postgres/typesense
1
0
Fork 0
mirror of https://github.com/typesense/typesense.git synced 2025-05-24 07:40:35 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Refactor NumericTrie::search_range.

Browse Source
This commit is contained in:
Harpreet Sangar 2023-06-01 12:52:09 +05:30
parent f2e27500c4
commit 5e36ea85f0
2 changed files with 73 additions and 50 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

67
src/numeric_range_trie.cpp
View File

@ -27,51 +27,70 @@ void NumericTrie::search_range(const int32_t& low, const bool& low_inclusive,
if (low < 0 && high >= 0) {
// Have to combine the results of >low from negative_trie and <high from positive_trie
uint32_t* negative_ids = nullptr;
uint32_t negative_ids_length = 0;
if (negative_trie != nullptr && !(low == -1 && !low_inclusive)) { // No need to search for (-1, ...
uint32_t* negative_ids = nullptr;
uint32_t negative_ids_length = 0;
auto abs_low = std::abs(low);
// Since we store absolute values, search_lesser would yield result for >low from negative_trie.
negative_trie->search_less_than(low_inclusive ? abs_low : abs_low - 1, negative_ids, negative_ids_length);
uint32_t* out = nullptr;
ids_length = ArrayUtils::or_scalar(negative_ids, negative_ids_length, ids, ids_length, &out);
delete [] negative_ids;
delete [] ids;
ids = out;
}
uint32_t* positive_ids = nullptr;
uint32_t positive_ids_length = 0;
if (positive_trie != nullptr && !(high == 0 && !high_inclusive)) { // No need to search for ..., 0)
uint32_t* positive_ids = nullptr;
uint32_t positive_ids_length = 0;
positive_trie->search_less_than(high_inclusive ? high : high - 1, positive_ids, positive_ids_length);
uint32_t* out = nullptr;
ids_length = ArrayUtils::or_scalar(positive_ids, positive_ids_length, ids, ids_length, &out);
delete [] positive_ids;
delete [] ids;
ids = out;
}
ids_length = ArrayUtils::or_scalar(negative_ids, negative_ids_length, positive_ids, positive_ids_length, &ids);
delete [] negative_ids;
delete [] positive_ids;
return;
} else if (low >= 0) {
// Search only in positive_trie
if (positive_trie == nullptr) {
return;
}
uint32_t* positive_ids = nullptr;
uint32_t positive_ids_length = 0;
if (positive_trie != nullptr) {
positive_trie->search_range(low_inclusive ? low : low + 1, high_inclusive ? high : high - 1,
positive_ids, positive_ids_length);
}
positive_trie->search_range(low_inclusive ? low : low + 1, high_inclusive ? high : high - 1,
positive_ids, positive_ids_length);
ids = positive_ids;
ids_length = positive_ids_length;
uint32_t* out = nullptr;
ids_length = ArrayUtils::or_scalar(positive_ids, positive_ids_length, ids, ids_length, &out);
delete [] positive_ids;
delete [] ids;
ids = out;
} else {
// Search only in negative_trie
if (negative_trie == nullptr) {
return;
}
uint32_t* negative_ids = nullptr;
uint32_t negative_ids_length = 0;
if (negative_trie != nullptr) {
// Since we store absolute values, switching low and high would produce the correct result.
auto abs_high = std::abs(high), abs_low = std::abs(low);
negative_trie->search_range(high_inclusive ? abs_high : abs_high + 1, low_inclusive ? abs_low : abs_low - 1,
negative_ids, negative_ids_length);
}
// Since we store absolute values, switching low and high would produce the correct result.
auto abs_high = std::abs(high), abs_low = std::abs(low);
negative_trie->search_range(high_inclusive ? abs_high : abs_high + 1, low_inclusive ? abs_low : abs_low - 1,
negative_ids, negative_ids_length);
ids = negative_ids;
ids_length = negative_ids_length;
uint32_t* out = nullptr;
ids_length = ArrayUtils::or_scalar(negative_ids, negative_ids_length, ids, ids_length, &out);
delete [] negative_ids;
delete [] ids;
ids = out;
}
}

56
test/numeric_range_trie_test.cpp
View File

@ -37,52 +37,51 @@ TEST_F(NumericRangeTrieTest, SearchRange) {
uint32_t ids_length = 0;
trie->search_range(32768, true, -32768, true, ids, ids_length);
std::unique_ptr<uint32_t[]> ids_guard(ids);
ASSERT_EQ(0, ids_length);
reset(ids, ids_length);
trie->search_range(-32768, true, 32768, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(pairs.size(), ids_length);
for (uint32_t i = 0; i < pairs.size(); i++) {
ASSERT_EQ(pairs[i].second, ids[i]);
}
reset(ids, ids_length);
trie->search_range(-32768, true, 32768, false, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(pairs.size() - 1, ids_length);
for (uint32_t i = 0; i < pairs.size() - 1; i++) {
ASSERT_EQ(pairs[i].second, ids[i]);
}
reset(ids, ids_length);
trie->search_range(-32768, true, 134217728, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(pairs.size(), ids_length);
for (uint32_t i = 0; i < pairs.size(); i++) {
ASSERT_EQ(pairs[i].second, ids[i]);
}
reset(ids, ids_length);
trie->search_range(-32768, true, 0, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(4, ids_length);
for (uint32_t i = 0; i < 4; i++) {
ASSERT_EQ(pairs[i].second, ids[i]);
}
reset(ids, ids_length);
trie->search_range(-32768, true, 0, false, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(4, ids_length);
for (uint32_t i = 0; i < 4; i++) {
ASSERT_EQ(pairs[i].second, ids[i]);
}
reset(ids, ids_length);
trie->search_range(-32768, false, 32768, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(pairs.size() - 1, ids_length);
for (uint32_t i = 0, j = 0; i < pairs.size(); i++) {
@ -90,93 +89,97 @@ TEST_F(NumericRangeTrieTest, SearchRange) {
ASSERT_EQ(pairs[i].second, ids[j++]);
}
reset(ids, ids_length);
trie->search_range(-134217728, true, 32768, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(pairs.size(), ids_length);
for (uint32_t i = 0; i < pairs.size(); i++) {
ASSERT_EQ(pairs[i].second, ids[i]);
}
reset(ids, ids_length);
trie->search_range(-134217728, true, 134217728, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(pairs.size(), ids_length);
for (uint32_t i = 0; i < pairs.size(); i++) {
ASSERT_EQ(pairs[i].second, ids[i]);
}
reset(ids, ids_length);
trie->search_range(-1, true, 32768, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(4, ids_length);
for (uint32_t i = 4, j = 0; i < pairs.size(); i++, j++) {
ASSERT_EQ(pairs[i].second, ids[j]);
}
reset(ids, ids_length);
trie->search_range(-1, false, 32768, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(4, ids_length);
for (uint32_t i = 4, j = 0; i < pairs.size(); i++, j++) {
ASSERT_EQ(pairs[i].second, ids[j]);
}
reset(ids, ids_length);
trie->search_range(-1, true, 0, true, ids, ids_length);
ASSERT_EQ(0, ids_length);
reset(ids, ids_length);
trie->search_range(-1, false, 0, false, ids, ids_length);
ASSERT_EQ(0, ids_length);
reset(ids, ids_length);
trie->search_range(8192, true, 32768, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(4, ids_length);
for (uint32_t i = 4, j = 0; i < ids_length; i++, j++) {
ASSERT_EQ(pairs[i].second, ids[j]);
}
reset(ids, ids_length);
trie->search_range(8192, true, 0x2000000, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(4, ids_length);
for (uint32_t i = 4, j = 0; i < ids_length; i++, j++) {
ASSERT_EQ(pairs[i].second, ids[j]);
}
reset(ids, ids_length);
trie->search_range(16384, true, 16384, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(1, ids_length);
ASSERT_EQ(56, ids[0]);
reset(ids, ids_length);
trie->search_range(16384, true, 16384, false, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(0, ids_length);
reset(ids, ids_length);
trie->search_range(16384, false, 16384, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(0, ids_length);
reset(ids, ids_length);
trie->search_range(16383, true, 16383, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(0, ids_length);
reset(ids, ids_length);
trie->search_range(8193, true, 16383, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(0, ids_length);
reset(ids, ids_length);
trie->search_range(-32768, true, -8192, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(4, ids_length);
for (uint32_t i = 0; i < ids_length; i++) {
ASSERT_EQ(pairs[i].second, ids[i]);
}
reset(ids, ids_length);
}
TEST_F(NumericRangeTrieTest, SearchGreaterThan) {
@ -439,7 +442,6 @@ TEST_F(NumericRangeTrieTest, MultivalueData) {
uint32_t ids_length = 0;
trie->search_less_than(0, false, ids, ids_length);
std::unique_ptr<uint32_t[]> ids_guard(ids);
std::vector<uint32_t> expected = {5, 8, 32, 35, 43};
@ -448,8 +450,8 @@ TEST_F(NumericRangeTrieTest, MultivalueData) {
ASSERT_EQ(expected[i], ids[i]);
}
reset(ids, ids_length);
trie->search_less_than(-16380, false, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(4, ids_length);
@ -458,8 +460,8 @@ TEST_F(NumericRangeTrieTest, MultivalueData) {
ASSERT_EQ(expected[i], ids[i]);
}
reset(ids, ids_length);
trie->search_less_than(16384, false, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(7, ids_length);
@ -468,8 +470,8 @@ TEST_F(NumericRangeTrieTest, MultivalueData) {
ASSERT_EQ(expected[i], ids[i]);
}
reset(ids, ids_length);
trie->search_greater_than(0, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(7, ids_length);
@ -478,8 +480,8 @@ TEST_F(NumericRangeTrieTest, MultivalueData) {
ASSERT_EQ(expected[i], ids[i]);
}
reset(ids, ids_length);
trie->search_greater_than(256, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(5, ids_length);
@ -488,8 +490,8 @@ TEST_F(NumericRangeTrieTest, MultivalueData) {
ASSERT_EQ(expected[i], ids[i]);
}
reset(ids, ids_length);
trie->search_greater_than(-32768, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(9, ids_length);
@ -498,8 +500,8 @@ TEST_F(NumericRangeTrieTest, MultivalueData) {
ASSERT_EQ(expected[i], ids[i]);
}
reset(ids, ids_length);
trie->search_range(-32768, true, 0, true, ids, ids_length);
ids_guard.reset(ids);
ASSERT_EQ(6, ids_length);
@ -507,6 +509,8 @@ TEST_F(NumericRangeTrieTest, MultivalueData) {
for (uint32_t i = 0; i < ids_length; i++) {
ASSERT_EQ(expected[i], ids[i]);
}
reset(ids, ids_length);
}
TEST_F(NumericRangeTrieTest, EmptyTrieOperations) {