feat: add additional log messages, added nudge to metadata (#2781)

Co-authored-by: Kim Gardner <kgardnr@gmail.com>

src/core/server/models/comment/revision.ts

@@ -1,12 +1,38 @@
 import { EncodedCommentActionCounts } from "coral-server/models/action/comment";
 
 export interface RevisionMetadata {
+  /**
+   * akismet is true when it was determined to be spam.
+   */
   akismet?: boolean;
+
+  /**
+   * linkCount is the number of links in a revision body that was detected.
+   */
   linkCount?: number;
+
+  /**
+   * perspective stores the detected properties from checking with the
+   * perspective model.
+   */
   perspective?: {
+    /**
+     * score is the percentage likelihood (in decimal form) that the comment
+     * matches the selected model.
+     */
     score: number;
+
+    /**
+     * model is the perspective model used to determine the above score.
+     */
     model: string;
   };
+
+  /**
+   * nudge when true indicates that the comment was written on the first try
+   * without a warning.
+   */
+  nudge?: boolean;
 }
 
 /**

src/core/server/services/comments/pipeline/phases/toxic.ts

@@ -12,7 +12,6 @@ import {
 import { LanguageCode } from "coral-common/helpers";
 import { Omit } from "coral-common/types";
 import { ToxicCommentError } from "coral-server/errors";
-import logger from "coral-server/logger";
 import { ACTION_TYPE } from "coral-server/models/action/comment";
 import { hasFeatureFlag } from "coral-server/models/tenant";
 import {
@@ -33,9 +32,12 @@ export const toxic: IntermediateModerationPhase = async ({
   nudge,
   log,
   htmlStripped,
+  // FEATURE_FLAG:DISABLE_WARN_USER_OF_TOXIC_COMMENT
+  comment: { body },
 }: Pick<
   ModerationPhaseContext,
-  "tenant" | "nudge" | "log" | "htmlStripped"
+  // FEATURE_FLAG:DISABLE_WARN_USER_OF_TOXIC_COMMENT
+  "tenant" | "nudge" | "log" | "htmlStripped" | "comment"
 >): Promise<IntermediatePhaseResult | void> => {
   if (!htmlStripped) {
     return;
@@ -91,7 +93,9 @@ export const toxic: IntermediateModerationPhase = async ({
   const timeout = ms("800ms");
 
   try {
-    logger.trace("checking comment toxicity");
+    // FEATURE_FLAG:DISABLE_WARN_USER_OF_TOXIC_COMMENT
+    log.info({ body }, "checking comment toxicity");
+    // log.trace("checking comment toxicity");
 
     // Pull the custom model out.
     const model = integration.model || TOXICITY_MODEL_DEFAULT;
@@ -116,11 +120,13 @@ export const toxic: IntermediateModerationPhase = async ({
 
     const isToxic = score > threshold;
     if (isToxic) {
-      log.trace({ score, isToxic, threshold, model }, "comment was toxic");
+      // FEATURE_FLAG:DISABLE_WARN_USER_OF_TOXIC_COMMENT
+      log.info({ score, isToxic, threshold, model }, "comment was toxic");
+      // log.trace({ score, isToxic, threshold, model }, "comment was toxic");
 
       // Throw an error if we're nudging instead of recording.
       if (nudge) {
-        // Only if the feature flag for disabling this behavior is not enabled.
+        // FEATURE_FLAG:DISABLE_WARN_USER_OF_TOXIC_COMMENT
         if (
           !hasFeatureFlag(
             tenant,
@@ -152,7 +158,9 @@ export const toxic: IntermediateModerationPhase = async ({
       };
     }
 
-    log.trace({ score, isToxic, threshold }, "comment was not toxic");
+    // FEATURE_FLAG:DISABLE_WARN_USER_OF_TOXIC_COMMENT
+    log.info({ score, isToxic, threshold }, "comment was not toxic");
+    // log.trace({ score, isToxic, threshold }, "comment was not toxic");
 
     return {
       metadata: {

src/core/server/services/comments/pipeline/pipeline.ts

@@ -81,7 +81,13 @@ export const compose = (
     status: GQLCOMMENT_STATUS.NONE,
     body: context.comment.body,
     actions: [],
-    metadata: context.comment.metadata || {},
+    metadata: {
+      // Merge in the passed comment metadata.
+      ...(context.comment.metadata || {}),
+
+      // Add the nudge to the comment metadata.
+      nudge: context.nudge,
+    },
     tags: [],
   };