more GRI corner-case tests

Author: mlin

bindings/python/genomicsqlite.py

@@ -12,10 +12,11 @@
 # One-time global initialization -- load the extension shared-library
 _DLL = find_library("genomicsqlite")
 assert _DLL, "couldn't locate genomicsqlite shared-library file"
-# open a dummy connection to :memory: just for setting up the extension
+# open a dummy connection to :memory: just for setting up the extension.
 _MEMCONN = sqlite3.connect(":memory:")
 _MEMCONN.enable_load_extension(True)
 _MEMCONN.load_extension(_DLL)
+# now that it's been loaded, the extension will automatically enable itself on any new connections.
 
 
 def _execute1(conn, sql, params=None):

include/genomicsqlite.h

@@ -99,7 +99,7 @@ int GenomicSQLiteOpen(const std::string &dbfile, sqlite3 **ppDb, std::string &er
  * (include SQLiteCpp/SQLiteCpp.h first)
  */
 #include <memory>
-std::unique_ptr<SQLite::Database> GenomicSQLiteOpen(const std::string &dbfile, int flags,
+std::unique_ptr<SQLite::Database> GenomicSQLiteOpen(const std::string &dbfile, int flags = 0,
                                                     const std::string &config_json = "{}");
 #endif
 
@@ -109,7 +109,7 @@ std::string GenomicSQLiteVacuumIntoSQL(const std::string &dbfile,
 std::string CreateGenomicRangeIndexSQL(const std::string &table, const std::string &rid,
                                        const std::string &beg, const std::string &end,
                                        int max_depth = -1);
-std::string GenomicRangeRowidsSQL(const std::string &indexed_table, sqlite3 *dbconn = nullptr,
+std::string GenomicRangeRowidsSQL(const std::string &indexed_table, sqlite3 *dbconn,
                                   const std::string &qrid = "?1", const std::string &qbeg = "?2",
                                   const std::string &qend = "?3");
 

test/test_gri.py

@@ -1,3 +1,4 @@
+import math
 import os
 import sqlite3
 import random
@@ -18,6 +19,17 @@
     1 + 16 + 256 + 4096 + 65536 + 1048576 + 16777216,
     1 + 16 + 256 + 4096 + 65336 + 1048576 + 16777216 + 268435456,
 ]
+POS_OFFSETS = [
+    0,
+    134217728,
+    8388608,
+    524288,
+    32768,
+    2048,
+    128,
+    8,
+    0,
+]
 
 
 def test_genomic_range_bin():
@@ -191,6 +203,39 @@ def fanout(query):
     assert fanout(query) == 9
 
 
+def test_boundaries():
+    # test abutting intervals near bin boundaries
+
+    con = sqlite3.connect(":memory:")
+    con.executescript("CREATE TABLE features(rid INTEGER, beg INTEGER, end INTEGER)")
+
+    insert = "INSERT INTO features(rid,beg,end) values(?,?,?)"
+    for depth in range(2, 9):
+        boundary = 3 * (16 ** (9 - depth)) + POS_OFFSETS[depth]
+        for ofs in range(-2, 3):
+            featlen = math.ceil(16 ** (9 - depth) / 2)
+            con.execute(insert, (42, boundary + ofs - featlen, boundary + ofs))
+            con.execute(insert, (42, boundary + ofs, boundary + ofs + featlen))
+
+    con.executescript(
+        genomicsqlite.create_genomic_range_index_sql(con, "features", "rid", "beg", "end")
+    )
+
+    query = genomicsqlite.genomic_range_rowids_sql(con, "features")[1:-1]
+    control = "SELECT _rowid_ FROM features NOT INDEXED WHERE rid = ? AND NOT (? > end OR ? < beg) ORDER BY _rowid_"
+    total_results = 0
+    for depth in range(2, 9):
+        boundary = 3 * (16 ** (9 - depth)) + POS_OFFSETS[depth]
+        for qlen in range(3):
+            for ofs in range(-3, 4):
+                tup = (42, boundary + ofs, boundary + ofs + qlen)
+                query_results = list(con.execute(query, tup))
+                control_results = list(con.execute(control, tup))
+                assert query_results == control_results
+                total_results += len(query_results)
+    assert total_results == 938
+
+
 def test_refseq():
     con = sqlite3.connect(":memory:")