diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2021-12-06 15:51:18 -0800 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2021-12-06 15:51:18 -0800 |
| commit | 25222cd3dcbf396c7d2a7372e32dd6e9ccae9ba4 (patch) | |
| tree | 7fdca7b0c80d51dcb894cbe4b4d60eca632a55b8 /fatcat_scholar/sim_pipeline.py | |
| parent | 1389c0fdf03b002b2a555f9a69755c4798b5530e (diff) | |
| download | fatcat-scholar-25222cd3dcbf396c7d2a7372e32dd6e9ccae9ba4.tar.gz fatcat-scholar-25222cd3dcbf396c7d2a7372e32dd6e9ccae9ba4.zip | |
SIM pipeline: fix bug w/r/t issues with no fatcat coverage at all
Diffstat (limited to 'fatcat_scholar/sim_pipeline.py')
| -rw-r--r-- | fatcat_scholar/sim_pipeline.py | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/fatcat_scholar/sim_pipeline.py b/fatcat_scholar/sim_pipeline.py index cceec30..e140816 100644 --- a/fatcat_scholar/sim_pipeline.py +++ b/fatcat_scholar/sim_pipeline.py @@ -162,7 +162,7 @@ class SimPipeline: self.issue_db.db.row_factory = sqlite3.Row cur = self.issue_db.db.cursor() for row in cur.execute( - "SELECT * FROM sim_issue LEFT JOIN sim_pub ON sim_issue.sim_pubid = sim_pub.sim_pubid WHERE sim_issue.release_count < 3" + "SELECT * FROM sim_issue LEFT JOIN sim_pub ON sim_issue.sim_pubid = sim_pub.sim_pubid WHERE sim_issue.release_count < 3 sim_issue.release_count IS NULL" ): if should_skip_item(row["issue_item"]): continue @@ -199,7 +199,7 @@ class SimPipeline: self.issue_db.db.row_factory = sqlite3.Row cur = self.issue_db.db.cursor() for row in cur.execute( - f"SELECT * FROM sim_issue LEFT JOIN sim_pub ON sim_issue.sim_pubid = sim_pub.sim_pubid WHERE sim_issue.release_count < {max_release_count}" + f"SELECT * FROM sim_issue LEFT JOIN sim_pub ON sim_issue.sim_pubid = sim_pub.sim_pubid WHERE sim_issue.release_count < {max_release_count} OR sim_issue.release_count IS NULL" ): if should_skip_item(row["issue_item"]): continue |