Merge pull request #382 from trustgraph-ai/fix/import-queues-not-working

Fix/import queues not working

trustgraph-flow/trustgraph/gateway/dispatch/document_embeddings_import.py

@@ -22,6 +22,9 @@ class DocumentEmbeddingsImport:
             pulsar_client, topic = queue, schema = DocumentEmbeddings
         )
 
+    async def start(self):
+        await self.publisher.start()
+
     async def destroy(self):
         self.running.stop()
 

trustgraph-flow/trustgraph/gateway/dispatch/graph_embeddings_import.py

@@ -22,6 +22,9 @@ class GraphEmbeddingsImport:
             pulsar_client, topic = queue, schema = GraphEmbeddings
         )
 
+    async def start(self):
+        await self.publisher.start()
+
     async def destroy(self):
         self.running.stop()
 

trustgraph-flow/trustgraph/gateway/dispatch/manager.py

@@ -160,6 +160,8 @@ class DispatcherManager:
             queue = qconfig,
         )
 
+        await dispatcher.start()
+
         return dispatcher
 
     async def process_flow_export(self, ws, running, params):

trustgraph-flow/trustgraph/gateway/dispatch/triples_import.py

@@ -22,6 +22,9 @@ class TriplesImport:
             pulsar_client, topic = queue, schema = Triples
         )
 
+    async def start(self):
+        await self.publisher.start()
+
     async def destroy(self):
         self.running.stop()
 

trustgraph-ocr/trustgraph/decoding/ocr/pdf_decoder.py

@@ -10,39 +10,42 @@ import pytesseract
 from pdf2image import convert_from_bytes
 
 from ... schema import Document, TextDocument, Metadata
-from ... schema import document_ingest_queue, text_ingest_queue
+from ... base import FlowProcessor, ConsumerSpec, ProducerSpec
-from ... log_level import LogLevel
-from ... base import ConsumerProducer
 
-module = "ocr"
+default_ident = "pdf-decoder"
 
-default_input_queue = document_ingest_queue
+class Processor(FlowProcessor):
-default_output_queue = text_ingest_queue
-default_subscriber = module
-
-class Processor(ConsumerProducer):
 
     def __init__(self, **params):
 
-        input_queue = params.get("input_queue", default_input_queue)
+        id = params.get("id", default_ident)
-        output_queue = params.get("output_queue", default_output_queue)
-        subscriber = params.get("subscriber", default_subscriber)
 
         super(Processor, self).__init__(
             **params | {
-                "input_queue": input_queue,
+                "id": id,
-                "output_queue": output_queue,
-                "subscriber": subscriber,
-                "input_schema": Document,
-                "output_schema": TextDocument,
             }
         )
 
+        self.register_specification(
+            ConsumerSpec(
+                name = "input",
+                schema = Document,
+                handler = self.on_message,
+            )
+        )
+
+        self.register_specification(
+            ProducerSpec(
+                name = "output",
+                schema = TextDocument,
+            )
+        )
+
         print("PDF OCR inited")
 
-    async def handle(self, msg):
+    async def on_message(self, msg, consumer, flow):
 
-        print("PDF message received")
+        print("PDF message received", flush=True)
 
         v = msg.value()
 
@@ -65,19 +68,15 @@ class Processor(ConsumerProducer):
                 text=text.encode("utf-8"),
             )
 
-            await self.send(r)
+            await flow("output").send(r)
 
         print("Done.", flush=True)
 
     @staticmethod
     def add_args(parser):
-
+        FlowProcessor.add_args(parser)
-        ConsumerProducer.add_args(
-            parser, default_input_queue, default_subscriber,
-            default_output_queue,
-        )
 
 def run():
 
-    Processor.launch(module, __doc__)
+    Processor.launch(default_ident, __doc__)