X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fmatrix-data.c;h=a699738ddb04ce8ddbc38d847e371f60fed306bd;hb=e48df05eeeb85838526a03a3371964f5f6b14321;hp=1532d1a1c784f7d07b3a766245f97dc1ad3715b0;hpb=bc963dae9be291ea0a7cccf189d13e00d3797cfd;p=pspp-builds.git diff --git a/src/matrix-data.c b/src/matrix-data.c index 1532d1a1..a699738d 100644 --- a/src/matrix-data.c +++ b/src/matrix-data.c @@ -18,15 +18,17 @@ 02111-1307, USA. */ #include -#include +#include "error.h" #include #include #include #include "algorithm.h" #include "alloc.h" +#include "case.h" #include "command.h" #include "data-in.h" -#include "dfm.h" +#include "dfm-read.h" +#include "dictionary.h" #include "error.h" #include "file-handle.h" #include "lexer.h" @@ -41,6 +43,19 @@ /* FIXME: /N subcommand not implemented. It should be pretty simple, too. */ +/* Different types of variables for MATRIX DATA procedure. Order is + important: these are used for sort keys. */ +enum + { + MXD_SPLIT, /* SPLIT FILE variables. */ + MXD_ROWTYPE, /* ROWTYPE_. */ + MXD_FACTOR, /* Factor variables. */ + MXD_VARNAME, /* VARNAME_. */ + MXD_CONTINUOUS, /* Continuous variables. */ + + MXD_COUNT + }; + /* Format type enums. */ enum format_type { @@ -101,7 +116,7 @@ static const char *content_names[PROX + 1] = struct matrix_data_pgm { struct pool *container; /* Arena used for all allocations. */ - struct file_handle *data_file; /* The data file to be read. */ + struct dfm_reader *reader; /* Data file to read. */ /* Format. */ enum format_type fmt; /* LIST or FREE. */ @@ -126,42 +141,44 @@ struct matrix_data_pgm int contents[EOC * 3 + 1]; /* Contents. */ int n_contents; /* Number of entries. */ - + /* Continuous variables. */ int n_continuous; /* Number of continuous variables. */ int first_continuous; /* Index into default_dict.var of first continuous variable. */ }; +/* Auxiliary data attached to MATRIX DATA variables. */ +struct mxd_var + { + int var_type; /* Variable type. */ + int sub_type; /* Subtype. */ + }; + static const struct case_source_class matrix_data_with_rowtype_source_class; static const struct case_source_class matrix_data_without_rowtype_source_class; -static int compare_variables_by_mxd_vartype (const void *pa, +static int compare_variables_by_mxd_var_type (const void *pa, const void *pb); static void read_matrices_without_rowtype (struct matrix_data_pgm *); static void read_matrices_with_rowtype (struct matrix_data_pgm *); static int string_to_content_type (char *, int *); - -#if DEBUGGING -static void debug_print (void); -#endif +static void attach_mxd_aux (struct variable *, int var_type, int sub_type); int cmd_matrix_data (void) { struct pool *pool; struct matrix_data_pgm *mx; - + struct file_handle *fh = NULL; + unsigned seen = 0; - lex_match_id ("MATRIX"); - lex_match_id ("DATA"); - discard_variables (); pool = pool_create (); mx = pool_alloc (pool, sizeof *mx); mx->container = pool; - mx->data_file = inline_file; + mx->reader = NULL; mx->fmt = LIST; mx->section = LOWER; mx->diag = DIAGONAL; @@ -222,9 +239,8 @@ cmd_matrix_data (void) if (strcmp (v[i], "ROWTYPE_")) { new_var = dict_create_var_assert (default_dict, v[i], 0); - new_var->p.mxd.vartype = MXD_CONTINUOUS; - new_var->p.mxd.subtype = i; - } + attach_mxd_aux (new_var, MXD_CONTINUOUS, i); + } else mx->explicit_rowtype = 1; free (v[i]); @@ -232,18 +248,15 @@ cmd_matrix_data (void) free (v); } - { - mx->rowtype_ = dict_create_var_assert (default_dict, - "ROWTYPE_", 8); - mx->rowtype_->p.mxd.vartype = MXD_ROWTYPE; - mx->rowtype_->p.mxd.subtype = 0; - } + mx->rowtype_ = dict_create_var_assert (default_dict, + "ROWTYPE_", 8); + attach_mxd_aux (mx->rowtype_, MXD_ROWTYPE, 0); } else if (lex_match_id ("FILE")) { lex_match ('='); - mx->data_file = fh_parse_file_handle (); - if (mx->data_file == NULL) + fh = fh_parse (); + if (fh == NULL) goto lossage; } else if (lex_match_id ("FORMAT")) @@ -302,10 +315,9 @@ cmd_matrix_data (void) mx->single_split = dict_create_var_assert (default_dict, tokid, 0); + attach_mxd_aux (mx->single_split, MXD_CONTINUOUS, 0); lex_get (); - mx->single_split->p.mxd.vartype = MXD_CONTINUOUS; - dict_set_split_vars (default_dict, &mx->single_split, 1); } else @@ -326,14 +338,16 @@ cmd_matrix_data (void) for (i = 0; i < split_cnt; i++) { - if (split[i]->p.mxd.vartype != MXD_CONTINUOUS) + struct mxd_var *mv = split[i]->aux; + assert (mv != NULL); + if (mv->var_type != MXD_CONTINUOUS) { msg (SE, _("Split variable %s is already another type."), tokid); goto lossage; } - split[i]->p.mxd.vartype = MXD_SPLIT; - split[i]->p.mxd.subtype = i; + var_clear_aux (split[i]); + attach_mxd_aux (split[i], MXD_SPLIT, i); } } } @@ -356,14 +370,17 @@ cmd_matrix_data (void) for (i = 0; i < mx->n_factors; i++) { - if (mx->factors[i]->p.mxd.vartype != MXD_CONTINUOUS) + struct variable *v = mx->factors[i]; + struct mxd_var *mv = v->aux; + assert (mv != NULL); + if (mv->var_type != MXD_CONTINUOUS) { msg (SE, _("Factor variable %s is already another type."), tokid); goto lossage; } - mx->factors[i]->p.mxd.vartype = MXD_FACTOR; - mx->factors[i]->p.mxd.subtype = i; + var_clear_aux (v); + attach_mxd_aux (v, MXD_FACTOR, i); } } } @@ -543,11 +560,8 @@ cmd_matrix_data (void) } /* Create VARNAME_. */ - { - mx->varname_ = dict_create_var_assert (default_dict, "VARNAME_", 8); - mx->varname_->p.mxd.vartype = MXD_VARNAME; - mx->varname_->p.mxd.subtype = 0; - } + mx->varname_ = dict_create_var_assert (default_dict, "VARNAME_", 8); + attach_mxd_aux (mx->varname_, MXD_VARNAME, 0); /* Sort the dictionary variables into the desired order for the system file output. */ @@ -556,7 +570,7 @@ cmd_matrix_data (void) size_t nv; dict_get_vars (default_dict, &v, &nv, 0); - qsort (v, nv, sizeof *v, compare_variables_by_mxd_vartype); + qsort (v, nv, sizeof *v, compare_variables_by_mxd_var_type); dict_reorder_vars (default_dict, v, nv); free (v); } @@ -578,7 +592,8 @@ cmd_matrix_data (void) for (i = 0; i < dict_get_var_cnt (default_dict); i++) { struct variable *v = dict_get_var (default_dict, i); - int type = v->p.mxd.vartype; + struct mxd_var *mv = v->aux; + int type = mv->var_type; assert (type >= 0 && type < MXD_COUNT); v->print = v->write = fmt_tab[type]; @@ -596,15 +611,17 @@ cmd_matrix_data (void) goto lossage; } -#if DEBUGGING - debug_print (); -#endif + mx->reader = dfm_open_reader (fh); + if (mx->reader == NULL) + goto lossage; if (mx->explicit_rowtype) read_matrices_with_rowtype (mx); else read_matrices_without_rowtype (mx); + dfm_close_reader (mx->reader); + pool_destroy (mx->container); return CMD_SUCCESS; @@ -661,121 +678,35 @@ string_to_content_type (char *s, int *collide) return -1; } -/* Compare two variables using p.mxd.vartype and p.mxd.subtype +/* Compare two variables using p.mxd.var_type and p.mxd.sub_type fields. */ static int -compare_variables_by_mxd_vartype (const void *a_, const void *b_) +compare_variables_by_mxd_var_type (const void *a_, const void *b_) { struct variable *const *pa = a_; struct variable *const *pb = b_; - const struct matrix_data_proc *a = &(*pa)->p.mxd; - const struct matrix_data_proc *b = &(*pb)->p.mxd; - - if (a->vartype != b->vartype) - return a->vartype > b->vartype ? 1 : -1; + const struct mxd_var *a = (*pa)->aux; + const struct mxd_var *b = (*pb)->aux; + + if (a->var_type != b->var_type) + return a->var_type > b->var_type ? 1 : -1; else - return a->subtype < b->subtype ? -1 : a->subtype > b->subtype; + return a->sub_type < b->sub_type ? -1 : a->sub_type > b->sub_type; } -#if DEBUGGING -/* Print out the command as input. */ +/* Attaches a struct mxd_var with the specific member values to + V. */ static void -debug_print (void) +attach_mxd_aux (struct variable *v, int var_type, int sub_type) { - printf ("MATRIX DATA\n\t/VARIABLES="); - - { - int i; - - for (i = 0; i < default_dict.nvar; i++) - printf ("%s ", default_dict.var[i]->name); - } - printf ("\n"); - - printf ("\t/FORMAT="); - if (fmt == LIST) - printf ("LIST"); - else if (fmt == FREE) - printf ("FREE"); - else - assert (0); - if (section == LOWER) - printf (" LOWER"); - else if (section == UPPER) - printf (" UPPER"); - else if (section == FULL) - printf (" FULL"); - else - assert (0); - if (diag == DIAGONAL) - printf (" DIAGONAL\n"); - else if (diag == NODIAGONAL) - printf (" NODIAGONAL\n"); - else - assert (0); - - if (dict_get_split_cnt (default_dict) != 0) - { - int i; - - printf ("\t/SPLIT="); - for (i = 0; i < dict_get_split_cnt (default_dict); i++) - printf ("%s ", dict_get_split_vars (default_dict)[i]->name); - if (single_split) - printf ("\t/* single split"); - printf ("\n"); - } + struct mxd_var *mv; - if (n_factors) - { - int i; - - printf ("\t/FACTORS="); - for (i = 0; i < n_factors; i++) - printf ("%s ", factors[i]->name); - printf ("\n"); - } - - if (cells != -1) - printf ("\t/CELLS=%d\n", cells); - - if (mx->pop_n != -1) - printf ("\t/N=%d\n", mx->pop_n); - - if (mx->n_contents) - { - int i; - int space = 0; - - printf ("\t/CONTENTS="); - for (i = 0; i < mx->n_contents; i++) - { - if (mx->contents[i] == LPAREN) - { - if (space) - printf (" "); - printf ("("); - space = 0; - } - else if (mx->contents[i] == RPAREN) - { - printf (")"); - space = 1; - } - else - { - - assert (mx->contents[i] >= 0 && mx->contents[i] <= PROX); - if (space) - printf (" "); - printf ("%s", content_names[mx->contents[i]]); - space = 1; - } - } - printf ("\n"); - } + assert (v->aux == NULL); + mv = xmalloc (sizeof *mv); + mv->var_type = var_type; + mv->sub_type = sub_type; + var_attach_aux (v, mv, var_dtor_free); } -#endif /* DEBUGGING */ /* Matrix tokenizer. */ @@ -786,6 +717,7 @@ enum matrix_token_type MSTR /* String. */ }; +/* A MATRIX DATA parsing token. */ struct matrix_token { enum matrix_token_type type; @@ -794,10 +726,10 @@ struct matrix_token int length; /* MSTR: tokstr length. */ }; -static int mget_token (struct matrix_token *, struct file_handle *); +static int mget_token (struct matrix_token *, struct dfm_reader *); #if DEBUGGING -#define mget_token(TOKEN, HANDLE) mget_token_dump(TOKEN, HANDLE) +#define mget_token(TOKEN, READER) mget_token_dump(TOKEN, READER) static void mdump_token (const struct matrix_token *token) @@ -817,35 +749,47 @@ mdump_token (const struct matrix_token *token) } static int -mget_token_dump (struct matrix_token *token, struct file_handle *data_file) +mget_token_dump (struct matrix_token *token, struct dfm_reader *reader) { - int result = (mget_token) (token, data_file); + int result = (mget_token) (token, reader); mdump_token (token); return result; } #endif -/* Return the current position in DATA_FILE. */ +/* Return the current position in READER. */ static const char * -context (struct file_handle *data_file) +context (struct dfm_reader *reader) { static char buf[32]; - int len; - char *p = dfm_get_record (data_file, &len); - - if (!p || !len) - strcpy (buf, "at end of line"); - else + + if (dfm_eof (reader)) + strcpy (buf, "at end of file"); + else { - char *cp = buf; - int n_copy = min (10, len); - cp = stpcpy (buf, "before `"); - while (n_copy && isspace ((unsigned char) *p)) - p++, n_copy++; - while (n_copy && !isspace ((unsigned char) *p)) - *cp++ = *p++, n_copy--; - *cp++ = '\''; - *cp = 0; + struct len_string line; + const char *sp; + + dfm_get_record (reader, &line); + sp = ls_c_str (&line); + while (sp < ls_end (&line) && isspace ((unsigned char) *sp)) + sp++; + if (sp >= ls_end (&line)) + strcpy (buf, "at end of line"); + else + { + char *dp; + size_t copy_cnt = 0; + + dp = stpcpy (buf, "before `"); + while (sp < ls_end (&line) && !isspace ((unsigned char) *sp) + && copy_cnt < 10) + { + *dp++ = *sp++; + copy_cnt++; + } + strcpy (dp, "'"); + } } return buf; @@ -853,70 +797,57 @@ context (struct file_handle *data_file) /* Is there at least one token left in the data file? */ static int -another_token (struct file_handle *data_file) +another_token (struct dfm_reader *reader) { - char *cp, *ep; - int len; - for (;;) { - cp = dfm_get_record (data_file, &len); - if (!cp) - return 0; + struct len_string line; + const char *cp; + + if (dfm_eof (reader)) + return 0; + dfm_get_record (reader, &line); - ep = cp + len; - while (isspace ((unsigned char) *cp) && cp < ep) + cp = ls_c_str (&line); + while (isspace ((unsigned char) *cp) && cp < ls_end (&line)) cp++; - if (cp < ep) - break; + if (cp < ls_end (&line)) + { + dfm_forward_columns (reader, cp - ls_c_str (&line)); + return 1; + } - dfm_fwd_record (data_file); + dfm_forward_record (reader); } - - dfm_set_record (data_file, cp); - - return 1; } -/* Parse a MATRIX DATA token from mx->data_file into TOKEN. */ +/* Parse a MATRIX DATA token from READER into TOKEN. */ static int -(mget_token) (struct matrix_token *token, struct file_handle *data_file) +(mget_token) (struct matrix_token *token, struct dfm_reader *reader) { - char *cp, *ep; - int len; + struct len_string line; int first_column; - - for (;;) - { - cp = dfm_get_record (data_file, &len); - if (!cp) - return 0; - - ep = cp + len; - while (isspace ((unsigned char) *cp) && cp < ep) - cp++; + char *cp; - if (cp < ep) - break; + if (!another_token (reader)) + return 0; - dfm_fwd_record (data_file); - } - - dfm_set_record (data_file, cp); - first_column = dfm_get_cur_col (data_file) + 1; + dfm_get_record (reader, &line); + first_column = dfm_column_start (reader); /* Three types of fields: quoted with ', quoted with ", unquoted. */ + cp = ls_c_str (&line); if (*cp == '\'' || *cp == '"') { int quote = *cp; token->type = MSTR; token->string = ++cp; - while (cp < ep && *cp != quote) + while (cp < ls_end (&line) && *cp != quote) cp++; token->length = cp - token->string; - if (cp < ep) + if (cp < ls_end (&line)) cp++; else msg (SW, _("Scope of string exceeds line.")); @@ -926,7 +857,8 @@ static int int is_num = isdigit ((unsigned char) *cp) || *cp == '.'; token->string = cp++; - while (cp < ep && !isspace ((unsigned char) *cp) && *cp != ',' + while (cp < ls_end (&line) + && !isspace ((unsigned char) *cp) && *cp != ',' && *cp != '-' && *cp != '+') { if (isdigit ((unsigned char) *cp)) @@ -962,56 +894,56 @@ static int token->type = MSTR; } - dfm_set_record (data_file, cp); + dfm_forward_columns (reader, cp - ls_c_str (&line)); return 1; } /* Forcibly skip the end of a line for content type CONTENT in - DATA_FILE. */ + READER. */ static int -force_eol (struct file_handle *data_file, const char *content) +force_eol (struct dfm_reader *reader, const char *content) { - char *cp; - int len; - - cp = dfm_get_record (data_file, &len); - if (!cp) + struct len_string line; + const char *cp; + + if (dfm_eof (reader)) return 0; - while (len && isspace (*cp)) - cp++, len--; + dfm_get_record (reader, &line); + + cp = ls_c_str (&line); + while (isspace ((unsigned char) *cp) && cp < ls_end (&line)) + cp++; - if (len) + if (cp < ls_end (&line)) { msg (SE, _("End of line expected %s while reading %s."), - context (data_file), content); + context (reader), content); return 0; } - dfm_fwd_record (data_file); - + dfm_forward_record (reader); return 1; } /* Back end, omitting ROWTYPE_. */ -/* MATRIX DATA data. */ -static double ***nr_data; - -/* Factor values. */ -static double *nr_factor_values; - -/* Largest-numbered cell that we have read in thus far, plus one. */ -static int max_cell_index; - -/* SPLIT FILE variable values. */ -static double *split_values; +struct nr_aux_data + { + struct matrix_data_pgm *mx; /* MATRIX DATA program. */ + double ***data; /* MATRIX DATA data. */ + double *factor_values; /* Factor values. */ + int max_cell_idx; /* Max-numbered cell that we have + read so far, plus one. */ + double *split_values; /* SPLIT FILE variable values. */ + }; -static int nr_read_splits (struct matrix_data_pgm *, int compare); -static int nr_read_factors (struct matrix_data_pgm *, int cell); -static void nr_output_data (struct matrix_data_pgm *, +static int nr_read_splits (struct nr_aux_data *, int compare); +static int nr_read_factors (struct nr_aux_data *, int cell); +static void nr_output_data (struct nr_aux_data *, struct ccase *, write_case_func *, write_case_data); static void matrix_data_read_without_rowtype (struct case_source *source, + struct ccase *, write_case_func *, write_case_data); @@ -1019,23 +951,25 @@ static void matrix_data_read_without_rowtype (struct case_source *source, static void read_matrices_without_rowtype (struct matrix_data_pgm *mx) { + struct nr_aux_data nr; + if (mx->cells == -1) mx->cells = 1; - - split_values = xmalloc (sizeof *split_values - * dict_get_split_cnt (default_dict)); - nr_factor_values = xmalloc (sizeof *nr_factor_values * mx->n_factors * mx->cells); - max_cell_index = 0; + + nr.mx = mx; + nr.data = NULL; + nr.factor_values = xmalloc (sizeof *nr.factor_values * mx->n_factors * mx->cells); + nr.max_cell_idx = 0; + nr.split_values = xmalloc (sizeof *nr.split_values + * dict_get_split_cnt (default_dict)); vfm_source = create_case_source (&matrix_data_without_rowtype_source_class, - mx); + default_dict, &nr); - procedure (NULL, NULL, NULL, NULL); + procedure (NULL, NULL); - free (split_values); - free (nr_factor_values); - - fh_close_handle (mx->data_file); + free (nr.split_values); + free (nr.factor_values); } /* Mirror data across the diagonal of matrix CP which contains @@ -1091,20 +1025,13 @@ fill_matrix (struct matrix_data_pgm *mx, int content, double *cp) If PER_FACTOR is nonzero, then factor information is read from the data file. Data is for cell number CELL. */ static int -nr_read_data_lines (struct matrix_data_pgm *mx, +nr_read_data_lines (struct nr_aux_data *nr, int per_factor, int cell, int content, int compare) { - /* Content type. */ - const int type = content_type[content]; - - /* Number of lines that must be parsed from the data file for this - content type. */ - int n_lines; - - /* Current position in vector or matrix. */ - double *cp; - - /* Counter. */ + struct matrix_data_pgm *mx = nr->mx; + const int type = content_type[content]; /* Content type. */ + int n_lines; /* Number of lines to parse from data file for this type. */ + double *cp; /* Current position in vector or matrix. */ int i; if (type != 1) @@ -1116,7 +1043,7 @@ nr_read_data_lines (struct matrix_data_pgm *mx, n_lines--; } - cp = nr_data[content][cell]; + cp = nr->data[content][cell]; if (type == 1 && mx->section == LOWER && mx->diag == NODIAGONAL) cp += mx->n_continuous; @@ -1124,9 +1051,9 @@ nr_read_data_lines (struct matrix_data_pgm *mx, { int n_cols; - if (!nr_read_splits (mx, 1)) + if (!nr_read_splits (nr, 1)) return 0; - if (per_factor && !nr_read_factors (mx, cell)) + if (per_factor && !nr_read_factors (nr, cell)) return 0; compare = 1; @@ -1155,6 +1082,7 @@ nr_read_data_lines (struct matrix_data_pgm *mx, break; default: assert (0); + abort (); } break; case 2: @@ -1162,6 +1090,7 @@ nr_read_data_lines (struct matrix_data_pgm *mx, break; default: assert (0); + abort (); } { @@ -1170,20 +1099,20 @@ nr_read_data_lines (struct matrix_data_pgm *mx, for (j = 0; j < n_cols; j++) { struct matrix_token token; - if (!mget_token (&token, mx->data_file)) + if (!mget_token (&token, mx->reader)) return 0; if (token.type != MNUM) { msg (SE, _("expecting value for %s %s"), dict_get_var (default_dict, j)->name, - context (mx->data_file)); + context (mx->reader)); return 0; } *cp++ = token.number; } if (mx->fmt != FREE - && !force_eol (mx->data_file, content_names[content])) + && !force_eol (mx->reader, content_names[content])) return 0; debug_printf (("\n")); } @@ -1192,7 +1121,7 @@ nr_read_data_lines (struct matrix_data_pgm *mx, cp += mx->n_continuous - n_cols; } - fill_matrix (mx, content, nr_data[content][cell]); + fill_matrix (mx, content, nr->data[content][cell]); return 1; } @@ -1201,21 +1130,23 @@ nr_read_data_lines (struct matrix_data_pgm *mx, writes them to the output file. Returns success. */ static void matrix_data_read_without_rowtype (struct case_source *source, + struct ccase *c, write_case_func *write_case, write_case_data wc_data) { - struct matrix_data_pgm *mx = source->aux; + struct nr_aux_data *nr = source->aux; + struct matrix_data_pgm *mx = nr->mx; { int *cp; - nr_data = pool_alloc (mx->container, (PROX + 1) * sizeof *nr_data); + nr->data = pool_alloc (mx->container, (PROX + 1) * sizeof *nr->data); { int i; for (i = 0; i <= PROX; i++) - nr_data[i] = NULL; + nr->data[i] = NULL; } for (cp = mx->contents; *cp != EOC; cp++) @@ -1232,12 +1163,12 @@ matrix_data_read_without_rowtype (struct case_source *source, int n_vectors = per_factor ? mx->cells : 1; int i; - nr_data[*cp] = pool_alloc (mx->container, - n_vectors * sizeof **nr_data); + nr->data[*cp] = pool_alloc (mx->container, + n_vectors * sizeof **nr->data); for (i = 0; i < n_vectors; i++) - nr_data[*cp][i] = pool_alloc (mx->container, - n_entries * sizeof ***nr_data); + nr->data[*cp][i] = pool_alloc (mx->container, + n_entries * sizeof ***nr->data); } } } @@ -1246,7 +1177,7 @@ matrix_data_read_without_rowtype (struct case_source *source, { int *bp, *ep, *np; - if (!nr_read_splits (mx, 0)) + if (!nr_read_splits (nr, 0)) return; for (bp = mx->contents; *bp != EOC; bp = np) @@ -1281,16 +1212,16 @@ matrix_data_read_without_rowtype (struct case_source *source, int *cp; for (cp = bp; cp < ep; cp++) - if (!nr_read_data_lines (mx, per_factor, i, *cp, cp != bp)) + if (!nr_read_data_lines (nr, per_factor, i, *cp, cp != bp)) return; } } } - nr_output_data (mx, write_case, wc_data); + nr_output_data (nr, c, write_case, wc_data); if (dict_get_split_cnt (default_dict) == 0 - || !another_token (mx->data_file)) + || !another_token (mx->reader)) return; } } @@ -1299,9 +1230,10 @@ matrix_data_read_without_rowtype (struct case_source *source, values read to the last values read and returns 1 if they're equal, 0 otherwise. */ static int -nr_read_splits (struct matrix_data_pgm *mx, int compare) +nr_read_splits (struct nr_aux_data *nr, int compare) { - static int just_read = 0; + struct matrix_data_pgm *mx = nr->mx; + static int just_read = 0; /* FIXME: WTF? */ size_t split_cnt; size_t i; @@ -1316,9 +1248,11 @@ nr_read_splits (struct matrix_data_pgm *mx, int compare) if (mx->single_split) { - if (!compare) - split_values[0] - = ++dict_get_split_vars (default_dict)[0]->p.mxd.subtype; + if (!compare) + { + struct mxd_var *mv = dict_get_split_vars (default_dict)[0]->aux; + nr->split_values[0] = ++mv->sub_type; + } return 1; } @@ -1329,21 +1263,22 @@ nr_read_splits (struct matrix_data_pgm *mx, int compare) for (i = 0; i < split_cnt; i++) { struct matrix_token token; - if (!mget_token (&token, mx->data_file)) + if (!mget_token (&token, mx->reader)) return 0; if (token.type != MNUM) { msg (SE, _("Syntax error expecting SPLIT FILE value %s."), - context (mx->data_file)); + context (mx->reader)); return 0; } if (!compare) - split_values[i] = token.number; - else if (split_values[i] != token.number) + nr->split_values[i] = token.number; + else if (nr->split_values[i] != token.number) { msg (SE, _("Expecting value %g for %s."), - split_values[i], dict_get_split_vars (default_dict)[i]->name); + nr->split_values[i], + dict_get_split_vars (default_dict)[i]->name); return 0; } } @@ -1355,20 +1290,21 @@ nr_read_splits (struct matrix_data_pgm *mx, int compare) values read to the last values read and returns 1 if they're equal, 0 otherwise. */ static int -nr_read_factors (struct matrix_data_pgm *mx, int cell) +nr_read_factors (struct nr_aux_data *nr, int cell) { + struct matrix_data_pgm *mx = nr->mx; int compare; if (mx->n_factors == 0) return 1; - assert (max_cell_index >= cell); - if (cell != max_cell_index) + assert (nr->max_cell_idx >= cell); + if (cell != nr->max_cell_idx) compare = 1; else { compare = 0; - max_cell_index++; + nr->max_cell_idx++; } { @@ -1377,22 +1313,22 @@ nr_read_factors (struct matrix_data_pgm *mx, int cell) for (i = 0; i < mx->n_factors; i++) { struct matrix_token token; - if (!mget_token (&token, mx->data_file)) + if (!mget_token (&token, mx->reader)) return 0; if (token.type != MNUM) { msg (SE, _("Syntax error expecting factor value %s."), - context (mx->data_file)); + context (mx->reader)); return 0; } if (!compare) - nr_factor_values[i + mx->n_factors * cell] = token.number; - else if (nr_factor_values[i + mx->n_factors * cell] != token.number) + nr->factor_values[i + mx->n_factors * cell] = token.number; + else if (nr->factor_values[i + mx->n_factors * cell] != token.number) { msg (SE, _("Syntax error expecting value %g for %s %s."), - nr_factor_values[i + mx->n_factors * cell], - mx->factors[i]->name, context (mx->data_file)); + nr->factor_values[i + mx->n_factors * cell], + mx->factors[i]->name, context (mx->reader)); return 0; } } @@ -1405,16 +1341,17 @@ nr_read_factors (struct matrix_data_pgm *mx, int cell) CP to the active file. */ static void dump_cell_content (struct matrix_data_pgm *mx, int content, double *cp, + struct ccase *c, write_case_func *write_case, write_case_data wc_data) { int type = content_type[content]; { - st_bare_pad_copy (temp_case->data[mx->rowtype_->fv].s, + st_bare_pad_copy (case_data_rw (c, mx->rowtype_->fv)->s, content_names[content], 8); if (type != 1) - memset (&temp_case->data[mx->varname_->fv].s, ' ', 8); + memset (case_data_rw (c, mx->varname_->fv)->s, ' ', 8); } { @@ -1428,11 +1365,11 @@ dump_cell_content (struct matrix_data_pgm *mx, int content, double *cp, for (j = 0; j < mx->n_continuous; j++) { int fv = dict_get_var (default_dict, mx->first_continuous + j)->fv; - temp_case->data[fv].f = *cp; + case_data_rw (c, fv)->f = *cp; cp++; } if (type == 1) - st_bare_pad_copy (temp_case->data[mx->varname_->fv].s, + st_bare_pad_copy (case_data_rw (c, mx->varname_->fv)->s, dict_get_var (default_dict, mx->first_continuous + i)->name, 8); @@ -1443,17 +1380,20 @@ dump_cell_content (struct matrix_data_pgm *mx, int content, double *cp, /* Finally dump out everything from nr_data[] to the output file. */ static void -nr_output_data (struct matrix_data_pgm *mx, +nr_output_data (struct nr_aux_data *nr, struct ccase *c, write_case_func *write_case, write_case_data wc_data) { + struct matrix_data_pgm *mx = nr->mx; + { struct variable *const *split; size_t split_cnt; size_t i; split_cnt = dict_get_split_cnt (default_dict); + split = dict_get_split_vars (default_dict); for (i = 0; i < split_cnt; i++) - temp_case->data[split[i]->fv].f = split_values[i]; + case_data_rw (c, split[i]->fv)->f = nr->split_values[i]; } if (mx->n_factors) @@ -1467,8 +1407,8 @@ nr_output_data (struct matrix_data_pgm *mx, for (factor = 0; factor < mx->n_factors; factor++) { - temp_case->data[mx->factors[factor]->fv].f - = nr_factor_values[factor + cell * mx->n_factors]; + case_data_rw (c, mx->factors[factor]->fv)->f + = nr->factor_values[factor + cell * mx->n_factors]; debug_printf (("f:%s ", mx->factors[factor]->name)); } } @@ -1479,11 +1419,11 @@ nr_output_data (struct matrix_data_pgm *mx, for (content = 0; content <= PROX; content++) if (mx->is_per_factor[content]) { - assert (nr_data[content] != NULL - && nr_data[content][cell] != NULL); + assert (nr->data[content] != NULL + && nr->data[content][cell] != NULL); - dump_cell_content (mx, content, nr_data[content][cell], - write_case, wc_data); + dump_cell_content (mx, content, nr->data[content][cell], + c, write_case, wc_data); } } } @@ -1496,21 +1436,18 @@ nr_output_data (struct matrix_data_pgm *mx, int factor; for (factor = 0; factor < mx->n_factors; factor++) - temp_case->data[mx->factors[factor]->fv].f = SYSMIS; + case_data_rw (c, mx->factors[factor]->fv)->f = SYSMIS; } for (content = 0; content <= PROX; content++) - if (!mx->is_per_factor[content] && nr_data[content] != NULL) - dump_cell_content (mx, content, nr_data[content][0], - write_case, wc_data); + if (!mx->is_per_factor[content] && nr->data[content] != NULL) + dump_cell_content (mx, content, nr->data[content][0], + c, write_case, wc_data); } } /* Back end, with ROWTYPE_. */ -/* Type of current row. */ -static int wr_content; - /* All the data for one set of factor values. */ struct factor_data { @@ -1520,19 +1457,26 @@ struct factor_data struct factor_data *next; }; -/* All the data, period. */ -struct factor_data *wr_data; - -/* Current factor. */ -struct factor_data *wr_current; +/* With ROWTYPE_ auxiliary data. */ +struct wr_aux_data + { + struct matrix_data_pgm *mx; /* MATRIX DATA program. */ + int content; /* Type of current row. */ + double *split_values; /* SPLIT FILE variable values. */ + struct factor_data *data; /* All the data. */ + struct factor_data *current; /* Current factor. */ + }; -static int wr_read_splits (struct matrix_data_pgm *, +static int wr_read_splits (struct wr_aux_data *, struct ccase *, + write_case_func *, write_case_data); +static int wr_output_data (struct wr_aux_data *, struct ccase *, write_case_func *, write_case_data); -static int wr_output_data (struct matrix_data_pgm *, write_case_func *, write_case_data); -static int wr_read_rowtype (const struct matrix_token *, struct file_handle *); -static int wr_read_factors (struct matrix_data_pgm *); -static int wr_read_indeps (struct matrix_data_pgm *); +static int wr_read_rowtype (struct wr_aux_data *, + const struct matrix_token *, struct dfm_reader *); +static int wr_read_factors (struct wr_aux_data *); +static int wr_read_indeps (struct wr_aux_data *); static void matrix_data_read_with_rowtype (struct case_source *, + struct ccase *, write_case_func *, write_case_data); @@ -1541,48 +1485,56 @@ static void matrix_data_read_with_rowtype (struct case_source *, static void read_matrices_with_rowtype (struct matrix_data_pgm *mx) { - wr_data = wr_current = NULL; - split_values = NULL; + struct wr_aux_data wr; + + wr.mx = mx; + wr.content = -1; + wr.split_values = NULL; + wr.data = NULL; + wr.current = NULL; mx->cells = 0; - vfm_source = create_case_source (&matrix_data_with_rowtype_source_class, mx); - - procedure (NULL, NULL, NULL, NULL); + vfm_source = create_case_source (&matrix_data_with_rowtype_source_class, + default_dict, &wr); + procedure (NULL, NULL); - free (split_values); - fh_close_handle (mx->data_file); + free (wr.split_values); } /* Read from the data file and write it to the active file. */ static void matrix_data_read_with_rowtype (struct case_source *source, + struct ccase *c, write_case_func *write_case, write_case_data wc_data) { - struct matrix_data_pgm *mx = source->aux; + struct wr_aux_data *wr = source->aux; + struct matrix_data_pgm *mx = wr->mx; do { - if (!wr_read_splits (mx, write_case, wc_data)) + if (!wr_read_splits (wr, c, write_case, wc_data)) return; - if (!wr_read_factors (mx)) + if (!wr_read_factors (wr)) return; - if (!wr_read_indeps (mx)) + if (!wr_read_indeps (wr)) return; } - while (another_token (mx->data_file)); + while (another_token (mx->reader)); - wr_output_data (mx, write_case, wc_data); + wr_output_data (wr, c, write_case, wc_data); } /* Read the split file variables. If they differ from the previous set of split variables then output the data. Returns success. */ static int -wr_read_splits (struct matrix_data_pgm *mx, +wr_read_splits (struct wr_aux_data *wr, + struct ccase *c, write_case_func *write_case, write_case_data wc_data) { + struct matrix_data_pgm *mx = wr->mx; int compare; size_t split_cnt; @@ -1590,39 +1542,38 @@ wr_read_splits (struct matrix_data_pgm *mx, if (split_cnt == 0) return 1; - if (split_values) + if (wr->split_values) compare = 1; else { compare = 0; - split_values = xmalloc (split_cnt * sizeof *split_values); + wr->split_values = xmalloc (split_cnt * sizeof *wr->split_values); } { int different = 0; - size_t split_cnt; int i; for (i = 0; i < split_cnt; i++) { struct matrix_token token; - if (!mget_token (&token, mx->data_file)) + if (!mget_token (&token, mx->reader)) return 0; if (token.type != MNUM) { msg (SE, _("Syntax error %s expecting SPLIT FILE value."), - context (mx->data_file)); + context (mx->reader)); return 0; } - if (compare && split_values[i] != token.number && !different) + if (compare && wr->split_values[i] != token.number && !different) { - if (!wr_output_data (mx, write_case, wc_data)) + if (!wr_output_data (wr, c, write_case, wc_data)) return 0; different = 1; mx->cells = 0; } - split_values[i] = token.number; + wr->split_values[i] = token.number; } } @@ -1668,20 +1619,24 @@ compare_factors (const void *a_, const void *b_, void *mx_) /* Write out the data for the current split file to the active file. */ static int -wr_output_data (struct matrix_data_pgm *mx, +wr_output_data (struct wr_aux_data *wr, + struct ccase *c, write_case_func *write_case, write_case_data wc_data) { + struct matrix_data_pgm *mx = wr->mx; + { struct variable *const *split; size_t split_cnt; size_t i; split_cnt = dict_get_split_cnt (default_dict); + split = dict_get_split_vars (default_dict); for (i = 0; i < split_cnt; i++) - temp_case->data[split[i]->fv].f = split_values[i]; + case_data_rw (c, split[i]->fv)->f = wr->split_values[i]; } - /* Sort the wr_data list. */ + /* Sort the wr->data list. */ { struct factor_data **factors; struct factor_data *iter; @@ -1689,12 +1644,12 @@ wr_output_data (struct matrix_data_pgm *mx, factors = xmalloc (sizeof *factors * mx->cells); - for (i = 0, iter = wr_data; iter; iter = iter->next, i++) + for (i = 0, iter = wr->data; iter; iter = iter->next, i++) factors[i] = iter; sort (factors, mx->cells, sizeof *factors, compare_factors, mx); - wr_data = factors[0]; + wr->data = factors[0]; for (i = 0; i < mx->cells - 1; i++) factors[i]->next = factors[i + 1]; factors[mx->cells - 1]->next = NULL; @@ -1706,14 +1661,14 @@ wr_output_data (struct matrix_data_pgm *mx, { struct factor_data *iter; - for (iter = wr_data; iter; iter = iter->next) + for (iter = wr->data; iter; iter = iter->next) { { int factor; for (factor = 0; factor < mx->n_factors; factor++) { - temp_case->data[mx->factors[factor]->fv].f + case_data_rw (c, mx->factors[factor]->fv)->f = iter->factors[factor]; debug_printf (("f:%s ", factors[factor]->name)); } @@ -1748,7 +1703,7 @@ wr_output_data (struct matrix_data_pgm *mx, fill_matrix (mx, content, iter->data[content]); dump_cell_content (mx, content, iter->data[content], - write_case, wc_data); + c, write_case, wc_data); } } } @@ -1757,26 +1712,27 @@ wr_output_data (struct matrix_data_pgm *mx, pool_destroy (mx->container); mx->container = pool_create (); - wr_data = wr_current = NULL; + wr->data = wr->current = NULL; return 1; } -/* Sets ROWTYPE_ based on the given TOKEN read from DATA_FILE. +/* Sets ROWTYPE_ based on the given TOKEN read from READER. Return success. */ static int -wr_read_rowtype (const struct matrix_token *token, - struct file_handle *data_file) +wr_read_rowtype (struct wr_aux_data *wr, + const struct matrix_token *token, + struct dfm_reader *reader) { - if (wr_content != -1) + if (wr->content != -1) { - msg (SE, _("Multiply specified ROWTYPE_ %s."), context (data_file)); + msg (SE, _("Multiply specified ROWTYPE_ %s."), context (reader)); return 0; } if (token->type != MSTR) { msg (SE, _("Syntax error %s expecting ROWTYPE_ string."), - context (data_file)); + context (reader)); return 0; } @@ -1790,12 +1746,12 @@ wr_read_rowtype (const struct matrix_token *token, for (cp = s; *cp; cp++) *cp = toupper ((unsigned char) *cp); - wr_content = string_to_content_type (s, NULL); + wr->content = string_to_content_type (s, NULL); } - if (wr_content == -1) + if (wr->content == -1) { - msg (SE, _("Syntax error %s."), context (data_file)); + msg (SE, _("Syntax error %s."), context (reader)); return 0; } @@ -1805,53 +1761,54 @@ wr_read_rowtype (const struct matrix_token *token, /* Read the factors for the current row. Select a set of factors and point wr_current to it. */ static int -wr_read_factors (struct matrix_data_pgm *mx) +wr_read_factors (struct wr_aux_data *wr) { + struct matrix_data_pgm *mx = wr->mx; double *factor_values = local_alloc (sizeof *factor_values * mx->n_factors); - wr_content = -1; + wr->content = -1; { int i; for (i = 0; i < mx->n_factors; i++) { struct matrix_token token; - if (!mget_token (&token, mx->data_file)) + if (!mget_token (&token, mx->reader)) goto lossage; if (token.type == MSTR) { - if (!wr_read_rowtype (&token, mx->data_file)) + if (!wr_read_rowtype (wr, &token, mx->reader)) goto lossage; - if (!mget_token (&token, mx->data_file)) + if (!mget_token (&token, mx->reader)) goto lossage; } if (token.type != MNUM) { msg (SE, _("Syntax error expecting factor value %s."), - context (mx->data_file)); + context (mx->reader)); goto lossage; } factor_values[i] = token.number; } } - if (wr_content == -1) + if (wr->content == -1) { struct matrix_token token; - if (!mget_token (&token, mx->data_file)) + if (!mget_token (&token, mx->reader)) goto lossage; - if (!wr_read_rowtype (&token, mx->data_file)) + if (!wr_read_rowtype (wr, &token, mx->reader)) goto lossage; } /* Try the most recent factor first as a simple caching mechanism. */ - if (wr_current) + if (wr->current) { int i; for (i = 0; i < mx->n_factors; i++) - if (factor_values[i] != wr_current->factors[i]) + if (factor_values[i] != wr->current->factors[i]) goto cache_miss; goto winnage; } @@ -1861,7 +1818,7 @@ cache_miss: { struct factor_data *iter; - for (iter = wr_data; iter; iter = iter->next) + for (iter = wr->data; iter; iter = iter->next) { int i; @@ -1869,7 +1826,7 @@ cache_miss: if (factor_values[i] != iter->factors[i]) goto next_item; - wr_current = iter; + wr->current = iter; goto winnage; next_item: ; @@ -1899,8 +1856,8 @@ cache_miss: } } - new->next = wr_data; - wr_data = wr_current = new; + new->next = wr->data; + wr->data = wr->current = new; mx->cells++; } @@ -1913,28 +1870,29 @@ lossage: return 0; } -/* Read the independent variables into wr_current. */ +/* Read the independent variables into wr->current. */ static int -wr_read_indeps (struct matrix_data_pgm *mx) +wr_read_indeps (struct wr_aux_data *wr) { - struct factor_data *c = wr_current; - const int type = content_type[wr_content]; - const int n_rows = c->n_rows[wr_content]; + struct matrix_data_pgm *mx = wr->mx; + struct factor_data *c = wr->current; + const int type = content_type[wr->content]; + const int n_rows = c->n_rows[wr->content]; double *cp; int n_cols; /* Allocate room for data if necessary. */ - if (c->data[wr_content] == NULL) + if (c->data[wr->content] == NULL) { int n_items = mx->n_continuous; if (type == 1) n_items *= mx->n_continuous; - c->data[wr_content] = pool_alloc (mx->container, + c->data[wr->content] = pool_alloc (mx->container, sizeof **c->data * n_items); } - cp = &c->data[wr_content][n_rows * mx->n_continuous]; + cp = &c->data[wr->content][n_rows * mx->n_continuous]; /* Figure out how much to read from this line. */ switch (type) @@ -1944,7 +1902,7 @@ wr_read_indeps (struct matrix_data_pgm *mx) if (n_rows > 0) { msg (SE, _("Duplicate specification for %s."), - content_names[wr_content]); + content_names[wr->content]); return 0; } if (type == 0) @@ -1956,7 +1914,7 @@ wr_read_indeps (struct matrix_data_pgm *mx) if (n_rows >= mx->n_continuous - (mx->section != FULL && mx->diag == NODIAGONAL)) { msg (SE, _("Too many rows of matrix data for %s."), - content_names[wr_content]); + content_names[wr->content]); return 0; } @@ -1981,12 +1939,14 @@ wr_read_indeps (struct matrix_data_pgm *mx) break; default: assert (0); + abort (); } break; default: assert (0); + abort (); } - c->n_rows[wr_content]++; + c->n_rows[wr->content]++; debug_printf ((" (c=%p,r=%d,n=%d)", c, n_rows + 1, n_cols)); @@ -1997,20 +1957,20 @@ wr_read_indeps (struct matrix_data_pgm *mx) for (j = 0; j < n_cols; j++) { struct matrix_token token; - if (!mget_token (&token, mx->data_file)) + if (!mget_token (&token, mx->reader)) return 0; if (token.type != MNUM) { msg (SE, _("Syntax error expecting value for %s %s."), dict_get_var (default_dict, mx->first_continuous + j)->name, - context (mx->data_file)); + context (mx->reader)); return 0; } *cp++ = token.number; } if (mx->fmt != FREE - && !force_eol (mx->data_file, content_names[wr_content])) + && !force_eol (mx->reader, content_names[wr->content])) return 0; debug_printf (("\n")); } @@ -2036,3 +1996,4 @@ matrix_data_without_rowtype_source_class = matrix_data_read_without_rowtype, NULL, }; +