The R Project SVN R-packages

Rev

Rev 3397 | Rev 3432 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
3255 mrmanese 1
#include "sqlite_dataframe.h"
3307 mrmanese 2
#include <math.h>
3
#include "Rmath.h"
3255 mrmanese 4
 
3358 mrmanese 5
/****************************************************************************
6
 * UTILITY FUNCTIONS
7
 ****************************************************************************/
3419 mrmanese 8
char *_create_svector1(SEXP name, const char *type, int * _namelen, int protect) {
3358 mrmanese 9
    int namelen, res;
3419 mrmanese 10
    char *iname = _create_sdf_skeleton1(name, &namelen, protect);
3255 mrmanese 11
 
3358 mrmanese 12
    if (iname == NULL) return NULL;
3255 mrmanese 13
 
3358 mrmanese 14
    sprintf(g_sql_buf[2], "create table [%s].sdf_data ([row name] text, "
15
            "V1 %s, primary key ([row name]))", iname, type);
16
    res = _sqlite_exec(g_sql_buf[2]);
17
    _sqlite_error(res);
3308 mrmanese 18
 
3358 mrmanese 19
    if (_namelen != NULL) *_namelen = namelen;
20
    return iname;
21
}
3255 mrmanese 22
 
3358 mrmanese 23
static SEXP _create_svector_sexp(const char *iname, const char *varname, 
24
        const char *type) {
25
    SEXP ret, value; int nprotected = 0;
3255 mrmanese 26
    PROTECT(ret = NEW_LIST(2)); nprotected++;
27
 
28
    /* set list names */
29
    PROTECT(value = NEW_CHARACTER(2)); nprotected++;
30
    SET_STRING_ELT(value, 0, mkChar("iname"));
31
    SET_STRING_ELT(value, 1, mkChar("varname"));
32
    SET_NAMES(ret, value);
33
 
34
    /* set list values */
35
    SET_VECTOR_ELT(ret, 0, mkString(iname));
36
    SET_VECTOR_ELT(ret, 1, mkString(varname));
37
 
3358 mrmanese 38
    /* set sexp class */
39
    if (strcmp(type, "ordered") == 0) {
40
        PROTECT(value = NEW_CHARACTER(3)); nprotected++;
41
        SET_VECTOR_ELT(value, 0, mkChar("sqlite.vector"));
42
        SET_VECTOR_ELT(value, 1, mkChar(type));
43
        SET_VECTOR_ELT(value, 2, mkChar("factor"));
44
    } else {
3255 mrmanese 45
        PROTECT(value = NEW_CHARACTER(2)); nprotected++;
3358 mrmanese 46
        SET_VECTOR_ELT(value, 0, mkChar("sqlite.vector"));
47
        SET_VECTOR_ELT(value, 1, mkChar(type));
48
        SET_CLASS(ret, value);
3255 mrmanese 49
    }
50
 
51
    UNPROTECT(nprotected);
3358 mrmanese 52
 
3255 mrmanese 53
    return ret;
54
}
55
 
3358 mrmanese 56
/* if ret == NULL, 3rd arg is the length of the vector created. otherwise
57
 * it is the index in the vector ret where we will put the result extracted
58
 * from ret */
3255 mrmanese 59
int _get_vector_index_typed_result(sqlite3_stmt *stmt, SEXP *ret, int idx_or_len) {
60
    int added = 1;
61
    if (*ret == NULL || *ret == R_NilValue) {
62
        const char *coltype = sqlite3_column_decltype(stmt, 0);
63
        if (sqlite3_column_type(stmt, 0) == SQLITE_NULL) {
64
            added = 0;
65
        }
66
 
67
        if (strcmp(coltype, "text") == 0) {
68
            PROTECT(*ret = NEW_CHARACTER(idx_or_len));
69
            if (added) 
3284 mrmanese 70
                SET_STRING_ELT(*ret, 0, mkChar((char *)sqlite3_column_text(stmt, 0)));
3255 mrmanese 71
        } else if (strcmp(coltype, "double") == 0) {
72
            PROTECT(*ret = NEW_NUMERIC(idx_or_len));
73
            if (added) REAL(*ret)[0] = sqlite3_column_double(stmt, 0);
74
        } else if (strcmp(coltype, "bit") == 0) {
75
            PROTECT(*ret = NEW_LOGICAL(idx_or_len));
76
            if (added) INTEGER(*ret)[0] = sqlite3_column_int(stmt, 0);
77
        } else if (strcmp(coltype, "integer") == 0 || 
78
                   strcmp(coltype, "int") == 0) {
79
            /* caller should just copy off the vars level attr for factors */
80
            PROTECT(*ret = NEW_INTEGER(idx_or_len));
81
            if (added) INTEGER(*ret)[0] = sqlite3_column_int(stmt, 0);
82
        } else added = 0;
83
 
84
        UNPROTECT(1);
85
    } else {
86
        const char *coltype = sqlite3_column_decltype(stmt, 0);
87
        if (sqlite3_column_type(stmt, 0) == SQLITE_NULL) {
88
            added = 0;
89
        } else if (strcmp(coltype, "text") == 0) {
90
            SET_STRING_ELT(*ret, idx_or_len, 
3284 mrmanese 91
                    mkChar((char *)sqlite3_column_text(stmt, 0)));
3255 mrmanese 92
        } else if (strcmp(coltype, "double") == 0) {
93
            REAL(*ret)[idx_or_len] = sqlite3_column_double(stmt, 0);
94
        } else if (strcmp(coltype, "bit") == 0) {
95
            INTEGER(*ret)[idx_or_len] = sqlite3_column_int(stmt, 0);
96
        } else if (strcmp(coltype, "integer") == 0 ||
97
                   strcmp(coltype, "int") == 0) {
98
            /* caller should just copy off the vars level attr for factors */
99
            INTEGER(*ret)[idx_or_len] = sqlite3_column_int(stmt, 0);
100
        } else added = 0;
101
    }
102
 
103
    return added;
104
}
105
 
3358 mrmanese 106
/****************************************************************************
107
 * SVEC FUNCTIONS
108
 ****************************************************************************/
109
SEXP sdf_get_variable(SEXP sdf, SEXP name) {
110
    if (!IS_CHARACTER(name)) {
111
        Rprintf("ERROR: argument is not a string.\n");
112
        return R_NilValue;
113
    }
114
 
115
    char *iname = SDF_INAME(sdf);
116
    char *varname = CHAR_ELT(name, 0);
117
 
3419 mrmanese 118
    if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;
3358 mrmanese 119
 
120
    /* check if sdf & varname w/in that sdf exists */
121
    sqlite3_stmt *stmt;
122
    sprintf(g_sql_buf[0], "select [%s] from [%s].sdf_data", varname, iname);
123
 
124
    int res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
125
 
126
    if (_sqlite_error(res)) return R_NilValue;
127
 
128
    const char *coltype = sqlite3_column_decltype(stmt, 0);
129
    sqlite3_finalize(stmt);
130
 
131
 
132
    SEXP ret, value, class = R_NilValue; int nprotected = 0;
133
    PROTECT(ret = NEW_LIST(2)); nprotected++;
134
 
135
    /* set list names */
136
    PROTECT(value = NEW_CHARACTER(2)); nprotected++;
137
    SET_STRING_ELT(value, 0, mkChar("iname"));
138
    SET_STRING_ELT(value, 1, mkChar("varname"));
139
    SET_NAMES(ret, value);
140
 
141
    /* set list values */
142
    SET_VECTOR_ELT(ret, 0, mkString(iname));
143
    SET_VECTOR_ELT(ret, 1, mkString(varname));
144
 
145
    /* set class */
146
    int type = -1;
147
    if (strcmp(coltype, "text") == 0) class = mkChar("character");
148
    else if (strcmp(coltype, "double") == 0) class = mkChar("numeric");
149
    else if (strcmp(coltype, "bit") == 0) class = mkChar("logical");
150
    else if (strcmp(coltype, "integer") == 0 || strcmp(coltype, "int") == 0) {
151
        /* determine if int, factor or ordered */
152
        type = _get_factor_levels1(iname, varname, ret);
153
        switch(type) {
154
            case VAR_INTEGER: class = mkChar("integer"); break;
155
            case VAR_FACTOR: class = mkChar("factor"); break;
156
            case VAR_ORDERED: class = mkChar("ordered");
157
        }
158
 
159
    }
160
 
161
    if (type != VAR_ORDERED) {
162
        PROTECT(value = NEW_CHARACTER(2)); nprotected++;
163
        SET_STRING_ELT(value, 0, mkChar("sqlite.vector"));
164
        SET_STRING_ELT(value, 1, class);
165
    } else {
166
        PROTECT(value = NEW_CHARACTER(3)); nprotected++;
167
        SET_STRING_ELT(value, 0, mkChar("sqlite.vector"));
168
        SET_STRING_ELT(value, 1, class);
169
        SET_STRING_ELT(value, 2, mkChar("factor"));
170
    }
171
    SET_CLASS(ret, value);
172
 
173
    UNPROTECT(nprotected);
174
    return ret;
175
 
176
}
177
 
178
 
3255 mrmanese 179
SEXP sdf_get_variable_length(SEXP svec) {
180
    char *iname = SDF_INAME(svec);
3419 mrmanese 181
    if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;
3255 mrmanese 182
 
183
    SEXP ret;
184
    PROTECT(ret = NEW_INTEGER(1));
3358 mrmanese 185
    INTEGER(ret)[0] = _get_row_count2(iname, 1);
3255 mrmanese 186
    UNPROTECT(1);
187
    return ret;
188
}
189
 
190
 
191
SEXP sdf_get_variable_index(SEXP svec, SEXP idx) {
192
    SEXP ret = R_NilValue, tmp;
193
    char *iname = SDF_INAME(svec), *varname = SVEC_VARNAME(svec);
3282 mrmanese 194
    int index, idxlen, i, retlen=0, res;
3358 mrmanese 195
    sqlite3_stmt *stmt;
3255 mrmanese 196
 
3419 mrmanese 197
    if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;
3308 mrmanese 198
 
3255 mrmanese 199
    idxlen = LENGTH(idx);
200
    if (idxlen < 1) return ret;
201
 
202
    sprintf(g_sql_buf[0], "select [%s] from [%s].sdf_data limit ?,1",
203
            varname, iname);
204
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
205
 
206
    /* get data based on index */
207
    if (IS_NUMERIC(idx)) {
208
        index = ((int) REAL(idx)[0]) - 1;
209
        if (index < 0 && idxlen == 1) return ret;
210
 
3281 mrmanese 211
        if (index >= 0) {
3255 mrmanese 212
            sqlite3_bind_int(stmt, 1, index);
213
            res = sqlite3_step(stmt);
214
            if (res == SQLITE_ROW) { 
215
                retlen = _get_vector_index_typed_result(stmt, &ret, idxlen);
216
            } 
217
        } 
218
 
3281 mrmanese 219
        if (index < 0 || res != SQLITE_ROW) {
3255 mrmanese 220
            /* something wrong w/ 1st idx, and it is quietly ignored. we make 
221
             * a "dummy" call to setup the SEXP */
3358 mrmanese 222
            sqlite3_reset(stmt);
3255 mrmanese 223
            sqlite3_bind_int(stmt, 1, 0);
3358 mrmanese 224
            res = sqlite3_step(stmt);
225
            if (res == SQLITE_ROW) {
226
                _get_vector_index_typed_result(stmt, &ret, idxlen - 1);
227
            }
3255 mrmanese 228
            retlen = 0;
229
        }
230
 
3358 mrmanese 231
 
3255 mrmanese 232
        if (idxlen > 1) {
233
            for (i = 1; i < idxlen; i++) {
234
                index = ((int) REAL(idx)[i]) - 1;
235
                if (index < 0) continue;
236
                sqlite3_reset(stmt);
237
                sqlite3_bind_int(stmt, 1, index);
238
                res = sqlite3_step(stmt);
239
                if (res == SQLITE_ROW)
240
                    retlen += _get_vector_index_typed_result(stmt, &ret, retlen); 
241
            }
242
        }
243
    } else if (IS_INTEGER(idx)) {
244
        /* similar to REAL (IS_NUMERIC) above, except that we don't have
245
         * to cast idx to int. can't refactor this out, sucks. */
246
        index = INTEGER(idx)[0] - 1;
247
        if (index < 0 && idxlen == 1) return ret;
248
 
3281 mrmanese 249
        if (index >= 0) {
3255 mrmanese 250
            sqlite3_bind_int(stmt, 1, index);
3281 mrmanese 251
            res = sqlite3_step(stmt);
252
            if (res == SQLITE_ROW) { 
253
                retlen = _get_vector_index_typed_result(stmt, &ret, idxlen);
254
            } 
255
        } 
256
 
257
        if (index < 0 || res != SQLITE_ROW) {
258
            /* something wrong w/ 1st idx, and it is quietly ignored. we make 
259
             * a "dummy" call to setup the SEXP */
3358 mrmanese 260
            sqlite3_reset(stmt);
3255 mrmanese 261
            sqlite3_bind_int(stmt, 1, 0);
3358 mrmanese 262
            res = sqlite3_step(stmt);
263
            if (res == SQLITE_ROW) {
264
                _get_vector_index_typed_result(stmt, &ret, idxlen - 1);
265
            }
3255 mrmanese 266
            retlen = 0;
267
        }
268
 
269
        if (idxlen > 1) {
270
            for (i = 1; i < idxlen; i++) {
271
                index = INTEGER(idx)[i] - 1;
272
                if (index < 0) continue;
273
                sqlite3_reset(stmt);
274
                sqlite3_bind_int(stmt, 1, index);
275
                sqlite3_step(stmt);
276
                retlen += _get_vector_index_typed_result(stmt, &ret, retlen); 
277
            }
278
        }
279
 
280
    } else if (IS_LOGICAL(idx)) {
281
        /* have to deal with recycling */
3358 mrmanese 282
        int veclen = _get_row_count2(iname, 1);
3255 mrmanese 283
 
284
        /* find if there is any TRUE element in the vector */
3281 mrmanese 285
        for (i = 0; i < idxlen && i < veclen; i++) {
3255 mrmanese 286
            if (LOGICAL(idx)[i]) {
287
                sqlite3_bind_int(stmt, 1, i);
288
                sqlite3_step(stmt);
289
                /* there are at least (idxlen-i) TRUE per cycle of the LOGICAL
290
                 * index. there are at least (veclen/idxlen) cycles (int div).
291
                 * at the last cycle, if (veclen%idxlen > 0), there will be
292
                 * at least (veclen%idxlen - i) if veclen%idxlen > i */
293
                retlen = (idxlen-i) * (veclen/idxlen);
294
                if (veclen%idxlen > i) retlen += (veclen%idxlen - i);
295
 
296
                /* create the vector */
297
                retlen = _get_vector_index_typed_result(stmt, &ret, retlen);
298
                break;
299
            }
300
        }
301
 
3281 mrmanese 302
        if (i < idxlen && i < veclen) {
3255 mrmanese 303
            for (i++; i < veclen; i++) {
304
                if (LOGICAL(idx)[i%idxlen]) {
305
                    sqlite3_reset(stmt);
306
                    sqlite3_bind_int(stmt, 1, i);
307
                    sqlite3_step(stmt);
308
                    retlen += _get_vector_index_typed_result(stmt, &ret, retlen);
309
                }
310
            }
311
        }
312
    }
313
 
314
    sqlite3_finalize(stmt);
315
 
316
    if (ret != R_NilValue) {
317
        ret = _shrink_vector(ret, retlen);
318
        tmp = GET_LEVELS(svec);
319
        if (tmp != R_NilValue) {
320
            SET_LEVELS(ret, duplicate(tmp));
321
            if (LENGTH(GET_CLASS(svec)) == 2) {
322
                SET_CLASS(ret, mkString("factor"));
323
            } else {
324
                PROTECT(tmp = NEW_CHARACTER(2));
325
                SET_STRING_ELT(tmp, 0, mkChar("ordered"));
326
                SET_STRING_ELT(tmp, 1, mkChar("factor"));
327
                UNPROTECT(1);
328
            }
329
        }
330
    }
331
 
332
    return ret;
333
}
334
 
3358 mrmanese 335
/* sqlite.vector.[<- */
336
SEXP sdf_set_variable_index(SEXP svec, SEXP idx, SEXP value) {
337
    int idx_len, val_len;
338
 
339
    /* match levels if factor or ordered */
340
    if (inherits(value, "ordered")) {
341
    } else if (inherits(value, "factor")) {
342
    }
343
 
344
    idx_len = LENGTH(idx);
345
    val_len = LENGTH(value);
346
 
347
    if (IS_NUMERIC(idx)) {
348
    } else if (IS_INTEGER(idx)) {
349
    } else if (IS_LOGICAL(idx)) {
350
    }
351
 
352
    return R_NilValue;
353
}
354
 
3419 mrmanese 355
SEXP sdf_variable_summary(SEXP svec, SEXP maxsum) {
356
    char *iname, *varname, *type;
357
    sqlite3_stmt *stmt;
358
    int nprotected = 0;
359
    SEXP ret, names;
360
 
361
    iname = SDF_INAME(svec);
362
    varname = SVEC_VARNAME(svec);
363
    USE_SDF1(iname, TRUE, FALSE);
364
 
365
    if ((inherits(svec, "ordered") && ((type = "ordered"))) ||
366
            (inherits(svec, "factor") && ((type = "factor")))) {
367
        int nrows, i, max_rows = INTEGER(maxsum)[0];
368
 
369
 
370
        sprintf(g_sql_buf[0], "[%s].[%s %s]", iname, type, varname);
371
        nrows = _get_row_count2(g_sql_buf[0], FALSE);
372
        if (nrows <= max_rows) max_rows = nrows;
373
        else { nrows = max_rows; max_rows--; }
374
 
375
        PROTECT(ret = NEW_INTEGER(nrows)); nprotected = 1;
376
        PROTECT(names = NEW_CHARACTER(nrows)); nprotected++;
377
 
378
        sprintf(g_sql_buf[0], "select [%s].[%s %s].label, count(*) from "
379
                "[%s].sdf_data join [%s].[%s %s] on [%s].sdf_data.[%s]=[%s].[%s %s].level "
380
                "group by [%s].sdf_data.[%s], [%s].[%s %s].level order by count(*) desc",
381
                iname, type, varname, iname, iname, type, varname, iname, varname,
382
                iname, type, varname, iname, varname, iname, type, varname);
383
        sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
384
 
385
        for (i = 0; i < max_rows; i++) {
386
            sqlite3_step(stmt);
387
            SET_STRING_ELT(names, i, mkChar((char *)sqlite3_column_text(stmt, 0)));
388
            INTEGER(ret)[i] = sqlite3_column_int(stmt, 1);
389
        }
390
 
391
        if (nrows > max_rows) {
392
            int others_sum = 0;
393
            SET_STRING_ELT(names, nrows-1, mkChar("(Others)"));
394
            while (sqlite3_step(stmt) == SQLITE_ROW) {
395
                others_sum += sqlite3_column_int(stmt, 1);
396
            }
397
            INTEGER(ret)[nrows-1] = others_sum;
398
        }
399
    } else if (inherits(svec, "logical")) {
400
        sprintf(g_sql_buf[0], "select count(*) from "
401
                "[%s].sdf_data group by [%s] order by [%s]", iname, varname, varname);
402
 
403
        PROTECT(names = NEW_CHARACTER(3)); nprotected = 1;
404
        PROTECT(ret = NEW_CHARACTER(3)); nprotected++;
405
 
406
        SET_STRING_ELT(names, 0, mkChar("Mode"));
407
        SET_STRING_ELT(names, 1, mkChar("FALSE"));
408
        SET_STRING_ELT(names, 2, mkChar("TRUE"));
409
 
410
        sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
411
        SET_STRING_ELT(ret, 0, mkChar("logical"));
412
        sqlite3_step(stmt);
413
        SET_STRING_ELT(ret, 1, mkChar((const char *)sqlite3_column_text(stmt, 0)));
414
        sqlite3_step(stmt);
415
        SET_STRING_ELT(ret, 2, mkChar((const char *)sqlite3_column_text(stmt, 0)));
416
 
417
    } else return R_NilValue;
418
 
419
    sqlite3_finalize(stmt);
420
    SET_NAMES(ret, names);
421
    SET_CLASS(ret, mkString("table"));
422
    UNPROTECT(nprotected);
423
    return ret;
424
}
425
 
426
 
3324 mrmanese 427
/* the global accumulator should be safe if we only do 1 cummulative or
428
 * aggregate at any time. it won't work for stuffs like "select max(col)-min(col)
429
 * from sdf_data". */
430
static double g_accumulator = 0.0; /* accumulator var for cumsum, cumprod, etc. */
431
static int g_start = 0;            /* flag for start of cummulation */
432
static int g_narm = 0;             /* for Summary group */
3255 mrmanese 433
 
3307 mrmanese 434
SEXP sdf_do_variable_math(SEXP func, SEXP vector, SEXP extra_args, SEXP _nargs) {
435
    char *iname, *iname_src, *varname_src, *funcname;
436
    int namelen, res, nargs;
437
    sqlite3_stmt *stmt;
3255 mrmanese 438
 
3307 mrmanese 439
    /* get data from arguments (function name and sqlite.vector stuffs) */
440
    funcname = CHAR_ELT(func, 0);
441
    iname_src = SDF_INAME(vector);
442
    varname_src = SVEC_VARNAME(vector);
443
    nargs = INTEGER(_nargs)[0];
3255 mrmanese 444
 
3419 mrmanese 445
    if (!USE_SDF1(iname_src, TRUE, TRUE)) return R_NilValue;
3308 mrmanese 446
 
3307 mrmanese 447
    /* check nargs */
448
    if (nargs > 2) {
449
        Rprintf("Error: Don't know how to handle Math functions w/ more than 2 args\n");
450
        return R_NilValue;
451
    }
452
 
453
    /* create a new sdf, with 1 column named V1 */
3419 mrmanese 454
    iname = _create_svector1(R_NilValue, "double", &namelen, TRUE);
3307 mrmanese 455
 
456
    /* insert into <newsdf>.col, row.names select func(col), rownames */
457
    sprintf(g_sql_buf[0], "insert into [%s].sdf_data([row name], V1) "
458
            "select [row name], %s([%s]) from [%s].sdf_data", iname, funcname,
459
            varname_src, iname_src);
460
 
461
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0); 
462
    if (_sqlite_error(res)) {
463
        sprintf(g_sql_buf[0], "detach %s", iname);
464
        _sqlite_exec(g_sql_buf[0]);
465
 
466
        /* we will return a string with the file name, and do file.remove
467
         * at R */
468
        iname[namelen] = '.';
469
        return mkString(iname);
470
    }
471
 
3324 mrmanese 472
    g_accumulator = 0.0;   /* initialize accumulator */
473
    g_start = 1;           /* flag that we are at start of accumulating */
3307 mrmanese 474
    sqlite3_step(stmt);
475
    sqlite3_finalize(stmt);
476
 
3419 mrmanese 477
    UNUSE_SDF2(iname);
478
    UNUSE_SDF2(iname_src);
479
 
3358 mrmanese 480
    return _create_svector_sexp(iname, "V1", "numeric");
481
}
3307 mrmanese 482
 
3358 mrmanese 483
SEXP sdf_do_variable_op(SEXP func, SEXP vector, SEXP op2) {
484
    char *iname = NULL, *iname_src, *varname_src, *funcname;
485
    int res, functype = -1, op2_len, svec_len, i;
486
    sqlite3_stmt *stmt, *stmt2;
3307 mrmanese 487
 
3358 mrmanese 488
    /* get data from arguments (function name and sqlite.vector stuffs) */
489
    funcname = CHAR_ELT(func, 0);
490
    iname_src = SDF_INAME(vector);
491
    varname_src = SVEC_VARNAME(vector);
3307 mrmanese 492
 
3419 mrmanese 493
    if (!USE_SDF1(iname_src, TRUE, TRUE)) return R_NilValue;
3358 mrmanese 494
    switch(funcname[0]) {
495
        case '+' :
496
        case '-' :
497
        case '*' :
498
        case '/' :
499
        case '^' :
500
        case '%' : /* %% and %/% */ 
501
            functype = 0; break;  /* output is REAL */
502
        case '&' :
503
        case '|' :
504
        case '!' :  /* ! and != */
505
            if (funcname[1] == 0) { functype = 1; break; }
506
        case '=' :  /* == */
507
        case '<' :  /* < and <= */
508
        case '>' :  /* > and >= */
509
            functype = 2; 
510
    }
3307 mrmanese 511
 
3358 mrmanese 512
    svec_len = _get_row_count2(iname_src, 1);
513
    if (functype == 0 || functype == 2 || (functype == 1 && funcname[0] != '!')) {
514
        char *insert_fmt_string1c, *insert_fmt_string1s;
515
        char *insert_fmt_string2c, *insert_fmt_string2s;
3419 mrmanese 516
        iname = _create_svector1(R_NilValue, (functype == 0) ? "double" : "bit", NULL, TRUE);
3307 mrmanese 517
 
3358 mrmanese 518
        if (functype == 1) { /* boolean binary operators */
519
            insert_fmt_string1s = "insert into [%s].sdf_data "
520
                        "select [row name], ([%s] != 0) %s (? != 0) from [%s].sdf_data";
521
            insert_fmt_string2s = "insert into [%s].sdf_data values(?, (? != 0) %s (? != 0))";
522
            /* no need for char version of fmt_string2, since %% only occurs for functype==0 */
523
            insert_fmt_string1c = insert_fmt_string1s;
524
            insert_fmt_string2c = insert_fmt_string2s;
525
        } else {
526
            insert_fmt_string1s = "insert into [%s].sdf_data "
527
                        "select [row name], [%s] %s ? from [%s].sdf_data";
528
            insert_fmt_string1c = "insert into [%s].sdf_data "
529
                        "select [row name], [%s] %c ? from [%s].sdf_data";
530
            /* fmt_string2c format operator from a char, used for %% and %/% */
531
            /* BUG: i'm casting to int before applying / & %, w/c is different
532
             * from R. added trunc to pass package test */
533
            insert_fmt_string2c = "insert into [%s].sdf_data values(?, trunc(? %c ?))";
534
            insert_fmt_string2s = "insert into [%s].sdf_data values(?, ? %s ?)";
535
        }
3307 mrmanese 536
 
3358 mrmanese 537
 
538
        if (IS_NUMERIC(op2)) {
539
            op2_len = LENGTH(op2);
540
 
541
            if (op2_len == 1) {
542
                if (funcname[0] == '%') {
543
                    sprintf(g_sql_buf[2], insert_fmt_string1c, iname, 
544
                            varname_src, funcname[1], iname_src);
545
                } else {
546
                    sprintf(g_sql_buf[2], insert_fmt_string1s, iname,
547
                            varname_src, funcname, iname_src);
548
                }
549
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
550
                _sqlite_error(res);
551
                sqlite3_bind_double(stmt, 1, REAL(op2)[0]);
552
                sqlite3_step(stmt);
3419 mrmanese 553
                sqlite3_finalize(stmt);
3358 mrmanese 554
            } else if (op2_len <= svec_len) {  /* recycle op2 */
555
                sprintf(g_sql_buf[2], "select [row name], [%s] from [%s].sdf_data",
556
                        varname_src, iname_src);
557
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
558
                _sqlite_error(res);
559
 
560
                if (funcname[0] == '%') {
561
                    /* sqlite does int div with "/" if both args are int. mod is % also.
562
                     * fortunately, in both cases the sqlite op is 2nd char of funcname */
563
                    sprintf(g_sql_buf[2], insert_fmt_string2c, iname, funcname[1]);  
3419 mrmanese 564
                    _sqlite_begin;
3358 mrmanese 565
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
566
                    _sqlite_error(res);
567
 
568
                    for (i = 0; i < svec_len; i++) {
569
                        sqlite3_step(stmt2);
570
 
571
                        sqlite3_reset(stmt);
572
                        sqlite3_bind_text(stmt, 1, (char *)sqlite3_column_text(stmt2, 0), -1, SQLITE_STATIC);
573
                        sqlite3_bind_int(stmt, 2, (int)sqlite3_column_double(stmt2, 1));
574
                        sqlite3_bind_int(stmt, 3, (int)REAL(op2)[i % op2_len]);
575
                        sqlite3_step(stmt);
576
                    }
3419 mrmanese 577
                    sqlite3_finalize(stmt);
578
                    sqlite3_finalize(stmt2);
579
                    _sqlite_commit;
3358 mrmanese 580
                } else { /* non-integer binary operation */
581
                    sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
3419 mrmanese 582
                    _sqlite_begin;
3358 mrmanese 583
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
584
                    _sqlite_error(res);
585
 
586
                    for (i = 0; i < svec_len; i++) {
587
                        sqlite3_step(stmt2);
588
 
589
                        sqlite3_reset(stmt);
590
                        sqlite3_bind_text(stmt, 1, (char *)sqlite3_column_text(stmt2, 0), -1, SQLITE_STATIC);
591
                        sqlite3_bind_double(stmt, 2, sqlite3_column_double(stmt2, 1));
592
                        sqlite3_bind_double(stmt, 3, REAL(op2)[i % op2_len]);
593
                        sqlite3_step(stmt);
594
                    }
3419 mrmanese 595
                    sqlite3_finalize(stmt);
596
                    sqlite3_finalize(stmt2);
597
                    _sqlite_commit;
3358 mrmanese 598
                }
599
            } else { /* op2_len > svec_len, recycle svec */
600
                sprintf(g_sql_buf[1], "select [%s] from [%s].sdf_data",
601
                        varname_src, iname_src);
602
 
603
                if (funcname[0] == '%') {
604
                    sprintf(g_sql_buf[2], insert_fmt_string2c, iname, funcname[1]);  
605
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
606
                    _sqlite_error(res);
607
                } else {
608
                    sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
609
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
610
                    _sqlite_error(res);
611
                }
612
 
3419 mrmanese 613
                i = 0; _sqlite_begin;
3358 mrmanese 614
                while (i < op2_len) {
615
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
616
                    _sqlite_error(res);
617
 
618
                    if (funcname[0] == '%') {
619
                        for ( ; i < op2_len && sqlite3_step(stmt) == SQLITE_ROW ; i++) {
620
                            sqlite3_step(stmt2);
621
 
622
                            sqlite3_reset(stmt);
623
                            /* simplify my life, just use ints for row names so that we
624
                             * don't have to worry about duplicates */
625
                            sqlite3_bind_int(stmt, 1, i); 
626
                            sqlite3_bind_int(stmt, 2, sqlite3_column_int(stmt2, 1));
627
                            sqlite3_bind_int(stmt, 3, (int)REAL(op2)[i % op2_len]);
628
                            sqlite3_step(stmt);
629
                        }
630
                    } else {
631
                        for ( ; i < op2_len && sqlite3_step(stmt) == SQLITE_ROW ; i++) {
632
                            sqlite3_step(stmt2);
633
 
634
                            sqlite3_reset(stmt);
635
                            sqlite3_bind_int(stmt, 1, i);
636
                            sqlite3_bind_double(stmt, 2, sqlite3_column_double(stmt2, 1));
637
                            sqlite3_bind_double(stmt, 3, REAL(op2)[i % op2_len]);
638
                            sqlite3_step(stmt);
639
                        }
640
                    }
641
 
642
                    /* recycle on svec if we loop again */
643
                    sqlite3_finalize(stmt2);
644
                }
3397 mrmanese 645
 
3419 mrmanese 646
                sqlite3_finalize(stmt);
647
                _sqlite_commit;
3358 mrmanese 648
            }
649
 
650
        } else if (IS_INTEGER(op2) || IS_LOGICAL(op2)) { /* I N T E G E R */
651
            /* we are taking advantage of the fact that logicals are stored as int.
652
             * if this becomes untrue in the future, then this is a bug */
653
            op2_len = LENGTH(op2);
654
 
655
            if (op2_len == 1) {
656
                if (funcname[0] == '%') {
657
                    sprintf(g_sql_buf[2], insert_fmt_string1c, iname, 
658
                            varname_src, funcname[1], iname_src);
659
                } else {
660
                    sprintf(g_sql_buf[2], insert_fmt_string1s, iname,
661
                            varname_src, funcname, iname_src);
662
                }
663
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
664
                _sqlite_error(res);
665
                sqlite3_bind_double(stmt, 1, (double)INTEGER(op2)[0]);
666
                sqlite3_step(stmt);
667
            } else if (op2_len <= svec_len) {
668
                sprintf(g_sql_buf[2], "select [row name], [%s] from [%s].sdf_data",
669
                        varname_src, iname_src);
670
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
671
                _sqlite_error(res);
672
 
673
                if (funcname[0] == '%') {
674
                    /* sqlite does int div with "/" if both args are int. mod is % also.
675
                     * fortunately, in both cases the sqlite op is 2nd char of funcname */
676
                    sprintf(g_sql_buf[2], insert_fmt_string2c, iname, funcname[1]);  
3419 mrmanese 677
                    _sqlite_begin;
3358 mrmanese 678
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
679
                    _sqlite_error(res);
680
 
681
                    for (i = 0; i < svec_len; i++) {
682
                        sqlite3_step(stmt2);
683
 
684
                        sqlite3_reset(stmt);
685
                        sqlite3_bind_text(stmt, 1, (char *)sqlite3_column_text(stmt2, 0), -1, SQLITE_STATIC);
686
                        sqlite3_bind_int(stmt, 2, sqlite3_column_int(stmt2, 1));
687
                        sqlite3_bind_int(stmt, 3, INTEGER(op2)[i % op2_len]);
688
                        sqlite3_step(stmt);
689
                    }
3419 mrmanese 690
                    _sqlite_commit;
3358 mrmanese 691
                } else {
692
                    sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
3419 mrmanese 693
                    _sqlite_begin;
3358 mrmanese 694
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
695
                    _sqlite_error(res);
696
 
697
                    for (i = 0; i < svec_len; i++) {
698
                        sqlite3_step(stmt2);
699
 
700
                        sqlite3_reset(stmt);
701
                        sqlite3_bind_text(stmt, 1, (char *)sqlite3_column_text(stmt2, 0), -1, SQLITE_STATIC);
702
                        sqlite3_bind_double(stmt, 2, sqlite3_column_double(stmt2, 1));
703
                        sqlite3_bind_double(stmt, 3, (double)INTEGER(op2)[i % op2_len]);
704
                        sqlite3_step(stmt);
705
                    }
3419 mrmanese 706
                    _sqlite_commit;
3358 mrmanese 707
                }
708
                sqlite3_finalize(stmt2);
709
            } else {
710
                sprintf(g_sql_buf[1], "select [%s] from [%s].sdf_data",
711
                        varname_src, iname_src);
712
 
713
                if (funcname[0] == '%') {
714
                    sprintf(g_sql_buf[2], insert_fmt_string2c, iname, funcname[1]);  
715
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
716
                    _sqlite_error(res);
717
                } else {
718
                    sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
719
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
720
                    _sqlite_error(res);
721
                }
722
 
3419 mrmanese 723
                i = 0; _sqlite_begin; 
3358 mrmanese 724
                while (i < op2_len) {
725
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
726
                    _sqlite_error(res);
727
 
728
                    if (funcname[0] == '%') {
729
                        for ( ; i < op2_len && sqlite3_step(stmt) == SQLITE_ROW ; i++) {
730
                            sqlite3_step(stmt2);
731
 
732
                            sqlite3_reset(stmt);
733
                            /* simplify my life, just use ints for row names so that we
734
                             * don't have to worry about duplicates */
735
                            sqlite3_bind_int(stmt, 1, i); 
736
                            sqlite3_bind_int(stmt, 2, sqlite3_column_int(stmt2, 1));
737
                            sqlite3_bind_int(stmt, 3, INTEGER(op2)[i % op2_len]);
738
                            sqlite3_step(stmt);
739
                        }
740
                    } else {
741
                        for ( ; i < op2_len && sqlite3_step(stmt) == SQLITE_ROW ; i++) {
742
                            sqlite3_step(stmt2);
743
 
744
                            sqlite3_reset(stmt);
745
                            sqlite3_bind_int(stmt, 1, i);
746
                            sqlite3_bind_double(stmt, 2, sqlite3_column_double(stmt2, 1));
747
                            sqlite3_bind_double(stmt, 3, (double)INTEGER(op2)[i % op2_len]);
748
                            sqlite3_step(stmt);
749
                        }
750
                    }
751
 
752
                    /* recycle on svec if we loop again */
753
                    sqlite3_finalize(stmt2);
754
                }
3419 mrmanese 755
                _sqlite_commit;
3358 mrmanese 756
            }
757
 
758
            sqlite3_finalize(stmt);
759
        } else if (inherits(op2, "sqlite.vector")) { 
760
            /* op2 is surely not a factor, as handled by the R wrapper */
761
            char *iname_op2, *varname_op2;
762
            sqlite3_stmt *stmt3;
763
            iname_op2 = SDF_INAME(op2);
764
            varname_op2 = SVEC_VARNAME(op2);
765
 
3419 mrmanese 766
            if (!USE_SDF1(iname_op2, TRUE, TRUE)) {
3358 mrmanese 767
                /* delete created sqlite.vector */
768
                Rprintf("Warning: detaching created result SDF %s\n", iname);
769
                sdf_detach_sdf(mkString(iname));
770
                return R_NilValue;
771
            }
3419 mrmanese 772
            _sqlite_begin;
3358 mrmanese 773
 
774
            op2_len = _get_row_count2(iname_op2, 1);
775
 
776
            sprintf(g_sql_buf[2], "select [%s] from [%s].sdf_data", varname_src, iname_src);
777
            res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
778
            _sqlite_error(res);
779
 
780
            sprintf(g_sql_buf[2], "select [%s] from [%s].sdf_data", varname_op2, iname_op2);
781
            res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt3, 0);
782
            _sqlite_error(res);
783
 
784
            if (funcname[0] == '%') {
785
                sprintf(g_sql_buf[2], insert_fmt_string2c, iname, funcname[1]);
786
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
787
                _sqlite_error(res);
788
 
789
                if (svec_len == op2_len) {
790
                    for (i = 0; i < svec_len; i++) {
791
                        sqlite3_step(stmt2); sqlite3_step(stmt3);
792
 
793
                        sqlite3_reset(stmt);
794
                        sqlite3_bind_int(stmt, 1, i);
795
                        sqlite3_bind_int(stmt, 2, sqlite3_column_int(stmt2, 0));
796
                        sqlite3_bind_int(stmt, 3, sqlite3_column_int(stmt3, 0));
797
                        sqlite3_step(stmt);
798
                    }
799
                } else if (svec_len < op2_len) {
800
                    i = 0;
801
                    while (i < op2_len) {
802
                        for ( ; sqlite3_step(stmt2) == SQLITE_ROW && i < op2_len; i++) {
803
                            sqlite3_step(stmt3);
804
 
805
                            sqlite3_reset(stmt);
806
                            sqlite3_bind_int(stmt, 1, i);
807
                            sqlite3_bind_int(stmt, 2, sqlite3_column_int(stmt2, 0));
808
                            sqlite3_bind_int(stmt, 3, sqlite3_column_int(stmt3, 0));
809
                            sqlite3_step(stmt);
810
                        }
811
                        sqlite3_reset(stmt2);
812
                    }
813
                } else {
814
                    i = 0;
815
                    while (i < svec_len) {
816
                        for ( ; sqlite3_step(stmt3) == SQLITE_ROW && i < svec_len; i++) {
817
                            sqlite3_step(stmt2);
818
 
819
                            sqlite3_reset(stmt);
820
                            sqlite3_bind_int(stmt, 1, i);
821
                            sqlite3_bind_int(stmt, 2, sqlite3_column_int(stmt2, 0));
822
                            sqlite3_bind_int(stmt, 3, sqlite3_column_int(stmt3, 0));
823
                            sqlite3_step(stmt);
824
                        }
825
                        sqlite3_reset(stmt3);
826
                    }
827
                }
3419 mrmanese 828
                _sqlite_commit;
3358 mrmanese 829
            } else { /* not an integer op %% or %/% */
830
                sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
831
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
832
                _sqlite_error(res);
833
 
834
                if (svec_len == op2_len) {
835
                    for (i = 0; i < svec_len; i++) {
836
                        sqlite3_step(stmt2); sqlite3_step(stmt3);
837
 
838
                        sqlite3_reset(stmt);
839
                        sqlite3_bind_int(stmt, 1, i);
840
                        sqlite3_bind_double(stmt, 2, sqlite3_column_double(stmt2, 0));
841
                        sqlite3_bind_double(stmt, 3, sqlite3_column_double(stmt3, 0));
842
                        sqlite3_step(stmt);
843
                    }
844
                } else if (svec_len < op2_len) { /* recycle svec */
845
                    i = 0;
846
                    while (i < op2_len) {
847
                        for ( ; sqlite3_step(stmt2) == SQLITE_ROW && i < op2_len; i++) {
848
                            sqlite3_step(stmt3);
849
 
850
                            sqlite3_reset(stmt);
851
                            sqlite3_bind_int(stmt, 1, i);
852
                            sqlite3_bind_double(stmt, 2, sqlite3_column_double(stmt2, 0));
853
                            sqlite3_bind_double(stmt, 3, sqlite3_column_double(stmt3, 0));
854
                            sqlite3_step(stmt);
855
                        }
856
                        sqlite3_reset(stmt2);
857
                    }
858
                } else { /* svec_len > op2_len, recycle op2 */
859
                    i = 0;
860
                    while (i < svec_len) {
861
                        for ( ; sqlite3_step(stmt3) == SQLITE_ROW && i < svec_len; i++) {
862
                            sqlite3_step(stmt2);
863
 
864
                            sqlite3_reset(stmt);
865
                            sqlite3_bind_int(stmt, 1, i);
866
                            sqlite3_bind_double(stmt, 2, sqlite3_column_double(stmt2, 0));
867
                            sqlite3_bind_double(stmt, 3, sqlite3_column_double(stmt3, 0));
868
                            sqlite3_step(stmt);
869
                        }
870
                        sqlite3_reset(stmt3);
871
                    }
872
                }
873
            }
874
 
875
            sqlite3_finalize(stmt);
876
            sqlite3_finalize(stmt2);
877
            sqlite3_finalize(stmt3);
3419 mrmanese 878
            _sqlite_commit;
879
 
880
            UNUSE_SDF2(iname_op2);
3358 mrmanese 881
        }
882
    } else if (functype == 1 && funcname[0] != '!') { /* unary not operator */
3419 mrmanese 883
        iname = _create_svector1(R_NilValue, "bit", NULL, TRUE);
3358 mrmanese 884
        sprintf(g_sql_buf[2], "insert into [%s].sdf_data "
885
                    "select [row name], [%s] == 0 from [%s].sdf_data", 
886
                    iname, varname_src, iname_src);
887
        res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
888
        _sqlite_error(res);
889
        sqlite3_step(stmt);
890
        sqlite3_finalize(stmt);
891
    }
892
 
3419 mrmanese 893
    if (iname != NULL) {
3358 mrmanese 894
        return _create_svector_sexp(iname, "V1", 
895
                (functype == 0) ? "numeric" : "logical");
3419 mrmanese 896
        UNUSE_SDF2(iname);
897
    }
3358 mrmanese 898
 
3419 mrmanese 899
    UNUSE_SDF2(iname_src);
900
 
3358 mrmanese 901
    return R_NilValue;
902
 
3307 mrmanese 903
}
904
 
3324 mrmanese 905
SEXP sdf_do_variable_summary(SEXP func, SEXP vector, SEXP na_rm) {
906
    char *iname_src, *varname_src, *funcname;
907
    int res;
908
    sqlite3_stmt *stmt;
909
    double _ret = NA_REAL; SEXP ret;
3307 mrmanese 910
 
3324 mrmanese 911
    /* get data from arguments (function name and sqlite.vector stuffs) */
912
    funcname = CHAR_ELT(func, 0);
913
    iname_src = SDF_INAME(vector);
914
    varname_src = SVEC_VARNAME(vector);
3307 mrmanese 915
 
3419 mrmanese 916
    if (!USE_SDF1(iname_src, TRUE, FALSE)) return R_NilValue;
3324 mrmanese 917
 
918
    g_narm = LOGICAL(na_rm)[0];
919
    if (strcmp(funcname, "range") == 0) {
920
        /* special handling for range. use min then max */
921
        g_start = 1;
922
        g_accumulator = 0.0;
923
        sprintf(g_sql_buf[0], "select min_df([%s]) from [%s].sdf_data", varname_src, iname_src);
924
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);
925
        if (_sqlite_error(res)) return R_NilValue;
926
        sqlite3_step(stmt); sqlite3_finalize(stmt);
927
        _ret = g_accumulator;
928
 
929
        if (R_IsNA(_ret) && !g_narm) {
930
            PROTECT(ret = NEW_NUMERIC(2));
931
            REAL(ret)[0] = REAL(ret)[1] = R_NaReal;
932
            goto __sdf_do_variable_summary_out;
933
        }
934
 
935
        g_start = 1;
936
        g_accumulator = 0.0;
937
        sprintf(g_sql_buf[0], "select max_df([%s]) from [%s].sdf_data", varname_src, iname_src);
938
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);
939
        if (_sqlite_error(res)) return R_NilValue;
940
        sqlite3_step(stmt); sqlite3_finalize(stmt);
941
 
942
        /* if there is NA, then the if above should have caught it already */
943
        PROTECT(ret = NEW_NUMERIC(2));
944
        REAL(ret)[0] = _ret;
945
        REAL(ret)[1] = g_accumulator;
946
    } else {
947
        g_start = 1;  /* we'll use these instead of sqlite3_aggregate_context */
948
        g_accumulator = 0.0;
949
        sprintf(g_sql_buf[0], "select %s_df([%s]) from [%s].sdf_data", funcname, varname_src, iname_src);
950
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);
951
        if (_sqlite_error(res)) return R_NilValue;
952
        res = sqlite3_step(stmt); sqlite3_finalize(stmt);
953
 
954
        if (strcmp(funcname, "all") == 0 || strcmp(funcname, "any") == 0) {
955
            PROTECT(ret = NEW_LOGICAL(1));
956
            if (R_IsNA(g_accumulator)) LOGICAL(ret)[0] = NA_INTEGER;
957
            else LOGICAL(ret)[0] = !(g_accumulator == 0);
958
        } else {
959
            PROTECT(ret = NEW_NUMERIC(1));
960
            REAL(ret)[0] = g_accumulator;
961
        }
962
    }
963
 
964
__sdf_do_variable_summary_out:
965
    UNPROTECT(1);
966
    return ret;
967
}
968
 
969
 
3358 mrmanese 970
SEXP sdf_sort_variable(SEXP svec, SEXP decreasing) {
971
    char *iname, *iname_src, *varname_src, *type;
972
    sqlite3_stmt *stmt;
973
    int res;
974
 
975
    iname_src = SDF_INAME(svec);
976
    varname_src = SVEC_VARNAME(svec);
977
 
3419 mrmanese 978
    if (!USE_SDF1(iname_src, TRUE, TRUE)) return R_NilValue;
3358 mrmanese 979
 
980
    /* determine type of svec */
981
    sprintf(g_sql_buf[0], "select [%s] from [%s].sdf_data limit 1", varname_src, iname_src);
982
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
983
    _sqlite_error(res);
984
    sqlite3_step(stmt);
985
    strcpy(g_sql_buf[0], sqlite3_column_decltype(stmt, 0));
986
    sqlite3_finalize(stmt);
987
 
988
    /* create a new vector of that type */
3419 mrmanese 989
    iname = _create_svector1(mkString("tmp-sort"), g_sql_buf[0], NULL, TRUE);
3358 mrmanese 990
 
991
    /* insert to new sdf ordered */
992
    sprintf(g_sql_buf[0], "insert into [%s].sdf_data "
993
            "select [row name], [%s] from [%s].sdf_data "
994
            "order by [%s] %s", iname, varname_src, iname_src, varname_src,
995
            (LOGICAL(decreasing)[0]) ? "desc" : "asc");
996
    res = _sqlite_exec(g_sql_buf[0]);
997
    _sqlite_error(res);
998
 
999
    if (inherits(svec, "factor")) { /* copy factor table to iname */
1000
        if (inherits(svec, "ordered")) type = "ordered";
1001
        else type = "factor";
1002
        _copy_factor_levels2(type, iname_src, varname_src, iname, "V1");
1003
    } else type = CHAR_ELT(GET_CLASS(svec), 1);
1004
 
3419 mrmanese 1005
    UNUSE_SDF2(iname_src);
1006
    UNUSE_SDF2(iname);
1007
 
3358 mrmanese 1008
    return _create_svector_sexp(iname, "V1", type);
1009
}
1010
 
3307 mrmanese 1011
/****************************************************************************
1012
 * VECTOR MATH/OPS/GROUP OPERATIONS
1013
 ****************************************************************************/
3324 mrmanese 1014
 
3307 mrmanese 1015
int __vecmath_checkarg(sqlite3_context *ctx, sqlite3_value *arg, double *value) {
1016
    int ret = 1;
1017
    if (sqlite3_value_type(arg) == SQLITE_NULL) { 
1018
        sqlite3_result_null(ctx); 
1019
        ret = 0;
1020
    } else {
1021
        if (sqlite3_value_type(arg) == SQLITE_INTEGER) 
1022
            *value = sqlite3_value_int(arg); 
1023
        else *value = sqlite3_value_double(arg); 
1024
    }
1025
    return ret;
1026
}
1027
 
1028
#define SQLITE_MATH_FUNC1(name, func) static void __vecmath_ ## name(\
1029
        sqlite3_context *ctx, int argc, sqlite3_value **argv) { \
1030
    double value; \
1031
    if (__vecmath_checkarg(ctx, argv[0], &value)) { \
1032
        sqlite3_result_double(ctx, func(value)); \
1033
    }  \
1034
}
1035
 
3324 mrmanese 1036
#define SQLITE_MATH_FUNC_CUM(name, func) static void __vecmath_ ## name(\
1037
        sqlite3_context *ctx, int argc, sqlite3_value **argv) { \
1038
    double value; \
1039
    if (__vecmath_checkarg(ctx, argv[0], &value)) { \
1040
        if (g_start) { g_start = 0; g_accumulator = value; } \
1041
        else g_accumulator = func(g_accumulator, value); \
1042
        sqlite3_result_double(ctx, g_accumulator); \
1043
    }  \
1044
}
1045
 
3307 mrmanese 1046
/* SQLITE_MATH_FUNC1(abs, abs)   in SQLite */
1047
SQLITE_MATH_FUNC1(sign, sign)   /* in R */
1048
SQLITE_MATH_FUNC1(sqrt, sqrt)
1049
SQLITE_MATH_FUNC1(floor, floor)
1050
SQLITE_MATH_FUNC1(ceiling, ceil)
1051
SQLITE_MATH_FUNC1(trunc, ftrunc) /* in R */
1052
/* SQLITE_MATH_FUNC1(round, )   in SQLite */
1053
/* SQLITE_MATH_FUNC1(signif, ) 2 arg */
1054
SQLITE_MATH_FUNC1(exp, exp)
1055
/* SQLITE_MATH_FUNC1(log, ) 2 arg */
1056
SQLITE_MATH_FUNC1(cos, cos)
1057
SQLITE_MATH_FUNC1(sin, sin)
1058
SQLITE_MATH_FUNC1(tan, tan)
1059
SQLITE_MATH_FUNC1(acos, acos)
1060
SQLITE_MATH_FUNC1(asin, asin)
1061
SQLITE_MATH_FUNC1(atan, atan)
1062
SQLITE_MATH_FUNC1(cosh, cosh)
1063
SQLITE_MATH_FUNC1(sinh, sinh)
1064
SQLITE_MATH_FUNC1(tanh, tanh)
1065
SQLITE_MATH_FUNC1(acosh, acosh)  /* nowhere in include?? */
1066
SQLITE_MATH_FUNC1(asinh, asinh)  /* nowhere in include?? */
1067
SQLITE_MATH_FUNC1(atanh, atanh)  /* nowhere in include?? */
1068
SQLITE_MATH_FUNC1(lgamma, lgammafn) /* in R */
1069
SQLITE_MATH_FUNC1(gamma, gammafn) /* in R */
1070
/* SQLITE_MATH_FUNC1(gammaCody, gammaCody)   * in R ?? */
1071
SQLITE_MATH_FUNC1(digamma, digamma) /* in R */    
1072
SQLITE_MATH_FUNC1(trigamma, trigamma) /* in R */
1073
 
3324 mrmanese 1074
#define SUM(a, b)  (a) + (b)
1075
#define PROD(a, b) (a) * (b)
1076
#define MIN(a, b) ((a) <= (b)) ? (a) : (b)
1077
#define MAX(a, b) ((a) >= (b)) ? (a) : (b)
1078
#define ALL(a, b) (((a) == 0) || ((b) == 0)) ? 0 : 1
1079
#define ANY(a, b) (((a) == 0) && ((b) == 0)) ? 0 : 1
1080
 
1081
SQLITE_MATH_FUNC_CUM(cumsum, SUM)
1082
SQLITE_MATH_FUNC_CUM(cumprod, PROD)
1083
SQLITE_MATH_FUNC_CUM(cummin, MIN)
1084
SQLITE_MATH_FUNC_CUM(cummax, MAX)
1085
 
1086
#define SQLITE_SUMMARY_FUNC(name, func) static void __vecsummary_ ## name(\
1087
        sqlite3_context *ctx, int argc, sqlite3_value **argv) { \
1088
    double value; \
1089
    if (!g_narm && R_IsNA(g_accumulator)) return; /* NA if na.rm=F & NA found */ \
1090
    if (sqlite3_value_type(argv[0]) != SQLITE_NULL) {  \
1091
        if (sqlite3_value_type(argv[0]) == SQLITE_INTEGER) {  \
1092
            int tmp = sqlite3_value_int(argv[0]); \
1093
            value = (tmp == NA_INTEGER) ? R_NaReal : tmp; \
1094
        } else value = sqlite3_value_double(argv[0]);  \
1095
        if (R_IsNA(value)) { \
1096
            if (!g_narm) g_accumulator = value; return; \
1097
        } else if (g_start) { g_start = 0; g_accumulator = value; } \
1098
        else g_accumulator = func(g_accumulator, value); \
1099
    } \
1100
}
1101
 
1102
SQLITE_SUMMARY_FUNC(all_df, ALL)
1103
SQLITE_SUMMARY_FUNC(any_df, ANY)
1104
SQLITE_SUMMARY_FUNC(sum_df, SUM)
1105
SQLITE_SUMMARY_FUNC(prod_df, PROD)
1106
SQLITE_SUMMARY_FUNC(min_df, MIN)
1107
SQLITE_SUMMARY_FUNC(max_df, MAX)
1108
 
1109
static void __vecsummary_finalize(sqlite3_context *ctx) {
1110
    /* g_accumulator already summarizes it. just return that */
1111
    sqlite3_result_double(ctx, g_accumulator);
1112
}
1113
 
3307 mrmanese 1114
#define VMENTRY1(func)  {#func, __vecmath_ ## func}
3324 mrmanese 1115
#define VSENTRY1(func)  {#func, __vecsummary_ ## func}
3307 mrmanese 1116
void __register_vector_math() {
1117
    int i, res;
1118
    static const struct {
1119
        char *name;
1120
        void (*func)(sqlite3_context*, int, sqlite3_value**);
1121
    } arr_func1[] = {
1122
        VMENTRY1(sign),
1123
        VMENTRY1(sqrt),
1124
        VMENTRY1(floor),
1125
        VMENTRY1(ceiling),
1126
        VMENTRY1(trunc),
1127
        VMENTRY1(exp),
1128
        VMENTRY1(cos),
1129
        VMENTRY1(sin),
1130
        VMENTRY1(tan),
1131
        VMENTRY1(acos),
1132
        VMENTRY1(asin),
1133
        VMENTRY1(atan),
1134
        VMENTRY1(cosh),
1135
        VMENTRY1(sinh),
1136
        VMENTRY1(tanh),
1137
        VMENTRY1(acosh),
1138
        VMENTRY1(asinh),
1139
        VMENTRY1(atanh),
1140
        VMENTRY1(lgamma),
1141
        VMENTRY1(gamma),
1142
        VMENTRY1(digamma),
3324 mrmanese 1143
        VMENTRY1(trigamma),
1144
        VMENTRY1(cumsum),
1145
        VMENTRY1(cumprod),
1146
        VMENTRY1(cummin),
1147
        VMENTRY1(cummax)
1148
    }, arr_sum1[] = {
1149
        VSENTRY1(all_df),  /* can't override sum, min, max */
1150
        VSENTRY1(any_df),
1151
        VSENTRY1(sum_df),
1152
        VSENTRY1(prod_df),
1153
        VSENTRY1(min_df),
1154
        VSENTRY1(max_df)
3307 mrmanese 1155
    };
1156
 
3324 mrmanese 1157
    int len = sizeof(arr_func1) / sizeof(arr_func1[0]);
3307 mrmanese 1158
 
3324 mrmanese 1159
    for (i = 0; i < len; i++) {
3307 mrmanese 1160
        res = sqlite3_create_function(g_workspace, arr_func1[i].name, 1, 
1161
                SQLITE_ANY, NULL, arr_func1[i].func, NULL, NULL);
1162
        _sqlite_error(res);
1163
    }
3324 mrmanese 1164
 
1165
    len = sizeof(arr_sum1) / sizeof(arr_sum1[0]);
1166
    for (i = 0; i < len; i++) {
1167
        res = sqlite3_create_function(g_workspace, arr_sum1[i].name, 1, 
1168
                SQLITE_ANY, NULL, NULL, arr_sum1[i].func, __vecsummary_finalize);
1169
        _sqlite_error(res);
1170
    }
3307 mrmanese 1171
}