obitools/obitools3
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 66 Wiki Activity

C: AVL trees: fixed a bug where storing the difference between 2 crc64

Browse Source 
values in an int64 would mess trees up resulting in failed data
dereplication
This commit is contained in:
Celine Mercier
2020-07-07 16:47:00 +02:00
parent 9100e14899
commit a3ebe5f118
2 changed files with 16 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
python/obitools3/commands/uniq.pyx
View File

@ -307,8 +307,8 @@ cdef uniq_sequences(View_NUC_SEQS view, View_NUC_SEQS o_view, ProgressBar pb, di
for x in categories : for x in categories :
catl.append(i_seq[x]) catl.append(i_seq[x])
unique_id = tuple(catl) + (i_seq_col[i],) #unique_id = tuple(catl) + (i_seq_col[i],)
#unique_id = tuple(catl) + (i_seq_col.get_line_idx(i),) unique_id = tuple(catl) + (i_seq_col.get_line_idx(i),)
#unique_id = tuple(i_seq[x] for x in categories) + (seq_col.get_line_idx(i),) # The line that cython can't read properly #unique_id = tuple(i_seq[x] for x in categories) + (seq_col.get_line_idx(i),) # The line that cython can't read properly
if unique_id in uniques: if unique_id in uniques:

16
src/obiavl.c
View File

@ -2259,7 +2259,13 @@ index_t obi_avl_add(OBIDMS_avl_p avl, Obi_blob_p value)
parent = next; parent = next;
// Compare the crc of the value with the crc of the current node // Compare the crc of the value with the crc of the current node
comp = (current_node->crc64) - crc; //comp = (current_node->crc64) - crc;
if ((current_node->crc64) == crc)
comp = 0;
else if ((current_node->crc64) > crc)
comp = 1;
else
comp = -1;
if (comp == 0) if (comp == 0)
{ // check if really same value { // check if really same value
@ -2354,7 +2360,13 @@ index_t obi_avl_find(OBIDMS_avl_p avl, Obi_blob_p value)
current_node = (avl->tree)+next; current_node = (avl->tree)+next;
// Compare the crc of the value with the crc of the current node // Compare the crc of the value with the crc of the current node
comp = (current_node->crc64) - crc; //comp = (current_node->crc64) - crc;
if ((current_node->crc64) == crc)
comp = 0;
else if ((current_node->crc64) > crc)
comp = 1;
else
comp = -1;
if (comp == 0) if (comp == 0)
{ // Check if really same value { // Check if really same value