Trace server improvements

openinference-streaming.patch

@@ -2,9 +2,10 @@ diff --git a/python/instrumentation/openinference-instrumentation-beeai/src/open
 index d3f3adc6..5497ff77 100644
 --- a/python/instrumentation/openinference-instrumentation-beeai/src/openinference/instrumentation/beeai/__init__.py
 +++ b/python/instrumentation/openinference-instrumentation-beeai/src/openinference/instrumentation/beeai/__init__.py
-@@ -1,18 +1,15 @@
+@@ -1,18 +1,16 @@
 -import contextlib
  import logging
++from enum import Enum
  from importlib.metadata import PackageNotFoundError, version
 -from typing import TYPE_CHECKING, Any, Callable, Collection, Generator
 +from typing import TYPE_CHECKING, Any, Callable, Collection
@@ -25,7 +26,7 @@ index d3f3adc6..5497ff77 100644
  from openinference.instrumentation import (
      OITracer,
      TraceConfig,
-@@ -32,13 +29,14 @@ except PackageNotFoundError:
+@@ -32,13 +30,14 @@ except PackageNotFoundError:
 
 
  class BeeAIInstrumentor(BaseInstrumentor):  # type: ignore
@@ -42,7 +43,7 @@ index d3f3adc6..5497ff77 100644
 
      def instrumentation_dependencies(self) -> Collection[str]:
          return _instruments
-@@ -70,39 +68,81 @@ class BeeAIInstrumentor(BaseInstrumentor):  # type: ignore
+@@ -70,39 +69,85 @@ class BeeAIInstrumentor(BaseInstrumentor):  # type: ignore
      def _uninstrument(self, **kwargs: Any) -> None:
          self._cleanup()
          self._processes.clear()
@@ -81,6 +82,13 @@ index d3f3adc6..5497ff77 100644
 +        node = processor.span
 +        _OTEL_TYPES = (bool, str, bytes, int, float)
 +        for key, value in node.attributes.items():
++            # Extract enum value if it's an enum
++            if isinstance(value, Enum):
++                value = value.value
++            # Extract enum values from arrays/tuples
++            elif isinstance(value, (list, tuple)):
++                value = type(value)(v.value if isinstance(v, Enum) else v for v in value)
++            # Convert to string if not an OTEL type
 +            if not isinstance(value, _OTEL_TYPES) and not (
 +                isinstance(value, (list, tuple)) and all(isinstance(v, _OTEL_TYPES) for v in value)
 +            ):
@@ -147,7 +155,7 @@ index d3f3adc6..5497ff77 100644
 
      @exception_handler
      async def _handler(self, data: Any, event: "EventMeta") -> None:
-@@ -118,10 +158,8 @@ class BeeAIInstrumentor(BaseInstrumentor):  # type: ignore
+@@ -118,10 +163,8 @@ class BeeAIInstrumentor(BaseInstrumentor):  # type: ignore
              if event.trace.parent_run_id and not parent:
                  raise ValueError(f"Parent run with ID {event.trace.parent_run_id} was not found!")
 
@@ -159,7 +167,7 @@ index d3f3adc6..5497ff77 100644
          else:
              node = self._processes[event.trace.run_id]
 
-@@ -129,8 +167,8 @@ class BeeAIInstrumentor(BaseInstrumentor):  # type: ignore
+@@ -129,8 +172,8 @@ class BeeAIInstrumentor(BaseInstrumentor):  # type: ignore
 
          if isinstance(data, RunContextFinishEvent):
              await node.end(data, event)

trace_server/server.py

@@ -269,6 +269,19 @@ def _b64_to_hex(value: str) -> str:
         return value
 
 
+_TRACE_ID_RE = re.compile(r"^[0-9a-fA-F]{32}$")
+_SPAN_ID_RE = re.compile(r"^[0-9a-fA-F]{16}$")
+
+
+def _normalize_hex_id(value: str, pattern: re.Pattern) -> str:
+    """Lowercase a hex trace/span ID so lookups are case-insensitive.
+
+    Only touches values that actually look like the expected hex ID, so
+    non-hex or malformed IDs are passed through untouched.
+    """
+    return value.lower() if pattern.match(value) else value
+
+
 def _normalize_protobuf_ids(otlp_data: dict) -> None:
     """Convert base64 trace/span IDs produced by MessageToDict to hex in-place."""
     for rs in otlp_data.get("resourceSpans") or []:
@@ -366,9 +379,9 @@ def _extract_spans(otlp_data: dict) -> list[SpanRow]:
                     status_code = _STATUS_CODE_NAMES.get(status_code, 0)
                 spans.append(
                     SpanRow(
-                        trace_id=span.get("traceId") or "",
-                        span_id=span.get("spanId") or "",
-                        parent_span_id=span.get("parentSpanId") or "",
+                        trace_id=_normalize_hex_id(span.get("traceId") or "", _TRACE_ID_RE),
+                        span_id=_normalize_hex_id(span.get("spanId") or "", _SPAN_ID_RE),
+                        parent_span_id=_normalize_hex_id(span.get("parentSpanId") or "", _SPAN_ID_RE),
                         name=name,
                         start_time=int(span.get("startTimeUnixNano") or 0),
                         end_time=int(span.get("endTimeUnixNano") or 0) or None,
@@ -488,6 +501,7 @@ def query_spans(issue: str, params: dict) -> list[dict]:
         trace_id = params.get("trace_id")
         if not trace_id:
             return []
+        trace_id = _normalize_hex_id(trace_id, _TRACE_ID_RE)
         query_bindings: list = [trace_id]
         subquery = "SELECT ? AS trace_id"
 
@@ -536,7 +550,7 @@ def query_spans(issue: str, params: dict) -> list[dict]:
 
     if trace_id := params.get("trace_id"):
         issue_conditions.append("si.trace_id = ?")
-        issue_bindings.append(trace_id)
+        issue_bindings.append(_normalize_hex_id(trace_id, _TRACE_ID_RE))
 
     if names := params.get("name"):
         name_list = [n.strip() for n in names.split(",")]