From 2abbcc53125eb1e9288698ca3a734016b601453c Mon Sep 17 00:00:00 2001 From: Julien Cristau Date: Wed, 18 May 2016 15:45:57 +0200 Subject: [PATCH] generate_releases: reduce number of sql queries for by-hash files Instead of doing two queries per filename and per hash function, store a set of known hashfiles up front and group the UPDATEs in a single query. Suggested by Colin Watson. Signed-off-by: Julien Cristau --- dak/generate_releases.py | 42 ++++++++++++++++++++++++++-------------- 1 file changed, 27 insertions(+), 15 deletions(-) diff --git a/dak/generate_releases.py b/dak/generate_releases.py index ae3577f3..dc46b24b 100755 --- a/dak/generate_releases.py +++ b/dak/generate_releases.py @@ -362,37 +362,49 @@ class ReleaseWriter(object): out.close() os.rename(outfile + '.new', outfile) + # Mark all by-hash files as obsolete. We will undo that for the ones + # we still reference later. query = """ UPDATE hashfile SET unreferenced = CURRENT_TIMESTAMP WHERE suite_id = :id AND unreferenced IS NULL""" session.execute(query, {'id': suite.suite_id}) if suite.byhash: + query = "SELECT path FROM hashfile WHERE suite_id = :id" + q = session.execute(query, {'id': suite.suite_id}) + known_hashfiles = set(row[0] for row in q) + updated = [] + new = [] + + # Update the hashfile table with new or updated files for filename in fileinfo: if not os.path.exists(filename): # probably an uncompressed index we didn't generate continue - + byhashdir = os.path.join(os.path.dirname(filename), 'by-hash') for h in hashes: field = h.release_field - hashfile = os.path.join(os.path.dirname(filename), 'by-hash', field, fileinfo[filename][field]) - query = "SELECT 1 FROM hashfile WHERE path = :p AND suite_id = :id" - q = session.execute( - query, - {'p': hashfile, 'id': suite.suite_id}) - if q.rowcount: - session.execute(''' - UPDATE hashfile SET unreferenced = NULL - WHERE path = :p and suite_id = :id''', - {'p': hashfile, 'id': suite.suite_id}) + hashfile = os.path.join(byhashdir, field, fileinfo[filename][field]) + if hashfile in known_hashfiles: + updated.append(hashfile) else: - session.execute(''' - INSERT INTO hashfile (path, suite_id) - VALUES (:p, :id)''', - {'p': hashfile, 'id': suite.suite_id}) + new.append(hashfile) + + if updated: + session.execute(""" + UPDATE hashfile SET unreferenced = NULL + WHERE path = ANY(:p) AND suite_id = :id""", + {'p': updated, 'id': suite.suite_id}) + if new: + session.execute(""" + INSERT INTO hashfile (path, suite_id) + VALUES (:p, :id)""", + [{'p': hashfile, 'id': suite.suite_id} for hashfile in new]) + session.commit() if suite.byhash: + # Create hardlinks in by-hash directories for filename in fileinfo: if not os.path.exists(filename): # probably an uncompressed index we didn't generate -- 2.39.5