Merge branch 'main' into feat/support-eb-dsm-extraction

Author: jeastham1993

.github/workflows/update-snapshots.yml

@@ -5,13 +5,15 @@ regions:
   - code: "us-west-2"
   - code: "af-south-1"
   - code: "ap-east-1"
+  - code: "ap-east-2"
   - code: "ap-south-1"
   - code: "ap-south-2"
   - code: "ap-southeast-1"
   - code: "ap-southeast-2"
   - code: "ap-southeast-3"
   - code: "ap-southeast-4"
   - code: "ap-southeast-5"
+  - code: "ap-southeast-6"
   - code: "ap-southeast-7"
   - code: "ap-northeast-1"
   - code: "ap-northeast-2"

ci/datasources/regions.yaml

@@ -28,4 +28,5 @@ if [ -d "dist" ]; then
 fi
 
 # Publish to pypi
+source venv/bin/activate
 poetry publish --build --username __token__ --password $PYPI_TOKEN

ci/publish_pypi.sh

@@ -75,6 +75,13 @@ def _resolve_env(self, key, default=None, cast=None, depends_on_tracing=False):
     add_span_pointers = _get_env("DD_BOTOCORE_ADD_SPAN_POINTERS", "true", as_bool)
     trace_extractor = _get_env("DD_TRACE_EXTRACTOR")
 
+    aws_service_representation_enabled = _get_env(
+        "DD_TRACE_AWS_SERVICE_REPRESENTATION_ENABLED", "true", as_bool
+    )
+    remove_integration_service_names_enabled = _get_env(
+        "DD_TRACE_REMOVE_INTEGRATION_SERVICE_NAMES_ENABLED", "false", as_bool
+    )
+
     enhanced_metrics_enabled = _get_env("DD_ENHANCED_METRICS", "true", as_bool)
 
     flush_in_thread = _get_env("DD_FLUSH_IN_THREAD", "false", as_bool)

datadog_lambda/config.py

@@ -43,6 +43,7 @@
     is_step_function_event,
     EventTypes,
     EventSubtypes,
+    resolve_multivalue_headers,
 )
 from datadog_lambda.durable import extract_context_from_durable_execution
 
@@ -207,7 +208,7 @@ def extract_context_from_http_event_or_context(
         if _is_context_complete(context):
             return context
 
-    headers = event.get("headers")
+    headers = resolve_multivalue_headers(event)
     context = propagator.extract(headers)
 
     if not _is_context_complete(context):
@@ -740,7 +741,9 @@ def extract_dd_trace_context(
         context = extract_context_from_request_header_or_context(
             event, lambda_context, event_source
         )
-    elif isinstance(event, (set, dict)) and "headers" in event:
+    elif isinstance(event, (set, dict)) and (
+        "headers" in event or "multiValueHeaders" in event
+    ):
         context = extract_context_from_http_event_or_context(
             event, lambda_context, event_source, decode_authorizer_context
         )
@@ -919,6 +922,9 @@ def create_inferred_span(
         elif event_source.equals(EventTypes.LAMBDA_FUNCTION_URL):
             logger.debug("Function URL event detected. Inferring a span")
             return create_inferred_span_from_lambda_function_url_event(event, context)
+        elif event_source.event_type == EventTypes.ALB:
+            logger.debug("ALB event detected. Inferring a span")
+            return create_inferred_span_from_alb_event(event, context)
         elif event_source.equals(
             EventTypes.API_GATEWAY, subtype=EventSubtypes.HTTP_API
         ):
@@ -982,11 +988,14 @@ def determine_service_name(
     if mapped_service:
         return mapped_service
 
+    # When integration service names are removed, inferred (synthetic) spans use
+    # the base service name (DD_SERVICE) instead of the AWS resource/instance
+    # representation.
+    if config.remove_integration_service_names_enabled and config.service:
+        return config.service
+
     # Check if AWS service representation is disabled
-    aws_service_representation = os.environ.get(
-        "DD_TRACE_AWS_SERVICE_REPRESENTATION_ENABLED", ""
-    ).lower()
-    if aws_service_representation in ("false", "0"):
+    if not config.aws_service_representation_enabled:
         return fallback
 
     # Use extracted_key if it exists and is not empty, otherwise use fallback
@@ -1034,6 +1043,54 @@ def create_inferred_span_from_lambda_function_url_event(event, context):
     return span
 
 
+def create_inferred_span_from_alb_event(event, context):
+    request_context = event.get("requestContext") or {}
+    elb = request_context.get("elb") or {}
+    target_group_arn = elb.get("targetGroupArn")
+
+    headers = resolve_multivalue_headers(event)
+    host = headers.get("host")
+    method = event.get("httpMethod")
+    path = event.get("path")
+    proto = headers.get("x-forwarded-proto", "http")
+
+    # ALB has no api id; key the service mapping off the load-balancer host and
+    # fall back to it when DD_TRACE_AWS_SERVICE_REPRESENTATION_ENABLED is on.
+    service_name = determine_service_name(service_mapping, host, "lambda_alb", host)
+
+    http_url = "%s://%s%s" % (proto, host, path) if host and path is not None else None
+    if method and path is not None:
+        resource = f"{method} {path}"
+    else:
+        resource = method or path
+
+    tags = {
+        "operation_name": "aws.alb",
+        "span.kind": "server",
+        "http.method": method,
+        "http.url": http_url,
+        "http.useragent": headers.get("user-agent"),
+        "endpoint": path,
+        "resource_names": resource,
+        "request_id": context.aws_request_id,
+        "target_group_arn": target_group_arn,
+    }
+    # Drop tags we couldn't derive so the span never carries malformed values.
+    tags = {key: value for key, value in tags.items() if value is not None}
+
+    tracer.set_tags(_dd_origin)
+    # ALB events carry no request timestamp (unlike API GW requestTimeEpoch /
+    # Function URL timeEpoch), so the span starts at handler time.
+    span = tracer.trace(
+        "aws.alb", service=service_name, resource=resource, span_type="http"
+    )
+    InferredSpanInfo.set_tags(tags, tag_source="self", synchronicity="sync")
+    if span:
+        span.set_tags(tags)
+        span.set_metric(InferredSpanInfo.METRIC, 1.0)
+    return span
+
+
 def is_api_gateway_invocation_async(event):
     hdrs = event.get("headers")
     if not hdrs:

datadog_lambda/tracing.py

@@ -293,6 +293,39 @@ def get_event_source_arn(source: _EventSource, event: dict, context: Any) -> str
     return event_source_arn
 
 
+def resolve_multivalue_headers(event):
+    """
+    Resolve request headers from single-value ``headers`` or ``multiValueHeaders``
+    (first value per key, matching datadog-lambda-js).
+    """
+    headers = event.get("headers")
+    if isinstance(headers, dict) and headers:
+        return headers
+
+    multi_value = event.get("multiValueHeaders")
+    if not isinstance(multi_value, dict):
+        return {}
+
+    resolved = {}
+    for key, value in multi_value.items():
+        if isinstance(value, list):
+            if value:
+                resolved[key] = value[0]
+        elif isinstance(value, str):
+            resolved[key] = value
+    return resolved
+
+
+def _get_header_case_insensitive(headers, name):
+    if not isinstance(headers, dict):
+        return None
+    name_lower = name.lower()
+    for key, value in headers.items():
+        if isinstance(key, str) and key.lower() == name_lower and value:
+            return value
+    return None
+
+
 def extract_http_tags(event):
     """
     Extracts HTTP facet tags from the triggering event
@@ -324,19 +357,32 @@ def extract_http_tags(event):
             path = apigateway_v2_http.get("path")
             method = apigateway_v2_http.get("method")
 
+    elif request_context and request_context.get("elb"):
+        # ALB events have no requestContext.stage; derive the URL from the
+        # forwarded host/proto headers and the top-level path.
+        alb_headers = resolve_multivalue_headers(event)
+        host = alb_headers.get("host")
+        if host:
+            proto = alb_headers.get("x-forwarded-proto", "http")
+            http_tags["http.url"] = proto + "://" + host
+
+        user_agent = alb_headers.get("user-agent")
+        if user_agent:
+            http_tags["http.useragent"] = user_agent
+
+        # ALB carries no route template, so use the request path as the route.
+        if path:
+            http_tags["http.route"] = path
+
     if path:
         if http_tags.get("http.url"):
             http_tags["http.url"] += path
     if method:
         http_tags["http.method"] = method
 
-    # Safely get headers
-    headers = event.get("headers", {})
-    if not isinstance(headers, dict):
-        headers = {}
-
-    if headers and headers.get("Referer"):
-        http_tags["http.referer"] = headers.get("Referer")
+    referer = _get_header_case_insensitive(resolve_multivalue_headers(event), "referer")
+    if referer:
+        http_tags["http.referer"] = referer
 
     # Try to get `routeKey` from API GW v2; otherwise try to get `resource` from API GW v1
     route = event.get("routeKey") or event.get("resource")
@@ -415,6 +461,8 @@ def is_step_function_event(event):
     The actual event must contain "Execution", "StateMachine", and "State" fields.
     """
     event = event.get("Payload", event)
+    if not isinstance(event, dict):
+        return False
 
     # JSONPath style
     if "Execution" in event and "StateMachine" in event and "State" in event:

datadog_lambda/trigger.py

@@ -1 +1 @@
-__version__ = "8.126.0.dev0"
+__version__ = "8.128.0.dev0"

datadog_lambda/version.py

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "datadog_lambda"
-version = "8.126.0.dev0"
+version = "8.128.0.dev0"
 description = "The Datadog AWS Lambda Library"
 authors = ["Datadog, Inc. <dev@datadoghq.com>"]
 license = "Apache-2.0"

pyproject.toml

@@ -43,9 +43,9 @@ bluebird@^3.7.2:
   integrity sha512-XpNj6GDQzdfW+r2Wnn7xiSAd7TM3jzkxGXBGTtWKuSXv1xUV+azxAm8jdWZN06QTQk+2N2XB9jRDkvbmQmcRtg==
 
 brace-expansion@^1.1.7:
-  version "1.1.14"
-  resolved "https://registry.yarnpkg.com/brace-expansion/-/brace-expansion-1.1.14.tgz#d9de602370d91347cd9ddad1224d4fd701eb348b"
-  integrity sha512-MWPGfDxnyzKU7rNOW9SP/c50vi3xrmrua/+6hfPbCS2ABNWfx24vPidzvC7krjU/RTo235sV776ymlsMtGKj8g==
+  version "1.1.18"
+  resolved "https://registry.yarnpkg.com/brace-expansion/-/brace-expansion-1.1.18.tgz#3ce74d89885136be1535341f8c3d4425c29a5cab"
+  integrity sha512-Edep/X9fGqVNmzKBVsDYIOtD+z1tuezV70LBjdCst9Tqu76lsnvRiZ6oTic1n+/BIwX6QDGAO94PN4N2SADvtw==
   dependencies:
     balanced-match "^1.0.0"
     concat-map "0.0.1"
@@ -366,7 +366,7 @@ log@^6.0.0:
     type "^2.7.3"
     uni-global "^1.0.0"
 
-minimatch@^3.1.1, minimatch@^3.1.5:
+minimatch@^3.1.1:
   version "3.1.5"
   resolved "https://registry.yarnpkg.com/minimatch/-/minimatch-3.1.5.tgz#580c88f8d5445f2bd6aa8f3cadefa0de79fbd69e"
   integrity sha512-VgjWUsnnT6n+NUk6eZq77zeFdpW2LWDzP6zFGrCbHXiYNul5Dzqk2HHQ5uFH2DNW5Xbp8+jVzaeNt94ssEEl4w==
@@ -564,10 +564,10 @@ shebang-regex@^1.0.0:
   resolved "https://registry.yarnpkg.com/shebang-regex/-/shebang-regex-1.0.0.tgz#da42f49740c0b42db2ca9728571cb190c98efea3"
   integrity sha512-wpoSFAxys6b2a2wHZ1XpDSgD7N9iVjg29Ph9uV/uaP9Ex/KXlkTZTeddxDPSYQpgvzKLGJke2UU0AzoGCjNIvQ==
 
-shell-quote@^1.8.1, shell-quote@^1.8.4:
-  version "1.8.4"
-  resolved "https://registry.yarnpkg.com/shell-quote/-/shell-quote-1.8.4.tgz#2edd9a4dcefc96649e2e2cb12f637b1f1d92a190"
-  integrity sha512-VsC6n6vz1ihYYyZZwX7YZSF5l5x36ca17OC+a69h94YqB7X6XLwf+5MOgynYir2SLFUbl8gIYvBo8K8RoNQ6bQ==
+shell-quote@^1.8.1:
+  version "1.10.0"
+  resolved "https://registry.yarnpkg.com/shell-quote/-/shell-quote-1.10.0.tgz#482033e192e4f5c07151521ffa03400ec71b1b0f"
+  integrity sha512-w1aiOKwKuRgtwAReIIj89puqg+I7GvX4IbLrvmhXbzQsj1+Zwi4VO3+fa6ZF91TWSjIxoEkKnMeHcLEODK5ZXA==
 
 split2@^3.1.1:
   version "3.2.2"