rename current version to latest version and use urlencode, and update release notes

Author: gary-huang

ddtrace/llmobs/_experiment.py

@@ -107,7 +107,7 @@ class Dataset:
     _id: str
     _records: List[DatasetRecord]
     _version: int
-    _current_version: int
+    _latest_version: int
     _dne_client: "LLMObsExperimentsClient"
     _new_records_by_record_id: Dict[str, DatasetRecordRaw]
     _updated_record_ids_to_new_fields: Dict[str, UpdatableDatasetRecord]
@@ -122,15 +122,15 @@ def __init__(
         dataset_id: str,
         records: List[DatasetRecord],
         description: str,
-        current_version: int,
+        latest_version: int,
         version: int,
         _dne_client: "LLMObsExperimentsClient",
     ) -> None:
         self.name = name
         self.project = project
         self.description = description
         self._id = dataset_id
-        self._current_version = current_version
+        self._latest_version = latest_version
         self._version = version
         self._dne_client = _dne_client
         self._records = records
@@ -171,10 +171,10 @@ def push(self) -> None:
                 record["record_id"] = record_id  # type: ignore
 
             # FIXME: we don't get version numbers in responses to deletion requests
-            self._current_version = new_version if new_version != -1 else self._current_version + 1
+            self._latest_version = new_version if new_version != -1 else self._latest_version + 1
             # no matter what the version was before the push, pushing will result in the dataset being on the current
             # version tracked by the backend
-            self._version = self._current_version
+            self._version = self._latest_version
         self._new_records_by_record_id = {}
         self._deleted_record_ids = []
         self._updated_record_ids_to_new_fields = {}
@@ -232,8 +232,8 @@ def url(self) -> str:
         return f"{_get_base_url()}/llm/datasets/{self._id}"
 
     @property
-    def current_version(self) -> int:
-        return self._current_version
+    def latest_version(self) -> int:
+        return self._latest_version
 
     @property
     def version(self) -> int:
@@ -448,7 +448,7 @@ def _run_task(self, jobs: int, raise_errors: bool = False, sample_size: Optional
                 dataset_id=self._dataset._id,
                 records=subset_records,
                 description=self._dataset.description,
-                current_version=self._dataset._current_version,
+                latest_version=self._dataset._latest_version,
                 version=self._dataset._version,
                 _dne_client=self._dataset._dne_client,
             )

ddtrace/llmobs/_writer.py

@@ -3,6 +3,7 @@
 import json
 import os
 import tempfile
+import urllib
 from typing import Any
 from typing import Dict
 from typing import List
@@ -400,7 +401,16 @@ def dataset_create(
         if dataset_id is None or dataset_id == "":
             raise ValueError(f"unexpected dataset state, invalid ID (is None: {dataset_id is None})")
         curr_version = response_data["data"]["attributes"]["current_version"]
-        return Dataset(dataset_name, project, dataset_id, [], description, curr_version, curr_version, _dne_client=self)
+        return Dataset(
+            name=dataset_name,
+            project=project,
+            dataset_id=dataset_id,
+            records=[],
+            description=description,
+            latest_version=curr_version,
+            version=curr_version,
+            _dne_client=self,
+        )
 
     @staticmethod
     def _get_record_json(record: Union[UpdatableDatasetRecord, DatasetRecordRaw], is_update: bool) -> JSONType:
@@ -484,12 +494,14 @@ def dataset_get_with_records(
         dataset_id = data[0]["id"]
 
         list_base_path = f"/api/unstable/llm-obs/v1/datasets/{dataset_id}/records"
+
+        url_options = {}
         if version:
-            list_base_path = f"{list_base_path}?filter[version]={version}"
+            url_options["filter[version]"] = version
 
         has_next_page = True
         class_records: List[DatasetRecord] = []
-        list_path = list_base_path
+        list_path = f"{list_base_path}?{urllib.parse.urlencode(url_options, safe='[]')}"
         page_num = 0
         while has_next_page:
             resp = self.request("GET", list_path, timeout=self.LIST_RECORDS_TIMEOUT)
@@ -514,17 +526,18 @@ def dataset_get_with_records(
             has_next_page = False
             if next_cursor:
                 has_next_page = True
-                list_path = f"{list_base_path}{'&' if version else '?'}page[cursor]={next_cursor}"
+                url_options["page[cursor]"] = next_cursor
+                list_path = f"{list_base_path}?{urllib.parse.urlencode(url_options, safe='[]')}"
                 logger.debug("next list records request path %s", list_path)
                 page_num += 1
         return Dataset(
-            dataset_name,
-            project,
-            dataset_id,
-            class_records,
-            dataset_description,
-            curr_version,
-            version or curr_version,
+            name=dataset_name,
+            project=project,
+            dataset_id=dataset_id,
+            records=class_records,
+            description=dataset_description,
+            latest_version=curr_version,
+            version=version or curr_version,
             _dne_client=self,
         )
 

releasenotes/notes/llmobs-dne-allow-versioned-dataset-pull-c7017f982b2c1f5b.yaml

@@ -3,3 +3,6 @@ features:
   - |
     LLM Observability: Previous dataset versions can be optionally pulled by passing the ``version`` 
     argument to ``LLMObs.pull_dataset``
+  - |
+    LLM Observability: Datasets have new properties ``version`` and ``latest_version`` to provide information on the
+    version of the dataset that is being worked with and the latest global version of the dataset, respectively