| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | /* | ||
| 2 | * Copyright (c) 2026 Tiger Data, Inc. | ||
| 3 | * Licensed under the PostgreSQL License. See LICENSE for details. | ||
| 4 | * | ||
| 5 | * vs_pg_rabitq.c - PostgreSQL functions for the rabitq type | ||
| 6 | * | ||
| 7 | * Type I/O, comparison operators, accessor functions, and encoding. | ||
| 8 | * Text format: {bits:f_add:f_rescale} where bits are 0/1 characters. | ||
| 9 | */ | ||
| 10 | |||
| 11 | #include <postgres.h> | ||
| 12 | |||
| 13 | #include <fmgr.h> | ||
| 14 | #include <lib/stringinfo.h> | ||
| 15 | #include <utils/array.h> | ||
| 16 | |||
| 17 | #include "quant/matrix.h" | ||
| 18 | #include "support_pg.h" | ||
| 19 | |||
| 20 | /* ---------------------------------------------------------------- | ||
| 21 | * Type I/O | ||
| 22 | * ---------------------------------------------------------------- */ | ||
| 23 | |||
| 24 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_in); | |
| 25 | |||
| 26 | Datum | ||
| 27 | 75 | vs_rabitq_in(PG_FUNCTION_ARGS) | |
| 28 | { | ||
| 29 | 75 | char *str = PG_GETARG_CSTRING(0); | |
| 30 | 75 | int32 typmod = PG_GETARG_INT32(2); | |
| 31 | 75 | uint8_t bits_buf[VS_RABITQ_BYTES(VEC32_MAX_DIM)]; | |
| 32 | 75 | int dim = 0; | |
| 33 | 75 | char *p = str; | |
| 34 | |||
| 35 | 75 | memset(bits_buf, 0, sizeof(bits_buf)); | |
| 36 | |||
| 37 | /* Skip whitespace */ | ||
| 38 |
2/2✓ Branch 0 taken 4 times.
✓ Branch 1 taken 75 times.
|
79 | while (*p == ' ' || *p == '\t') |
| 39 | 4 | p++; | |
| 40 | |||
| 41 | /* Expect opening '{' */ | ||
| 42 |
2/2✓ Branch 0 taken 1 times.
✓ Branch 1 taken 74 times.
|
75 | if (*p != '{') |
| 43 |
1/2✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
|
1 | ereport(ERROR, |
| 44 | (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), | ||
| 45 | errmsg("rabitq must start with \"{\""))); | ||
| 46 | 74 | p++; | |
| 47 | |||
| 48 | /* Parse bit characters */ | ||
| 49 |
4/4✓ Branch 0 taken 574 times.
✓ Branch 1 taken 72 times.
✓ Branch 2 taken 573 times.
✓ Branch 3 taken 1 times.
|
646 | while (*p && *p != ':' && *p != '}') |
| 50 | { | ||
| 51 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 573 times.
|
573 | if (dim >= VEC32_MAX_DIM) |
| 52 | ✗ | ereport(ERROR, | |
| 53 | (errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED), | ||
| 54 | errmsg("rabitq cannot have more than %d dimensions", | ||
| 55 | VEC32_MAX_DIM))); | ||
| 56 | |||
| 57 |
2/2✓ Branch 0 taken 261 times.
✓ Branch 1 taken 312 times.
|
573 | if (*p == '1') |
| 58 | 261 | bits_buf[dim / 8] |= (1 << (dim % 8)); | |
| 59 |
2/2✓ Branch 0 taken 1 times.
✓ Branch 1 taken 311 times.
|
312 | else if (*p != '0') |
| 60 |
1/2✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
|
1 | ereport(ERROR, |
| 61 | (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), | ||
| 62 | errmsg("invalid bit character '%c' in rabitq", *p))); | ||
| 63 | |||
| 64 | 572 | dim++; | |
| 65 | 572 | p++; | |
| 66 | } | ||
| 67 | |||
| 68 |
2/2✓ Branch 0 taken 1 times.
✓ Branch 1 taken 72 times.
|
73 | if (dim < 1) |
| 69 |
1/2✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
|
1 | ereport(ERROR, |
| 70 | (errcode(ERRCODE_DATA_EXCEPTION), | ||
| 71 | errmsg("rabitq must have at least 1 dimension"))); | ||
| 72 | |||
| 73 | /* Expect ':' separator */ | ||
| 74 |
2/2✓ Branch 0 taken 1 times.
✓ Branch 1 taken 71 times.
|
72 | if (*p != ':') |
| 75 |
1/2✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
|
1 | ereport(ERROR, |
| 76 | (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), | ||
| 77 | errmsg("expected \":\" after bits in rabitq"))); | ||
| 78 | 71 | p++; | |
| 79 | |||
| 80 | /* Parse f_add */ | ||
| 81 | 71 | char *end; | |
| 82 | 71 | errno = 0; | |
| 83 | 71 | float f_add = strtof(p, &end); | |
| 84 |
2/4✓ Branch 0 taken 71 times.
✗ Branch 1 not taken.
✗ Branch 2 not taken.
✓ Branch 3 taken 71 times.
|
71 | if (end == p || errno == ERANGE) |
| 85 | ✗ | ereport(ERROR, | |
| 86 | (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), | ||
| 87 | errmsg("invalid f_add value in rabitq: \"%s\"", str))); | ||
| 88 | 71 | vs_pg_check_value_finite(f_add); | |
| 89 | 70 | p = end; | |
| 90 | |||
| 91 | /* Expect ':' separator */ | ||
| 92 |
2/2✓ Branch 0 taken 1 times.
✓ Branch 1 taken 69 times.
|
70 | if (*p != ':') |
| 93 |
1/2✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
|
1 | ereport(ERROR, |
| 94 | (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), | ||
| 95 | errmsg("expected \":\" after f_add in rabitq"))); | ||
| 96 | 69 | p++; | |
| 97 | |||
| 98 | /* Parse f_rescale */ | ||
| 99 | 69 | errno = 0; | |
| 100 | 69 | float f_rescale = strtof(p, &end); | |
| 101 |
2/4✓ Branch 0 taken 69 times.
✗ Branch 1 not taken.
✗ Branch 2 not taken.
✓ Branch 3 taken 69 times.
|
69 | if (end == p || errno == ERANGE) |
| 102 | ✗ | ereport(ERROR, | |
| 103 | (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), | ||
| 104 | errmsg("invalid f_rescale value in rabitq: \"%s\"", str))); | ||
| 105 | 69 | vs_pg_check_value_finite(f_rescale); | |
| 106 | 68 | p = end; | |
| 107 | |||
| 108 | /* Expect closing '}' */ | ||
| 109 |
2/2✓ Branch 0 taken 1 times.
✓ Branch 1 taken 67 times.
|
68 | if (*p != '}') |
| 110 |
1/2✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
|
1 | ereport(ERROR, |
| 111 | (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), | ||
| 112 | errmsg("rabitq must end with \"}\""))); | ||
| 113 | 67 | p++; | |
| 114 | |||
| 115 | /* Check no trailing content */ | ||
| 116 |
2/2✓ Branch 0 taken 4 times.
✓ Branch 1 taken 67 times.
|
71 | while (*p == ' ' || *p == '\t') |
| 117 | 4 | p++; | |
| 118 |
2/2✓ Branch 0 taken 1 times.
✓ Branch 1 taken 66 times.
|
67 | if (*p != '\0') |
| 119 |
1/2✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
|
1 | ereport(ERROR, |
| 120 | (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), | ||
| 121 | errmsg("unexpected characters after \"}\" in " | ||
| 122 | "rabitq"))); | ||
| 123 | |||
| 124 | 66 | vs_pg_check_expected_dim(dim, typmod); | |
| 125 | |||
| 126 | 66 | RaBitQVector *result = vs_pg_rabitq_alloc(dim); | |
| 127 | 66 | result->f_add = f_add; | |
| 128 | 66 | result->f_rescale = f_rescale; | |
| 129 | 66 | memcpy(result->bits, bits_buf, VS_RABITQ_BYTES(dim)); | |
| 130 | |||
| 131 | 66 | PG_RETURN_RABITQ_P(result); | |
| 132 | } | ||
| 133 | |||
| 134 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_out); | |
| 135 | |||
| 136 | Datum | ||
| 137 | 16 | vs_rabitq_out(PG_FUNCTION_ARGS) | |
| 138 | { | ||
| 139 | 16 | RaBitQVector *v = PG_GETARG_RABITQ_P(0); | |
| 140 | 16 | StringInfoData buf; | |
| 141 | |||
| 142 | 16 | initStringInfo(&buf); | |
| 143 | 16 | appendStringInfoChar(&buf, '{'); | |
| 144 | |||
| 145 |
2/2✓ Branch 1 taken 121 times.
✓ Branch 2 taken 16 times.
|
137 | for (int i = 0; i < v->dim; i++) |
| 146 | 121 | appendStringInfoChar( | |
| 147 |
2/2✓ Branch 0 taken 60 times.
✓ Branch 1 taken 61 times.
|
121 | &buf, (v->bits[i / 8] >> (i % 8)) & 1 ? '1' : '0'); |
| 148 | |||
| 149 | 16 | appendStringInfo(&buf, ":%.8g:%.8g}", v->f_add, v->f_rescale); | |
| 150 | |||
| 151 | 16 | PG_RETURN_CSTRING(buf.data); | |
| 152 | } | ||
| 153 | |||
| 154 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_typmod_in); | |
| 155 | |||
| 156 | Datum | ||
| 157 | 4 | vs_rabitq_typmod_in(PG_FUNCTION_ARGS) | |
| 158 | { | ||
| 159 | 4 | ArrayType *ta = PG_GETARG_ARRAYTYPE_P(0); | |
| 160 | 4 | int n; | |
| 161 | 4 | int32 *tl = ArrayGetIntegerTypmods(ta, &n); | |
| 162 | |||
| 163 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 4 times.
|
4 | if (n != 1) |
| 164 | ✗ | ereport(ERROR, | |
| 165 | (errcode(ERRCODE_INVALID_PARAMETER_VALUE), | ||
| 166 | errmsg("invalid type modifier"))); | ||
| 167 | |||
| 168 | 4 | int dim = tl[0]; | |
| 169 | |||
| 170 | 4 | vs_pg_check_dim_valid(dim); | |
| 171 | |||
| 172 | 4 | PG_RETURN_INT32(dim); | |
| 173 | } | ||
| 174 | |||
| 175 | /* ---------------------------------------------------------------- | ||
| 176 | * Accessor functions | ||
| 177 | * ---------------------------------------------------------------- */ | ||
| 178 | |||
| 179 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_dims); | |
| 180 | |||
| 181 | Datum | ||
| 182 | 4 | vs_rabitq_dims(PG_FUNCTION_ARGS) | |
| 183 | { | ||
| 184 | 4 | RaBitQVector *v = PG_GETARG_RABITQ_P(0); | |
| 185 | 4 | PG_RETURN_INT32(v->dim); | |
| 186 | } | ||
| 187 | |||
| 188 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_f_add); | |
| 189 | |||
| 190 | Datum | ||
| 191 | 2 | vs_rabitq_f_add(PG_FUNCTION_ARGS) | |
| 192 | { | ||
| 193 | 2 | RaBitQVector *v = PG_GETARG_RABITQ_P(0); | |
| 194 | 2 | PG_RETURN_FLOAT8((double)v->f_add); | |
| 195 | } | ||
| 196 | |||
| 197 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_f_rescale); | |
| 198 | |||
| 199 | Datum | ||
| 200 | 2 | vs_rabitq_f_rescale(PG_FUNCTION_ARGS) | |
| 201 | { | ||
| 202 | 2 | RaBitQVector *v = PG_GETARG_RABITQ_P(0); | |
| 203 | 2 | PG_RETURN_FLOAT8((double)v->f_rescale); | |
| 204 | } | ||
| 205 | |||
| 206 | /* ---------------------------------------------------------------- | ||
| 207 | * Comparison functions (for btree opclass) | ||
| 208 | * | ||
| 209 | * Compare dim first, then memcmp on bits. f_add/f_rescale are | ||
| 210 | * intentionally excluded -- bits are the quantized identity. | ||
| 211 | * ---------------------------------------------------------------- */ | ||
| 212 | |||
| 213 | static int | ||
| 214 | 17 | vs_rabitq_cmp_internal(RaBitQVector *a, RaBitQVector *b) | |
| 215 | { | ||
| 216 |
2/2✓ Branch 0 taken 1 times.
✓ Branch 1 taken 16 times.
|
17 | if (a->dim != b->dim) |
| 217 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 1 times.
|
1 | return (a->dim < b->dim) ? -1 : 1; |
| 218 | |||
| 219 | 16 | int nbytes = VS_RABITQ_BYTES(a->dim); | |
| 220 | 16 | return memcmp(a->bits, b->bits, nbytes); | |
| 221 | } | ||
| 222 | |||
| 223 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_cmp); | |
| 224 | |||
| 225 | Datum | ||
| 226 | 9 | vs_rabitq_cmp(PG_FUNCTION_ARGS) | |
| 227 | { | ||
| 228 | 9 | RaBitQVector *a = PG_GETARG_RABITQ_P(0); | |
| 229 | 9 | RaBitQVector *b = PG_GETARG_RABITQ_P(1); | |
| 230 | 9 | PG_RETURN_INT32(vs_rabitq_cmp_internal(a, b)); | |
| 231 | } | ||
| 232 | |||
| 233 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_lt); | |
| 234 | |||
| 235 | Datum | ||
| 236 | 2 | vs_rabitq_lt(PG_FUNCTION_ARGS) | |
| 237 | { | ||
| 238 | 2 | RaBitQVector *a = PG_GETARG_RABITQ_P(0); | |
| 239 | 2 | RaBitQVector *b = PG_GETARG_RABITQ_P(1); | |
| 240 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 2 times.
|
2 | PG_RETURN_BOOL(vs_rabitq_cmp_internal(a, b) < 0); |
| 241 | } | ||
| 242 | |||
| 243 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_le); | |
| 244 | |||
| 245 | Datum | ||
| 246 | 1 | vs_rabitq_le(PG_FUNCTION_ARGS) | |
| 247 | { | ||
| 248 | 1 | RaBitQVector *a = PG_GETARG_RABITQ_P(0); | |
| 249 | 1 | RaBitQVector *b = PG_GETARG_RABITQ_P(1); | |
| 250 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 1 times.
|
1 | PG_RETURN_BOOL(vs_rabitq_cmp_internal(a, b) <= 0); |
| 251 | } | ||
| 252 | |||
| 253 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_eq); | |
| 254 | |||
| 255 | Datum | ||
| 256 | 2 | vs_rabitq_eq(PG_FUNCTION_ARGS) | |
| 257 | { | ||
| 258 | 2 | RaBitQVector *a = PG_GETARG_RABITQ_P(0); | |
| 259 | 2 | RaBitQVector *b = PG_GETARG_RABITQ_P(1); | |
| 260 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 2 times.
|
2 | PG_RETURN_BOOL(vs_rabitq_cmp_internal(a, b) == 0); |
| 261 | } | ||
| 262 | |||
| 263 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_ne); | |
| 264 | |||
| 265 | Datum | ||
| 266 | 1 | vs_rabitq_ne(PG_FUNCTION_ARGS) | |
| 267 | { | ||
| 268 | 1 | RaBitQVector *a = PG_GETARG_RABITQ_P(0); | |
| 269 | 1 | RaBitQVector *b = PG_GETARG_RABITQ_P(1); | |
| 270 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 1 times.
|
1 | PG_RETURN_BOOL(vs_rabitq_cmp_internal(a, b) != 0); |
| 271 | } | ||
| 272 | |||
| 273 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_ge); | |
| 274 | |||
| 275 | Datum | ||
| 276 | 1 | vs_rabitq_ge(PG_FUNCTION_ARGS) | |
| 277 | { | ||
| 278 | 1 | RaBitQVector *a = PG_GETARG_RABITQ_P(0); | |
| 279 | 1 | RaBitQVector *b = PG_GETARG_RABITQ_P(1); | |
| 280 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 1 times.
|
1 | PG_RETURN_BOOL(vs_rabitq_cmp_internal(a, b) >= 0); |
| 281 | } | ||
| 282 | |||
| 283 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_gt); | |
| 284 | |||
| 285 | Datum | ||
| 286 | 1 | vs_rabitq_gt(PG_FUNCTION_ARGS) | |
| 287 | { | ||
| 288 | 1 | RaBitQVector *a = PG_GETARG_RABITQ_P(0); | |
| 289 | 1 | RaBitQVector *b = PG_GETARG_RABITQ_P(1); | |
| 290 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 1 times.
|
1 | PG_RETURN_BOOL(vs_rabitq_cmp_internal(a, b) > 0); |
| 291 | } | ||
| 292 | |||
| 293 | /* ---------------------------------------------------------------- | ||
| 294 | * Cast function (typmod enforcement) | ||
| 295 | * ---------------------------------------------------------------- */ | ||
| 296 | |||
| 297 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq); | |
| 298 | |||
| 299 | Datum | ||
| 300 | 6 | vs_rabitq(PG_FUNCTION_ARGS) | |
| 301 | { | ||
| 302 | 6 | RaBitQVector *v = PG_GETARG_RABITQ_P(0); | |
| 303 | 6 | int32 typmod = PG_GETARG_INT32(1); | |
| 304 | |||
| 305 | 6 | vs_pg_check_expected_dim(v->dim, typmod); | |
| 306 | |||
| 307 | 5 | PG_RETURN_RABITQ_P(v); | |
| 308 | } | ||
| 309 | |||
| 310 | /* ---------------------------------------------------------------- | ||
| 311 | * Encode function | ||
| 312 | * ---------------------------------------------------------------- */ | ||
| 313 | |||
| 314 | 9 | PG_FUNCTION_INFO_V1(vs_rabitq_encode_pg); | |
| 315 | |||
| 316 | Datum | ||
| 317 | 5 | vs_rabitq_encode_pg(PG_FUNCTION_ARGS) | |
| 318 | { | ||
| 319 | 5 | Vec32 *input = PG_GETARG_VEC32_P(0); | |
| 320 | 5 | Vec32 *centroid = PG_GETARG_VEC32_P(1); | |
| 321 | 5 | RaBitQParamsPG *params = PG_GETARG_RABITQ_PARAMS_P(2); | |
| 322 | |||
| 323 | 5 | vs_pg_check_dims_match(input->dim, centroid->dim); | |
| 324 | 4 | vs_pg_check_dims_match(input->dim, params->dim); | |
| 325 | |||
| 326 | 4 | int dim = input->dim; | |
| 327 | |||
| 328 | /* Build RaBitQParams from the PG varlena's pre-computed matrix */ | ||
| 329 | 4 | RaBitQParams *rparams = | |
| 330 | 4 | vs_rabitq_create_from_matrix(dim, params->seed, params->P); | |
| 331 | |||
| 332 | 4 | RaBitQVector *result = vs_pg_rabitq_alloc(dim); | |
| 333 | |||
| 334 | 4 | Vec32Ref input_ref = Vec32ToRef(input); | |
| 335 | 4 | Vec32Ref centroid_ref = Vec32ToRef(centroid); | |
| 336 | 4 | RaBitQData *data = VS_RABITQ_DATA(result); | |
| 337 | |||
| 338 | 4 | int ret = vs_rabitq_encode_into(rparams, input_ref, centroid_ref, data); | |
| 339 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 4 times.
|
4 | if (ret != 0) |
| 340 | ✗ | ereport(ERROR, | |
| 341 | (errcode(ERRCODE_INTERNAL_ERROR), | ||
| 342 | errmsg("rabitq encoding failed"))); | ||
| 343 | |||
| 344 | 4 | PG_RETURN_RABITQ_P(result); | |
| 345 | } | ||
| 346 |