From 5d7652105966517c0230ccf8a7368f1f746cb0bc Mon Sep 17 00:00:00 2001
From: LukeZekes <41444778+LukeZekes@users.noreply.github.com>
Date: Tue, 18 Mar 2025 13:41:39 -0500
Subject: [PATCH] Change in max_sim function for newer postgresql versions
 (#54)

* add filename option for text documents

* fix: duplicated env variables

* Include chunk score in query response

* don't initialize colpali store if one is not provided

* Add score to ChunkSource

* Add huggingface cache volume to avoid redownloads

* Replace <~> in maximum_similarity calculation

* Add dev_mode setup to default configuration

* Dump with databridge documents (no pdfs)

* Add dev_mode setup to default configuration

* Comment docker instructions for api host

* Use empty dump file
---
 core/vector_store/multi_vector_store.py |  5 ++++-
 databridge.toml                         |  2 +-
 docker-compose.yml                      | 23 ++++++++++++++++++++++-
 dockerfile                              |  4 ++++
 dump.sql                                |  0
 init.sql                                |  4 +++-
 postgres.dockerfile                     |  3 +++
 quick_setup.py                          |  5 ++++-
 8 files changed, 41 insertions(+), 5 deletions(-)
 create mode 100644 dump.sql

diff --git a/core/vector_store/multi_vector_store.py b/core/vector_store/multi_vector_store.py
index 692830c..9eecf8d 100644
--- a/core/vector_store/multi_vector_store.py
+++ b/core/vector_store/multi_vector_store.py
@@ -132,7 +132,10 @@ class MultiVectorStore(BaseVectorStore):
                             SELECT unnest(document) AS document
                         ),
                         similarities AS (
-                            SELECT query_number, 1 - ((document <~> query) / bit_length(query)) AS similarity FROM queries CROSS JOIN documents
+                            SELECT 
+                                query_number, 
+                                1.0 - (bit_count(document # query)::float / greatest(bit_length(query), 1)::float) AS similarity
+                            FROM queries CROSS JOIN documents
                         ),
                         max_similarities AS (
                             SELECT MAX(similarity) AS max_similarity FROM similarities GROUP BY query_number
diff --git a/databridge.toml b/databridge.toml
index 1e91590..6e14c74 100644
--- a/databridge.toml
+++ b/databridge.toml
@@ -1,5 +1,5 @@
 [api]
-host = "localhost"
+host = "localhost" # Needs to be "0.0.0.0" for docker
 port = 8000
 reload = true
 
diff --git a/docker-compose.yml b/docker-compose.yml
index a95934e..89678c1 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -1,6 +1,20 @@
 version: '3.8'
 
+x-ollama-check: &ollama-check
+  # This command reads the toml file and checks if any provider is set to "ollama"
+  command: >
+    /bin/sh -c '
+    grep -q "provider *= *\"ollama\"" databridge.toml &&
+    echo "true" > /tmp/needs_ollama ||
+    echo "false" > /tmp/needs_ollama'
+
 services:
+  config-check:
+    image: alpine
+    volumes:
+      - ./databridge.toml:/databridge.toml
+    <<: *ollama-check
+
   databridge:
     build: .
     ports:
@@ -16,11 +30,15 @@ services:
       - ./storage:/app/storage
       - ./logs:/app/logs
       - ./databridge.toml:/app/databridge.toml
+      - huggingface_cache:/root/.cache/huggingface
     depends_on:
       postgres:
         condition: service_healthy
+      config-check:
+        condition: service_completed_successfully
       ollama:
         condition: service_started
+        required: false
     networks:
       - databridge-network
     env_file:
@@ -51,6 +69,8 @@ services:
 
   ollama:
     image: ollama/ollama:latest
+    profiles:
+      - ollama
     ports:
       - "11434:11434"
     volumes:
@@ -66,4 +86,5 @@ networks:
 
 volumes:
   postgres_data:
-  ollama_data: 
+  ollama_data:
+  huggingface_cache: 
diff --git a/dockerfile b/dockerfile
index 5856a70..d475474 100644
--- a/dockerfile
+++ b/dockerfile
@@ -65,6 +65,10 @@ reload = false\n\
 \n\
 [auth]\n\
 jwt_algorithm = "HS256"\n\
+dev_mode = true\n\
+dev_entity_id = "dev_user"\n\
+dev_entity_type = "developer"\n\
+dev_permissions = ["read", "write", "admin"]\n\
 \n\
 [completion]\n\
 provider = "ollama"\n\
diff --git a/dump.sql b/dump.sql
new file mode 100644
index 0000000..e69de29
diff --git a/init.sql b/init.sql
index 698a840..7792c76 100644
--- a/init.sql
+++ b/init.sql
@@ -60,7 +60,9 @@ CREATE OR REPLACE FUNCTION max_sim(document bit[], query bit[]) RETURNS double p
         SELECT unnest(document) AS document
     ),
     similarities AS (
-        SELECT query_number, 1 - ((document <~> query) / bit_length(query)) AS similarity
+        SELECT 
+            query_number, 
+            1.0 - (bit_count(document # query)::float / greatest(bit_length(query), 1)::float) AS similarity
         FROM queries CROSS JOIN documents
     ),
     max_similarities AS (
diff --git a/postgres.dockerfile b/postgres.dockerfile
index ca92690..4b72bf3 100644
--- a/postgres.dockerfile
+++ b/postgres.dockerfile
@@ -20,3 +20,6 @@ RUN apk del git build-base clang llvm postgresql-dev \
 
 # Copy initialization scripts
 COPY init.sql /docker-entrypoint-initdb.d/
+
+# Copy data dump
+COPY dump.sql /tmp/dump.sql
\ No newline at end of file
diff --git a/quick_setup.py b/quick_setup.py
index 498757d..081025a 100644
--- a/quick_setup.py
+++ b/quick_setup.py
@@ -328,7 +328,10 @@ def setup_postgres():
                         SELECT unnest(document_bits) AS document
                     ),
                     similarities AS (
-                        SELECT query_number, 1 - ((document <~> query) / bit_length(query)) AS similarity FROM queries CROSS JOIN documents
+                        SELECT 
+                            query_number, 
+                            1.0 - (bit_count(document # query)::float / greatest(bit_length(query), 1)::float) AS similarity 
+                        FROM queries CROSS JOIN documents
                     ),
                     max_similarities AS (
                         SELECT MAX(similarity) AS max_similarity FROM similarities GROUP BY query_number