Allow clients to tag requests with a source_tag (#324)

## Problem
Need to allow clients to include a `source_tag` to identify the source
of their requests.

## Solution
Allow clients to specify a `source_tag` field in the client constructor,
that will be used to identify the traffic source, if applicable.

Example:
```python
from pinecone import Pinecone

pc = Pinecone(api_key='foo', source_tag='bar')

pc.list_indexes()
```

This would cause the user-agent to get a value like:
```
User-Agent: 'python-client-3.1.0 (urllib3:2.0.7); source_tag=bar'
```

## Type of Change

- [x] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing
functionality to not work as expected)

## Testing

- [ ] Tests are passing
- [ ] Verify source_tag included in user-agent in control plane and data
plane (REST and gRPC)

Author: ssmith-pc

pinecone/config/config.py

@@ -5,6 +5,7 @@
 from pinecone.config.openapi import OpenApiConfigFactory
 from pinecone.core.client.configuration import Configuration as OpenApiConfiguration
 from pinecone.utils import normalize_host
+from pinecone.utils.constants import SOURCE_TAG
 
 class Config(NamedTuple):
     api_key: str = ""
@@ -14,6 +15,7 @@ class Config(NamedTuple):
     ssl_ca_certs: Optional[str] = None
     ssl_verify: Optional[bool] = None
     additional_headers: Optional[Dict[str, str]] = {}
+    source_tag: Optional[str] = None
 
 class ConfigBuilder:
     """
@@ -46,13 +48,14 @@ def build(
         api_key = api_key or kwargs.pop("api_key", None) or os.getenv("PINECONE_API_KEY")
         host = host or kwargs.pop("host", None)
         host = normalize_host(host)
+        source_tag = kwargs.pop(SOURCE_TAG, None)
 
         if not api_key:
             raise PineconeConfigurationError("You haven't specified an Api-Key.")
         if not host:
             raise PineconeConfigurationError("You haven't specified a host.")
 
-        return Config(api_key, host, proxy_url, proxy_headers, ssl_ca_certs, ssl_verify, additional_headers)
+        return Config(api_key, host, proxy_url, proxy_headers, ssl_ca_certs, ssl_verify, additional_headers, source_tag)
 
     @staticmethod
     def build_openapi_config(

pinecone/control/pinecone.py

@@ -633,5 +633,6 @@ def Index(self, name: str = '', host: str = '', **kwargs):
             api_key=api_key,
             pool_threads=pt,
             openapi_config=openapi_config,
+            source_tag=self.config.source_tag,
             **kwargs
         )

pinecone/grpc/base.py

@@ -14,6 +14,7 @@
 from .utils import _generate_request_id
 from .config import GRPCClientConfig
 from pinecone.utils.constants import MAX_MSG_SIZE, REQUEST_ID, CLIENT_VERSION
+from pinecone.utils.user_agent import get_user_agent_grpc
 from pinecone.exceptions import PineconeException
 
 _logger = logging.getLogger(__name__)
@@ -77,7 +78,8 @@ def __init__(
             }
         )
 
-        self._channel = channel or self._gen_channel()
+        options = {"grpc.primary_user_agent": get_user_agent_grpc(config)}
+        self._channel = channel or self._gen_channel(options=options)
         self.stub = self.stub_class(self._channel)
 
     @property

pinecone/grpc/pinecone.py

@@ -120,13 +120,10 @@ def Index(self, name: str = '', host: str = '', **kwargs):
         if name == '' and host == '':
             raise ValueError("Either name or host must be specified")
 
-        if host != '':
-            # Use host if it is provided
-            config = ConfigBuilder.build(api_key=self.config.api_key, host=host)
-            return GRPCIndex(index_name=name, config=config, **kwargs)
-
-        if name != '':
-            # Otherwise, get host url from describe_index using the index name
-            index_host = self.index_host_store.get_host(self.index_api, self.config, name)
-            config = ConfigBuilder.build(api_key=self.config.api_key, host=index_host)
-            return GRPCIndex(index_name=name, config=config, **kwargs)
+        # Use host if it is provided, otherwise get host from describe_index
+        index_host = host or self.index_host_store.get_host(self.index_api, self.config, name)
+
+        config = ConfigBuilder.build(api_key=self.config.api_key,
+                                     host=index_host,
+                                     source_tag=self.config.source_tag)
+        return GRPCIndex(index_name=name, config=config, **kwargs)

pinecone/utils/constants.py

@@ -28,3 +28,5 @@ class NodeType(str, enum.Enum):
 
 REQUIRED_VECTOR_FIELDS = {"id", "values"}
 OPTIONAL_VECTOR_FIELDS = {"sparse_values", "metadata"}
+
+SOURCE_TAG = "source_tag"

pinecone/utils/setup_openapi_client.py

@@ -6,7 +6,7 @@ def setup_openapi_client(api_klass, config, openapi_config, pool_threads):
         configuration=openapi_config, 
         pool_threads=pool_threads
     )
-    api_client.user_agent = get_user_agent()
+    api_client.user_agent = get_user_agent(config)
     extra_headers = config.additional_headers or {}
     for key, value in extra_headers.items():
         api_client.set_default_header(key, value)

pinecone/utils/user_agent.py

@@ -1,9 +1,29 @@
 import urllib3
 
 from .version import __version__
+from .constants import SOURCE_TAG
+import re
 
-def get_user_agent():
-    client_id = f"python-client-{__version__}"
+def _build_source_tag_field(source_tag):
+    # normalize source tag
+    # 1. Lowercase
+    # 2. Limit charset to [a-z0-9_ ]
+    # 3. Trim left/right whitespace
+    # 4. Condense multiple spaces to one, and replace with underscore
+    tag = source_tag.lower()
+    tag = re.sub(r'[^a-z0-9_ ]', '', tag)
+    tag = tag.strip()
+    tag = "_".join(tag.split())
+    return f"{SOURCE_TAG}={tag}"
+
+def _get_user_agent(client_id, config):
     user_agent_details = {"urllib3": urllib3.__version__}
     user_agent = "{} ({})".format(client_id, ", ".join([f"{k}:{v}" for k, v in user_agent_details.items()]))
-    return user_agent
+    user_agent += f"; {_build_source_tag_field(config.source_tag)}" if config.source_tag else ""
+    return user_agent
+
+def get_user_agent(config):
+    return _get_user_agent(f"python-client-{__version__}", config)
+
+def get_user_agent_grpc(config):
+    return _get_user_agent(f"python-client[grpc]-{__version__}", config)

tests/unit/test_config_builder.py

@@ -21,6 +21,17 @@ def test_build_merges_key_and_host_when_openapi_config_provided(self):
         assert config.host == "https://my-controller-host"
         assert config.additional_headers == {}
 
+    def test_build_with_source_tag(self):
+        config = ConfigBuilder.build(
+            api_key="my-api-key", 
+            host="https://my-controller-host", 
+            source_tag="my-source-tag",
+        )
+        assert config.api_key == "my-api-key"
+        assert config.host == "https://my-controller-host"
+        assert config.additional_headers == {}
+        assert config.source_tag == "my-source-tag"
+
     def test_build_errors_when_no_api_key_is_present(self):
         with pytest.raises(PineconeConfigurationError) as e:
             ConfigBuilder.build()

tests/unit/test_control.py

@@ -1,5 +1,6 @@
 import pytest
-from pinecone import Pinecone, PodSpec, ServerlessSpec
+import re
+from pinecone import ConfigBuilder, Pinecone, PodSpec, ServerlessSpec
 from pinecone.core.client.models import IndexList, IndexModel
 from pinecone.core.client.api.manage_indexes_api import ManageIndexesApi
 from pinecone.core.client.configuration import Configuration as OpenApiConfiguration
@@ -41,6 +42,15 @@ def test_overwrite_useragent(self):
         assert p.index_api.api_client.default_headers['User-Agent'] == 'test-user-agent'
         assert len(p.index_api.api_client.default_headers) == 1
 
+    def test_set_source_tag_in_useragent(self):
+        p = Pinecone(api_key="123-456-789", source_tag="test_source_tag")
+        assert re.search(r"source_tag=test_source_tag", p.index_api.api_client.user_agent) is not None
+
+    def test_set_source_tag_in_useragent_via_config(self):
+        config = ConfigBuilder.build(api_key='YOUR_API_KEY', host='https://my-host', source_tag='my_source_tag')
+        p = Pinecone(config=config)
+        assert re.search(r"source_tag=my_source_tag", p.index_api.api_client.user_agent) is not None
+
     @pytest.mark.parametrize("timeout_value, describe_index_responses, expected_describe_index_calls, expected_sleep_calls", [
         # When timeout=None, describe_index is called until ready
         (None, [{ "status": {"ready": False}}, {"status": {"ready": True}}], 2, 1),

tests/unit/test_index_initialization.py

@@ -1,5 +1,6 @@
 import pytest
-from pinecone import Pinecone
+import re
+from pinecone import ConfigBuilder, Pinecone
 
 class TestIndexClientInitialization():
     @pytest.mark.parametrize(
@@ -50,4 +51,15 @@ def test_overwrite_useragent(self):
         )
         assert len(index._vector_api.api_client.default_headers) == 1
         assert 'User-Agent' in index._vector_api.api_client.default_headers
-        assert index._vector_api.api_client.default_headers['User-Agent'] == 'test-user-agent'
+        assert index._vector_api.api_client.default_headers['User-Agent'] == 'test-user-agent'
+
+    def test_set_source_tag(self):
+        pc = Pinecone(api_key="123-456-789", source_tag="test_source_tag")
+        index = pc.Index(host='myhost')
+        assert re.search(r"source_tag=test_source_tag", pc.index_api.api_client.user_agent) is not None
+
+    def test_set_source_tag_via_config(self):
+        config = ConfigBuilder.build(api_key='YOUR_API_KEY', host='https://my-host', source_tag='my_source_tag')
+        pc = Pinecone(config=config)
+        index = pc.Index(host='myhost')
+        assert re.search(r"source_tag=my_source_tag", pc.index_api.api_client.user_agent) is not None