Skip to content

Commit

Permalink
Merge pull request #803 from techmatters/CHI-3162-link-transcripts-dir
Browse files Browse the repository at this point in the history
CHI-3162 Link redacted transcripts directory contacts within ai s3 buckets
  • Loading branch information
mythilytm authored Jan 16, 2025
2 parents b19edf3 + f6152df commit 7ab3a8d
Show file tree
Hide file tree
Showing 3 changed files with 22 additions and 13 deletions.
13 changes: 10 additions & 3 deletions .github/workflows/generate-ai-training-ecs.yml
Original file line number Diff line number Diff line change
Expand Up @@ -49,12 +49,11 @@ jobs:
- name: Run New ECS task
id: run-task
run: |
CLUSTER="${{ inputs.environment }}-ecs-cluster"
CLUSTER="development-ecs-cluster"
FAMILY="${{ 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"
Expand All @@ -66,7 +65,15 @@ jobs:
--overrides '{
"containerOverrides": [{
"name": "'"$FAMILY"'",
"command": ["sh", "-c", "'"$COMMAND"'"],
"command": [
"npm",
"run",
"start:generate-ai-training-set",
"${{ inputs.environment }}",
"${{ inputs.helpline-shortcodes }}",
"tl-aselo-ai-${{ inputs.environment }}-us-east-1",
"tl-aselo-ai-${{ inputs.environment }}-us-east-1"
],
"environment": [
{"name": "SSM_REGION", "value": "us-east-1"},
{"name": "AWS_REGION", "value": "us-east-1"}
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -52,15 +52,18 @@ export type TrainingSetContact = {
export const streamTrainingSetContacts = async (
accountSid: HrmAccountId,
): Promise<ReadableStream> => {
const qs = new QueryStream(
pgp.as.format(SELECT_CATEGORIES_SUMMARY_AND_TRANSCRIPTS_SQL, { accountSid }),
[],
{ highWaterMark: HIGH_WATER_MARK },
);
const formattedQuery = pgp.as.format(SELECT_CATEGORIES_SUMMARY_AND_TRANSCRIPTS_SQL, {
accountSid,
});

const qs = new QueryStream(formattedQuery, [], { highWaterMark: HIGH_WATER_MARK });
// Expose the readable stream to the caller as a promise for further pipelining
return new Promise(resolve => {
return new Promise((resolve, reject) => {
db.stream(qs, resultStream => {
resolve(resultStream);
}).catch(error => {
console.error('Error streaming contacts:', error);
reject(error);
});
});
};
7 changes: 3 additions & 4 deletions hrm-domain/scheduled-tasks/generate-ai-training-set/index.ts
Original file line number Diff line number Diff line change
Expand Up @@ -60,10 +60,9 @@ export const generate = async (
environment: 'development' | 'staging' | 'production',
hlShortCodes: string[],
targetBucket: string,
sourceBucket?: string,
sourceBucket: string,
) => {
const accountSidMappings = await lookupAccountSids(environment, hlShortCodes);
console.log('Account SIDs found:');
accountSidMappings.forEach(({ accountSid, shortCode }) => {
console.log(`Account SID for ${shortCode}: ${accountSid}`);
});
Expand All @@ -90,8 +89,8 @@ export const generate = async (
sourceBucket,
);
} catch (error) {
console.info(
`No transcript found for contact ${trainingSetContact.contactId} in ${shortCode} bucket. Skipping...`,
console.debug(
`No transcript found for contact ${trainingSetContact.contactId} in ${trainingSetContact.transcriptBucket}/${trainingSetContact.transcriptKey} . Skipping...`,
);
callback();
return;
Expand Down

0 comments on commit 7ab3a8d

Please sign in to comment.