Merge pull request #1831 from boazsender/automated-transcriptions

Add sentence retreival and storage to transcription webhook.

Author: Mephistic

functions/src/events/scrapeEvents.ts

@@ -281,7 +281,7 @@ const shouldScrapeVideo = async (EventId: number) => {
   if (!eventData) {
     return false
   }
-  if (!eventData.videoFetchedAt) {
+  if (!eventData.videoURL) {
     return withinCutoff(new Date(Hearing.check(eventData).startsAt.toDate()))
   }
   return false

functions/src/webhooks/transcription.ts

@@ -19,6 +19,7 @@ export const transcription = functions
         const transcript = await assembly.transcripts.get(
           req.body.transcript_id
         )
+
         if (transcript && transcript.webhook_auth) {
           // If there is a transcript and the transcript has an auth property,
           // look for an event (aka Hearing) in the DB with a matching ID.
@@ -67,6 +68,10 @@ export const transcription = functions
             if (authenticatedEventIds.length === 1) {
               // If there is one authenticated event, pull out the parts we want to
               // save and try to save them in the db.
+
+              const { paragraphs } = await assembly.transcripts.paragraphs(
+                transcript.id
+              )
               const { id, text, audio_url, utterances } = transcript
               try {
                 const transcriptionInDb = await db
@@ -102,6 +107,24 @@ export const transcription = functions
                   await writer.close()
                 }
 
+                if (paragraphs) {
+                  const writer = db.bulkWriter()
+                  for (let paragraph of paragraphs) {
+                    const { confidence, start, end, text } = paragraph
+
+                    writer.set(
+                      db
+                        .collection("transcriptions")
+                        .doc(`${transcript.id}`)
+                        .collection("paragraphs")
+                        .doc(),
+                      { confidence, start, end, text }
+                    )
+                  }
+
+                  await writer.close()
+                }
+
                 // Delete the hashed webhook auth token from our db now that
                 // we're done.
                 for (const index in authenticatedEventIds) {