chore(configuration)!: change parseConfigFile() to *not* add default values, nor merge *_list fields

experimental/packages/configuration/src/FileConfigFactory.ts

@@ -10,11 +10,9 @@ import * as yaml from 'yaml';
 import type { ConfigFactory } from './IConfigFactory';
 import { substituteEnvVars } from './utils';
 import type {
-  BatchLogRecordProcessor,
-  BatchSpanProcessor,
-  CardinalityLimits,
+  AttributeNameValue,
   ConfigurationModel,
-  PeriodicMetricReader,
+  TextMapPropagator,
 } from './generated/types';
 // Pre-compiled ajv validator — eliminates runtime schema compilation on cold start.
 // Generated by `npm run generate:config`; do not import ajv directly here.
@@ -74,191 +72,101 @@ export function parseConfigFile(): ConfigurationModel {
   // Strip file_format from output — it's a meta-field, not a config value
   delete (data as Record<string, unknown>)['file_format'];
 
-  applyConfigDefaults(data);
-  mergeAttributesList(data);
-  mergeCompositeList(data);
-  applyBatchProcessorDefaults(data);
-  applyPeriodicReaderDefaults(data);
-  applyOtlpHttpEncodingDefaults(data);
-
   return data;
 }
 
 /**
- * Apply default encoding ('protobuf') to all otlp_http exporters in the config
- * where encoding was not explicitly specified. The spec defines protobuf as the
- * default encoding for OTLP HTTP exporters.
+ * Merge attributes from `resource.attributes` and `resource.attributes_list`
+ * (comma-separated key=value pairs) into an `AttributeNameValue[]` array.
+ * Entries already in `resource.attributes` taking precedence.
+ *
+ * Note: The returned array might not be a copy, so it should not be mutated.
  */
-function applyOtlpHttpEncodingDefaults(data: ConfigurationModel): void {
-  const applyEncoding = (
-    exporter: { encoding?: string | null } | null | undefined
-  ) => {
-    if (exporter && exporter.encoding == null) {
-      exporter.encoding = 'protobuf';
-    }
-  };
-
-  for (const processor of data.tracer_provider?.processors ?? []) {
-    applyEncoding(processor.batch?.exporter?.otlp_http);
-    applyEncoding(processor.simple?.exporter?.otlp_http);
-  }
-
-  for (const reader of data.meter_provider?.readers ?? []) {
-    applyEncoding(reader.periodic?.exporter?.otlp_http);
+export function mergeResourceAttributesConfig(
+  attributes?: AttributeNameValue[],
+  attributes_list?: string | null
+): AttributeNameValue[] | undefined {
+  if (!attributes_list) {
+    return attributes;
   }
 
-  for (const processor of data.logger_provider?.processors ?? []) {
-    applyEncoding(processor.batch?.exporter?.otlp_http);
-    applyEncoding(processor.simple?.exporter?.otlp_http);
-  }
-}
+  const mergedAttrs = attributes ? attributes.slice() : [];
+  const existingNames = new Set(mergedAttrs.map(a => a.name));
 
-/**
- * Merge resource.attributes_list (comma-separated key=value pairs) into
- * resource.attributes, with entries already in attributes taking precedence.
- *
- * Per the spec, `,` and `=` in keys and values MUST be percent-encoded, and
- * other characters MAY be percent-encoded. On any parse or decode error, the
- * entire attributes_list is discarded and a warning is emitted.
- * See https://opentelemetry.io/docs/specs/otel/resource/sdk/#specifying-resource-information-via-an-environment-variable
- */
-function mergeAttributesList(data: ConfigurationModel): void {
-  const resource = data.resource;
-  const list = resource?.attributes_list;
-  if (typeof list !== 'string' || !list.trim()) return;
-
-  const decoded: Array<{ key: string; value: string }> = [];
-  for (const pair of list.split(',')) {
+  // TODO: handle attribute limits, if any
+  let discard = false;
+  const listAttrs: Record<string, string> = {};
+  for (const pair of attributes_list.split(',')) {
     if (pair.trim() === '') continue;
 
-    // Per spec, `=` must be percent-encoded in keys/values, so a valid entry
-    // splits into exactly two parts.
     const parts = pair.split('=');
     if (parts.length !== 2) {
       diag.warn(
         `Invalid format for resource.attributes_list entry "${pair}": expected key=value with '=' percent-encoded in keys/values. Discarding all entries.`
       );
-      return;
+      discard = true;
+      break;
     }
 
-    const rawKey = parts[0].trim();
+    const rawName = parts[0].trim();
     const rawValue = parts[1].trim();
-    if (rawKey === '') {
+    if (rawName === '') {
       diag.warn(
-        `Empty attribute key in resource.attributes_list entry "${pair}". Discarding all entries.`
+        `Empty attribute name in resource.attributes_list entry "${pair}". Discarding all entries.`
       );
-      return;
+      discard = true;
+      break;
     }
 
     try {
-      decoded.push({
-        key: decodeURIComponent(rawKey),
-        value: decodeURIComponent(rawValue),
-      });
+      listAttrs[decodeURIComponent(rawName)] = decodeURIComponent(rawValue);
     } catch (e) {
       diag.warn(
         `Failed to percent-decode resource.attributes_list entry "${pair}", discarding all entries: ${e}`
       );
-      return;
+      discard = true;
+      break;
     }
   }
 
-  if (resource!.attributes == null) {
-    resource!.attributes = [];
-  }
-
-  const existingKeys = new Set(
-    resource!.attributes.map((a: { name: string }) => a.name)
-  );
-
-  for (const { key, value } of decoded) {
-    if (!existingKeys.has(key)) {
-      resource!.attributes.push({ name: key, value, type: 'string' });
+  if (!discard) {
+    for (const [name, value] of Object.entries(listAttrs)) {
+      if (!existingNames.has(name)) {
+        mergedAttrs.push({ name, value, type: 'string' });
+      }
     }
   }
+
+  return mergedAttrs;
 }
 
 /**
  * Merge propagator.composite_list (comma-separated propagator names) into
  * propagator.composite, with entries already in composite taking precedence.
+ * Merge TextMapPropagator configs from `propagator.composite` and `propagator.composite_list`
+ * (comma-separated names) into a `TextMapPropagator[]` array.
+ * Entries already in `propagator.composite` taking precedence.
+ *
+ * Note: The returned array might not be a copy, so it should not be mutated.
  */
-function mergeCompositeList(data: ConfigurationModel): void {
-  const propagator = data.propagator;
-  const list = propagator?.composite_list;
-  if (typeof list !== 'string' || !list.trim()) return;
-
-  if (propagator!.composite == null) {
-    propagator!.composite = [];
+export function mergePropagatorCompositeConfig(
+  composite?: TextMapPropagator[],
+  composite_list?: string | null
+): TextMapPropagator[] | undefined {
+  if (!composite_list) {
+    return composite;
   }
 
-  const existingNames = new Set(
-    (propagator!.composite as Array<Record<string, unknown>>).map(
-      entry => Object.keys(entry)[0]
-    )
-  );
+  const mergedComposite = composite ? composite.slice() : [];
+  const existingNames = new Set(mergedComposite.map(c => Object.keys(c)[0]));
 
-  for (const name of list.split(',')) {
+  for (const name of composite_list.split(',')) {
     const trimmed = name.trim();
     if (trimmed && !existingNames.has(trimmed)) {
-      (propagator!.composite as Array<Record<string, unknown>>).push({
-        [trimmed]: {},
-      });
+      mergedComposite.push({ [trimmed]: null });
+      existingNames.add(trimmed);
     }
   }
-}
 
-/**
- * Apply spec-defined defaults for batch span/log-record processor fields that
- * are not encoded in the JSON schema.
- */
-function applyBatchProcessorDefaults(data: ConfigurationModel): void {
-  const applyDefaults = (
-    batch: BatchSpanProcessor | BatchLogRecordProcessor
-  ) => {
-    if (batch.schedule_delay == null) batch.schedule_delay = 5000;
-    if (batch.export_timeout == null) batch.export_timeout = 30000;
-    if (batch.max_queue_size == null) batch.max_queue_size = 2048;
-    if (batch.max_export_batch_size == null) batch.max_export_batch_size = 512;
-  };
-
-  for (const processor of data.tracer_provider?.processors ?? []) {
-    if (processor.batch) applyDefaults(processor.batch);
-  }
-  for (const processor of data.logger_provider?.processors ?? []) {
-    if (processor.batch) applyDefaults(processor.batch);
-  }
-}
-
-/**
- * Apply spec-defined defaults for periodic metric reader fields.
- */
-function applyPeriodicReaderDefaults(data: ConfigurationModel): void {
-  for (const reader of data.meter_provider?.readers ?? []) {
-    const periodic = reader.periodic as PeriodicMetricReader | undefined;
-    if (!periodic) continue;
-    if (periodic.interval == null) periodic.interval = 60000;
-    if (periodic.timeout == null) periodic.timeout = 30000;
-    if (periodic.cardinality_limits == null) {
-      periodic.cardinality_limits = { default: 2000 } as CardinalityLimits;
-    }
-  }
-}
-
-/**
- * Apply spec-defined defaults that are not encoded in the JSON schema.
- * Both FileConfigFactory and EnvironmentConfigFactory apply the same defaults
- * so consumers see consistent behaviour regardless of config source.
- */
-function applyConfigDefaults(data: ConfigurationModel): void {
-  if (data.disabled == null) {
-    data.disabled = false;
-  }
-  if (data.log_level == null) {
-    data.log_level = 'info';
-  }
-  if (data.attribute_limits == null) {
-    data.attribute_limits = { attribute_count_limit: 128 };
-  } else if (data.attribute_limits.attribute_count_limit == null) {
-    data.attribute_limits.attribute_count_limit = 128;
-  }
+  return mergedComposite;
 }

experimental/packages/configuration/src/index.ts

@@ -19,5 +19,10 @@ export type {
   HttpTls as HttpTlsConfigModel,
   GrpcTls as GrpcTlsConfigModel,
   SeverityNumber as SeverityNumberConfigModel,
+  TextMapPropagator as TextMapPropagatorConfigModel,
 } from './generated/types';
 export { createConfigFactory } from './ConfigFactory';
+export {
+  mergeResourceAttributesConfig,
+  mergePropagatorCompositeConfig,
+} from './FileConfigFactory';