Better tracking of failed requests + logging context exclude

package.json

@@ -1,6 +1,6 @@
 {
   "name": "browsertrix-crawler",
-  "version": "1.0.0-beta.6",
+  "version": "1.0.0-beta.7",
   "main": "browsertrix-crawler",
   "type": "module",
   "repository": "https://github.com/webrecorder/browsertrix-crawler",

src/crawler.ts

@@ -189,7 +189,8 @@ export class Crawler {
     const debugLogging = this.params.logging.includes("debug");
     logger.setDebugLogging(debugLogging);
     logger.setLogLevel(this.params.logLevel);
-    logger.setContext(this.params.context);
+    logger.setContext(this.params.logContext);
+    logger.setExcludeContext(this.params.logExcludeContext);
 
     // if automatically restarts on error exit code,
     // exit with 0 from fatal by default, to avoid unnecessary restart

src/util/argParser.ts

@@ -15,7 +15,11 @@ import {
 import { ScopedSeed } from "./seeds.js";
 import { interpolateFilename } from "./storage.js";
 import { screenshotTypes } from "./screenshots.js";
-import { LOG_CONTEXT_TYPES, logger } from "./logger.js";
+import {
+  DEFAULT_EXCLUDE_LOG_CONTEXTS,
+  LOG_CONTEXT_TYPES,
+  logger,
+} from "./logger.js";
 
 // ============================================================================
 class ArgParser {
@@ -225,6 +229,14 @@ class ArgParser {
         coerce,
       },
 
+      logExcludeContext: {
+        describe: "Comma-separated list of contexts to NOT include in logs",
+        type: "array",
+        default: DEFAULT_EXCLUDE_LOG_CONTEXTS,
+        choices: LOG_CONTEXT_TYPES,
+        coerce,
+      },
+
       text: {
         describe:
           "Extract initial (default) or final text to pages.jsonl or WARC resource record(s)",

src/util/logger.ts

@@ -26,6 +26,7 @@ export const LOG_CONTEXT_TYPES = [
   "general",
   "worker",
   "recorder",
+  "recorderNetwork",
   "writer",
   "state",
   "redis",
@@ -51,13 +52,20 @@ export const LOG_CONTEXT_TYPES = [
 
 export type LogContext = (typeof LOG_CONTEXT_TYPES)[number];
 
+export const DEFAULT_EXCLUDE_LOG_CONTEXTS: LogContext[] = [
+  "recorderNetwork",
+  "jsError",
+  "screencast",
+];
+
 // ===========================================================================
 class Logger {
   logStream: Writable | null = null;
   debugLogging = false;
   logErrorsToRedis = false;
   logLevels: string[] = [];
-  contexts: string[] = [];
+  contexts: LogContext[] = [];
+  excludeContexts: LogContext[] = [];
   crawlState?: RedisCrawlState | null = null;
   fatalExitCode = 17;
 
@@ -81,18 +89,22 @@ class Logger {
     this.logLevels = logLevels;
   }
 
-  setContext(contexts: string[]) {
+  setContext(contexts: LogContext[]) {
     this.contexts = contexts;
   }
 
+  setExcludeContext(contexts: LogContext[]) {
+    this.excludeContexts = contexts;
+  }
+
   setCrawlState(crawlState: RedisCrawlState) {
     this.crawlState = crawlState;
   }
 
   logAsJSON(
     message: string,
     dataUnknown: unknown,
-    context: string,
+    context: LogContext,
     logLevel = "info",
   ) {
     // eslint-disable-next-line @typescript-eslint/no-explicit-any
@@ -110,6 +122,12 @@ class Logger {
       }
     }
 
+    if (this.excludeContexts.length) {
+      if (this.excludeContexts.indexOf(context) >= 0) {
+        return;
+      }
+    }
+
     const dataToLog = {
       timestamp: new Date().toISOString(),
       logLevel: logLevel,