fix setting base_dir to full paths when not using file system. (#1096)

* fix setting base_dir to full paths when not using file system.

* add general resolve_path

.semversioner/next-release/patch-20240904173227165702.json

@@ -0,0 +1,4 @@
+{
+  "type": "patch",
+  "description": "fix setting base_dir to full paths when not using file system."
+}

graphrag/config/__init__.py

@@ -68,7 +68,7 @@ from .models import (
     UmapConfig,
 )
 from .read_dotenv import read_dotenv
-from .resolve_timestamp_path import resolve_timestamp_path
+from .resolve_path import resolve_path
 
 __all__ = [
     "ApiKeyMissingError",
@@ -127,6 +127,6 @@ __all__ = [
     "load_config",
     "load_config_from_file",
     "read_dotenv",
-    "resolve_timestamp_path",
+    "resolve_path",
     "search_for_config_in_root_dir",
 ]

graphrag/config/load_config.py

@@ -7,8 +7,9 @@ from pathlib import Path
 
 from .config_file_loader import load_config_from_file, search_for_config_in_root_dir
 from .create_graphrag_config import create_graphrag_config
+from .enums import ReportingType, StorageType
 from .models.graph_rag_config import GraphRagConfig
-from .resolve_timestamp_path import resolve_timestamp_path
+from .resolve_path import resolve_path
 
 
 def load_config(
@@ -47,19 +48,19 @@ def load_config(
     else:
         config = create_graphrag_config(root_dir=str(root))
 
-    if run_id:
-        config.storage.base_dir = str(
-            resolve_timestamp_path((root / config.storage.base_dir).resolve(), run_id)
+    config.storage.base_dir = str(
+        resolve_path(
+            config.storage.base_dir,
+            root if config.storage.type == StorageType.file else None,
+            run_id,
         )
-        config.reporting.base_dir = str(
-            resolve_timestamp_path((root / config.reporting.base_dir).resolve(), run_id)
-        )
-    else:
-        config.storage.base_dir = str(
-            resolve_timestamp_path((root / config.storage.base_dir).resolve())
-        )
-        config.reporting.base_dir = str(
-            resolve_timestamp_path((root / config.reporting.base_dir).resolve())
+    )
+    config.reporting.base_dir = str(
+        resolve_path(
+            config.reporting.base_dir,
+            root if config.reporting.type == ReportingType.file else None,
+            run_id,
         )
+    )
 
     return config

graphrag/config/resolve_path.py

@@ -79,9 +79,9 @@ def _resolve_timestamp_path_with_dir(
     return _resolve_timestamp_path_with_value(path, timestamp_dirs[0].name)
 
 
-def resolve_timestamp_path(
+def _resolve_timestamp_path(
     path: str | Path,
-    pattern_or_timestamp_value: re.Pattern[str] | str = re.compile(r"^\d{8}-\d{6}$"),
+    pattern_or_timestamp_value: re.Pattern[str] | str | None = None,
 ) -> Path:
     r"""Timestamp path resolver.
 
@@ -110,6 +110,43 @@ def resolve_timestamp_path(
         If the parent directory expecting to contain timestamp directories does not exist or is not a directory.
         Or if no timestamp directories are found in the parent directory that match the pattern.
     """
+    if not pattern_or_timestamp_value:
+        pattern_or_timestamp_value = re.compile(r"^\d{8}-\d{6}$")
     if isinstance(pattern_or_timestamp_value, str):
         return _resolve_timestamp_path_with_value(path, pattern_or_timestamp_value)
     return _resolve_timestamp_path_with_dir(path, pattern_or_timestamp_value)
+
+
+def resolve_path(
+    path_to_resolve: Path | str,
+    root_dir: Path | str | None = None,
+    pattern_or_timestamp_value: re.Pattern[str] | str | None = None,
+) -> Path:
+    """Resolve the path.
+
+    Resolves any timestamp variables by either using the provided timestamp value if string or
+    by looking up the latest available timestamp directory that matches the given pattern.
+    Resolves the path against the root directory if provided.
+
+    Parameters
+    ----------
+    path_to_resolve : Path | str
+        The path to resolve.
+    root_dir : Path | str | None default=None
+        The root directory to resolve the path from, if provided.
+    pattern_or_timestamp_value : re.Pattern[str] | str, default=None
+        The pattern to use to match the timestamp directories or the timestamp value to use.
+        If a string is provided, the path will be resolved with the given string value.
+        Otherwise, the path will be resolved with the latest available timestamp directory
+        that matches the given pattern.
+
+    Returns
+    -------
+    Path
+        The resolved path.
+    """
+    if root_dir:
+        path_to_resolve = (Path(root_dir) / path_to_resolve).resolve()
+    else:
+        path_to_resolve = Path(path_to_resolve)
+    return _resolve_timestamp_path(path_to_resolve, pattern_or_timestamp_value)

graphrag/query/cli.py

@@ -9,7 +9,7 @@ from pathlib import Path
 
 import pandas as pd
 
-from graphrag.config import load_config, resolve_timestamp_path
+from graphrag.config import load_config, resolve_path
 from graphrag.index.progress import PrintProgressReporter
 
 from . import api
@@ -34,9 +34,7 @@ def run_global_search(
     config = load_config(root, config_filepath)
 
     if data_dir:
-        config.storage.base_dir = str(
-            resolve_timestamp_path((root / data_dir).resolve())
-        )
+        config.storage.base_dir = str(resolve_path(data_dir, root))
 
     data_path = Path(config.storage.base_dir).resolve()
 
@@ -112,9 +110,7 @@ def run_local_search(
     config = load_config(root, config_filepath)
 
     if data_dir:
-        config.storage.base_dir = str(
-            resolve_timestamp_path((root / data_dir).resolve())
-        )
+        config.storage.base_dir = str(resolve_path(data_dir, root))
 
     data_path = Path(config.storage.base_dir).resolve()
 

tests/unit/config/test_resolve_path.py

@@ -0,0 +1,42 @@
+# Copyright (c) 2024 Microsoft Corporation.
+# Licensed under the MIT License
+
+from pathlib import Path
+
+from graphrag.config.resolve_path import resolve_path
+
+
+def test_resolve_path_no_timestamp_with_run_id():
+    path = Path("path/to/data")
+    result = resolve_path(path, pattern_or_timestamp_value="20240812-121000")
+    assert result == path
+
+
+def test_resolve_path_no_timestamp_without_run_id():
+    path = Path("path/to/data")
+    result = resolve_path(path)
+    assert result == path
+
+
+def test_resolve_path_with_timestamp_and_run_id():
+    path = Path("some/path/${timestamp}/data")
+    expected = Path("some/path/20240812/data")
+    result = resolve_path(path, pattern_or_timestamp_value="20240812")
+    assert result == expected
+
+
+def test_resolve_path_with_timestamp_and_inferred_directory():
+    cwd = Path(__file__).parent
+    path = cwd / "fixtures/timestamp_dirs/${timestamp}/data"
+    expected = cwd / "fixtures/timestamp_dirs/20240812-120000/data"
+    result = resolve_path(path)
+    assert result == expected
+
+
+def test_resolve_path_absolute():
+    cwd = Path(__file__).parent
+    path = "fixtures/timestamp_dirs/${timestamp}/data"
+    expected = cwd / "fixtures/timestamp_dirs/20240812-120000/data"
+    result = resolve_path(path, cwd)
+    assert result == expected
+    assert result.is_absolute()

tests/unit/config/test_resolve_timestamp_path.py

@@ -1,33 +0,0 @@
-# Copyright (c) 2024 Microsoft Corporation.
-# Licensed under the MIT License
-
-from pathlib import Path
-
-from graphrag.config.resolve_timestamp_path import resolve_timestamp_path
-
-
-def test_resolve_timestamp_path_no_timestamp_with_run_id():
-    path = Path("path/to/data")
-    result = resolve_timestamp_path(path, "20240812-121000")
-    assert result == path
-
-
-def test_resolve_timestamp_path_no_timestamp_without_run_id():
-    path = Path("path/to/data")
-    result = resolve_timestamp_path(path)
-    assert result == path
-
-
-def test_resolve_timestamp_path_with_timestamp_and_run_id():
-    path = Path("some/path/${timestamp}/data")
-    expected = Path("some/path/20240812/data")
-    result = resolve_timestamp_path(path, "20240812")
-    assert result == expected
-
-
-def test_resolve_timestamp_path_with_timestamp_and_inferred_directory():
-    cwd = Path(__file__).parent
-    path = cwd / "fixtures/timestamp_dirs/${timestamp}/data"
-    expected = cwd / "fixtures/timestamp_dirs/20240812-120000/data"
-    result = resolve_timestamp_path(path)
-    assert result == expected