change to sourceBucket

techmatters · Dec 30, 2024 · 06d6113 · 06d6113
1 parent bb99194
commit 06d6113
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 12 deletions.
diff --git a/.github/workflows/generate-ai-training-ecs.yml b/.github/workflows/generate-ai-training-ecs.yml
@@ -15,6 +15,9 @@
 name: Generate AI Training Set Task
 
 on:
+  push:
+    branches:
+      - CHI-3162-link-transcripts-dir
   workflow_dispatch:
     inputs:
       environment:
@@ -49,14 +52,14 @@ jobs:
       - name: Run New ECS task
         id: run-task
         run: |
-          CLUSTER="${{ inputs.environment }}-ecs-cluster"
-          FAMILY="${{ inputs.environment }}-hrm-scheduled-task"
-          SUBNETS=$(if [ ${{ inputs.environment }} == "development" ]; then echo "subnet-034e5c652dbad09dd"; else echo "subnet-00b4e88c0ea178f3d"; fi)
-          SECURITY_GROUPS=($(if [ ${{ inputs.environment}} == "development" ]; then echo "sg-09194f9a648baf082 sg-047498d4c7b2cedd8 sg-06d6458accc0ec5ed sg-0ace4338c75e5d3de"; 
-            else echo "sg-02fd053fe7b4660ed sg-0e7119c4423f2c7b0 sg-066d0055caa7cbe90 sg-0c7fb74e11cb64ff8"; fi))
-          COMMAND="npm run start:generate-ai-training-set ${{ inputs.environment }} ${{ inputs.helpline-shortcodes }} tl-aselo-ai-${{ inputs.environment }}-us-east-1"
-          TASK_ROLE="arn:aws:iam::${{ secrets.AWS_ACCOUNT_ID }}:role/${{ inputs.environment }}-ecsTaskRole"
-          EXEC_ROLE="arn:aws:iam::${{ secrets.AWS_ACCOUNT_ID }}:role/${{ inputs.environment }}-ecsTaskExecutionRole"
+          CLUSTER="development-ecs-cluster"
+          FAMILY="development-hrm-scheduled-task"
+          SUBNETS="subnet-034e5c652dbad09dd"
+          SECURITY_GROUPS="sg-09194f9a648baf082,sg-047498d4c7b2cedd8,sg-06d6458accc0ec5ed,sg-0ace4338c75e5d3de"
+          COMMAND="npm run start:generate-ai-training-set development as tl-aselo-ai-development tl-aselo-ai-development"
+          TASK_ROLE="arn:aws:iam::712893914485:role/development-ecsTaskRole"
+          EXEC_ROLE="arn:aws:iam::712893914485:role/development-ecsTaskExecutionRole"
+
 
           aws ecs run-task \
             --cluster $CLUSTER \

diff --git a/hrm-domain/scheduled-tasks/generate-ai-training-set/index.ts b/hrm-domain/scheduled-tasks/generate-ai-training-set/index.ts
@@ -90,16 +90,19 @@ export const generate = async (
               sourceBucket,
             );
           } catch (error) {
-            console.info(
-              `No transcript found for contact ${trainingSetContact.contactId} in ${shortCode} bucket. Skipping...`,
-            );
+            // console.info(
+            //   `No transcript found for contact ${trainingSetContact.contactId} in ${shortCode} bucket. Skipping...`,
+            // );
             callback();
             return;
           }
+          console.log(
+            `Uploading contact ${trainingSetDoc.contactId} from ${sourceBucket}`,
+          );
           await uploadTrainingSetDocument(
             trainingSetDoc.contactId,
             JSON.stringify(trainingSetDoc),
-            targetBucket,
+            sourceBucket,
             shortCode,
           );
 

diff --git a/hrm-domain/scheduled-tasks/generate-ai-training-set/trainingSetDocument.ts b/hrm-domain/scheduled-tasks/generate-ai-training-set/trainingSetDocument.ts
@@ -45,6 +45,8 @@ export const attachTranscript = async (
     ? `${shortCode.toLowerCase()}/${trainingSetContact.transcriptKey}`
     : trainingSetContact.transcriptKey;
 
+  console.log(`Reading transcript from ${readBucket}/${readKey}`);
+
   const transcriptDocJson = await getS3Object({
     key: readKey,
     bucket: readBucket,