The R Project SVN R-packages

Rev

Rev 3432 | Rev 3446 | 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". */
3432 mrmanese 430
static long double g_accumulator = 0.0; /* accumulator var for cumsum, cumprod, etc. */
3324 mrmanese 431
static int g_start = 0;            /* flag for start of cummulation */
432
static int g_narm = 0;             /* for Summary group */
3255 mrmanese 433
 
3434 mrmanese 434
SEXP sdf_do_variable_math(SEXP func, SEXP vector, SEXP other_args) {
3307 mrmanese 435
    char *iname, *iname_src, *varname_src, *funcname;
3434 mrmanese 436
    int namelen, res;
3307 mrmanese 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);
3255 mrmanese 443
 
3419 mrmanese 444
    if (!USE_SDF1(iname_src, TRUE, TRUE)) return R_NilValue;
3308 mrmanese 445
 
3307 mrmanese 446
    /* create a new sdf, with 1 column named V1 */
3419 mrmanese 447
    iname = _create_svector1(R_NilValue, "double", &namelen, TRUE);
3307 mrmanese 448
 
449
    /* insert into <newsdf>.col, row.names select func(col), rownames */
3434 mrmanese 450
    if (strcmp(funcname, "round") == 0 || strcmp(funcname, "signif") == 0) {
451
        double digits = REAL(_getListElement(other_args, "digits"))[0];
452
        sprintf(g_sql_buf[0], "insert into [%s].sdf_data([row name], V1) "
453
                "select [row name], %s([%s],?) from [%s].sdf_data", iname, funcname,
454
                varname_src, iname_src);
455
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0); 
456
        if (_sqlite_error(res)) goto vecmath_prepare_error;
457
        res = sqlite3_bind_double(stmt, 1, digits);
458
    } else if (strcmp(funcname, "log") == 0) {
459
        double base = REAL(_getListElement(other_args, "base"))[0];
460
        sprintf(g_sql_buf[0], "insert into [%s].sdf_data([row name], V1) "
461
                "select [row name], %s([%s],?) from [%s].sdf_data", iname, funcname,
462
                varname_src, iname_src);
463
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0); 
464
        if (_sqlite_error(res)) goto vecmath_prepare_error;
465
        res = sqlite3_bind_double(stmt, 1, base);
466
    } else {
467
        sprintf(g_sql_buf[0], "insert into [%s].sdf_data([row name], V1) "
468
                "select [row name], %s([%s]) from [%s].sdf_data", iname, funcname,
469
                varname_src, iname_src);
470
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0); 
471
    }
3307 mrmanese 472
 
473
    if (_sqlite_error(res)) {
3434 mrmanese 474
vecmath_prepare_error:
3307 mrmanese 475
        sprintf(g_sql_buf[0], "detach %s", iname);
476
        _sqlite_exec(g_sql_buf[0]);
477
 
478
        /* we will return a string with the file name, and do file.remove
479
         * at R */
480
        iname[namelen] = '.';
481
        return mkString(iname);
482
    }
483
 
3324 mrmanese 484
    g_accumulator = 0.0;   /* initialize accumulator */
485
    g_start = 1;           /* flag that we are at start of accumulating */
3307 mrmanese 486
    sqlite3_step(stmt);
487
    sqlite3_finalize(stmt);
488
 
3419 mrmanese 489
    UNUSE_SDF2(iname);
490
    UNUSE_SDF2(iname_src);
491
 
3358 mrmanese 492
    return _create_svector_sexp(iname, "V1", "numeric");
493
}
3307 mrmanese 494
 
3434 mrmanese 495
SEXP sdf_do_variable_op(SEXP func, SEXP vector, SEXP op2, SEXP arg_reversed) {
3358 mrmanese 496
    char *iname = NULL, *iname_src, *varname_src, *funcname;
3434 mrmanese 497
    int res, functype = -1, op2_len, svec_len, i, reversed;
3358 mrmanese 498
    sqlite3_stmt *stmt, *stmt2;
3307 mrmanese 499
 
3358 mrmanese 500
    /* get data from arguments (function name and sqlite.vector stuffs) */
501
    funcname = CHAR_ELT(func, 0);
502
    iname_src = SDF_INAME(vector);
503
    varname_src = SVEC_VARNAME(vector);
3434 mrmanese 504
    reversed = LOGICAL(arg_reversed)[0];
3307 mrmanese 505
 
3419 mrmanese 506
    if (!USE_SDF1(iname_src, TRUE, TRUE)) return R_NilValue;
3358 mrmanese 507
    switch(funcname[0]) {
508
        case '+' :
509
        case '-' :
510
        case '*' :
511
        case '/' :
512
        case '^' :
513
        case '%' : /* %% and %/% */ 
514
            functype = 0; break;  /* output is REAL */
515
        case '&' :
516
        case '|' :
517
        case '!' :  /* ! and != */
518
            if (funcname[1] == 0) { functype = 1; break; }
519
        case '=' :  /* == */
520
        case '<' :  /* < and <= */
521
        case '>' :  /* > and >= */
522
            functype = 2; 
523
    }
3307 mrmanese 524
 
3358 mrmanese 525
    svec_len = _get_row_count2(iname_src, 1);
526
    if (functype == 0 || functype == 2 || (functype == 1 && funcname[0] != '!')) {
527
        char *insert_fmt_string1c, *insert_fmt_string1s;
528
        char *insert_fmt_string2c, *insert_fmt_string2s;
3434 mrmanese 529
        int vec_idx, sdf_idx;
530
 
3419 mrmanese 531
        iname = _create_svector1(R_NilValue, (functype == 0) ? "double" : "bit", NULL, TRUE);
3307 mrmanese 532
 
3434 mrmanese 533
        /* insert_fmt_string prefixes:
534
         * 1 - op2 is an ordinary vector, op2_len = 1, straightforward insert-select
535
         * 2 - op2 is an ordinary vector, op2_len > 1, may need recycling, loop both sides
536
         * s - operator will be printed as string
537
         * c - operator is either INTDIV or RMOD (int division, R modulo). initially, I
538
         *     thought I could cast both op to int then use / and % of sqlite, which is just
539
         *     the 2nd character of the funcname. however, R's INTDIV and casts to int 
540
         *     after division (e.g. 3.5 %/% 1.5 == 2). RMOD operates on doubles too,
541
         *     which is the remainder of the largest int multiple of "divisor"
542
         *     (e.g. 3.5 %% 1.5 = 0.5)
543
         */ 
3358 mrmanese 544
        if (functype == 1) { /* boolean binary operators */
3434 mrmanese 545
            if (!reversed) {
546
                insert_fmt_string1s = "insert into [%s].sdf_data "
547
                            "select [row name], ([%s] != 0) %s (? != 0) from [%s].sdf_data";
548
            } else {
549
                insert_fmt_string1s = "insert into [%s].sdf_data "
550
                            "select [row name], (? != 0) %s ([%s] != 0) from [%s].sdf_data";
551
            }
3358 mrmanese 552
            insert_fmt_string2s = "insert into [%s].sdf_data values(?, (? != 0) %s (? != 0))";
553
            /* no need for char version of fmt_string2, since %% only occurs for functype==0 */
554
            insert_fmt_string1c = insert_fmt_string1s;
555
            insert_fmt_string2c = insert_fmt_string2s;
556
        } else {
3434 mrmanese 557
            if (!reversed) {
558
                insert_fmt_string1s = "insert into [%s].sdf_data "
559
                            "select [row name], [%s] %s ? from [%s].sdf_data";
560
                insert_fmt_string1c = "insert into [%s].sdf_data "
561
                            "select [row name], %s([%s],?) from [%s].sdf_data";
562
            } else {
563
                insert_fmt_string1s = "insert into [%s].sdf_data "
564
                            "select [row name], ? %s [%s] from [%s].sdf_data";
565
                insert_fmt_string1c = "insert into [%s].sdf_data "
566
                            "select [row name], %s(?,[%s]) from [%s].sdf_data";
567
            }
3358 mrmanese 568
            /* fmt_string2c format operator from a char, used for %% and %/% */
3434 mrmanese 569
            insert_fmt_string2c = "insert into [%s].sdf_data values(?, %s(?,?))";
3358 mrmanese 570
            insert_fmt_string2s = "insert into [%s].sdf_data values(?, ? %s ?)";
571
        }
3307 mrmanese 572
 
3434 mrmanese 573
        /* for 2* insert statements, the values below specifies the index of bind().
574
         * if !reversed, then e1 is svec, e2 is anything. otherwise, e2 is the svec */
575
        if (reversed) { sdf_idx = 3; vec_idx = 2; }
576
        else { sdf_idx = 2; vec_idx = 3; }
3358 mrmanese 577
 
578
        if (IS_NUMERIC(op2)) {
579
            op2_len = LENGTH(op2);
580
 
581
            if (op2_len == 1) {
582
                if (funcname[0] == '%') {
583
                    sprintf(g_sql_buf[2], insert_fmt_string1c, iname, 
3434 mrmanese 584
                            (funcname[1] == '%') ? "r_mod" : "r_intdiv", varname_src, iname_src);
3358 mrmanese 585
                } else {
586
                    sprintf(g_sql_buf[2], insert_fmt_string1s, iname,
587
                            varname_src, funcname, iname_src);
588
                }
589
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
590
                _sqlite_error(res);
591
                sqlite3_bind_double(stmt, 1, REAL(op2)[0]);
592
                sqlite3_step(stmt);
3419 mrmanese 593
                sqlite3_finalize(stmt);
3358 mrmanese 594
            } else if (op2_len <= svec_len) {  /* recycle op2 */
595
                sprintf(g_sql_buf[2], "select [row name], [%s] from [%s].sdf_data",
596
                        varname_src, iname_src);
597
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
598
                _sqlite_error(res);
599
 
600
                if (funcname[0] == '%') {
3434 mrmanese 601
                    sprintf(g_sql_buf[2], insert_fmt_string2c, iname, 
602
                            (funcname[1] == '%') ? "r_mod" : "r_intdiv");  
3419 mrmanese 603
                    _sqlite_begin;
3358 mrmanese 604
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
605
                    _sqlite_error(res);
606
 
607
                    for (i = 0; i < svec_len; i++) {
608
                        sqlite3_step(stmt2);
609
 
610
                        sqlite3_reset(stmt);
611
                        sqlite3_bind_text(stmt, 1, (char *)sqlite3_column_text(stmt2, 0), -1, SQLITE_STATIC);
3434 mrmanese 612
                        sqlite3_bind_int(stmt, sdf_idx, (int)sqlite3_column_double(stmt2, 1));
613
                        sqlite3_bind_int(stmt, vec_idx, (int)REAL(op2)[i % op2_len]);
3358 mrmanese 614
                        sqlite3_step(stmt);
615
                    }
3419 mrmanese 616
                    sqlite3_finalize(stmt);
617
                    sqlite3_finalize(stmt2);
618
                    _sqlite_commit;
3358 mrmanese 619
                } else { /* non-integer binary operation */
620
                    sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
3419 mrmanese 621
                    _sqlite_begin;
3358 mrmanese 622
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
623
                    _sqlite_error(res);
624
 
625
                    for (i = 0; i < svec_len; i++) {
626
                        sqlite3_step(stmt2);
627
 
628
                        sqlite3_reset(stmt);
629
                        sqlite3_bind_text(stmt, 1, (char *)sqlite3_column_text(stmt2, 0), -1, SQLITE_STATIC);
3434 mrmanese 630
                        sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 1));
631
                        sqlite3_bind_double(stmt, vec_idx, REAL(op2)[i % op2_len]);
3358 mrmanese 632
                        sqlite3_step(stmt);
633
                    }
3419 mrmanese 634
                    sqlite3_finalize(stmt);
635
                    sqlite3_finalize(stmt2);
636
                    _sqlite_commit;
3358 mrmanese 637
                }
638
            } else { /* op2_len > svec_len, recycle svec */
639
                sprintf(g_sql_buf[1], "select [%s] from [%s].sdf_data",
640
                        varname_src, iname_src);
641
 
642
                if (funcname[0] == '%') {
3434 mrmanese 643
                    sprintf(g_sql_buf[2], insert_fmt_string2c, iname, 
644
                            (funcname[1] == '%') ? "r_mod" : "r_intdiv");
3358 mrmanese 645
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
646
                    _sqlite_error(res);
647
                } else {
648
                    sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
649
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
650
                    _sqlite_error(res);
651
                }
3434 mrmanese 652
 
3419 mrmanese 653
                i = 0; _sqlite_begin;
3358 mrmanese 654
                while (i < op2_len) {
655
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
656
                    _sqlite_error(res);
657
 
658
                    if (funcname[0] == '%') {
659
                        for ( ; i < op2_len && sqlite3_step(stmt) == SQLITE_ROW ; i++) {
660
                            sqlite3_step(stmt2);
661
 
662
                            sqlite3_reset(stmt);
663
                            /* simplify my life, just use ints for row names so that we
664
                             * don't have to worry about duplicates */
665
                            sqlite3_bind_int(stmt, 1, i); 
3434 mrmanese 666
                            sqlite3_bind_int(stmt, sdf_idx, sqlite3_column_int(stmt2, 1));
667
                            sqlite3_bind_int(stmt, vec_idx, (int)REAL(op2)[i % op2_len]);
3358 mrmanese 668
                            sqlite3_step(stmt);
669
                        }
670
                    } else {
671
                        for ( ; i < op2_len && sqlite3_step(stmt) == SQLITE_ROW ; i++) {
672
                            sqlite3_step(stmt2);
673
 
674
                            sqlite3_reset(stmt);
675
                            sqlite3_bind_int(stmt, 1, i);
3434 mrmanese 676
                            sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 1));
677
                            sqlite3_bind_double(stmt, vec_idx, REAL(op2)[i % op2_len]);
3358 mrmanese 678
                            sqlite3_step(stmt);
679
                        }
680
                    }
681
 
682
                    /* recycle on svec if we loop again */
683
                    sqlite3_finalize(stmt2);
684
                }
3397 mrmanese 685
 
3419 mrmanese 686
                sqlite3_finalize(stmt);
687
                _sqlite_commit;
3358 mrmanese 688
            }
689
 
690
        } else if (IS_INTEGER(op2) || IS_LOGICAL(op2)) { /* I N T E G E R */
691
            /* we are taking advantage of the fact that logicals are stored as int.
692
             * if this becomes untrue in the future, then this is a bug */
693
            op2_len = LENGTH(op2);
694
 
695
            if (op2_len == 1) {
696
                if (funcname[0] == '%') {
697
                    sprintf(g_sql_buf[2], insert_fmt_string1c, iname, 
3434 mrmanese 698
                            varname_src, (funcname[1] == '%') ? "r_mod" : "r_intdiv", iname_src);
3358 mrmanese 699
                } else {
700
                    sprintf(g_sql_buf[2], insert_fmt_string1s, iname,
701
                            varname_src, funcname, iname_src);
702
                }
703
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
704
                _sqlite_error(res);
705
                sqlite3_bind_double(stmt, 1, (double)INTEGER(op2)[0]);
706
                sqlite3_step(stmt);
707
            } else if (op2_len <= svec_len) {
708
                sprintf(g_sql_buf[2], "select [row name], [%s] from [%s].sdf_data",
709
                        varname_src, iname_src);
710
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
711
                _sqlite_error(res);
712
 
713
                if (funcname[0] == '%') {
714
                    /* sqlite does int div with "/" if both args are int. mod is % also.
715
                     * fortunately, in both cases the sqlite op is 2nd char of funcname */
3434 mrmanese 716
                    sprintf(g_sql_buf[2], insert_fmt_string2c, iname, 
717
                            (funcname[1] == '%') ? "r_mod" : "r_intdiv");
3419 mrmanese 718
                    _sqlite_begin;
3358 mrmanese 719
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
720
                    _sqlite_error(res);
721
 
722
                    for (i = 0; i < svec_len; i++) {
723
                        sqlite3_step(stmt2);
724
 
725
                        sqlite3_reset(stmt);
726
                        sqlite3_bind_text(stmt, 1, (char *)sqlite3_column_text(stmt2, 0), -1, SQLITE_STATIC);
3434 mrmanese 727
                        sqlite3_bind_int(stmt, sdf_idx, sqlite3_column_int(stmt2, 1));
728
                        sqlite3_bind_int(stmt, vec_idx, INTEGER(op2)[i % op2_len]);
3358 mrmanese 729
                        sqlite3_step(stmt);
730
                    }
3419 mrmanese 731
                    _sqlite_commit;
3358 mrmanese 732
                } else {
733
                    sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
3419 mrmanese 734
                    _sqlite_begin;
3358 mrmanese 735
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
736
                    _sqlite_error(res);
737
 
738
                    for (i = 0; i < svec_len; i++) {
739
                        sqlite3_step(stmt2);
740
 
741
                        sqlite3_reset(stmt);
742
                        sqlite3_bind_text(stmt, 1, (char *)sqlite3_column_text(stmt2, 0), -1, SQLITE_STATIC);
3434 mrmanese 743
                        sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 1));
744
                        sqlite3_bind_double(stmt, vec_idx, (double)INTEGER(op2)[i % op2_len]);
3358 mrmanese 745
                        sqlite3_step(stmt);
746
                    }
3419 mrmanese 747
                    _sqlite_commit;
3358 mrmanese 748
                }
749
                sqlite3_finalize(stmt2);
750
            } else {
751
                sprintf(g_sql_buf[1], "select [%s] from [%s].sdf_data",
752
                        varname_src, iname_src);
753
 
754
                if (funcname[0] == '%') {
3434 mrmanese 755
                    sprintf(g_sql_buf[2], insert_fmt_string2c, iname,
756
                            (funcname[1] == '%') ? "r_mod" : "r_intdiv");
3358 mrmanese 757
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
758
                    _sqlite_error(res);
759
                } else {
760
                    sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
761
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
762
                    _sqlite_error(res);
763
                }
764
 
3419 mrmanese 765
                i = 0; _sqlite_begin; 
3358 mrmanese 766
                while (i < op2_len) {
767
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
768
                    _sqlite_error(res);
769
 
770
                    if (funcname[0] == '%') {
771
                        for ( ; i < op2_len && sqlite3_step(stmt) == SQLITE_ROW ; i++) {
772
                            sqlite3_step(stmt2);
773
 
774
                            sqlite3_reset(stmt);
775
                            /* simplify my life, just use ints for row names so that we
776
                             * don't have to worry about duplicates */
777
                            sqlite3_bind_int(stmt, 1, i); 
3434 mrmanese 778
                            sqlite3_bind_int(stmt, sdf_idx, sqlite3_column_int(stmt2, 1));
779
                            sqlite3_bind_int(stmt, vec_idx, INTEGER(op2)[i % op2_len]);
3358 mrmanese 780
                            sqlite3_step(stmt);
781
                        }
782
                    } else {
783
                        for ( ; i < op2_len && sqlite3_step(stmt) == SQLITE_ROW ; i++) {
784
                            sqlite3_step(stmt2);
785
 
786
                            sqlite3_reset(stmt);
787
                            sqlite3_bind_int(stmt, 1, i);
3434 mrmanese 788
                            sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 1));
789
                            sqlite3_bind_double(stmt, vec_idx, (double)INTEGER(op2)[i % op2_len]);
3358 mrmanese 790
                            sqlite3_step(stmt);
791
                        }
792
                    }
793
 
794
                    /* recycle on svec if we loop again */
795
                    sqlite3_finalize(stmt2);
796
                }
3419 mrmanese 797
                _sqlite_commit;
3358 mrmanese 798
            }
799
 
800
            sqlite3_finalize(stmt);
801
        } else if (inherits(op2, "sqlite.vector")) { 
802
            /* op2 is surely not a factor, as handled by the R wrapper */
3434 mrmanese 803
            /* even though it is impossible for reversed to be FALSE, still use
804
             * sdf_idx and vec_idx so that code would be less confusing */
3358 mrmanese 805
            char *iname_op2, *varname_op2;
806
            sqlite3_stmt *stmt3;
807
            iname_op2 = SDF_INAME(op2);
808
            varname_op2 = SVEC_VARNAME(op2);
809
 
3419 mrmanese 810
            if (!USE_SDF1(iname_op2, TRUE, TRUE)) {
3358 mrmanese 811
                /* delete created sqlite.vector */
812
                Rprintf("Warning: detaching created result SDF %s\n", iname);
813
                sdf_detach_sdf(mkString(iname));
814
                return R_NilValue;
815
            }
3419 mrmanese 816
            _sqlite_begin;
3358 mrmanese 817
 
818
            op2_len = _get_row_count2(iname_op2, 1);
819
 
820
            sprintf(g_sql_buf[2], "select [%s] from [%s].sdf_data", varname_src, iname_src);
821
            res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, 0);
822
            _sqlite_error(res);
823
 
824
            sprintf(g_sql_buf[2], "select [%s] from [%s].sdf_data", varname_op2, iname_op2);
825
            res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt3, 0);
826
            _sqlite_error(res);
827
 
828
            if (funcname[0] == '%') {
3434 mrmanese 829
                sprintf(g_sql_buf[2], insert_fmt_string2c, iname,
830
                            (funcname[1] == '%') ? "r_mod" : "r_intdiv");
3358 mrmanese 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);
3434 mrmanese 840
                        sqlite3_bind_int(stmt, sdf_idx, sqlite3_column_int(stmt2, 0));
841
                        sqlite3_bind_int(stmt, vec_idx, sqlite3_column_int(stmt3, 0));
3358 mrmanese 842
                        sqlite3_step(stmt);
843
                    }
844
                } else if (svec_len < op2_len) {
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);
3434 mrmanese 852
                            sqlite3_bind_int(stmt, sdf_idx, sqlite3_column_int(stmt2, 0));
853
                            sqlite3_bind_int(stmt, vec_idx, sqlite3_column_int(stmt3, 0));
3358 mrmanese 854
                            sqlite3_step(stmt);
855
                        }
856
                        sqlite3_reset(stmt2);
857
                    }
858
                } else {
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);
3434 mrmanese 866
                            sqlite3_bind_int(stmt, sdf_idx, sqlite3_column_int(stmt2, 0));
867
                            sqlite3_bind_int(stmt, vec_idx, sqlite3_column_int(stmt3, 0));
3358 mrmanese 868
                            sqlite3_step(stmt);
869
                        }
870
                        sqlite3_reset(stmt3);
871
                    }
872
                }
3419 mrmanese 873
                _sqlite_commit;
3358 mrmanese 874
            } else { /* not an integer op %% or %/% */
875
                sprintf(g_sql_buf[2], insert_fmt_string2s, iname, funcname);
876
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
877
                _sqlite_error(res);
878
 
879
                if (svec_len == op2_len) {
880
                    for (i = 0; i < svec_len; i++) {
881
                        sqlite3_step(stmt2); sqlite3_step(stmt3);
882
 
883
                        sqlite3_reset(stmt);
884
                        sqlite3_bind_int(stmt, 1, i);
3434 mrmanese 885
                        sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 0));
886
                        sqlite3_bind_double(stmt, vec_idx, sqlite3_column_double(stmt3, 0));
3358 mrmanese 887
                        sqlite3_step(stmt);
888
                    }
889
                } else if (svec_len < op2_len) { /* recycle svec */
890
                    i = 0;
891
                    while (i < op2_len) {
892
                        for ( ; sqlite3_step(stmt2) == SQLITE_ROW && i < op2_len; i++) {
893
                            sqlite3_step(stmt3);
894
 
895
                            sqlite3_reset(stmt);
896
                            sqlite3_bind_int(stmt, 1, i);
3434 mrmanese 897
                            sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 0));
898
                            sqlite3_bind_double(stmt, vec_idx, sqlite3_column_double(stmt3, 0));
3358 mrmanese 899
                            sqlite3_step(stmt);
900
                        }
901
                        sqlite3_reset(stmt2);
902
                    }
903
                } else { /* svec_len > op2_len, recycle op2 */
904
                    i = 0;
905
                    while (i < svec_len) {
906
                        for ( ; sqlite3_step(stmt3) == SQLITE_ROW && i < svec_len; i++) {
907
                            sqlite3_step(stmt2);
908
 
909
                            sqlite3_reset(stmt);
910
                            sqlite3_bind_int(stmt, 1, i);
3434 mrmanese 911
                            sqlite3_bind_double(stmt, sdf_idx, sqlite3_column_double(stmt2, 0));
912
                            sqlite3_bind_double(stmt, vec_idx, sqlite3_column_double(stmt3, 0));
3358 mrmanese 913
                            sqlite3_step(stmt);
914
                        }
915
                        sqlite3_reset(stmt3);
916
                    }
917
                }
918
            }
919
 
920
            sqlite3_finalize(stmt);
921
            sqlite3_finalize(stmt2);
922
            sqlite3_finalize(stmt3);
3419 mrmanese 923
            _sqlite_commit;
924
 
925
            UNUSE_SDF2(iname_op2);
3358 mrmanese 926
        }
927
    } else if (functype == 1 && funcname[0] != '!') { /* unary not operator */
3419 mrmanese 928
        iname = _create_svector1(R_NilValue, "bit", NULL, TRUE);
3358 mrmanese 929
        sprintf(g_sql_buf[2], "insert into [%s].sdf_data "
930
                    "select [row name], [%s] == 0 from [%s].sdf_data", 
931
                    iname, varname_src, iname_src);
932
        res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
933
        _sqlite_error(res);
934
        sqlite3_step(stmt);
935
        sqlite3_finalize(stmt);
936
    }
937
 
3419 mrmanese 938
    if (iname != NULL) {
3358 mrmanese 939
        return _create_svector_sexp(iname, "V1", 
940
                (functype == 0) ? "numeric" : "logical");
3419 mrmanese 941
        UNUSE_SDF2(iname);
942
    }
3358 mrmanese 943
 
3419 mrmanese 944
    UNUSE_SDF2(iname_src);
945
 
3358 mrmanese 946
    return R_NilValue;
947
 
3307 mrmanese 948
}
949
 
3324 mrmanese 950
SEXP sdf_do_variable_summary(SEXP func, SEXP vector, SEXP na_rm) {
951
    char *iname_src, *varname_src, *funcname;
952
    int res;
953
    sqlite3_stmt *stmt;
954
    double _ret = NA_REAL; SEXP ret;
3307 mrmanese 955
 
3324 mrmanese 956
    /* get data from arguments (function name and sqlite.vector stuffs) */
957
    funcname = CHAR_ELT(func, 0);
958
    iname_src = SDF_INAME(vector);
959
    varname_src = SVEC_VARNAME(vector);
3307 mrmanese 960
 
3419 mrmanese 961
    if (!USE_SDF1(iname_src, TRUE, FALSE)) return R_NilValue;
3324 mrmanese 962
 
963
    g_narm = LOGICAL(na_rm)[0];
964
    if (strcmp(funcname, "range") == 0) {
965
        /* special handling for range. use min then max */
966
        g_start = 1;
967
        g_accumulator = 0.0;
968
        sprintf(g_sql_buf[0], "select min_df([%s]) from [%s].sdf_data", varname_src, iname_src);
969
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);
970
        if (_sqlite_error(res)) return R_NilValue;
971
        sqlite3_step(stmt); sqlite3_finalize(stmt);
972
        _ret = g_accumulator;
973
 
974
        if (R_IsNA(_ret) && !g_narm) {
975
            PROTECT(ret = NEW_NUMERIC(2));
976
            REAL(ret)[0] = REAL(ret)[1] = R_NaReal;
977
            goto __sdf_do_variable_summary_out;
978
        }
979
 
980
        g_start = 1;
981
        g_accumulator = 0.0;
982
        sprintf(g_sql_buf[0], "select max_df([%s]) from [%s].sdf_data", varname_src, iname_src);
983
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);
984
        if (_sqlite_error(res)) return R_NilValue;
985
        sqlite3_step(stmt); sqlite3_finalize(stmt);
986
 
987
        /* if there is NA, then the if above should have caught it already */
988
        PROTECT(ret = NEW_NUMERIC(2));
989
        REAL(ret)[0] = _ret;
990
        REAL(ret)[1] = g_accumulator;
991
    } else {
992
        g_start = 1;  /* we'll use these instead of sqlite3_aggregate_context */
993
        g_accumulator = 0.0;
994
        sprintf(g_sql_buf[0], "select %s_df([%s]) from [%s].sdf_data", funcname, varname_src, iname_src);
995
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);
996
        if (_sqlite_error(res)) return R_NilValue;
997
        res = sqlite3_step(stmt); sqlite3_finalize(stmt);
998
 
999
        if (strcmp(funcname, "all") == 0 || strcmp(funcname, "any") == 0) {
1000
            PROTECT(ret = NEW_LOGICAL(1));
1001
            if (R_IsNA(g_accumulator)) LOGICAL(ret)[0] = NA_INTEGER;
1002
            else LOGICAL(ret)[0] = !(g_accumulator == 0);
1003
        } else {
1004
            PROTECT(ret = NEW_NUMERIC(1));
1005
            REAL(ret)[0] = g_accumulator;
1006
        }
1007
    }
1008
 
1009
__sdf_do_variable_summary_out:
1010
    UNPROTECT(1);
1011
    return ret;
1012
}
1013
 
1014
 
3358 mrmanese 1015
SEXP sdf_sort_variable(SEXP svec, SEXP decreasing) {
1016
    char *iname, *iname_src, *varname_src, *type;
1017
    sqlite3_stmt *stmt;
1018
    int res;
1019
 
1020
    iname_src = SDF_INAME(svec);
1021
    varname_src = SVEC_VARNAME(svec);
1022
 
3419 mrmanese 1023
    if (!USE_SDF1(iname_src, TRUE, TRUE)) return R_NilValue;
3358 mrmanese 1024
 
1025
    /* determine type of svec */
1026
    sprintf(g_sql_buf[0], "select [%s] from [%s].sdf_data limit 1", varname_src, iname_src);
1027
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
1028
    _sqlite_error(res);
1029
    sqlite3_step(stmt);
1030
    strcpy(g_sql_buf[0], sqlite3_column_decltype(stmt, 0));
1031
    sqlite3_finalize(stmt);
1032
 
1033
    /* create a new vector of that type */
3419 mrmanese 1034
    iname = _create_svector1(mkString("tmp-sort"), g_sql_buf[0], NULL, TRUE);
3358 mrmanese 1035
 
1036
    /* insert to new sdf ordered */
1037
    sprintf(g_sql_buf[0], "insert into [%s].sdf_data "
1038
            "select [row name], [%s] from [%s].sdf_data "
1039
            "order by [%s] %s", iname, varname_src, iname_src, varname_src,
1040
            (LOGICAL(decreasing)[0]) ? "desc" : "asc");
1041
    res = _sqlite_exec(g_sql_buf[0]);
1042
    _sqlite_error(res);
1043
 
1044
    if (inherits(svec, "factor")) { /* copy factor table to iname */
1045
        if (inherits(svec, "ordered")) type = "ordered";
1046
        else type = "factor";
1047
        _copy_factor_levels2(type, iname_src, varname_src, iname, "V1");
1048
    } else type = CHAR_ELT(GET_CLASS(svec), 1);
1049
 
3419 mrmanese 1050
    UNUSE_SDF2(iname_src);
1051
    UNUSE_SDF2(iname);
1052
 
3358 mrmanese 1053
    return _create_svector_sexp(iname, "V1", type);
1054
}
1055
 
3307 mrmanese 1056
/****************************************************************************
1057
 * VECTOR MATH/OPS/GROUP OPERATIONS
1058
 ****************************************************************************/
3324 mrmanese 1059
 
3434 mrmanese 1060
R_INLINE int __vecmath_checkarg(sqlite3_context *ctx, sqlite3_value *arg, double *value) {
3307 mrmanese 1061
    int ret = 1;
1062
    if (sqlite3_value_type(arg) == SQLITE_NULL) { 
1063
        sqlite3_result_null(ctx); 
1064
        ret = 0;
1065
    } else {
1066
        if (sqlite3_value_type(arg) == SQLITE_INTEGER) 
1067
            *value = sqlite3_value_int(arg); 
1068
        else *value = sqlite3_value_double(arg); 
1069
    }
1070
    return ret;
1071
}
1072
 
1073
#define SQLITE_MATH_FUNC1(name, func) static void __vecmath_ ## name(\
1074
        sqlite3_context *ctx, int argc, sqlite3_value **argv) { \
1075
    double value; \
1076
    if (__vecmath_checkarg(ctx, argv[0], &value)) { \
1077
        sqlite3_result_double(ctx, func(value)); \
1078
    }  \
1079
}
1080
 
3434 mrmanese 1081
#define SQLITE_MATH_FUNC2(name, func) static void __vecmath_ ## name(\
1082
        sqlite3_context *ctx, int argc, sqlite3_value **argv) { \
1083
    double value1, value2; \
1084
    if (__vecmath_checkarg(ctx, argv[0], &value1) && \
1085
        __vecmath_checkarg(ctx, argv[1], &value2)) { \
1086
        sqlite3_result_double(ctx, func((long double)value1, (long double)value2)); \
1087
    }  \
1088
}
1089
 
3324 mrmanese 1090
#define SQLITE_MATH_FUNC_CUM(name, func) static void __vecmath_ ## name(\
1091
        sqlite3_context *ctx, int argc, sqlite3_value **argv) { \
1092
    double value; \
1093
    if (__vecmath_checkarg(ctx, argv[0], &value)) { \
1094
        if (g_start) { g_start = 0; g_accumulator = value; } \
1095
        else g_accumulator = func(g_accumulator, value); \
1096
        sqlite3_result_double(ctx, g_accumulator); \
1097
    }  \
1098
}
1099
 
3434 mrmanese 1100
#define LOGBASE(a, b) log(a)/log(b)
1101
 
3307 mrmanese 1102
/* SQLITE_MATH_FUNC1(abs, abs)   in SQLite */
1103
SQLITE_MATH_FUNC1(sign, sign)   /* in R */
1104
SQLITE_MATH_FUNC1(sqrt, sqrt)
1105
SQLITE_MATH_FUNC1(floor, floor)
1106
SQLITE_MATH_FUNC1(ceiling, ceil)
1107
SQLITE_MATH_FUNC1(trunc, ftrunc) /* in R */
3434 mrmanese 1108
/*SQLITE_MATH_FUNC2(round, fprec)  2 arg, in SQLite, but override with R's version */
1109
SQLITE_MATH_FUNC2(signif, fround) /* 2 arg, in R */
3307 mrmanese 1110
SQLITE_MATH_FUNC1(exp, exp)
3434 mrmanese 1111
SQLITE_MATH_FUNC2(log, LOGBASE) /* 2 arg */
3307 mrmanese 1112
SQLITE_MATH_FUNC1(cos, cos)
1113
SQLITE_MATH_FUNC1(sin, sin)
1114
SQLITE_MATH_FUNC1(tan, tan)
1115
SQLITE_MATH_FUNC1(acos, acos)
1116
SQLITE_MATH_FUNC1(asin, asin)
1117
SQLITE_MATH_FUNC1(atan, atan)
1118
SQLITE_MATH_FUNC1(cosh, cosh)
1119
SQLITE_MATH_FUNC1(sinh, sinh)
1120
SQLITE_MATH_FUNC1(tanh, tanh)
1121
SQLITE_MATH_FUNC1(acosh, acosh)  /* nowhere in include?? */
1122
SQLITE_MATH_FUNC1(asinh, asinh)  /* nowhere in include?? */
1123
SQLITE_MATH_FUNC1(atanh, atanh)  /* nowhere in include?? */
1124
SQLITE_MATH_FUNC1(lgamma, lgammafn) /* in R */
1125
SQLITE_MATH_FUNC1(gamma, gammafn) /* in R */
1126
/* SQLITE_MATH_FUNC1(gammaCody, gammaCody)   * in R ?? */
1127
SQLITE_MATH_FUNC1(digamma, digamma) /* in R */    
1128
SQLITE_MATH_FUNC1(trigamma, trigamma) /* in R */
1129
 
3324 mrmanese 1130
#define SUM(a, b)  (a) + (b)
1131
#define PROD(a, b) (a) * (b)
1132
#define MIN(a, b) ((a) <= (b)) ? (a) : (b)
1133
#define MAX(a, b) ((a) >= (b)) ? (a) : (b)
1134
#define ALL(a, b) (((a) == 0) || ((b) == 0)) ? 0 : 1
1135
#define ANY(a, b) (((a) == 0) && ((b) == 0)) ? 0 : 1
1136
 
1137
SQLITE_MATH_FUNC_CUM(cumsum, SUM)
1138
SQLITE_MATH_FUNC_CUM(cumprod, PROD)
1139
SQLITE_MATH_FUNC_CUM(cummin, MIN)
1140
SQLITE_MATH_FUNC_CUM(cummax, MAX)
1141
 
1142
#define SQLITE_SUMMARY_FUNC(name, func) static void __vecsummary_ ## name(\
1143
        sqlite3_context *ctx, int argc, sqlite3_value **argv) { \
1144
    double value; \
1145
    if (!g_narm && R_IsNA(g_accumulator)) return; /* NA if na.rm=F & NA found */ \
1146
    if (sqlite3_value_type(argv[0]) != SQLITE_NULL) {  \
1147
        if (sqlite3_value_type(argv[0]) == SQLITE_INTEGER) {  \
1148
            int tmp = sqlite3_value_int(argv[0]); \
1149
            value = (tmp == NA_INTEGER) ? R_NaReal : tmp; \
1150
        } else value = sqlite3_value_double(argv[0]);  \
1151
        if (R_IsNA(value)) { \
1152
            if (!g_narm) g_accumulator = value; return; \
1153
        } else if (g_start) { g_start = 0; g_accumulator = value; } \
1154
        else g_accumulator = func(g_accumulator, value); \
1155
    } \
1156
}
1157
 
1158
SQLITE_SUMMARY_FUNC(all_df, ALL)
1159
SQLITE_SUMMARY_FUNC(any_df, ANY)
1160
SQLITE_SUMMARY_FUNC(sum_df, SUM)
1161
SQLITE_SUMMARY_FUNC(prod_df, PROD)
1162
SQLITE_SUMMARY_FUNC(min_df, MIN)
1163
SQLITE_SUMMARY_FUNC(max_df, MAX)
1164
 
1165
static void __vecsummary_finalize(sqlite3_context *ctx) {
1166
    /* g_accumulator already summarizes it. just return that */
1167
    sqlite3_result_double(ctx, g_accumulator);
1168
}
1169
 
3434 mrmanese 1170
static void __r_modulo(sqlite3_context *ctx, int argc, sqlite3_value **argv) {
1171
    if (sqlite3_value_type(argv[0]) == SQLITE_NULL ||
1172
        sqlite3_value_type(argv[1]) == SQLITE_NULL) { 
1173
        sqlite3_result_null(ctx); 
1174
    } else {
1175
        double v1, v2, q, tmp;
1176
        if (sqlite3_value_type(argv[0]) == SQLITE_INTEGER && 
1177
            sqlite3_value_type(argv[1]) == SQLITE_INTEGER) {
1178
            int i1, i2;
1179
            i1 = sqlite3_value_int(argv[0]);
1180
            i2 = sqlite3_value_int(argv[1]);
1181
            if (i1 > 0 && i2 > 0) { 
1182
                sqlite3_result_int(ctx, i1 % i2); return;
1183
            }
1184
            v1 = i1; v2 = i2;
1185
        } else {
1186
            v1 = sqlite3_value_double(argv[0]);
1187
            v2 = sqlite3_value_double(argv[1]);
1188
        }
1189
        /* copied from myfmod() in src/main/arithmetic.c */
1190
        q = v1 / v2; 
1191
        if (v2 == 0) sqlite3_result_double(ctx, R_NaN);
1192
        tmp = v1 - floor(q) * v2;
1193
        /* checking omitted */
1194
        q = floor(tmp/v2);
1195
        sqlite3_result_double(ctx, tmp - q*v2);
1196
 
1197
    }
1198
}
1199
 
1200
static void __r_intdiv(sqlite3_context *ctx, int argc, sqlite3_value **argv) {
1201
    if (sqlite3_value_type(argv[0]) == SQLITE_NULL ||
1202
        sqlite3_value_type(argv[1]) == SQLITE_NULL) { 
1203
        sqlite3_result_null(ctx); 
1204
    } else {
1205
        double v1, v2;
1206
        if (sqlite3_value_type(argv[0]) == SQLITE_INTEGER && 
1207
            sqlite3_value_type(argv[1]) == SQLITE_INTEGER) {
1208
            int i1, i2;
1209
            i1 = sqlite3_value_int(argv[0]);
1210
            i2 = sqlite3_value_int(argv[1]);
1211
            if (i1 == NA_INTEGER || i2 == NA_INTEGER) {
1212
                sqlite3_result_int(ctx, NA_INTEGER); return;
1213
            } else if (i2 == 0) {
1214
                sqlite3_result_int(ctx, 0); return;
1215
            }
1216
            v1 = i1; v2 = i2;
1217
        } else {
1218
            v1 = sqlite3_value_double(argv[0]);
1219
            v2 = sqlite3_value_double(argv[1]);
1220
        }
1221
        /* copied from IDIVOP cases in src/main/arithmetic.c */
1222
        sqlite3_result_double(ctx, floor(v1/v2));
1223
    }
1224
}
3307 mrmanese 1225
#define VMENTRY1(func)  {#func, __vecmath_ ## func}
3324 mrmanese 1226
#define VSENTRY1(func)  {#func, __vecsummary_ ## func}
3307 mrmanese 1227
void __register_vector_math() {
1228
    int i, res;
1229
    static const struct {
1230
        char *name;
1231
        void (*func)(sqlite3_context*, int, sqlite3_value**);
1232
    } arr_func1[] = {
1233
        VMENTRY1(sign),
1234
        VMENTRY1(sqrt),
1235
        VMENTRY1(floor),
1236
        VMENTRY1(ceiling),
1237
        VMENTRY1(trunc),
1238
        VMENTRY1(exp),
1239
        VMENTRY1(cos),
1240
        VMENTRY1(sin),
1241
        VMENTRY1(tan),
1242
        VMENTRY1(acos),
1243
        VMENTRY1(asin),
1244
        VMENTRY1(atan),
1245
        VMENTRY1(cosh),
1246
        VMENTRY1(sinh),
1247
        VMENTRY1(tanh),
1248
        VMENTRY1(acosh),
1249
        VMENTRY1(asinh),
1250
        VMENTRY1(atanh),
1251
        VMENTRY1(lgamma),
1252
        VMENTRY1(gamma),
1253
        VMENTRY1(digamma),
3324 mrmanese 1254
        VMENTRY1(trigamma),
1255
        VMENTRY1(cumsum),
1256
        VMENTRY1(cumprod),
1257
        VMENTRY1(cummin),
1258
        VMENTRY1(cummax)
3434 mrmanese 1259
    }, arr_func2[] =  {
1260
        {"r_mod", __r_modulo},
1261
        {"r_intdiv", __r_intdiv},
1262
        /*VMENTRY1(round),*/
1263
        VMENTRY1(signif),
1264
        VMENTRY1(log)
3324 mrmanese 1265
    }, arr_sum1[] = {
1266
        VSENTRY1(all_df),  /* can't override sum, min, max */
1267
        VSENTRY1(any_df),
1268
        VSENTRY1(sum_df),
1269
        VSENTRY1(prod_df),
1270
        VSENTRY1(min_df),
1271
        VSENTRY1(max_df)
3307 mrmanese 1272
    };
1273
 
3324 mrmanese 1274
    int len = sizeof(arr_func1) / sizeof(arr_func1[0]);
3307 mrmanese 1275
 
3324 mrmanese 1276
    for (i = 0; i < len; i++) {
3307 mrmanese 1277
        res = sqlite3_create_function(g_workspace, arr_func1[i].name, 1, 
1278
                SQLITE_ANY, NULL, arr_func1[i].func, NULL, NULL);
1279
        _sqlite_error(res);
1280
    }
3324 mrmanese 1281
 
3434 mrmanese 1282
    len = sizeof(arr_func2) / sizeof(arr_func2[0]);
1283
    for (i = 0; i < len; i++) {
1284
        res = sqlite3_create_function(g_workspace, arr_func2[i].name, 2, 
1285
                SQLITE_ANY, NULL, arr_func2[i].func, NULL, NULL);
1286
        _sqlite_error(res);
1287
    }
1288
 
3324 mrmanese 1289
    len = sizeof(arr_sum1) / sizeof(arr_sum1[0]);
1290
    for (i = 0; i < len; i++) {
1291
        res = sqlite3_create_function(g_workspace, arr_sum1[i].name, 1, 
1292
                SQLITE_ANY, NULL, NULL, arr_sum1[i].func, __vecsummary_finalize);
1293
        _sqlite_error(res);
1294
    }
3307 mrmanese 1295
}