Adding rdftab compatibility (#4)

Author: cmungall

src/rdf_sql_bulkloader/cli.py

@@ -34,8 +34,12 @@ def main(verbose: int, quiet: bool):
 @click.option(
     "--force/--no-force", default=False, show_default=True, help="Recreates db if already present"
 )
+@click.option(
+    "--rdftab-compatibility/--no-rdftab-compatibility", default=True, show_default=True,
+    help="Creates a statements table, compatible with rdftab"
+)
 @click.argument("files", nargs=-1)
-def load_sqlite(files, output, force: bool):
+def load_sqlite(files, output, force: bool, rdftab_compatibility: bool):
     """Run the rdf-sql-bulkloader's demo command."""
     output_path = Path(output)
     if output_path.exists():
@@ -44,6 +48,7 @@ def load_sqlite(files, output, force: bool):
         else:
             raise ValueError(f"Path exists {output_path}")
     loader = SqliteBulkloader(output)
+    loader.rdftab_compatibility = rdftab_compatibility
     if len(files) > 1:
         logging.warning("Blank nodes may be shared TODO FIX ME")
     for file in files:

src/rdf_sql_bulkloader/loaders/bulkloader.py

@@ -89,6 +89,7 @@ class BulkLoader(ABC):
     prefix_map: PREFIX_MAP = None
     converter: Converter = None
     index_statements = False
+    rdftab_compatibility = True
 
     def __post_init__(self):
         if self.prefix_map is None:

src/rdf_sql_bulkloader/loaders/sqlite3_bulkloader.py

@@ -1,3 +1,4 @@
+import logging
 import sqlite3
 from dataclasses import dataclass
 from typing import Any
@@ -6,6 +7,17 @@
 
 COLS = ["subject", "predicate", "object", "value", "datatype", "language"]
 
+RDFTAB_INSERT = """
+CREATE TABLE statements AS 
+SELECT
+ subject AS stanza,
+ subject,
+ predicate,
+ value,
+ datatype,
+ language
+FROM statement
+"""
 
 @dataclass
 class SqliteBulkloader(BulkLoader):
@@ -18,8 +30,10 @@ def bulkload(self, path: str):
         tuples = []
         for t in self.statements(path):
             tuples.append(t)
-        print(len(tuples))
+        logging.info(f"Loaded {len(tuples)} loaded")
         colstr = ",".join(COLS)
         qs = ",".join(["?" for _ in COLS])
         con.executemany(f"insert into statement({colstr}) values ({qs})", tuples)
+        if self.rdftab_compatibility:
+            con.execute(RDFTAB_INSERT)
         con.commit()

tests/test_sqlite3_bulkloader.py

@@ -36,15 +36,22 @@ class TestSqlit3BulkLoader(unittest.TestCase):
 
     def test_bulkload(self):
         loader = SqliteBulkloader(path=":memory:")
+        loader.rdftab_compatibility = True
         loader.bulkload(TEST_INPUT_OWL)
         con = loader.connection
         cur = con.cursor()
         cur.execute("select * from statement WHERE subject=:subject", {"subject": NUCLEUS})
         stmts = list(cur.fetchall())
-        for s in stmts:
-            print(s)
+        #for s in stmts:
+        #    print(s)
         for case in CASES:
             self.assertIn(case, stmts)
+        # rdftab
+        cur.execute("select * from statements WHERE subject=:subject", {"subject": NUCLEUS})
+        stmts = list(cur.fetchall())
+        for s in stmts:
+            print(s)
+        assert len(stmts) > 1
 
     def test_lang_tags(self):
         """