Add GDB debugging to integration test + fix

ratschlab · Oct 16, 2024 · 7535d81 · 7535d81
1 parent 90d6840
commit 7535d81
Show file tree

Hide file tree

Showing 11 changed files with 27 additions and 39 deletions.
diff --git a/metagraph/integration_tests/base.py b/metagraph/integration_tests/base.py
@@ -6,7 +6,12 @@
 
 script_path = os.path.dirname(os.path.realpath(__file__))
 
-METAGRAPH = f'{os.getcwd()}/metagraph'
+USE_GDB = False
+METAGRAPH_EXE = f'{os.getcwd()}/metagraph'
+GDB_PREFIX = 'gdb -ex run -ex quit --args ' if USE_GDB else ''
+METAGRAPH = GDB_PREFIX + METAGRAPH_EXE
+DNA_MODE = os.readlink(METAGRAPH_EXE).endswith("_DNA")
+PROTEIN_MODE = os.readlink(METAGRAPH_EXE).endswith("_Protein")
 
 TEST_DATA_DIR = os.path.join(script_path, '..', 'tests', 'data')
 

diff --git a/metagraph/integration_tests/test_align.py b/metagraph/integration_tests/test_align.py
@@ -6,14 +6,11 @@
 import glob
 import os
 
-from base import TestingBase, METAGRAPH, TEST_DATA_DIR, NUM_THREADS
+from base import PROTEIN_MODE, DNA_MODE, TestingBase, METAGRAPH, TEST_DATA_DIR, NUM_THREADS
 
 
 """Test graph construction and alignment"""
 
-DNA_MODE = os.readlink(METAGRAPH).endswith("_DNA")
-PROTEIN_MODE = os.readlink(METAGRAPH).endswith("_Protein")
-
 graph_file_extension = {'succinct': '.dbg',
                         'bitmap': '.bitmapdbg',
                         'hash': '.orhashdbg',

diff --git a/metagraph/integration_tests/test_annotate.py b/metagraph/integration_tests/test_annotate.py
@@ -6,13 +6,11 @@
 import filecmp
 import glob
 import os
-from base import TestingBase, METAGRAPH, TEST_DATA_DIR, NUM_THREADS
+from base import PROTEIN_MODE, TestingBase, METAGRAPH, TEST_DATA_DIR, NUM_THREADS
 
 
 """Test graph annotation"""
 
-PROTEIN_MODE = os.readlink(METAGRAPH).endswith("_Protein")
-
 graph_file_extension = {'succinct': '.dbg',
                         'bitmap': '.bitmapdbg',
                         'hash': '.orhashdbg',

diff --git a/metagraph/integration_tests/test_api.py b/metagraph/integration_tests/test_api.py
@@ -13,10 +13,7 @@
 from concurrent.futures import Future
 from parameterized import parameterized, parameterized_class
 
-from base import TestingBase, METAGRAPH, TEST_DATA_DIR
-
-PROTEIN_MODE = os.readlink(METAGRAPH).endswith("_Protein")
-
+from base import PROTEIN_MODE, TestingBase, METAGRAPH, TEST_DATA_DIR
 
 class TestAPIBase(TestingBase):
     @classmethod

diff --git a/metagraph/integration_tests/test_assemble.py b/metagraph/integration_tests/test_assemble.py
@@ -6,14 +6,12 @@
 import gzip
 import itertools
 from helpers import get_test_class_name
-from base import TestingBase, graph_file_extension, METAGRAPH, TEST_DATA_DIR, NUM_THREADS
+from base import PROTEIN_MODE, TestingBase, graph_file_extension, METAGRAPH, TEST_DATA_DIR, NUM_THREADS
 from test_query import anno_file_extension, GRAPH_TYPES, ANNO_TYPES, product
 
 
 """Test graph assemble"""
 
-PROTEIN_MODE = os.readlink(METAGRAPH).endswith("_Protein")
-
 gfa_tests = {
     'compacted': {
         'fasta_path': TEST_DATA_DIR + '/transcripts_100.fa',

diff --git a/metagraph/integration_tests/test_build.py b/metagraph/integration_tests/test_build.py
@@ -5,14 +5,11 @@
 from tempfile import TemporaryDirectory
 import glob
 import os
-from base import TestingBase, METAGRAPH, TEST_DATA_DIR
+from base import PROTEIN_MODE, DNA_MODE, TestingBase, METAGRAPH, TEST_DATA_DIR
 
 
 """Test graph construction"""
 
-PROTEIN_MODE = os.readlink(METAGRAPH).endswith("_Protein")
-DNA_MODE = os.readlink(METAGRAPH).endswith("_DNA")
-
 graph_file_extension = {'succinct': '.dbg',
                         'bitmap': '.bitmapdbg',
                         'hash': '.orhashdbg',

diff --git a/metagraph/integration_tests/test_build_weighted.py b/metagraph/integration_tests/test_build_weighted.py
@@ -7,13 +7,11 @@
 import glob
 import os
 import gzip
-from base import TestingBase, METAGRAPH, TEST_DATA_DIR
+from base import PROTEIN_MODE, TestingBase, METAGRAPH, TEST_DATA_DIR
 
 
 """Test graph construction"""
 
-PROTEIN_MODE = os.readlink(METAGRAPH).endswith("_Protein")
-
 graph_file_extension = {'succinct': '.dbg',
                         'bitmap': '.bitmapdbg',
                         'hash': '.orhashdbg',

diff --git a/metagraph/integration_tests/test_clean.py b/metagraph/integration_tests/test_clean.py
@@ -7,13 +7,11 @@
 import glob
 import os
 import gzip
-from base import TestingBase, METAGRAPH, TEST_DATA_DIR, NUM_THREADS
+from base import PROTEIN_MODE, TestingBase, METAGRAPH, TEST_DATA_DIR, NUM_THREADS
 
 
 """Test graph construction"""
 
-PROTEIN_MODE = os.readlink(METAGRAPH).endswith("_Protein")
-
 graph_file_extension = {'succinct': '.dbg',
                         'bitmap': '.bitmapdbg',
                         'hash': '.orhashdbg',

diff --git a/metagraph/integration_tests/test_query.py b/metagraph/integration_tests/test_query.py
@@ -8,15 +8,12 @@
 import os
 import numpy as np
 from helpers import get_test_class_name
-from base import TestingBase, METAGRAPH, TEST_DATA_DIR, graph_file_extension
+from base import PROTEIN_MODE, DNA_MODE, TestingBase, METAGRAPH, TEST_DATA_DIR, graph_file_extension
 import hashlib
 
 
 """Test graph construction"""
 
-DNA_MODE = os.readlink(METAGRAPH).endswith("_DNA")
-PROTEIN_MODE = os.readlink(METAGRAPH).endswith("_Protein")
-
 anno_file_extension = {'column': '.column.annodbg',
                        'column_coord': '.column_coord.annodbg',
                        'brwt_coord': '.brwt_coord.annodbg',

diff --git a/metagraph/integration_tests/test_transform_anno.py b/metagraph/integration_tests/test_transform_anno.py
@@ -12,8 +12,6 @@
 
 """Test operations on annotation columns"""
 
-DNA_MODE = os.readlink(METAGRAPH).endswith("_DNA")
-PROTEIN_MODE = os.readlink(METAGRAPH).endswith("_Protein")
 TEST_DATA_DIR = os.path.dirname(os.path.realpath(__file__)) + '/../tests/data'
 
 NUM_THREADS = 4

diff --git a/metagraph/src/annotation/row_diff_builder.cpp b/metagraph/src/annotation/row_diff_builder.cpp
@@ -418,8 +418,7 @@ void build_pred_succ(const graph::DeBruijnGraph &graph,
                   outfbase);
 
     std::optional<sdsl::bit_vector> dummy;
-    auto* succinct = dynamic_cast<graph::DBGSuccinct const*>(&graph);
-    if (succinct) {
+    if (auto* succinct = dynamic_cast<graph::DBGSuccinct const*>(&graph)) {
         dummy = succinct->get_boss().mark_all_dummy_edges(num_threads);
     }
 
@@ -449,16 +448,20 @@ void build_pred_succ(const graph::DeBruijnGraph &graph,
         std::vector<bool> pred_boundary_buf;
 
         for (node_index i = start; i < std::min(start + BS, graph.max_index() + 1); ++i) {
-            if (graph.in_graph(i)) {
+            if (graph.in_graph(i) && !(dummy && (*dummy)[i])) { // Legacy check for DBGSuccinct
                 if(!graph.has_no_outgoing(i)) {
                     auto j = row_diff_successor(graph, i, rd_succ);
-                    succ_buf.push_back(to_row(j));
-                    succ_boundary_buf.push_back(0);
+                    if(!dummy || !(*dummy)[j]) { // Legacy check for DBGSuccinct
+                        succ_buf.push_back(to_row(j));
+                        succ_boundary_buf.push_back(0);
+                    }
                 }
                 if(rd_succ[i]) {
                     graph.adjacent_incoming_nodes(i, [&](auto pred) {
-                        pred_buf.push_back(to_row(pred));
-                        pred_boundary_buf.push_back(0);
+                        if(!dummy || !(*dummy)[pred]) { // Legacy check for DBGSuccinct
+                            pred_buf.push_back(to_row(pred));
+                            pred_boundary_buf.push_back(0);
+                        }
                     });
                 }
             }
@@ -509,8 +512,10 @@ void assign_anchors(const graph::DeBruijnGraph &graph,
         sum_and_call_counts(count_vectors_dir, row_reduction_extension, "row reduction",
             [&](int32_t count) {
                 // check if the reduction is negative
-                if (count < 0)
+                if (count < 0) {
+                    logger->error("anchoring {}", to_node(i));
                     anchors_bv[to_node(i)] = true;
+                }
                 i++;
             }
         );