[DE-558] ArangoSearch column cache

CHANGELOG.md

@@ -1,6 +1,8 @@
 main
 ----
 
+* Added cache and primaryKeyCache parameters to the inverted index API.
+
 * Added allow_retry query parameter, making it possible to retry fetching the latest batch from a cursor.
 
 * Added OverloadControlDatabase, enabling the client to react effectively to potential server overloads.

arango/collection.py

@@ -1342,6 +1342,8 @@ def add_inverted_index(
         includeAllFields: Optional[bool] = None,
         trackListPositions: Optional[bool] = None,
         searchField: Optional[bool] = None,
+        primaryKeyCache: Optional[bool] = None,
+        cache: Optional[bool] = None,
     ) -> Result[Json]:
         """Create a new inverted index, introduced in version 3.10.
 
@@ -1351,22 +1353,31 @@ def add_inverted_index(
         :type name: str | None
         :param inBackground: Do not hold the collection lock.
         :type inBackground: bool | None
-        :param parallelism:
+        :param parallelism: The number of threads to use for indexing the fields.
         :type parallelism: int | None
-        :param primarySort:
-        :type primarySort: Json | None
-        :param storedValues:
+        :param primarySort: Primary sort order to enable an AQL optimization.
+        :type primarySort: Optional[Json]
+        :param storedValues: An array of objects with paths to additional
+            attributes to store in the index.
         :type storedValues: Sequence[Json] | None
-        :param analyzer:
-        :type analyzer: str | None
-        :param features:
+        :param analyzer: Analyzer to use by default.
+        :type analyzer: Optional[str]
+        :param features: List of Analyzer features.
         :type features: Sequence[str] | None
-        :param includeAllFields:
+        :param includeAllFields: This option only applies if you use the
+            inverted index in search-alias views.
         :type includeAllFields: bool | None
-        :param trackListPositions:
+        :param trackListPositions: This option only applies if you use the
+            inverted index in search-alias views, and searchField is true.
         :type trackListPositions: bool | None
-        :param searchField:
+        :param searchField: This option only applies if you use the inverted
+            index in search-alias views
         :type searchField: bool | None
+        :param primaryKeyCache: Always cache the primary key column in memory.
+        :type primaryKeyCache: bool | None
+        :param cache: Always cache the field normalization values in memory
+            for all fields by default.
+        :type cache: bool | None
         :return: New index details.
         :rtype: dict
         :raise arango.exceptions.IndexCreateError: If create fails.
@@ -1395,6 +1406,10 @@ def add_inverted_index(
             data["searchField"] = searchField
         if fields is not None:
             data["fields"] = fields
+        if primaryKeyCache is not None:
+            data["primaryKeyCache"] = primaryKeyCache
+        if cache is not None:
+            data["cache"] = cache
 
         return self._add_index(data)
 

arango/formatter.py

@@ -67,6 +67,40 @@ def format_index(body: Json) -> Json:
         result["legacyPolygons"] = body["legacyPolygons"]
     if "estimates" in body:
         result["estimates"] = body["estimates"]
+    if "analyzer" in body:
+        result["analyzer"] = body["analyzer"]
+    if "cleanupIntervalStep" in body:
+        result["cleanup_interval_step"] = body["cleanupIntervalStep"]
+        if "commitIntervalMsec" in body:
+            result["commit_interval_msec"] = body["commitIntervalMsec"]
+    if "consolidationIntervalMsec" in body:
+        result["consolidation_interval_msec"] = body["consolidationIntervalMsec"]
+    if "consolidationPolicy" in body:
+        result["consolidation_policy"] = format_view_consolidation_policy(
+            body["consolidationPolicy"]
+        )
+    if "features" in body:
+        result["features"] = body["features"]
+    if "includeAllFields" in body:
+        result["include_all_fields"] = body["includeAllFields"]
+    if "primarySort" in body:
+        result["primary_sort"] = body["primarySort"]
+    if "searchField" in body:
+        result["search_field"] = body["searchField"]
+    if "trackListPositions" in body:
+        result["track_list_positions"] = body["trackListPositions"]
+    if "version" in body:
+        result["version"] = body["version"]
+    if "cache" in body:
+        result["cache"] = body["cache"]
+    if "primaryKeyCache" in body:
+        result["primaryKeyCache"] = body["primaryKeyCache"]
+    if "writebufferIdle" in body:
+        result["writebuffer_idle"] = body["writebufferIdle"]
+    if "writebufferActive" in body:
+        result["writebuffer_active"] = body["writebufferActive"]
+    if "writebufferSizeMax" in body:
+        result["writebuffer_max_size"] = body["writebufferSizeMax"]
 
     return verify_format(body, result)
 
@@ -902,6 +936,12 @@ def format_view(body: Json) -> Json:
     if "indexes" in body:
         result["indexes"] = body["indexes"]
 
+    # Introduced in 3.9.6 EE
+    if "primaryKeyCache" in body:
+        result["primaryKeyCache"] = body["primaryKeyCache"]
+    if "primarySortCache" in body:
+        result["primarySortCache"] = body["primarySortCache"]
+
     # Introduced in 3.12 EE
     if "optimizeTopK" in body:
         result["optimizeTopK"] = body["optimizeTopK"]

tests/test_index.py

@@ -1,3 +1,6 @@
+import pytest
+from packaging import version
+
 from arango.exceptions import (
     IndexCreateError,
     IndexDeleteError,
@@ -197,6 +200,34 @@ def test_add_ttl_index(icol):
     icol.delete_index(result["id"])
 
 
+def test_add_inverted_index(icol, enterprise, db_version):
+    if db_version < version.parse("3.10.0"):
+        pytest.skip("Inverted indexes are not supported before 3.10.0")
+
+    parameters = dict(
+        fields=[{"name": "attr1", "cache": True}],
+        name="c0_cached",
+        storedValues=[{"fields": ["a"], "compression": "lz4", "cache": True}],
+        includeAllFields=True,
+        analyzer="identity",
+        primarySort={"cache": True, "fields": [{"field": "a", "direction": "asc"}]},
+    )
+    expected_keys = ["primary_sort", "analyzer", "include_all_fields", "search_field"]
+
+    if enterprise and db_version >= version.parse("3.10.2"):
+        parameters["cache"] = True
+        parameters["primaryKeyCache"] = True
+        expected_keys.extend(["cache", "primaryKeyCache"])
+
+    result = icol.add_inverted_index(**parameters)
+    assert result["id"] in extract("id", icol.indexes())
+
+    for key in expected_keys:
+        assert key in result
+
+    icol.delete_index(result["id"])
+
+
 def test_delete_index(icol, bad_col):
     old_indexes = set(extract("id", icol.indexes()))
     icol.add_hash_index(["attr3", "attr4"], unique=True)

tests/test_view.py

@@ -189,6 +189,8 @@ def test_arangosearch_view_properties(db, col, enterprise, db_version):
             }
         )
 
+        if db_version >= version.parse("3.9.6"):
+            params.update({"primarySortCache": True, "primaryKeyCache": True})
         if db_version >= version.parse("3.10.3"):
             params.update({"storedValues": ["attr1", "attr2"]})