From 043db73ee6c06a9338112b58056d1c2bb3c370f0 Mon Sep 17 00:00:00 2001
From: jp170na <jan.ptak@student.tuke.sk>
Date: Sun, 11 Oct 2026 00:21:47 +0200
Subject: [PATCH] Refactor GraphRAG into standalone tool

---
 Dockerfile                                                  | 1 +
 tools/__init__.py                                           | 0
 tools/graphrag/__init__.py                                  | 0
 {scripts => tools/graphrag}/build_graphrag.py               | 6 +++---
 .../graphrag/evaluation}/graphrag_questions.json            | 0
 {scripts => tools/graphrag}/graph_db.py                     | 0
 {scripts => tools/graphrag}/graph_search.py                 | 2 +-
 {scripts => tools/graphrag}/graphrag.py                     | 0
 {scripts => tools/graphrag}/neo4j_smoke.py                  | 2 +-
 tools/graphrag/tests/__init__.py                            | 0
 {test => tools/graphrag/tests}/test_graph_search.py         | 2 +-
 {test => tools/graphrag/tests}/test_graphrag.py             | 2 +-
 12 files changed, 8 insertions(+), 7 deletions(-)
 create mode 100644 tools/__init__.py
 create mode 100644 tools/graphrag/__init__.py
 rename {scripts => tools/graphrag}/build_graphrag.py (99%)
 rename {evaluation/json_files => tools/graphrag/evaluation}/graphrag_questions.json (100%)
 rename {scripts => tools/graphrag}/graph_db.py (100%)
 rename {scripts => tools/graphrag}/graph_search.py (99%)
 rename {scripts => tools/graphrag}/graphrag.py (100%)
 rename {scripts => tools/graphrag}/neo4j_smoke.py (82%)
 create mode 100644 tools/graphrag/tests/__init__.py
 rename {test => tools/graphrag/tests}/test_graph_search.py (99%)
 rename {test => tools/graphrag/tests}/test_graphrag.py (99%)

diff --git a/Dockerfile b/Dockerfile
index e026b59..b214c1d 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -26,6 +26,7 @@ RUN python -m pip install --no-cache-dir \
 
 COPY app ./app
 COPY scripts ./scripts
+COPY tools ./tools
 
 RUN mkdir -p data
 
diff --git a/tools/__init__.py b/tools/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/tools/graphrag/__init__.py b/tools/graphrag/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/scripts/build_graphrag.py b/tools/graphrag/build_graphrag.py
similarity index 99%
rename from scripts/build_graphrag.py
rename to tools/graphrag/build_graphrag.py
index af12487..29896f3 100644
--- a/scripts/build_graphrag.py
+++ b/tools/graphrag/build_graphrag.py
@@ -5,11 +5,11 @@ import json
 from pathlib import Path
 from typing import Any
 
-from scripts.graph_db import (
+from .graph_db import (
     create_neo4j_driver,
     get_neo4j_settings,
 )
-from scripts.graphrag import (
+from .graphrag import (
     GRAPH_SCOPE,
     build_graph_payload,
 )
@@ -17,7 +17,7 @@ from scripts.graphrag import (
 
 PROJECT_ROOT = Path(
     __file__
-).resolve().parents[1]
+).resolve().parents[2]
 
 DOCUMENTS_FILE = (
     PROJECT_ROOT
diff --git a/evaluation/json_files/graphrag_questions.json b/tools/graphrag/evaluation/graphrag_questions.json
similarity index 100%
rename from evaluation/json_files/graphrag_questions.json
rename to tools/graphrag/evaluation/graphrag_questions.json
diff --git a/scripts/graph_db.py b/tools/graphrag/graph_db.py
similarity index 100%
rename from scripts/graph_db.py
rename to tools/graphrag/graph_db.py
diff --git a/scripts/graph_search.py b/tools/graphrag/graph_search.py
similarity index 99%
rename from scripts/graph_search.py
rename to tools/graphrag/graph_search.py
index a6bd3a1..72da71e 100644
--- a/scripts/graph_search.py
+++ b/tools/graphrag/graph_search.py
@@ -1283,7 +1283,7 @@ def search_graph(
     limit: int = DEFAULT_LIMIT,
     max_entities: int = DEFAULT_MAX_ENTITIES,
 ) -> dict[str, Any]:
-    from scripts.graph_db import (
+    from .graph_db import (
         create_neo4j_driver,
         get_neo4j_settings,
     )
diff --git a/scripts/graphrag.py b/tools/graphrag/graphrag.py
similarity index 100%
rename from scripts/graphrag.py
rename to tools/graphrag/graphrag.py
diff --git a/scripts/neo4j_smoke.py b/tools/graphrag/neo4j_smoke.py
similarity index 82%
rename from scripts/neo4j_smoke.py
rename to tools/graphrag/neo4j_smoke.py
index 41ae918..aad2b7b 100644
--- a/scripts/neo4j_smoke.py
+++ b/tools/graphrag/neo4j_smoke.py
@@ -1,4 +1,4 @@
-from scripts.graph_db import verify_neo4j_connection
+from .graph_db import verify_neo4j_connection
 
 
 def main() -> None:
diff --git a/tools/graphrag/tests/__init__.py b/tools/graphrag/tests/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/test/test_graph_search.py b/tools/graphrag/tests/test_graph_search.py
similarity index 99%
rename from test/test_graph_search.py
rename to tools/graphrag/tests/test_graph_search.py
index fe241c6..44c7bc7 100644
--- a/test/test_graph_search.py
+++ b/tools/graphrag/tests/test_graph_search.py
@@ -4,7 +4,7 @@ from typing import Any
 
 import pytest
 
-import scripts.graph_search as graph_search
+import tools.graphrag.graph_search as graph_search
 
 
 class FakeSession:
diff --git a/test/test_graphrag.py b/tools/graphrag/tests/test_graphrag.py
similarity index 99%
rename from test/test_graphrag.py
rename to tools/graphrag/tests/test_graphrag.py
index d03a23d..0d1c3d3 100644
--- a/test/test_graphrag.py
+++ b/tools/graphrag/tests/test_graphrag.py
@@ -1,6 +1,6 @@
 from __future__ import annotations
 
-from scripts.graphrag import (
+from tools.graphrag.graphrag import (
     build_graph_payload,
     extract_explicit_proposal_title,
     extract_works,