commit 8e8772fb11b89e0cda3c553e053eeb3c8a41a5b9
parent dbb2bb78577af39dd3df7f903f1b3dbf7154bfa2
Author: walther chen <walther.chen@gmail.com>
Date: Wed, 19 Mar 2025 16:41:40 +0700
update for new pool in 0.7.0, remove some unneeded copies, fix some test mem leaks
Diffstat:
3 files changed, 31 insertions(+), 26 deletions(-)
diff --git a/ba1b.c3 b/ba1b.c3
@@ -23,7 +23,7 @@ fn int main(String[] args) {
}
fn String[] frequent_words(String text, int k, Allocator alloc= allocator::heap()) {
- @pool(alloc) {
+ @pool() {
FrequencyTable freq_map = util::frequency_table(text, k, allocator::temp());
int[] counts = freq_map.tvalues();
diff --git a/ba1d.c3 b/ba1d.c3
@@ -23,7 +23,7 @@ fn int main(String[] args) {
fn int[] pattern_matching(String pattern, String genome, Allocator alloc = allocator::heap()) {
if (pattern.len == 0 || genome.len == 0) return {};
List{int} res;
- @pool(alloc) {
+ @pool() {
res.tinit();
for (int i = 0; i <= genome.len - pattern.len; i += 1) {
if (genome[i:pattern.len] == pattern) {
diff --git a/util-frequency.c3 b/util-frequency.c3
@@ -23,19 +23,21 @@ fn String reverse_complement(String pattern, Allocator alloc = allocator::heap()
}
fn void test_reverse_complement() @test {
- assert(reverse_complement("AAAACCCGGT") == "ACCGGGTTTT");
+ assert(reverse_complement("AAAACCCGGT", tmem()) == "ACCGGGTTTT");
}
alias FrequencyTable = HashMap{String, int};
// counts of kmers in a text.
+// Lifetime of result is same as that of `text` param.
fn FrequencyTable frequency_table(String text, int k, Allocator alloc = allocator::heap()) {
FrequencyTable kmer_counts;
kmer_counts.init(allocator: alloc);
for (int i = 0; i <= text.len - k; i += 1) {
// TODO get_or_update? Annoying to get twice
- // Modified std so that hashmap doesn't copy keys
- String key = text[i:k].copy(alloc);
+ // Modified std so that hashmap doesn't copy keys, but these
+ // keys lifetime is the same as `text`.
+ String key = text[i:k];
kmer_counts.@get_or_set(key, 0);
if (try count = kmer_counts.get_ref(key)) {
*count += 1;
@@ -44,6 +46,7 @@ fn FrequencyTable frequency_table(String text, int k, Allocator alloc = allocato
return kmer_counts;
}
+// Lifetime of result is that of the `genome` param.
fn String[] clump_finding(
String genome,
int k,
@@ -53,19 +56,19 @@ fn String[] clump_finding(
{
if (k == 0 || genome.len == 0) return {};
HashMap{String, char} clumps; // a set
- @pool(alloc) {
+ @pool() {
clumps.tinit();
for (int i = 0; i <= genome.len - region_len; i += 1) {
- // TODO Could use another pool here, but there appears to be a bug with nested pools
- FrequencyTable freq_map = frequency_table(genome[i:region_len], k, tmem());
- freq_map.@each(; String kmer, int count) {
- if (count >= clump_threshold) {
- // TODO can probably just use a growable array on small inputs
- // std hashmap modified to not copy keys, so have to copy out kmer from
- // frequency table
- String key = kmer.copy(alloc);
- clumps.set(key, 0);
- }
+ @pool() {
+ FrequencyTable freq_map = frequency_table(genome[i:region_len], k, tmem());
+ freq_map.@each(; String kmer, int count) {
+ if (count >= clump_threshold) {
+ // TODO can probably just use a growable array on small inputs
+ // std hashmap modified to not copy keys, lifetime of kmer is that of
+ // the `genome` param
+ clumps.set(kmer, 0);
+ }
+ };
};
}
return clumps.keys(alloc);
@@ -101,7 +104,7 @@ fn int[] approximate_pattern_matching(
{
if (pattern.len == 0 || genome.len == 0) return {};
List{int} res;
- @pool(alloc) {
+ @pool() {
res.tinit();
for (int i = 0; i <= genome.len - pattern.len; i += 1) {
// TODO more efficient to break early when calculating hamming distance
@@ -155,7 +158,7 @@ fn String[] _frequent_words_with_mismatches(
bool with_rc,
Allocator alloc= allocator::heap())
{
- @pool(alloc) {
+ @pool() {
if (with_rc) {
text = text.tconcat(reverse_complement(text, allocator::temp()));
}
@@ -165,7 +168,7 @@ fn String[] _frequent_words_with_mismatches(
String pattern = text[i:k];
String[] neighborhood = neighbors(pattern, d, allocator::temp());
foreach (neighbor : neighborhood) {
- freq_map.@get_or_set(neighbor, 0); // don't need to clone, keys copied into table
+ freq_map.@get_or_set(neighbor, 0);
if (try int* count = freq_map.get_ref(neighbor)) {
*count += 1;
}
@@ -230,8 +233,10 @@ fn void test_frequent_words_with_mismatches() @test {
},
};
foreach (t : tests) {
- String[] matches = frequent_words_with_mismatches(t.text, t.k, t.d);
- test::expect_equal_slices_sorted(t.expected, matches);
+ @pool() {
+ String[] matches = frequent_words_with_mismatches(t.text, t.k, t.d, tmem());
+ test::expect_equal_slices_sorted(t.expected, matches);
+ };
}
}
fn void test_frequent_words_with_mismatches_and_rc() @test {
@@ -244,7 +249,7 @@ fn void test_frequent_words_with_mismatches_and_rc() @test {
},
};
foreach (t : tests) {
- String[] matches = frequent_words_with_mismatches_and_rc(t.text, t.k, t.d);
+ String[] matches = frequent_words_with_mismatches_and_rc(t.text, t.k, t.d, tmem());
test::expect_equal_slices_sorted(t.expected, matches);
}
}
@@ -257,11 +262,11 @@ fn String[] neighbors(String pattern, int d, Allocator alloc = allocator::heap()
{
if (d == 0) return {pattern};
if (pattern.len == 0) return {};
- @pool(alloc) {
+ @pool() {
DynamicArenaAllocator neighborhood_arena;
DynamicArenaAllocator suffix_neighborhood_arena;
- neighborhood_arena.init(1024, allocator::temp());
- suffix_neighborhood_arena.init(1024, allocator::temp());
+ neighborhood_arena.init(tmem(), 1024);
+ suffix_neighborhood_arena.init(tmem(), 1024);
List{String} neighborhood;
List{String} suffix_neighborhood;
neighborhood.init_with_array(&neighborhood_arena, {"A", "C", "G", "T"});
@@ -352,7 +357,7 @@ fn void test_neighbors() @test {
},
};
foreach (t : tests) {
- String[] matches = neighbors(t.text, t.d);
+ String[] matches = neighbors(t.text, t.d, tmem());
test::expect_equal_slices_sorted(t.expected, matches);
}
}
