X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fdata-io%2Fget-data.c;h=a37e3f0326119808e49075542f89afbe6241ba9e;hb=885ac3ed70c6cf89e951792de1b0c991b309b5c4;hp=daec18f41da340c9d0e5b9c882eae4e3c2fbc53d;hpb=261eabd9113876a12994f7b419cbf3cf1ee98453;p=pspp diff --git a/src/language/data-io/get-data.c b/src/language/data-io/get-data.c index daec18f41d..a37e3f0326 100644 --- a/src/language/data-io/get-data.c +++ b/src/language/data-io/get-data.c @@ -1,6 +1,6 @@ /* PSPP - a program for statistical analysis. Copyright (C) 2007, 2008, 2009, 2010, 2011, 2012, - 2013 Free Software Foundation, Inc. + 2013, 2015, 2016 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -57,16 +57,29 @@ static int parse_get_psql (struct lexer *lexer, struct dataset *); int cmd_get_data (struct lexer *lexer, struct dataset *ds) { - struct spreadsheet_read_options opts; char *tok = NULL; - lex_force_match (lexer, T_SLASH); + struct spreadsheet_read_options opts; + + opts.sheet_name = NULL; + opts.sheet_index = -1; + opts.cell_range = NULL; + opts.read_names = false; + opts.asw = -1; + + if (! lex_force_match (lexer, T_SLASH)) + goto error; if (!lex_force_match_id (lexer, "TYPE")) goto error; - lex_force_match (lexer, T_EQUALS); + if (!lex_force_match (lexer, T_EQUALS)) + goto error; + + const char *s = lex_tokcstr (lexer); + + if (s) + tok = strdup (s); - tok = strdup (lex_tokcstr (lexer)); if (lex_match_id (lexer, "TXT")) { free (tok); @@ -77,44 +90,38 @@ cmd_get_data (struct lexer *lexer, struct dataset *ds) free (tok); return parse_get_psql (lexer, ds); } - else if (lex_match_id (lexer, "GNM") || + else if (lex_match_id (lexer, "GNM") || lex_match_id (lexer, "ODS")) { char *filename = NULL; - struct casereader *reader = NULL; - struct dictionary *dict = NULL; - if (!parse_spreadsheet (lexer, &filename, &opts)) goto error; - if ( 0 == strncasecmp (tok, "GNM", 3)) - { - struct spreadsheet *spreadsheet = gnumeric_probe (filename, true); - if (spreadsheet == NULL) - goto error; - reader = gnumeric_make_reader (spreadsheet, &opts); - dict = spreadsheet->dict; - } + struct spreadsheet *spreadsheet = NULL; + if (0 == strncasecmp (tok, "GNM", 3)) + spreadsheet = gnumeric_probe (filename, true); else if (0 == strncasecmp (tok, "ODS", 3)) - { - struct spreadsheet *spreadsheet = ods_probe (filename, true); - if (spreadsheet == NULL) - goto error; - reader = ods_make_reader (spreadsheet, &opts); - dict = spreadsheet->dict; - ods_destroy (spreadsheet); - } + spreadsheet = ods_probe (filename, true); + if (spreadsheet == NULL) + { + msg (SE, _("error reading file `%s'"), filename); + free (filename); + goto error; + } free (filename); + struct casereader *reader = spreadsheet_make_reader (spreadsheet, &opts); if (reader) { - dataset_set_dict (ds, dict); + dataset_set_dict (ds, dict_clone (spreadsheet->dict)); dataset_set_source (ds, reader); free (tok); destroy_spreadsheet_read_info (&opts); + spreadsheet_unref (spreadsheet); return CMD_SUCCESS; } + spreadsheet_unref (spreadsheet); } else msg (SE, _("Unsupported TYPE %s."), tok); @@ -136,12 +143,14 @@ parse_get_psql (struct lexer *lexer, struct dataset *ds) psql.bsize = -1; ds_init_empty (&psql.sql); - lex_force_match (lexer, T_SLASH); + if (! lex_force_match (lexer, T_SLASH)) + goto error; if (!lex_force_match_id (lexer, "CONNECT")) goto error; - lex_force_match (lexer, T_EQUALS); + if (! lex_force_match (lexer, T_EQUALS)) + goto error; if (!lex_force_string (lexer)) goto error; @@ -150,28 +159,34 @@ parse_get_psql (struct lexer *lexer, struct dataset *ds) lex_get (lexer); - while (lex_match (lexer, T_SLASH) ) + while (lex_match (lexer, T_SLASH)) { - if ( lex_match_id (lexer, "ASSUMEDSTRWIDTH")) + if (lex_match_id (lexer, "ASSUMEDSTRWIDTH")) { lex_match (lexer, T_EQUALS); - psql.str_width = lex_integer (lexer); - lex_get (lexer); + if (lex_force_int_range (lexer, "ASSUMEDSTRWIDTH", 1, 32767)) + { + psql.str_width = lex_integer (lexer); + lex_get (lexer); + } } - else if ( lex_match_id (lexer, "BSIZE")) + else if (lex_match_id (lexer, "BSIZE")) { lex_match (lexer, T_EQUALS); - psql.bsize = lex_integer (lexer); - lex_get (lexer); + if (lex_force_int_range (lexer, "BSIZE", 1, INT_MAX)) + { + psql.bsize = lex_integer (lexer); + lex_get (lexer); + } } - else if ( lex_match_id (lexer, "UNENCRYPTED")) + else if (lex_match_id (lexer, "UNENCRYPTED")) { psql.allow_clear = true; } else if (lex_match_id (lexer, "SQL")) { lex_match (lexer, T_EQUALS); - if ( ! lex_force_string (lexer) ) + if (! lex_force_string (lexer)) goto error; ds_put_substring (&psql.sql, lex_tokss (lexer)); @@ -182,7 +197,7 @@ parse_get_psql (struct lexer *lexer, struct dataset *ds) struct dictionary *dict = NULL; struct casereader *reader = psql_open_reader (&psql, &dict); - if ( reader ) + if (reader) { dataset_set_dict (ds, dict); dataset_set_source (ds, reader); @@ -203,7 +218,7 @@ parse_get_psql (struct lexer *lexer, struct dataset *ds) } static bool -parse_spreadsheet (struct lexer *lexer, char **filename, +parse_spreadsheet (struct lexer *lexer, char **filename, struct spreadsheet_read_options *opts) { opts->sheet_index = 1; @@ -212,12 +227,14 @@ parse_spreadsheet (struct lexer *lexer, char **filename, opts->read_names = true; opts->asw = -1; - lex_force_match (lexer, T_SLASH); + if (! lex_force_match (lexer, T_SLASH)) + goto error; if (!lex_force_match_id (lexer, "FILE")) goto error; - lex_force_match (lexer, T_EQUALS); + if (! lex_force_match (lexer, T_EQUALS)) + goto error; if (!lex_force_string (lexer)) goto error; @@ -226,20 +243,23 @@ parse_spreadsheet (struct lexer *lexer, char **filename, lex_get (lexer); - while (lex_match (lexer, T_SLASH) ) + while (lex_match (lexer, T_SLASH)) { - if ( lex_match_id (lexer, "ASSUMEDSTRWIDTH")) + if (lex_match_id (lexer, "ASSUMEDSTRWIDTH")) { lex_match (lexer, T_EQUALS); - opts->asw = lex_integer (lexer); - lex_get (lexer); + if (lex_force_int_range (lexer, "ASSUMEDSTRWIDTH", 1, 32767)) + { + opts->asw = lex_integer (lexer); + lex_get (lexer); + } } else if (lex_match_id (lexer, "SHEET")) { lex_match (lexer, T_EQUALS); if (lex_match_id (lexer, "NAME")) { - if ( ! lex_force_string (lexer) ) + if (! lex_force_string (lexer)) goto error; opts->sheet_name = ss_xstrdup (lex_tokss (lexer)); @@ -249,12 +269,9 @@ parse_spreadsheet (struct lexer *lexer, char **filename, } else if (lex_match_id (lexer, "INDEX")) { + if (!lex_force_int_range (lexer, "INDEX", 1, INT_MAX)) + goto error; opts->sheet_index = lex_integer (lexer); - if (opts->sheet_index <= 0) - { - msg (SE, _("The sheet index must be greater than or equal to 1")); - goto error; - } lex_get (lexer); } else @@ -274,7 +291,7 @@ parse_spreadsheet (struct lexer *lexer, char **filename, } else if (lex_match_id (lexer, "RANGE")) { - if ( ! lex_force_string (lexer) ) + if (! lex_force_string (lexer)) goto error; opts->cell_range = ss_xstrdup (lex_tokss (lexer)); @@ -291,7 +308,7 @@ parse_spreadsheet (struct lexer *lexer, char **filename, { lex_match (lexer, T_EQUALS); - if ( lex_match_id (lexer, "ON")) + if (lex_match_id (lexer, "ON")) { opts->read_names = true; } @@ -355,11 +372,13 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) enum data_parser_type type; bool has_type; - lex_force_match (lexer, T_SLASH); + if (! lex_force_match (lexer, T_SLASH)) + goto error; if (!lex_force_match_id (lexer, "FILE")) goto error; - lex_force_match (lexer, T_EQUALS); + if (! lex_force_match (lexer, T_EQUALS)) + goto error; fh = fh_parse (lexer, FH_REF_FILE | FH_REF_INLINE, NULL); if (fh == NULL) goto error; @@ -369,6 +388,7 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) data_parser_set_type (parser, DP_DELIMITED); data_parser_set_span (parser, false); data_parser_set_quotes (parser, ss_empty ()); + data_parser_set_quote_escape (parser, true); data_parser_set_empty_line_has_field (parser, true); for (;;) @@ -399,7 +419,7 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) DP_DELIMITED, &has_type); else { - lex_error_expecting (lexer, "FIXED", "DELIMITED", NULL_SENTINEL); + lex_error_expecting (lexer, "FIXED", "DELIMITED"); goto error; } if (!ok) @@ -408,13 +428,8 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) else if (lex_match_id (lexer, "FIRSTCASE")) { lex_match (lexer, T_EQUALS); - if (!lex_force_int (lexer)) + if (!lex_force_int_range (lexer, "FIRSTCASE", 1, INT_MAX)) goto error; - if (lex_integer (lexer) < 1) - { - msg (SE, _("Value of FIRSTCASE must be 1 or greater.")); - goto error; - } data_parser_set_skip (parser, lex_integer (lexer) - 1); lex_get (lexer); } @@ -437,7 +452,7 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) } else { - lex_error_expecting (lexer, "LINE", "VARIABLES", NULL_SENTINEL); + lex_error_expecting (lexer, "LINE", "VARIABLES"); goto error; } } @@ -446,13 +461,8 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) if (!set_type (parser, "FIXCASE", DP_FIXED, &has_type)) goto error; lex_match (lexer, T_EQUALS); - if (!lex_force_int (lexer)) + if (!lex_force_int_range (lexer, "FIXCASE", 1, INT_MAX)) goto error; - if (lex_integer (lexer) < 1) - { - msg (SE, _("Value of FIXCASE must be at least 1.")); - goto error; - } data_parser_set_records (parser, lex_integer (lexer)); lex_get (lexer); } @@ -461,33 +471,22 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) lex_match (lexer, T_EQUALS); if (lex_match (lexer, T_ALL)) { - data_parser_set_case_limit (parser, -1); - data_parser_set_case_percent (parser, 100); + /* Nothing to do. */ } else if (lex_match_id (lexer, "FIRST")) { if (!lex_force_int (lexer)) goto error; - if (lex_integer (lexer) < 1) - { - msg (SE, _("Value of FIRST must be at least 1.")); - goto error; - } - data_parser_set_case_limit (parser, lex_integer (lexer)); lex_get (lexer); } else if (lex_match_id (lexer, "PERCENT")) { if (!lex_force_int (lexer)) goto error; - if (lex_integer (lexer) < 1 || lex_integer (lexer) > 100) - { - msg (SE, _("Value of PERCENT must be between 1 and 100.")); - goto error; - } - data_parser_set_case_percent (parser, lex_integer (lexer)); lex_get (lexer); } + msg (SW, _("Ignoring obsolete IMPORTCASES subcommand. (N OF CASES " + "or SAMPLE may be used to substitute.)")); } else if (lex_match_id_n (lexer, "DELIMITERS", 4)) { @@ -541,14 +540,11 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) data_parser_set_quotes (parser, lex_tokss (lexer)); lex_get (lexer); } - else if (settings_get_syntax () == ENHANCED - && lex_match_id (lexer, "ESCAPE")) - data_parser_set_quote_escape (parser, true); else if (lex_match_id (lexer, "VARIABLES")) break; else { - lex_error_expecting (lexer, "VARIABLES", NULL_SENTINEL); + lex_error_expecting (lexer, "VARIABLES"); goto error; } } @@ -564,48 +560,42 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) while (type == DP_FIXED && lex_match (lexer, T_SLASH)) { - if (!lex_force_int (lexer)) + if (!lex_force_int_range (lexer, NULL, record, + data_parser_get_records (parser))) goto error; - if (lex_integer (lexer) < record) - { - msg (SE, _("The record number specified, %ld, is at or " - "before the previous record, %d. Data " - "fields must be listed in order of " - "increasing record number."), - lex_integer (lexer), record); - goto error; - } - if (lex_integer (lexer) > data_parser_get_records (parser)) - { - msg (SE, _("The record number specified, %ld, exceeds " - "the number of records per case specified " - "on FIXCASE, %d."), - lex_integer (lexer), data_parser_get_records (parser)); - goto error; - } record = lex_integer (lexer); lex_get (lexer); } + const char * tstr = lex_tokcstr (lexer); + if (tstr == NULL) + { + lex_error (lexer, NULL); + goto error; + } + name = xstrdup (tstr); if (!lex_force_id (lexer) - || !dict_id_is_valid (dict, lex_tokcstr (lexer), true)) - goto error; - name = xstrdup (lex_tokcstr (lexer)); + || !dict_id_is_valid (dict, name, true)) + { + goto error; + } lex_get (lexer); - if (type == DP_DELIMITED) { if (!parse_format_specifier (lexer, &input) - || !fmt_check_input (&input)) - goto error; - - output = fmt_for_output_from_input (&input); + || !fmt_check_input (&input)) + { + goto error; + } + output = fmt_for_output_from_input (&input, + settings_get_fmt_settings ()); } else { char fmt_type_name[FMT_TYPE_LEN_MAX + 1]; enum fmt_type fmt_type; - int w, d; + uint16_t w; + uint8_t d; if (!parse_column_range (lexer, 0, &fc, &lc, NULL)) goto error; @@ -618,14 +608,12 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) msg (SE, _("Unknown format type `%s'."), fmt_type_name); goto error; } - /* Compose input format. */ input.type = fmt_type; input.w = lc - fc + 1; input.d = 0; if (!fmt_check_input (&input)) goto error; - /* Compose output format. */ if (w != 0) { @@ -636,9 +624,9 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) goto error; } else - output = fmt_for_output_from_input (&input); + output = fmt_for_output_from_input (&input, + settings_get_fmt_settings ()); } - v = dict_create_var (dict, name, fmt_var_width (&input)); if (v == NULL) { @@ -646,7 +634,6 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) goto error; } var_set_both_formats (v, &output); - if (type == DP_DELIMITED) data_parser_add_delimited_field (parser, &input, var_get_case_index (v), @@ -663,14 +650,14 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) if (reader == NULL) goto error; - data_parser_make_active_file (parser, ds, reader, dict); + data_parser_make_active_file (parser, ds, reader, dict, NULL, NULL); fh_unref (fh); free (encoding); return CMD_SUCCESS; error: data_parser_destroy (parser); - dict_destroy (dict); + dict_unref (dict); fh_unref (fh); free (name); free (encoding); @@ -678,7 +665,7 @@ parse_get_txt (struct lexer *lexer, struct dataset *ds) } -static void +static void destroy_spreadsheet_read_info (struct spreadsheet_read_options *opts) { free (opts->cell_range);