aboutsummaryrefslogtreecommitdiff
path: root/trygvis/eda/cli/add_to_db.py
diff options
context:
space:
mode:
Diffstat (limited to 'trygvis/eda/cli/add_to_db.py')
-rw-r--r--trygvis/eda/cli/add_to_db.py33
1 files changed, 22 insertions, 11 deletions
diff --git a/trygvis/eda/cli/add_to_db.py b/trygvis/eda/cli/add_to_db.py
index 84f8d74..cb035a4 100644
--- a/trygvis/eda/cli/add_to_db.py
+++ b/trygvis/eda/cli/add_to_db.py
@@ -3,16 +3,27 @@ import trygvis.eda.cli as cli
def run(files, args):
def load(g):
- s = 0
- for f in files:
- cli.info("Adding %s" % f)
- pre = len(g)
- g.load(f, format="turtle")
- post = len(g)
- diff = post - pre
- s += diff
- cli.info("Loaded %d tuples" % diff)
-
- cli.info("Done. Loaded %d tuples" % s)
+ agg = 0
+ file_count = len(files)
+ cli.info("Loading %d files" % file_count)
+ for i, f in enumerate(files):
+ cli.info("Adding file #%d/%d: %s" % (i + 1, file_count, f))
+
+ tmp = cli.create_graph()
+ tmp.load(f, format="turtle")
+ size = len(tmp)
+ # g.add_graph(tmp)
+
+ for idx, t in enumerate(tmp.triples((None, None, None))):
+ g.add(t)
+
+ if idx % 10000 == 0:
+ g.commit()
+ g.commit()
+
+ agg += size
+ cli.info("Loaded %d tuples" % size)
+
+ cli.info("Done. Loaded %d tuples (including duplicates)" % agg)
cli.with_database(load)