feat(integrations): Add integration for qdrant

sentry_sdk/consts.py

@@ -426,6 +426,8 @@ class OP:
     COHERE_CHAT_COMPLETIONS_CREATE = "ai.chat_completions.create.cohere"
     COHERE_EMBEDDINGS_CREATE = "ai.embeddings.create.cohere"
     DB = "db"
+    DB_QDRANT_GRPC = "db.qdrant.grpc"
+    DB_QDRANT_REST = "db.qdrant.rest"
     DB_REDIS = "db.redis"
     EVENT_DJANGO = "event.django"
     FUNCTION = "function"

sentry_sdk/integrations/qdrant/__init__.py

@@ -0,0 +1,40 @@
+from sentry_sdk.integrations import DidNotEnable
+
+try:
+    from qdrant_client.http import ApiClient, AsyncApiClient
+    import grpc
+except ImportError:
+    raise DidNotEnable("Qdrant client not installed")
+
+from sentry_sdk.integrations import Integration
+from sentry_sdk.integrations.qdrant.consts import _IDENTIFIER
+from sentry_sdk.integrations.qdrant.qdrant import (
+    _sync_api_client_send_inner,
+    _async_api_client_send_inner,
+    _wrap_channel_sync,
+    _wrap_channel_async,
+)
+
+
+class QdrantIntegration(Integration):
+    identifier = _IDENTIFIER
+
+    def __init__(self, mute_children_http_spans=True):
+        # type: (bool) -> None
+        self.mute_children_http_spans = mute_children_http_spans
+
+    @staticmethod
+    def setup_once():
+        # type: () -> None
+
+        # hooks for the REST client
+        ApiClient.send_inner = _sync_api_client_send_inner(ApiClient.send_inner)
+        AsyncApiClient.send_inner = _async_api_client_send_inner(
+            AsyncApiClient.send_inner
+        )
+
+        # hooks for the gRPC client
+        grpc.secure_channel = _wrap_channel_sync(grpc.secure_channel)
+        grpc.insecure_channel = _wrap_channel_sync(grpc.insecure_channel)
+        grpc.aio.secure_channel = _wrap_channel_async(grpc.aio.secure_channel)
+        grpc.aio.insecure_channel = _wrap_channel_async(grpc.aio.insecure_channel)

sentry_sdk/integrations/qdrant/consts.py

@@ -0,0 +1,121 @@
+from sentry_sdk.integrations.qdrant.path_matching import PathTrie
+
+SPAN_ORIGIN = "auto.db.qdrant"
+
+# created from https://github.com/qdrant/qdrant/blob/master/docs/redoc/v1.11.x/openapi.json
+# only used for qdrants REST API. gRPC is using other identifiers
+_PATH_TO_OPERATION_ID = {
+    "/collections/{collection_name}/shards": {"put": "create_shard_key"},
+    "/collections/{collection_name}/shards/delete": {"post": "delete_shard_key"},
+    "/": {"get": "root"},
+    "/telemetry": {"get": "telemetry"},
+    "/metrics": {"get": "metrics"},
+    "/locks": {"post": "post_locks", "get": "get_locks"},
+    "/healthz": {"get": "healthz"},
+    "/livez": {"get": "livez"},
+    "/readyz": {"get": "readyz"},
+    "/issues": {"get": "get_issues", "delete": "clear_issues"},
+    "/cluster": {"get": "cluster_status"},
+    "/cluster/recover": {"post": "recover_current_peer"},
+    "/cluster/peer/{peer_id}": {"delete": "remove_peer"},
+    "/collections": {"get": "get_collections"},
+    "/collections/{collection_name}": {
+        "get": "get_collection",
+        "put": "create_collection",
+        "patch": "update_collection",
+        "delete": "delete_collection",
+    },
+    "/collections/aliases": {"post": "update_aliases"},
+    "/collections/{collection_name}/index": {"put": "create_field_index"},
+    "/collections/{collection_name}/exists": {"get": "collection_exists"},
+    "/collections/{collection_name}/index/{field_name}": {
+        "delete": "delete_field_index"
+    },
+    "/collections/{collection_name}/cluster": {
+        "get": "collection_cluster_info",
+        "post": "update_collection_cluster",
+    },
+    "/collections/{collection_name}/aliases": {"get": "get_collection_aliases"},
+    "/aliases": {"get": "get_collections_aliases"},
+    "/collections/{collection_name}/snapshots/upload": {
+        "post": "recover_from_uploaded_snapshot"
+    },
+    "/collections/{collection_name}/snapshots/recover": {
+        "put": "recover_from_snapshot"
+    },
+    "/collections/{collection_name}/snapshots": {
+        "get": "list_snapshots",
+        "post": "create_snapshot",
+    },
+    "/collections/{collection_name}/snapshots/{snapshot_name}": {
+        "delete": "delete_snapshot",
+        "get": "get_snapshot",
+    },
+    "/snapshots": {"get": "list_full_snapshots", "post": "create_full_snapshot"},
+    "/snapshots/{snapshot_name}": {
+        "delete": "delete_full_snapshot",
+        "get": "get_full_snapshot",
+    },
+    "/collections/{collection_name}/shards/{shard_id}/snapshots/upload": {
+        "post": "recover_shard_from_uploaded_snapshot"
+    },
+    "/collections/{collection_name}/shards/{shard_id}/snapshots/recover": {
+        "put": "recover_shard_from_snapshot"
+    },
+    "/collections/{collection_name}/shards/{shard_id}/snapshots": {
+        "get": "list_shard_snapshots",
+        "post": "create_shard_snapshot",
+    },
+    "/collections/{collection_name}/shards/{shard_id}/snapshots/{snapshot_name}": {
+        "delete": "delete_shard_snapshot",
+        "get": "get_shard_snapshot",
+    },
+    "/collections/{collection_name}/points/{id}": {"get": "get_point"},
+    "/collections/{collection_name}/points": {
+        "post": "get_points",
+        "put": "upsert_points",
+    },
+    "/collections/{collection_name}/points/delete": {"post": "delete_points"},
+    "/collections/{collection_name}/points/vectors": {"put": "update_vectors"},
+    "/collections/{collection_name}/points/vectors/delete": {"post": "delete_vectors"},
+    "/collections/{collection_name}/points/payload": {
+        "post": "set_payload",
+        "put": "overwrite_payload",
+    },
+    "/collections/{collection_name}/points/payload/delete": {"post": "delete_payload"},
+    "/collections/{collection_name}/points/payload/clear": {"post": "clear_payload"},
+    "/collections/{collection_name}/points/batch": {"post": "batch_update"},
+    "/collections/{collection_name}/points/scroll": {"post": "scroll_points"},
+    "/collections/{collection_name}/points/search": {"post": "search_points"},
+    "/collections/{collection_name}/points/search/batch": {
+        "post": "search_batch_points"
+    },
+    "/collections/{collection_name}/points/search/groups": {
+        "post": "search_point_groups"
+    },
+    "/collections/{collection_name}/points/recommend": {"post": "recommend_points"},
+    "/collections/{collection_name}/points/recommend/batch": {
+        "post": "recommend_batch_points"
+    },
+    "/collections/{collection_name}/points/recommend/groups": {
+        "post": "recommend_point_groups"
+    },
+    "/collections/{collection_name}/points/discover": {"post": "discover_points"},
+    "/collections/{collection_name}/points/discover/batch": {
+        "post": "discover_batch_points"
+    },
+    "/collections/{collection_name}/points/count": {"post": "count_points"},
+    "/collections/{collection_name}/points/query": {"post": "query_points"},
+    "/collections/{collection_name}/points/query/batch": {"post": "query_batch_points"},
+    "/collections/{collection_name}/points/query/groups": {
+        "post": "query_points_groups"
+    },
+}
+
+_DISALLOWED_PROTO_FIELDS = {"data", "keyword"}
+
+_DISALLOWED_REST_FIELDS = {"nearest", "value"}
+
+_IDENTIFIER = "qdrant"
+
+_qdrant_trie = PathTrie(_PATH_TO_OPERATION_ID)

sentry_sdk/integrations/qdrant/path_matching.py

@@ -0,0 +1,144 @@
+from typing import Any, Dict, Optional, List
+
+
+class TrieNode:
+    def __init__(self, is_placeholder=False):
+        """
+        Initializes a TrieNode.
+
+        :param is_placeholder: Indicates if this node represents a placeholder (wildcard).
+        """
+        self.children = {}  # type: Dict[str, 'TrieNode']
+        self.operation_ids = {}  # type: Dict[str, str]
+        self.is_placeholder = is_placeholder  # type: bool
+
+    @classmethod
+    def from_dict(cls, data, parent_path=""):
+        # type: (Dict[str, Any], str) -> 'TrieNode'
+        """
+        Recursively constructs a TrieNode from a nested dictionary.
+
+        :param data: Nested dictionary mapping path segments to either nested dictionaries
+                     or dictionaries of HTTP methods to operation IDs.
+        :param parent_path: The accumulated path from the root to the current node.
+        :return: Root TrieNode of the constructed trie.
+        """
+        node = cls()
+        for path, methods in data.items():
+            segments = PathTrie.split_path(path)
+            current = node
+            for segment in segments:
+                is_placeholder = segment.startswith("{") and segment.endswith("}")
+                key = "*" if is_placeholder else segment
+
+                if key not in current.children:
+                    current.children[key] = TrieNode(is_placeholder=is_placeholder)
+                current = current.children[key]
+
+            if isinstance(methods, dict):
+                for method, operation_id in methods.items():
+                    current.operation_ids[method.lower()] = operation_id
+
+        return node
+
+    def to_dict(self, current_path=""):
+        # type: (str) -> Dict[str, Any]
+        """
+        Serializes the TrieNode and its children back to a nested dictionary.
+
+        :param current_path: The accumulated path from the root to the current node.
+        :return: Nested dictionary representing the trie.
+        """
+        result = {}  # type: Dict[str, Any]
+        if self.operation_ids:
+            path_key = current_path or "/"
+            result[path_key] = self.operation_ids.copy()
+
+        for segment, child in self.children.items():
+            # replace wildcard '*' back to placeholder format if necessary.
+            # allows for TrieNode.from_dict(TrieNode.to_dict()) to be idempotent.
+            display_segment = "{placeholder}" if child.is_placeholder else segment
+            new_path = (
+                f"{current_path}/{display_segment}"
+                if current_path
+                else f"/{display_segment}"
+            )
+            child_dict = child.to_dict(new_path)
+            result.update(child_dict)
+
+        return result
+
+
+class PathTrie:
+    WILDCARD = "*"  # type: str
+
+    def __init__(self, data=None):
+        # type: (Optional[Dict[str, Any]]) -> None
+        """
+        Initializes the PathTrie with optional initial data.
+
+        :param data: Optional nested dictionary to initialize the trie.
+        """
+        self.root = TrieNode.from_dict(data or {})  # type: TrieNode
+
+    def insert(self, path, method, operation_id):
+        # type: (str, str, str) -> None
+        """
+        Inserts a path into the trie with its corresponding HTTP method and operation ID.
+
+        :param path: The API path (e.g., '/users/{user_id}/posts').
+        :param method: HTTP method (e.g., 'GET', 'POST').
+        :param operation_id: The operation identifier associated with the path and method.
+        """
+        current = self.root
+        segments = self.split_path(path)
+
+        for segment in segments:
+            is_placeholder = self._is_placeholder(segment)
+            key = self.WILDCARD if is_placeholder else segment
+
+            if key not in current.children:
+                current.children[key] = TrieNode(is_placeholder=is_placeholder)
+            current = current.children[key]
+
+        current.operation_ids[method.lower()] = operation_id
+
+    def match(self, path, method):
+        # type: (str, str) -> Optional[str]
+        """
+        Matches a given path and HTTP method to its corresponding operation ID.
+
+        :param path: The API path to match.
+        :param method: HTTP method to match.
+        :return: The operation ID if a match is found; otherwise, None.
+        """
+        current = self.root
+        segments = self.split_path(path)
+
+        for segment in segments:
+            if segment in current.children:
+                current = current.children[segment]
+            elif self.WILDCARD in current.children:
+                current = current.children[self.WILDCARD]
+            else:
+                return None
+
+        return current.operation_ids.get(method.lower())
+
+    def to_dict(self):
+        # type: () -> Dict[str, Any]
+        return self.root.to_dict()
+
+    @staticmethod
+    def split_path(path):
+        # type: (str) -> List[str]
+        return [segment for segment in path.strip("/").split("/") if segment]
+
+    @staticmethod
+    def _is_placeholder(segment):
+        # type: (str) -> bool
+        return segment.startswith("{") and segment.endswith("}")
+
+    def __repr__(self):
+        # type: () -> str
+        return f"PathTrie({self.to_dict()})"