move generate_contents_information to utils.py

[dak.git] / daklib / database.py

diff --git a/daklib/database.py b/daklib/database.py
index 1f659606b667f3238a6fb169f8228b4e934f01e0..9cefc38189c8df6333b9db5affc0a8a4014b4139 100755 (executable)
--- a/daklib/database.py
+++ b/daklib/database.py
@@ -45,6 +45,8 @@ suite_version_cache = {}
 suite_bin_version_cache = {}
 content_path_id_cache = {}
 content_file_id_cache = {}
+insert_contents_file_cache = {}
+cache_preloaded = False
 
 ################################################################################
 
@@ -250,24 +252,50 @@ def get_suite_version(source, suite, arch):
 
     return version
 
-def get_latest_binary_version_id(binary, suite, arch):
+def get_latest_binary_version_id(binary, section, suite, arch):
     global suite_bin_version_cache
-    cache_key = "%s_%s" % (binary, suite)
+    cache_key = "%s_%s_%s_%s" % (binary, section, suite, arch)
+    cache_key_all = "%s_%s_%s_%s" % (binary, section, suite, get_architecture_id("all"))
 
+    # Check for the cache hit for its arch, then arch all
     if suite_bin_version_cache.has_key(cache_key):
         return suite_bin_version_cache[cache_key]
+    if suite_bin_version_cache.has_key(cache_key_all):
+        return suite_bin_version_cache[cache_key_all]
+    if cache_preloaded == True:
+        return # package does not exist
 
-    q = projectB.query("SELECT b.id, b.version FROM binaries b JOIN bin_associations ba ON (b.id = ba.bin) WHERE b.package = '%s' AND b.architecture = '%d' AND ba.suite = '%d'" % (binary, int(arch), int(suite)))
+    q = projectB.query("SELECT DISTINCT b.id FROM binaries b JOIN bin_associations ba ON (b.id = ba.bin) JOIN override o ON (o.package=b.package) WHERE b.package = '%s' AND b.architecture = '%d' AND ba.suite = '%d' AND o.section = '%d'" % (binary, int(arch), int(suite), int(section)))
 
-    highest_bid, highest_version = None, None
+    if not q.getresult():
+        return False
 
-    for bi in q.getresult():
-        if highest_version == None or apt_pkg.VersionCompare(bi[1], highest_version) == 1:
-             highest_bid = bi[0]
-             highest_version = bi[1]
+    highest_bid = q.getresult()[0][0]
 
+    suite_bin_version_cache[cache_key] = highest_bid
     return highest_bid
 
+def preload_binary_id_cache():
+    global suite_bin_version_cache, cache_preloaded
+
+    # Get suite info
+    q = projectB.query("SELECT id FROM suite")
+    suites = q.getresult()
+
+    # Get arch mappings
+    q = projectB.query("SELECT id FROM architecture")
+    arches = q.getresult()
+
+    for suite in suites:
+        for arch in arches:
+            q = projectB.query("SELECT DISTINCT b.id, b.package, o.section FROM binaries b JOIN bin_associations ba ON (b.id = ba.bin) JOIN override o ON (o.package=b.package) WHERE b.architecture = '%d' AND ba.suite = '%d'" % (int(arch[0]), int(suite[0])))
+
+            for bi in q.getresult():
+                cache_key = "%s_%s_%s_%s" % (bi[1], bi[2], suite[0], arch[0])
+                suite_bin_version_cache[cache_key] = int(bi[0])
+
+    cache_preloaded = True
+
 ################################################################################
 
 def get_or_set_maintainer_id (maintainer):
@@ -429,12 +457,8 @@ def get_or_set_contents_file_id(file):
         q = projectB.query(sql_select)
         if not q.getresult():
             # since this can be called within a transaction, we can't use currval
-            q = projectB.query("SELECT nextval('content_file_names_id_seq')")
-            file_id = int(q.getresult()[0][0])
-            projectB.query("INSERT INTO content_file_names VALUES ('%d', '%s')" % (file_id, file))
-            content_file_id_cache[file] =  file_id
-        else:
-            content_file_id_cache[file] = int(q.getresult()[0][0])
+            q = projectB.query("INSERT INTO content_file_names VALUES (DEFAULT, '%s') RETURNING id" % (file))
+        content_file_id_cache[file] = int(q.getresult()[0][0])
     return content_file_id_cache[file]
 
 ################################################################################
@@ -447,18 +471,21 @@ def get_or_set_contents_path_id(path):
         q = projectB.query(sql_select)
         if not q.getresult():
             # since this can be called within a transaction, we can't use currval
-            q = projectB.query("SELECT nextval('content_file_names_id_seq')")
-            path_id = int(q.getresult()[0][0])
-            projectB.query("INSERT INTO content_file_paths VALUES ('%d', '%s')" % ( path_id, path))
-            content_path_id_cache[path] = path_id
-        else:
-            content_path_id_cache[path] = int(q.getresult()[0][0])
-
+            q = projectB.query("INSERT INTO content_file_paths VALUES (DEFAULT, '%s') RETURNING id" % (path))
+        content_path_id_cache[path] = int(q.getresult()[0][0])
     return content_path_id_cache[path]
 
 ################################################################################
 
 def insert_content_path(bin_id, fullpath):
+    global insert_contents_file_cache
+    cache_key = "%s_%s" % (bin_id, fullpath)
+
+    # have we seen this contents before?
+    # probably only revelant during package import
+    if insert_contents_file_cache.has_key(cache_key):
+        return
+
     # split the path into basename, and pathname
     (path, file)  = os.path.split(fullpath)
 
@@ -466,6 +493,12 @@ def insert_content_path(bin_id, fullpath):
     file_id = get_or_set_contents_file_id(file)
     path_id = get_or_set_contents_path_id(path)
 
+    # Determine if we're inserting a duplicate row
+    q = projectB.query("SELECT 1 FROM content_associations WHERE binary_pkg = '%d' AND filepath = '%d' AND filename = '%d'" % (int(bin_id), path_id, file_id))
+    if q.getresult():
+        # Yes we are, return without doing the insert
+        return
+
     # Put them into content_assiocations
     projectB.query("INSERT INTO content_associations VALUES (DEFAULT, '%d', '%d', '%d')" % (bin_id, path_id, file_id))
     return