schema: identify hash values by an integerhashid

This one is a bit more complex, than the other transformations, because
the new hashvalue table has to be cleaned with a trigger. During a test
import the -wal file exploded. The resulting db is similar in size to
the original.

1 files changed, 10 insertions, 3 deletions

diff --git a/readyaml.py b/readyaml.py
index bb8ac54..007ed96 100755
--- a/readyaml.py
+++ b/readyaml.py
@@ -45,9 +45,16 @@ def readyaml(db, stream):
         cur.execute("INSERT INTO content (pid, filename, size) VALUES (?, ?, ?);",
                     (pid, entry["name"], entry["size"]))
         cid = cur.lastrowid
-        cur.executemany("INSERT INTO hash (cid, function, hash) VALUES (?, ?, ?);",
-                        ((cid, func, hexhash)
-                         for func, hexhash in entry["hashes"].items()))
+        for func, hexhash in entry["hashes"].items():
+            cur.execute("SELECT id FROM hashvalue WHERE hash = ?;", (hexhash,))
+            row = cur.fetchone()
+            if row:
+                hid = row[0]
+            else:
+                cur.execute("INSERT INTO hashvalue (hash) VALUES (?);", (hexhash,))
+                hid = cur.lastrowid
+            cur.execute("INSERT INTO hash (cid, function, hid) VALUES (?, ?, ?);",
+                        (cid, func, hid))
     raise ValueError("missing commit block")
 
 def main():