         }
 366
 367                         case GNM_SRL_COMMENT: {
 368                                 GnmComment *comment = sheet_get_comment (item->ep.sheet, &item->ep.eval);
 369                                 txt = g_strdup (cell_comment_text_get (comment));
 370                                 locus_type = _("comment");
 371                                 break;
 372                         }
 373                         default:
 374                                 ; /* Probably should not happen.  */
 375                         }
 376
 377                         if (ssgrep_print_filenames)
 378                                 g_print ("%s:", arg);
 379
 380                         if (ssgrep_print_type)
 381                                 g_print ("%s:", locus_type);
 382
 383                         if (ssgrep_print_locus)
 384                                 g_print ("%s!%s:",
 385                                          item->ep.sheet->name_quoted,
 386                                          cellpos_as_string (&item->ep.eval));
 387
 388                         if (txt) {
 389                                 g_print ("%s\n", txt);
 390                                 g_free (txt);
 391                         } else
 392                                 g_print ("\n");
 393                 }
 394         }
 395
 396         gnm_search_filter_matching_free (matches);
 397         gnm_search_collect_cells_free (cells);
 398         g_object_unref (search);
 399         g_object_unref (wb);
 400 }
 401
 402 /* simple stripped down hash of lower case target, only used for string table
 403  * searches */
 404 static void
 405 add_target (GHashTable *ssgrep_targets, char const *target)
 406 {
 407         char *orig = g_strstrip (g_strdup (target));
 408         char *clean = g_utf8_strdown (orig, -1);
 409         g_hash_table_insert (ssgrep_targets, clean, orig);
 410 }
 411
 412 int
 413 main (int argc, char const **argv)
 414 {
 415         GHashTable      *ssgrep_targets;
 416         GOErrorInfo     *plugin_errs;
 417         GOIOContext     *ioc;
 418         GOCmdContext    *cc;
 419         GOptionContext  *ocontext;
 420         GError          *error = NULL;
 421         int              i, N;
 422         const char *argv_stdin[] = { "fd://1", NULL };
 423
 424         /* No code before here, we need to init threads */
 425         argv = gnm_pre_parse_init (argc, argv);
 426
 427         gnm_conf_set_persistence (FALSE);
 428
 429         ocontext = g_option_context_new (_("PATTERN INFILE..."));
 430         g_option_context_add_main_entries (ocontext, ssgrep_options, GETTEXT_PACKAGE);
 431         g_option_context_add_group        (ocontext, gnm_get_option_group ());
 432         g_option_context_parse (ocontext, &argc, (gchar ***)&argv, &error);
 433         g_option_context_free (ocontext);
 434
 435         if (error) {
 436                 g_printerr (_("%s\nRun '%s --help' to see a full list of available command line options.\n"),
 437                             error->message, g_get_prgname ());
 438                 g_error_free (error);
 439                 return 1;
 440         }
 441
 442         if (ssgrep_show_version) {
 443                 g_printerr (_("version '%s'\ndatadir := '%s'\nlibdir := '%s'\n"),
 444                             GNM_VERSION_FULL, gnm_sys_data_dir (), gnm_sys_lib_dir ());
 445                 return 0;
 446         }
 447
 448         gnm_init ();
 449
 450         ssgrep_targets = g_hash_table_new_full (g_str_hash, g_str_equal, g_free, g_free);
 451         if (ssgrep_pattern_file) {
 452                 char *uri = go_shell_arg_to_uri (ssgrep_pattern_file);
 453                 GsfInput         *input;
 454                 GsfInputTextline *textline;
 455                 GError           *err = NULL;
 456                 const unsigned char *line;
 457                 GString *pat;
 458
 459                 input = go_file_open (uri, &err);
 460                 g_free (uri);
 461
 462                 if (!input) {
 463                         g_printerr (_("%s: Cannot read %s: %s\n"),
 464                                     g_get_prgname (), ssgrep_pattern_file, err->message);
 465                         g_error_free (err);
 466                         return 1;
 467                 }
 468
 469                 textline = (GsfInputTextline *)gsf_input_textline_new (input);
 470                 g_object_unref (input);
 471
 472                 pat = g_string_new (NULL);
 473                 while (NULL != (line = gsf_input_textline_ascii_gets (textline))) {
 474                         if (pat->len)
 475                                 g_string_append_c (pat, '|');
 476
 477                         if (ssgrep_fixed_strings)
 478                                 go_regexp_quote (pat, line);
 479                         else
 480                                 g_string_append (pat, line);
 481
 482                         add_target (ssgrep_targets, line);
 483                 }
 484
 485                 ssgrep_pattern = g_string_free (pat, FALSE);
 486
 487                 g_object_unref (textline);
 488
 489                 i = 1;
 490                 N = argc - i;
 491         } else {
 492                 if (argc < 2) {
 493                         g_printerr (_("%s: Missing pattern\n"), g_get_prgname ());
 494                         return 1;
 495                 }
 496
 497                 if (ssgrep_fixed_strings) {
 498                         GString *pat = g_string_new (NULL);
 499                         go_regexp_quote (pat, argv[1]);
 500                         ssgrep_pattern = g_string_free (pat, FALSE);
 501                 } else
 502                         ssgrep_pattern = g_strdup (argv[1]);
 503                 add_target (ssgrep_targets, argv[1]);
 504
 505                 i = 2;
 506                 N = argc - i;
 507         }
 508
 509         if (argv[i] == NULL) {
 510                 argv = argv_stdin;
 511                 i = 0;
 512                 N = 1;
 513         }
 514
 515         cc = gnm_cmd_context_stderr_new ();
 516         gnm_plugins_init (GO_CMD_CONTEXT (cc));
 517         go_plugin_db_activate_plugin_list (
 518                 go_plugins_get_available_plugins (), &plugin_errs);
 519         if (plugin_errs) {
 520                 /* FIXME: What do we want to do here? */
 521                 go_error_info_free (plugin_errs);
 522         }
 523
 524         ioc = go_io_context_new (cc);
 525         go_io_context_set_num_files (ioc, N);
 526         go_component_set_default_command_context (cc);
 527
 528         if (ssgrep_print_filenames == (gboolean)2)
 529                 ssgrep_print_filenames = (N > 1);
 530
 531         for (; argv[i]; i++) {
 532                 const char *arg = argv[i];
 533                 char *uri = go_shell_arg_to_uri (arg);
 534                 go_io_context_processing_file (ioc, uri);
 535                 ssgrep (arg, uri, ioc, ssgrep_targets, ssgrep_pattern);
 536                 g_free (uri);
 537         }
 538
 539         g_hash_table_destroy (ssgrep_targets);
 540
 541         go_component_set_default_command_context (NULL);
 542         g_object_unref (ioc);
 543
 544         g_object_unref (cc);
 545         gnm_shutdown ();
 546         gnm_pre_parse_shutdown ();
 547
 548         /* This special case matches what "man grep" says.  */
 549         if (ssgrep_quiet && ssgrep_any_matches)
 550                 return 0;
 551
 552         if (ssgrep_error)
 553                 return 2;
 554
 555         return ssgrep_any_matches ? 0 : 1;
 556 }