stList *getComponents(stList *edges) {
/*
* Gets a list of connected components, each connected component
* being represented as a list of the edges, such that each edge is in exactly one
* connected component. Allows for multi-graphs (multiple edges connecting two nodes).
*/
stHash *nodesToEdges = getNodesToEdgesHash(edges);
/*
* Traverse the edges greedily
*/
stList *components =
stList_construct3(0, (void(*)(void *)) stList_destruct);
stList *nodes = stHash_getKeys(nodesToEdges);
while (stList_length(nodes) > 0) {
stIntTuple *node = stList_pop(nodes);
stList *edges = stHash_search(nodesToEdges, node);
if (edges != NULL) { //We have a component to build
stSortedSet *component = stSortedSet_construct();
stHash_remove(nodesToEdges, node);
for (int64_t i = 0; i < stList_length(edges); i++) {
stIntTuple *edge = stList_get(edges, i);
getComponentsP(nodesToEdges, stIntTuple_get(edge, 0),
component);
getComponentsP(nodesToEdges, stIntTuple_get(edge, 1),
component);
}
stList_append(components, stSortedSet_getList(component));
//Cleanup
stSortedSet_destruct(component);
stList_destruct(edges);
}
stIntTuple_destruct(node);
}
assert(stHash_size(nodesToEdges) == 0);
stHash_destruct(nodesToEdges);
stList_destruct(nodes);
return components;
}
int64_t stSet_size(stSet *set) {
return stHash_size(set->hash);
}
static void test_newBlockHashFromBlock_0(CuTest *testCase) {
stList *orderList = stList_construct3(0, free);
stHash *observedHash = createBlockHashFromString("a score=0 test=0\n"
"s reference.chr0 0 13 + 158545518 gcagctgaaaaca\n"
"s name.chr1 0 10 + 100 ATGT---ATGCCG\n"
"s name2.chr1 0 10 + 100 ATGT---ATGCCG\n"
"s name3.chr1 5 13 - 100 ATGTgggATGCCG\n",
orderList);
CuAssertTrue(testCase, stHash_size(observedHash) == 4);
row_t *key = NULL;
key = stHash_search(observedHash, "reference");
CuAssertTrue(testCase, key != NULL);
CuAssertTrue(testCase, strcmp(key->name, "reference.chr0") == 0);
CuAssertTrue(testCase, strcmp(key->prevName, "reference.chr0") == 0);
CuAssertTrue(testCase, key->multipleNames == false);
CuAssertTrue(testCase, key->start == 0);
CuAssertTrue(testCase, key->length == 13);
CuAssertTrue(testCase, key->sourceLength == 158545518);
CuAssertTrue(testCase, key->prevRightPos == 12);
CuAssertTrue(testCase, key->strand == '+');
CuAssertTrue(testCase, key->prevStrand == '+');
CuAssertTrue(testCase, strcmp(key->sequence, "gcagctgaaaaca") == 0);
// row 2
key = stHash_search(observedHash, "name");
CuAssertTrue(testCase, key != NULL);
CuAssertTrue(testCase, strcmp(key->name, "name.chr1") == 0);
CuAssertTrue(testCase, strcmp(key->prevName, "name.chr1") == 0);
CuAssertTrue(testCase, key->multipleNames == false);
CuAssertTrue(testCase, key->start == 0);
CuAssertTrue(testCase, key->length == 10);
CuAssertTrue(testCase, key->sourceLength == 100);
CuAssertTrue(testCase, key->prevRightPos == 9);
CuAssertTrue(testCase, key->strand == '+');
CuAssertTrue(testCase, key->prevStrand == '+');
CuAssertTrue(testCase, strcmp(key->sequence, "ATGT---ATGCCG") == 0);
// row 3
key = stHash_search(observedHash, "name2");
CuAssertTrue(testCase, key != NULL);
CuAssertTrue(testCase, strcmp(key->name, "name2.chr1") == 0);
CuAssertTrue(testCase, strcmp(key->prevName, "name2.chr1") == 0);
CuAssertTrue(testCase, key->multipleNames == false);
CuAssertTrue(testCase, key->start == 0);
CuAssertTrue(testCase, key->length == 10);
CuAssertTrue(testCase, key->sourceLength == 100);
CuAssertTrue(testCase, key->prevRightPos == 9);
CuAssertTrue(testCase, key->strand == '+');
CuAssertTrue(testCase, key->prevStrand == '+');
CuAssertTrue(testCase, strcmp(key->sequence, "ATGT---ATGCCG") == 0);
// row 4
key = stHash_search(observedHash, "name3");
CuAssertTrue(testCase, key != NULL);
CuAssertTrue(testCase, strcmp(key->name, "name3.chr1") == 0);
CuAssertTrue(testCase, strcmp(key->prevName, "name3.chr1") == 0);
CuAssertTrue(testCase, key->multipleNames == false);
CuAssertTrue(testCase, key->start == 5);
CuAssertTrue(testCase, key->length == 13);
CuAssertTrue(testCase, key->sourceLength == 100);
CuAssertTrue(testCase, key->prevRightPos == 17);
CuAssertTrue(testCase, key->strand == '-');
CuAssertTrue(testCase, key->prevStrand == '-');
CuAssertTrue(testCase, strcmp(key->sequence, "ATGTgggATGCCG") == 0);
stList_destruct(orderList);
stHash_destruct(observedHash);
}