Create embed.py

services/embed_service/embed.py

@@ -0,0 +1,41 @@
+import os
+from ..utilities.transcripts import VTTTranscriptLoader, DocumentEmbedder
+from ..utilities.setup import get_files
+
+
+
+class EmbeddingService:
+    def __init__(self, conf):
+        self.keys = get_files.get_keys()
+        self.conf = conf
+
+    def __enter__(self):
+        print("Start Embedding Service")
+        return self
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        print("Exiting Embedding Service")
+    
+    def get_transcripts(self, files):
+        # Get filepaths and load them in document format
+        filepaths = [file.name for file in files]
+        loader = VTTTranscriptLoader(filepaths)
+        results = loader.load()
+            
+        return results
+    
+    def run(self, files):
+        # gets the files, cleans them, and loads them into pinecone
+        results = self.get_transcripts(files)
+        
+        doc_embedder =  DocumentEmbedder(
+            api_keys=self.keys,
+            files=results,
+            embedding=self.conf["embeddings"]["embedding"],
+            index_name=self.conf["embeddings"]["index_name"],
+        )
+        
+        # uploads them into pinecone
+        doc_embedder.embed()
+        
+        return "complete"