fix(instr-undici): fix user agent extraction and handle of multiple values on headers

plugins/node/instrumentation-undici/src/types.ts

@@ -22,9 +22,10 @@ export interface UndiciRequest {
   path: string;
   /**
    * Serialized string of headers in the form `name: value\r\n` for v5
-   * Array of strings v6
+   * Array of strings `[key1, value1, key2, value2]`, where values are
+   * `string | string[]` for v6
    */
-  headers: string | string[];
+  headers: string | (string | string[])[];
   /**
    * Helper method to add headers (from v6)
    */

plugins/node/instrumentation-undici/src/undici.ts

@@ -165,6 +165,50 @@
     });
   }
 
+  private parseRequestHeaders(request: UndiciRequest) {
+    const result = new Map<string, string | string[]>();
+
+    if (Array.isArray(request.headers)) {
+      // headers are an array [k1, v2, k2, v2] (undici v6+)
+      // values could be string or a string[] for multiple values
+      for (let i = 0; i < request.headers.length; i += 2) {
+        const key = request.headers[i];
+        const value = request.headers[i + 1];
+
+        // Key should always be a string, but the types don't know that, and let's be safe
+        if (typeof key === 'string') {
+          result.set(key.toLowerCase(), value);
+        }
+      }
+    } else if (typeof request.headers === 'string') {
+      // headers are a raw string (undici v5)
+      // headers could be repeated in several lines for multiple values
+      const headers = request.headers.split('\r\n');
+      for (const line of headers) {
+        if (!line) {
+          continue;
+        }
+        const colonIndex = line.indexOf(':');
+        if (colonIndex === -1) {
+          // Invalid header? Probably this can't happen, but again let's be safe.
+          continue;
+        }
+        const key = line.substring(0, colonIndex).toLowerCase();
+        const value = line.substring(colonIndex + 1).trim();
+        const allValues = result.get(key);
+
+        if (allValues && Array.isArray(allValues)) {
+          allValues.push(value);
+        } else if (allValues) {
+          result.set(key, [allValues, value]);
+        } else {
+          result.set(key, value);
+        }
+      }
+    }
+    return result;
+  }
+
   // This is the 1st message we receive for each request (fired after request creation). Here we will
   // create the span and populate some atttributes, then link the span to the request for further
   // span processing
@@ -218,24 +262,13 @@
     }
 
     // Get user agent from headers
-    let userAgent;
-    if (Array.isArray(request.headers)) {
-      const idx = request.headers.findIndex(
-        h => h.toLowerCase() === 'user-agent'
-      );
-      if (idx >= 0) {
-        userAgent = request.headers[idx + 1];
-      }
-    } else if (typeof request.headers === 'string') {
-      const headers = request.headers.split('\r\n');
-      const uaHeader = headers.find(h =>
-        h.toLowerCase().startsWith('user-agent')
-      );
-      userAgent =
-        uaHeader && uaHeader.substring(uaHeader.indexOf(':') + 1).trim();
-    }
+    const headersMap = this.parseRequestHeaders(request);
+    const userAgentValues = headersMap.get('user-agent');
 
-    if (userAgent) {
+    if (userAgentValues) {
+      const userAgent = Array.isArray(userAgentValues)
+        ? userAgentValues[userAgentValues.length - 1]
+        : userAgentValues;
       attributes[SemanticAttributes.USER_AGENT_ORIGINAL] = userAgent;
     }
 
@@ -329,27 +362,14 @@
       const headersToAttribs = new Set(
         config.headersToSpanAttributes.requestHeaders.map(n => n.toLowerCase())
       );
+      const headersMap = this.parseRequestHeaders(request);
 
-      // headers could be in form
-      // ['name: value', ...] for v5
-      // ['name', 'value', ...] for v6
-      const rawHeaders = Array.isArray(request.headers)
-        ? request.headers
-        : request.headers.split('\r\n');
-      rawHeaders.forEach((h, idx) => {
-        const sepIndex = h.indexOf(':');
-        const hasSeparator = sepIndex !== -1;
-        const name = (
-          hasSeparator ? h.substring(0, sepIndex) : h
-        ).toLowerCase();
-        const value = hasSeparator
-          ? h.substring(sepIndex + 1)
-          : rawHeaders[idx + 1];
-
+      for (const [name, value] of headersMap.entries()) {
         if (headersToAttribs.has(name)) {
-          spanAttributes[`http.request.header.${name}`] = value.trim();
+          const attrValue = Array.isArray(value) ? value.join(', ') : value;
+          spanAttributes[`http.request.header.${name}`] = attrValue.trim();
         }
-      });
+      }
     }
 
     span.setAttributes(spanAttributes);

plugins/node/instrumentation-undici/test/undici.test.ts

@@ -797,43 +797,87 @@ describe('UndiciInstrumentation `undici` tests', function () {
       });
     });
 
-    it('should not report an user-agent if it was not defined', async function () {
-      let spans = memoryExporter.getFinishedSpans();
-      assert.strictEqual(spans.length, 0);
+    const userAgentRequests: Array<{
+      name: string;
+      headers: Record<string, string | string[] | undefined>;
+      expectedUserAgent: string | undefined;
+    }> = [
+      {
+        name: 'no user-agent',
+        headers: { 'foo-client': 'bar' },
+        expectedUserAgent: undefined,
+      },
+      {
+        name: 'a user-agent',
+        headers: { 'foo-client': 'bar', 'user-agent': 'custom' },
+        expectedUserAgent: 'custom',
+      },
+      {
+        name: 'explicitly-undefined user-agent',
+        headers: { 'foo-client': 'bar', 'user-agent': undefined },
+        expectedUserAgent: undefined,
+      },
+      // contra the spec, but we shouldn't crash
+      {
+        name: 'multiple user-agents',
+        headers: {
+          'foo-client': 'bar',
+          'user-agent': ['agent', 'other-agent'],
+        },
+        expectedUserAgent: 'other-agent',
+      },
+      {
+        name: 'another header with value user-agent',
+        headers: { 'foo-client': 'user-agent', 'user-agent': 'custom' },
+        expectedUserAgent: 'custom',
+      },
+      {
+        name: 'another header with multiple values',
+        headers: { 'foo-client': ['one', 'two'], 'user-agent': 'custom' },
+        expectedUserAgent: 'custom',
+      },
+      {
+        name: 'another header with explicitly-undefined value',
+        headers: { 'foo-client': undefined, 'user-agent': 'custom' },
+        expectedUserAgent: 'custom',
+      },
+    ];
 
-      // Do some requests
-      const headers = {
-        'foo-client': 'bar',
-      };
+    for (const testCase of userAgentRequests) {
+      it(`should report the correct user-agent when the request has ${testCase.name}`, async function () {
+        let spans = memoryExporter.getFinishedSpans();
+        assert.strictEqual(spans.length, 0);
 
-      const queryRequestUrl = `${protocol}://${hostname}:${mockServer.port}/?query=test`;
-      const queryResponse = await undici.request(queryRequestUrl, { headers });
-      await consumeResponseBody(queryResponse.body);
+        const queryRequestUrl = `${protocol}://${hostname}:${mockServer.port}/?query=test`;
+        const queryResponse = await undici.request(queryRequestUrl, {
+          headers: testCase.headers,
+        });
+        await consumeResponseBody(queryResponse.body);
 
-      assert.ok(
-        queryResponse.headers['propagation-error'] == null,
-        'propagation is set for instrumented requests'
-      );
+        assert.ok(
+          queryResponse.headers['propagation-error'] == null,
+          'propagation is set for instrumented requests'
+        );
 
-      spans = memoryExporter.getFinishedSpans();
-      const span = spans[0];
-      assert.ok(span, 'a span is present');
-      assert.strictEqual(spans.length, 1);
-      assertSpan(span, {
-        hostname: 'localhost',
-        httpStatusCode: queryResponse.statusCode,
-        httpMethod: 'GET',
-        path: '/',
-        query: '?query=test',
-        reqHeaders: headers,
-        resHeaders: queryResponse.headers,
+        spans = memoryExporter.getFinishedSpans();
+        const span = spans[0];
+        assert.ok(span, 'a span is present');
+        assert.strictEqual(spans.length, 1);
+        assertSpan(span, {
+          hostname: 'localhost',
+          httpStatusCode: queryResponse.statusCode,
+          httpMethod: 'GET',
+          path: '/',
+          query: '?query=test',
+          reqHeaders: testCase.headers,
+          resHeaders: queryResponse.headers,
+        });
+        assert.strictEqual(
+          span.attributes['user_agent.original'],
+          testCase.expectedUserAgent
+        );
       });
-      assert.strictEqual(
-        span.attributes['user_agent.original'],
-        undefined,
-        'user-agent is undefined'
-      );
-    });
+    }
 
     it('should create valid span if request.path is a full URL', async function () {
       let spans = memoryExporter.getFinishedSpans();

plugins/node/instrumentation-undici/test/utils/assertSpan.ts

@@ -171,7 +171,7 @@ export const assertSpan = (
     if (userAgent) {
       assert.strictEqual(
         span.attributes[SemanticAttributes.USER_AGENT_ORIGINAL],
-        userAgent
+        Array.isArray(userAgent) ? userAgent[userAgent.length - 1] : userAgent
       );
     }
   }