Merge pull request #388 from pawan-privado/pawan/ai-vecdb

Add support for AI and vector DB SDKs

rules/sinks/storages/astradb/java.yml

@@ -0,0 +1,13 @@
+
+#  Sink rule for ThirdParty SDK
+#  The id follows a format : "ThirdParties.SDK.<THIRD_PARTY_ORGANISATION>.<SUB_ORGANISATION_IF_APPLICABLE>"
+
+sinks:
+
+  - id: ThirdParties.SDK.AstraDB
+    name: AstraDB
+    domains:
+      - "docs.datastax.com"
+    patterns:
+      - "(?i)(com[.]datastax[.]astra).*"
+    tags:

rules/sinks/storages/astradb/javascript.yml

@@ -0,0 +1,13 @@
+
+#  Sink rule for ThirdParty SDK
+#  The id follows a format : "ThirdParties.SDK.<THIRD_PARTY_ORGANISATION>.<SUB_ORGANISATION_IF_APPLICABLE>"
+
+sinks:
+
+  - id: ThirdParties.SDK.AstraDB
+    name: AstraDB
+    domains:
+      - "datastax.com"
+    patterns:
+      - "@datastax\\/astra-db-ts"
+    tags:

rules/sinks/storages/astradb/python.yml

@@ -0,0 +1,15 @@
+sinks:
+  - id: Storages.AstraDB.Initialize
+    name: AstraDB DB
+    domains:
+      - datastax.com
+    patterns:
+      - "(?i).*(?:langchain.*vectorstores.*.AstraDB).*"
+
+  - id: Storages.AstraDB.ReadAndWrite
+    name: Astra DB
+    domains:
+      - datastax.com
+    patterns:
+      - (?i).*(astrapy).*(?:create_collection|insert_many|vector_find).*
+    tags:

rules/sinks/storages/chromadb/javascript.yml

@@ -0,0 +1,12 @@
+
+# Sink Rules for storage database ChromaDB - trychroma.com
+
+sinks:
+
+  - id: Storages.Chroma.ReadAndWrite
+    name: Chroma VectorDB
+    domains:
+      - trychroma.com
+    patterns:
+      - "(?i)(chromadb).*"
+    tags:

rules/sinks/storages/chromadb/python.yml

@@ -0,0 +1,33 @@
+
+# Sink Rules for vector DB ChromaDB
+
+sinks:
+  - id: Storages.Chroma.Initialize
+    name: Chroma DB
+    domains:
+      - trychroma.com
+    patterns:
+      - "(?i).*(?:langchain.*vectorstores.*.Chroma).*"
+
+  - id: Storages.Chroma.Client
+    name: Chroma DB
+    domains:
+      - trychroma.com
+    patterns:
+      - "(?i).*(?:chromadb).*(?:HttpClient).*"
+
+  - id: Storages.Chroma.Read
+    name: Chroma DB (Read)
+    domains:
+      - trychroma.com
+    patterns:
+      - "(?i).*(?:chromadb).*(?:search|query|get_collection).*"
+    tags:
+
+  - id: Storages.Chroma.Write
+    name: Chroma DB (Write)
+    domains:
+      - trychroma.com
+    patterns:
+      - "(?i).*(?:chromadb).*(?:add|insert|do_bulk_insert|update|upsert|create_collection).*"
+    tags:

rules/sinks/storages/cosmosdb/python.yaml

@@ -2,6 +2,13 @@
 # Sink Rules for storage database Azure Cosmos DB - https://azure.microsoft.com/en-us/services/cosmos-db/
 
 sinks:
+  - id: Storages.AzureCosmosDb.Initialize
+    name: Azure Cosmos DB
+    domains:
+      - azure.microsoft.com
+      - microsoft.com
+    patterns:
+      - "(?i).*(?:langchain.*vectorstores.*.AzureCosmosDBVectorSearch).*"
 
   - id: Storages.AzureCosmosDb
     name: Azure Cosmos DB
@@ -10,4 +17,4 @@ sinks:
       - microsoft.com
     patterns:
       - "(?i).*(?:azure.*cosmos).*(?:create.{0,1}item|read.{0,1}item|delete.{0,1}item|query).*"
-    tags:
+    tags:

rules/sinks/storages/deeplake/python.yml

@@ -0,0 +1,18 @@
+
+# Sink Rules for vector DB Deeplake
+
+sinks:
+  - id: Storages.Deeplake.Initialize
+    name: Deeplake DB
+    domains:
+      - activeloop.ai
+    patterns:
+      - "(?i).*(?:langchain.*vectorstores.*.DeepLake).*"
+
+  - id: Storages.Deeplake.ReadAndWrite
+    name: Deeplake DB
+    domains:
+      - activeloop.ai
+    patterns:
+      - (?i).*(deeplake).*(?:load|add|search).*
+    tags:

rules/sinks/storages/elasticsearch/python.yaml

@@ -2,6 +2,12 @@
 # Sink Rules for storage database Elasticsearch - https://www.elastic.co/elasticsearch/
 
 sinks:
+  - id: Storages.Elasticsearch.Initialize
+    name: Elasticsearch(Initialize)
+    domains:
+      - elastic.co
+    patterns:
+      - "(?i).*(?:langchain.*vectorstores.*.ElasticsearchStore).*"
 
   - id: Storages.Elasticsearch.Read
     name: Elasticsearch(Read)

rules/sinks/storages/lance/python.yml

@@ -0,0 +1,26 @@
+
+# Sink Rules for vector DB ChromaDB
+
+sinks:
+  - id: Storages.Lance.Initialize
+    name: Lance DB
+    domains:
+      - https://lancedb.com/
+    patterns:
+      - "(?i).*(?:langchain.*vectorstores.*.LanceDB).*"
+
+  - id: Storages.Lance.Read
+    name: Lance DB (Write)
+    domains:
+      - https://lancedb.com/
+    patterns:
+      - "(?i).*(?:lancedb.db).*(?:open_table|search).*"
+    tags:
+
+  - id: Storages.Lance.Write
+    name: Lance DB (Write)
+    domains:
+      - https://lancedb.com/
+    patterns:
+      - "(?i).*(?:lancedb.db).*(?:add|create_table|create_index|delete|drop_table).*"
+    tags:

rules/sinks/storages/milvus/java.yml

@@ -0,0 +1,13 @@
+
+#  Sink rule for ThirdParty SDK
+#  The id follows a format : "ThirdParties.SDK.<THIRD_PARTY_ORGANISATION>.<SUB_ORGANISATION_IF_APPLICABLE>"
+
+sinks:
+
+  - id: ThirdParties.SDK.Milvus
+    name: Milvus
+    domains:
+      - "milvus.io"
+    patterns:
+      - "(?i)(io[.]milvus).*"
+    tags:

rules/sinks/storages/milvus/javascript.yml

@@ -0,0 +1,13 @@
+
+#  Sink rule for ThirdParty SDK
+#  The id follows a format : "ThirdParties.SDK.<THIRD_PARTY_ORGANISATION>.<SUB_ORGANISATION_IF_APPLICABLE>"
+
+sinks:
+
+  - id: ThirdParties.SDK.Milvus
+    name: Milvus
+    domains:
+      - "milvus.io"
+    patterns:
+      - "@zilliz\\/milvus2-sdk-node"
+    tags:

rules/sinks/storages/milvus/python.yml

@@ -0,0 +1,26 @@
+
+# Sink Rules for vector DB Milvus
+
+sinks:
+  - id: Storages.Milvus.Initialize
+    name: Milvus DB
+    domains:
+      - milvus.io
+    patterns:
+      - "(?i).*(?:langchain.*vectorstores.*.Milvus).*"
+
+  - id: Storages.Milvus.Read
+    name: Milvus DB (Read)
+    domains:
+      - milvus.io
+    patterns:
+      - "(?i).*(?:pymilvus).*(?:search|query).*"
+    tags:
+
+  - id: Storages.Milvus.Write
+    name: Milvus DB (Write)
+    domains:
+      - milvus.io
+    patterns:
+      - "(?i).*(?:pymilvus).*(?:insert|do_bulk_insert|upsert).*"
+    tags:

rules/sinks/storages/mongodb/python.yaml

@@ -2,6 +2,12 @@
 # Sink Rules for storage database MongoDB - https://www.mongodb.com/
 
 sinks:
+  - id: Storages.MongoDB.Initialize
+    name: MongoDB DB
+    domains:
+      - mongodb.com
+    patterns:
+      - "(?i).*(?:langchain.*vectorstores.*.MongoDBAtlasVectorSearch).*"
 
   - id: Storages.MongoDB.Read
     name: MongoDB(Read)
@@ -19,4 +25,4 @@ sinks:
     patterns:
       - "(?i).*(pymongo|flask_pymongo|motor.*MotorClient).*(?:update|replace)one.*"
       - "(?i).*(pymongo|flask_pymongo|motor.*MotorClient).*(?:insert|delete|write|upload|remove|update|bulk|drop).*"
-    tags:
+    tags:

rules/sinks/storages/opensearch/python.yml

@@ -0,0 +1,10 @@
+
+# Sink Rules for vector DB Opensearch
+
+sinks:
+  - id: Storages.OpenSearch.Initialize
+    name: OpenSearch DB
+    domains:
+      - https://opensearch.org/
+    patterns:
+      - "(?i).*(?:langchain.*vectorstores.*.OpenSearchVectorSearch).*"

rules/sinks/storages/pinecone/javascript.yml

@@ -0,0 +1,12 @@
+
+# Sink Rules for storage database Pinecone - pinecone.io
+
+sinks:
+
+  - id: Storages.Pinecone.ReadAndWrite
+    name: Pinecone VectorDB
+    domains:
+      - pinecone.io
+    patterns:
+      - "(?i)(pinecone-database|pinecone).*"
+    tags:

rules/sinks/storages/pinecone/python.yml

@@ -0,0 +1,26 @@
+
+# Sink Rules for vector DB Pinecone
+
+sinks:
+  - id: Storages.Pinecone.Initialize
+    name: Pinecone DB
+    domains:
+      - pinecone.io
+    patterns:
+      - "(?i).*(?:langchain.*.Pinecone).*"
+
+  - id: Storages.Pinecone.Read
+    name: Pinecone DB (Read)
+    domains:
+      - pinecone.io
+    patterns:
+      - "(?i).*(?:pinecone).*(?:fetch|query).*"
+    tags:
+
+  - id: Storages.Pinecone.Write
+    name: Pinecone DB (Write)
+    domains:
+      - pinecone.io
+    patterns:
+      - "(?i).*(?:pinecone).*(?:upsert|update).*"
+    tags:

rules/sinks/storages/qdrant/javascript.yml

@@ -0,0 +1,12 @@
+
+# Sink Rules for storage database Qdrant - qdrant.tech
+
+sinks:
+
+  - id: Storages.Qdrant.ReadAndWrite
+    name: Qdrant DB
+    domains:
+      - qdrant.tech
+    patterns:
+      - "@qdrant\\/js-client-rest"
+    tags:

rules/sinks/storages/qdrant/python.yml

@@ -0,0 +1,26 @@
+
+# Sink Rules for vector DB Qdrant
+
+sinks:
+  - id: Storages.Qdrant.Initialize
+    name: Qdrant DB
+    domains:
+      - qdrant.tech
+    patterns:
+      - "(?i).*(?:langchain.*vectorstores.*.Qdrant).*"
+
+  - id: Storages.Qdrant.Read
+    name: Qdrant DB (Read)
+    domains:
+      - qdrant.tech
+    patterns:
+      - "(?i).*(?:qdrant_client).*(?:query|search).*"
+    tags:
+
+  - id: Storages.Qdrant.Write
+    name: Qdrant DB (Write)
+    domains:
+      - qdrant.tech
+    patterns:
+      - "(?i).*(?:qdrant_client).*(?:upsert|add).*"
+    tags:

rules/sinks/storages/redis/python.yaml

@@ -1,8 +1,15 @@
 sinks:
+  - id: Storages.Redis.Initialize
+    name: Redis DB
+    domains:
+      - redis.io
+    patterns:
+      - "(?i).*(?:langchain.*vectorstores.*.Redis).*"
+
   - id: Storages.Redis.Read
     name: Redis DB
     domains:
       - redis.io
     patterns:
       - (?i).*(redisdb).*
-    tags:
+    tags:

rules/sinks/storages/vald/java.yml

@@ -0,0 +1,13 @@
+
+#  Sink rule for ThirdParty SDK
+#  The id follows a format : "ThirdParties.SDK.<THIRD_PARTY_ORGANISATION>.<SUB_ORGANISATION_IF_APPLICABLE>"
+
+sinks:
+
+  - id: ThirdParties.SDK.Vald
+    name: Vald
+    domains:
+      - "vald.vdaas.org"
+    patterns:
+      - "(?i)(org[.]vdaas[.]vald).*"
+    tags:

rules/sinks/storages/vald/javascript.yml

@@ -0,0 +1,13 @@
+
+#  Sink rule for ThirdParty SDK
+#  The id follows a format : "ThirdParties.SDK.<THIRD_PARTY_ORGANISATION>.<SUB_ORGANISATION_IF_APPLICABLE>"
+
+sinks:
+
+  - id: ThirdParties.SDK.Vald
+    name: Vald
+    domains:
+      - "vald.vdaas.org"
+    patterns:
+      - "(?i)(vald-client-node).*"
+    tags:

rules/sinks/storages/vald/python.yml

@@ -0,0 +1,15 @@
+sinks:
+  - id: Storages.Vald.Initialize
+    name: Vald DB
+    domains:
+      - vald.vdaas.org
+    patterns:
+      - "(?i).*(?:langchain.*vectorstores.*.Vald).*"
+
+  - id: Storages.Vald.ReadAndWrite
+    name: Vald DB
+    domains:
+      - vald.vdaas.org
+    patterns:
+      - (?i).*(vald).*(?:Insert|Update|Upsert|Search|Remove).*
+    tags: