summaryrefslogtreecommitdiff
path: root/storage/migrations
diff options
context:
space:
mode:
Diffstat (limited to 'storage/migrations')
-rw-r--r--storage/migrations/004_populate_fts.up.sql30
-rw-r--r--storage/migrations/005_drop_unused_embeddings.down.sql87
-rw-r--r--storage/migrations/005_drop_unused_embeddings.up.sql32
3 files changed, 123 insertions, 26 deletions
diff --git a/storage/migrations/004_populate_fts.up.sql b/storage/migrations/004_populate_fts.up.sql
index 1d1b16a..1068bf7 100644
--- a/storage/migrations/004_populate_fts.up.sql
+++ b/storage/migrations/004_populate_fts.up.sql
@@ -1,26 +1,4 @@
--- Populate FTS table with existing embeddings
-DELETE FROM fts_embeddings;
-
-INSERT INTO fts_embeddings (slug, raw_text, filename, embedding_size)
-SELECT slug, raw_text, filename, 384 FROM embeddings_384;
-
-INSERT INTO fts_embeddings (slug, raw_text, filename, embedding_size)
-SELECT slug, raw_text, filename, 768 FROM embeddings_768;
-
-INSERT INTO fts_embeddings (slug, raw_text, filename, embedding_size)
-SELECT slug, raw_text, filename, 1024 FROM embeddings_1024;
-
-INSERT INTO fts_embeddings (slug, raw_text, filename, embedding_size)
-SELECT slug, raw_text, filename, 1536 FROM embeddings_1536;
-
-INSERT INTO fts_embeddings (slug, raw_text, filename, embedding_size)
-SELECT slug, raw_text, filename, 2048 FROM embeddings_2048;
-
-INSERT INTO fts_embeddings (slug, raw_text, filename, embedding_size)
-SELECT slug, raw_text, filename, 3072 FROM embeddings_3072;
-
-INSERT INTO fts_embeddings (slug, raw_text, filename, embedding_size)
-SELECT slug, raw_text, filename, 4096 FROM embeddings_4096;
-
-INSERT INTO fts_embeddings (slug, raw_text, filename, embedding_size)
-SELECT slug, raw_text, filename, 5120 FROM embeddings_5120; \ No newline at end of file
+-- Populate FTS table with existing embeddings (incremental - only inserts missing rows)
+-- Only use 768 embeddings as that's what we use
+INSERT OR IGNORE INTO fts_embeddings (slug, raw_text, filename, embedding_size)
+SELECT slug, raw_text, filename, 768 FROM embeddings_768; \ No newline at end of file
diff --git a/storage/migrations/005_drop_unused_embeddings.down.sql b/storage/migrations/005_drop_unused_embeddings.down.sql
new file mode 100644
index 0000000..063cb88
--- /dev/null
+++ b/storage/migrations/005_drop_unused_embeddings.down.sql
@@ -0,0 +1,87 @@
+-- Recreate unused embedding tables (for rollback)
+CREATE TABLE IF NOT EXISTS embeddings_384 (
+ id INTEGER PRIMARY KEY AUTOINCREMENT,
+ embeddings BLOB NOT NULL,
+ slug TEXT NOT NULL,
+ raw_text TEXT NOT NULL,
+ filename TEXT NOT NULL,
+ created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
+);
+
+CREATE TABLE IF NOT EXISTS embeddings_1024 (
+ id INTEGER PRIMARY KEY AUTOINCREMENT,
+ embeddings BLOB NOT NULL,
+ slug TEXT NOT NULL,
+ raw_text TEXT NOT NULL,
+ filename TEXT NOT NULL,
+ created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
+);
+
+CREATE TABLE IF NOT EXISTS embeddings_1536 (
+ id INTEGER PRIMARY KEY AUTOINCREMENT,
+ embeddings BLOB NOT NULL,
+ slug TEXT NOT NULL,
+ raw_text TEXT NOT NULL,
+ filename TEXT NOT NULL,
+ created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
+);
+
+CREATE TABLE IF NOT EXISTS embeddings_2048 (
+ id INTEGER PRIMARY KEY AUTOINCREMENT,
+ embeddings BLOB NOT NULL,
+ slug TEXT NOT NULL,
+ raw_text TEXT NOT NULL,
+ filename TEXT NOT NULL,
+ created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
+);
+
+CREATE TABLE IF NOT EXISTS embeddings_3072 (
+ id INTEGER PRIMARY KEY AUTOINCREMENT,
+ embeddings BLOB NOT NULL,
+ slug TEXT NOT NULL,
+ raw_text TEXT NOT NULL,
+ filename TEXT NOT NULL,
+ created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
+);
+
+CREATE TABLE IF NOT EXISTS embeddings_4096 (
+ id INTEGER PRIMARY KEY AUTOINCREMENT,
+ embeddings BLOB NOT NULL,
+ slug TEXT NOT NULL,
+ raw_text TEXT NOT NULL,
+ filename TEXT NOT NULL,
+ created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
+);
+
+CREATE TABLE IF NOT EXISTS embeddings_5120 (
+ id INTEGER PRIMARY KEY AUTOINCREMENT,
+ embeddings BLOB NOT NULL,
+ slug TEXT NOT NULL,
+ raw_text TEXT NOT NULL,
+ filename TEXT NOT NULL,
+ created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
+);
+
+CREATE INDEX IF NOT EXISTS idx_embeddings_384_filename ON embeddings_384(filename);
+CREATE INDEX IF NOT EXISTS idx_embeddings_1024_filename ON embeddings_1024(filename);
+CREATE INDEX IF NOT EXISTS idx_embeddings_1536_filename ON embeddings_1536(filename);
+CREATE INDEX IF NOT EXISTS idx_embeddings_2048_filename ON embeddings_2048(filename);
+CREATE INDEX IF NOT EXISTS idx_embeddings_3072_filename ON embeddings_3072(filename);
+CREATE INDEX IF NOT EXISTS idx_embeddings_4096_filename ON embeddings_4096(filename);
+CREATE INDEX IF NOT EXISTS idx_embeddings_5120_filename ON embeddings_5120(filename);
+
+CREATE INDEX IF NOT EXISTS idx_embeddings_384_slug ON embeddings_384(slug);
+CREATE INDEX IF NOT EXISTS idx_embeddings_1024_slug ON embeddings_1024(slug);
+CREATE INDEX IF NOT EXISTS idx_embeddings_1536_slug ON embeddings_1536(slug);
+CREATE INDEX IF NOT EXISTS idx_embeddings_2048_slug ON embeddings_2048(slug);
+CREATE INDEX IF NOT EXISTS idx_embeddings_3072_slug ON embeddings_3072(slug);
+CREATE INDEX IF NOT EXISTS idx_embeddings_4096_slug ON embeddings_4096(slug);
+CREATE INDEX IF NOT EXISTS idx_embeddings_5120_slug ON embeddings_5120(slug);
+
+CREATE INDEX IF NOT EXISTS idx_embeddings_384_created_at ON embeddings_384(created_at);
+CREATE INDEX IF NOT EXISTS idx_embeddings_1024_created_at ON embeddings_1024(created_at);
+CREATE INDEX IF NOT EXISTS idx_embeddings_1536_created_at ON embeddings_1536(created_at);
+CREATE INDEX IF NOT EXISTS idx_embeddings_2048_created_at ON embeddings_2048(created_at);
+CREATE INDEX IF NOT EXISTS idx_embeddings_3072_created_at ON embeddings_3072(created_at);
+CREATE INDEX IF NOT EXISTS idx_embeddings_4096_created_at ON embeddings_4096(created_at);
+CREATE INDEX IF NOT EXISTS idx_embeddings_5120_created_at ON embeddings_5120(created_at);
diff --git a/storage/migrations/005_drop_unused_embeddings.up.sql b/storage/migrations/005_drop_unused_embeddings.up.sql
new file mode 100644
index 0000000..f26e30f
--- /dev/null
+++ b/storage/migrations/005_drop_unused_embeddings.up.sql
@@ -0,0 +1,32 @@
+-- Drop unused embedding tables (we only use 768)
+DROP INDEX IF EXISTS idx_embeddings_384_filename;
+DROP INDEX IF EXISTS idx_embeddings_1024_filename;
+DROP INDEX IF EXISTS idx_embeddings_1536_filename;
+DROP INDEX IF EXISTS idx_embeddings_2048_filename;
+DROP INDEX IF EXISTS idx_embeddings_3072_filename;
+DROP INDEX IF EXISTS idx_embeddings_4096_filename;
+DROP INDEX IF EXISTS idx_embeddings_5120_filename;
+
+DROP INDEX IF EXISTS idx_embeddings_384_slug;
+DROP INDEX IF EXISTS idx_embeddings_1024_slug;
+DROP INDEX IF EXISTS idx_embeddings_1536_slug;
+DROP INDEX IF EXISTS idx_embeddings_2048_slug;
+DROP INDEX IF EXISTS idx_embeddings_3072_slug;
+DROP INDEX IF EXISTS idx_embeddings_4096_slug;
+DROP INDEX IF EXISTS idx_embeddings_5120_slug;
+
+DROP INDEX IF EXISTS idx_embeddings_384_created_at;
+DROP INDEX IF EXISTS idx_embeddings_1024_created_at;
+DROP INDEX IF EXISTS idx_embeddings_1536_created_at;
+DROP INDEX IF EXISTS idx_embeddings_2048_created_at;
+DROP INDEX IF EXISTS idx_embeddings_3072_created_at;
+DROP INDEX IF EXISTS idx_embeddings_4096_created_at;
+DROP INDEX IF EXISTS idx_embeddings_5120_created_at;
+
+DROP TABLE IF EXISTS embeddings_384;
+DROP TABLE IF EXISTS embeddings_1024;
+DROP TABLE IF EXISTS embeddings_1536;
+DROP TABLE IF EXISTS embeddings_2048;
+DROP TABLE IF EXISTS embeddings_3072;
+DROP TABLE IF EXISTS embeddings_4096;
+DROP TABLE IF EXISTS embeddings_5120;