jaxl-python 0.0.7__tar.gz → 0.0.9__tar.gz

{jaxl_python-0.0.7 → jaxl_python-0.0.9}/.gitignore

@@ -26,6 +26,8 @@ _scm_version.py
 *.log
 .DS_Store
 license.py
+schema.py
+schema.yaml
 
 jaxl-api-credentials*.json
 docs

{jaxl_python-0.0.7 → jaxl_python-0.0.9}/PKG-INFO

@@ -1,7 +1,7 @@
 Metadata-Version: 2.4
 Name: jaxl-python
-Version: 0.0.7
-Summary: Jaxl CLI & Jaxl Python API Client.
+Version: 0.0.9
+Summary: Official Python SDK and CLI for interfacing with the Jaxl API. Manage calls, messages, IVRs, devices, teams, payments, campaigns, streaming transcriptions and AI Agents. Built on OpenAPI, it offers both generated API clients and Pythonic wrappers for ease of use.
 Home-page: https://github.com/jaxl-innovations-private-limited/jaxl-python
 Download-URL: https://github.com/jaxl-innovations-private-limited/jaxl-python/archive/main.zip
 Author: Jaxl Innovations Private Limited

{jaxl_python-0.0.7 → jaxl_python-0.0.9}/examples/README.md

@@ -35,6 +35,29 @@ pip install -U jaxl-python[grout]
 
 ## Run
 
+View available flags:
+
+```bash
+jaxl apps run -h
+usage: jaxl apps run [-h] [--app APP] [--host HOST] [--port PORT] [--transcribe] [--transcribe-model-size TRANSCRIBE_MODEL_SIZE] [--transcribe-language TRANSCRIBE_LANGUAGE]
+                     [--transcribe-device TRANSCRIBE_DEVICE]
+
+options:
+  -h, --help            show this help message and exit
+  --app APP             Dotted path to Jaxl SDK App module to run e.g. examples.app:JaxlApp
+  --host HOST           Defaults to 127.0.0.1
+  --port PORT           Defaults to 9919
+  --transcribe          This flag is required to enable realtime transcription pipeline
+  --transcribe-model-size TRANSCRIBE_MODEL_SIZE
+                        Options are: tiny, base, small, medium, large
+  --transcribe-language TRANSCRIBE_LANGUAGE
+                        Options are: auto, ar, bg, bn, cs, da, de, el, en, es, et, fi, fil, fr, gu, he, hi, hr, hu, id, it, ja, kn, ko, lt, lv, ml, mr, nl, no, pa, pl, pt, ro, ru, sk, sl, sr, sv, ta, te, th, tr, uk, ur, vi, zh
+  --transcribe-device TRANSCRIBE_DEVICE
+                        Options are: auto, cpu, cuda, cuda:N, mps
+```
+
+Run your Jaxl SDK app:
+
 ```bash
 jaxl apps run --app <Module:ClassName>
 ```
@@ -43,9 +66,10 @@ jaxl apps run --app <Module:ClassName>
 
 You will need to expose your IVR app publicly so that Jaxl servers can reach your app.
 
-In a separate terminal, start `grout` to get a public URL:
+In a separate terminal, start [`grout`](https://github.com/abhinavsingh/proxy.py?tab=readme-ov-file#grout-ngrok-alternative) to get a public URL:
 
 ```bash
+pip install -U jaxl-python[grout]
 grout http://127.0.0.1:9919
 ```
 
@@ -94,19 +118,23 @@ PYTHONPATH=. jaxl apps run --app examples:JaxlAppStreamingAudioChunk
 ### Realtime Streaming Speech Segments
 
 ```bash
+pip install -U jaxl-python[silence]
 PYTHONPATH=. jaxl apps run --app examples:JaxlAppStreamingSpeechSegment
 ```
 
 ### Realtime Streaming Transcriptions per Speech Segment
 
 ```bash
-PYTHONPATH=. jaxl apps run --app examples:JaxlAppStreamingTranscription
+pip install -U jaxl-python[silence,transcribe]
+PYTHONPATH=. jaxl apps run --app examples:JaxlAppStreamingTranscription --transcribe
 ```
 
 ### AI Agent: Realtime Transcriptions STT ➡️ LLM/MCP ➡️ TTS
 
 ```bash
-PYTHONPATH=. jaxl apps run --app examples:JaxlAppStreamingAIAgent
+pip install -U jaxl-python[silence,transcribe]
+export JAXL_OLLAMA_URL=https://<llm.domain>/api/chat
+PYTHONPATH=. jaxl apps run --app examples:JaxlAppStreamingAIAgent --transcribe
 ```
 
 ## Production

{jaxl_python-0.0.7 → jaxl_python-0.0.9}/examples/streaming_aiagent.py

@@ -10,7 +10,7 @@ with or without modification, is strictly prohibited.
 import asyncio
 import logging
 import os
-from typing import Any, Dict, Optional
+from typing import Any, Dict, List, Optional
 
 from jaxl.api.base import (
     HANDLER_RESPONSE,
@@ -57,11 +57,12 @@ class JaxlAppStreamingAIAgent(BaseJaxlApp):
                 "content": _get_system_prompt("Example Company", "example.com"),
             },
         ]
+        self._chunks: List[str] = []
 
     async def handle_setup(self, req: JaxlWebhookRequest) -> HANDLER_RESPONSE:
         return JaxlWebhookResponse(
             prompt=["Welcome to AI agent demo"],
-            num_characters=1,
+            num_characters=-1,
         )
 
     async def handle_teardown(self, req: JaxlWebhookRequest) -> HANDLER_RESPONSE:
@@ -75,7 +76,7 @@ class JaxlAppStreamingAIAgent(BaseJaxlApp):
         req: JaxlStreamRequest,
         transcription: Dict[str, Any],
         num_inflight_transcribe_requests: int,
-    ) -> HANDLER_RESPONSE:
+    ) -> None:
         logging.debug(
             "📝 %s %d",
             transcription["text"],
@@ -90,22 +91,32 @@ class JaxlAppStreamingAIAgent(BaseJaxlApp):
             )
             self._ctask.cancel()
             self._ctask = None
-        self._ctask = asyncio.create_task(self._chat_with_llm(transcription["text"]))
+        self._ctask = asyncio.create_task(
+            self._chat_with_llm(req, transcription["text"])
+        )
         return None
 
-    async def _chat_with_llm(self, transcription: str) -> None:
+    async def _chat_with_llm(self, req: JaxlStreamRequest, transcription: str) -> None:
         url = os.environ.get("JAXL_OLLAMA_URL", None)
         assert url is not None
         self._messages.append({"role": "user", "content": transcription})
 
+        async def _on_llm_response_chunk(response: Optional[Dict[str, Any]]) -> None:
+            await self._on_llm_response_chunk(req, response)
+
         logger.debug("💬 %s", transcription)
         await self.chat_with_ollama(
-            on_response_chunk_callback=self._on_llm_response_chunk,
+            on_response_chunk_callback=_on_llm_response_chunk,
             url=url,
             messages=self._messages,
         )
 
-    async def _on_llm_response_chunk(self, response: Optional[Dict[str, Any]]) -> None:
+    async def _on_llm_response_chunk(
+        self,
+        req: JaxlStreamRequest,
+        response: Optional[Dict[str, Any]],
+    ) -> None:
+        assert req.state
         if response is None:
             logger.warning("❌ %s", "Unable to get agent response")
             self._ctask = None
@@ -113,5 +124,8 @@ class JaxlAppStreamingAIAgent(BaseJaxlApp):
         if response["done"]:
             logger.debug("🎭 %s", "End of agent response")
             self._ctask = None
+            reply = "".join(self._chunks)
+            await self.tts(req.state.call_id, prompts=[reply])
+            self._chunks = []
             return
-        logger.debug("🕵️ %s", response["message"]["content"])
+        self._chunks.append(response["message"]["content"])

{jaxl_python-0.0.7 → jaxl_python-0.0.9}/examples/streaming_audio_chunks.py

@@ -7,6 +7,8 @@ Redistribution and use in source and binary forms,
 with or without modification, is strictly prohibited.
 """
 
+import logging
+
 from jaxl.api.base import (
     HANDLER_RESPONSE,
     BaseJaxlApp,
@@ -21,13 +23,13 @@ class JaxlAppStreamingAudioChunk(BaseJaxlApp):
     async def handle_setup(self, req: JaxlWebhookRequest) -> HANDLER_RESPONSE:
         return JaxlWebhookResponse(
             prompt=["Welcome to streaming audio chunk demo"],
-            num_characters=1,
+            num_characters=-1,
         )
 
     async def handle_audio_chunk(
         self,
         req: JaxlStreamRequest,
         slin16: bytes,
-    ) -> HANDLER_RESPONSE:
-        print(f"Received {len(slin16)} bytes of raw audio")
+    ) -> None:
+        logging.info(f"Received {len(slin16)} bytes of raw audio")
         return None

{jaxl_python-0.0.7 → jaxl_python-0.0.9}/examples/streaming_speech_segments.py

@@ -7,6 +7,7 @@ Redistribution and use in source and binary forms,
 with or without modification, is strictly prohibited.
 """
 
+import logging
 from typing import List
 
 from jaxl.api.base import (
@@ -23,13 +24,13 @@ class JaxlAppStreamingSpeechSegment(BaseJaxlApp):
     async def handle_setup(self, req: JaxlWebhookRequest) -> HANDLER_RESPONSE:
         return JaxlWebhookResponse(
             prompt=["Welcome to streaming speech audio segments demo"],
-            num_characters=1,
+            num_characters=-1,
         )
 
     async def handle_speech_segment(
         self,
         req: JaxlStreamRequest,
         slin16s: List[bytes],
-    ) -> HANDLER_RESPONSE:
-        print(f"Received {len(slin16s)} chunks in speech segment")
+    ) -> None:
+        logging.info(f"Received {len(slin16s)} chunks in speech segment")
         return None

{jaxl_python-0.0.7 → jaxl_python-0.0.9}/examples/streaming_transcriptions.py

@@ -7,6 +7,7 @@ Redistribution and use in source and binary forms,
 with or without modification, is strictly prohibited.
 """
 
+import logging
 from typing import Any, Dict
 
 from jaxl.api.base import (
@@ -23,7 +24,7 @@ class JaxlAppStreamingTranscription(BaseJaxlApp):
     async def handle_setup(self, req: JaxlWebhookRequest) -> HANDLER_RESPONSE:
         return JaxlWebhookResponse(
             prompt=["Welcome to streaming transcriptions demo"],
-            num_characters=1,
+            num_characters=-1,
         )
 
     async def handle_transcription(
@@ -31,6 +32,8 @@ class JaxlAppStreamingTranscription(BaseJaxlApp):
         req: JaxlStreamRequest,
         transcription: Dict[str, Any],
         num_inflight_transcribe_requests: int,
-    ) -> HANDLER_RESPONSE:
-        print(transcription["text"], num_inflight_transcribe_requests)
+    ) -> None:
+        assert req.state
+        logging.info(transcription["text"], num_inflight_transcribe_requests)
+        await self.tts(req.state.call_id, prompts=[transcription["text"]])
         return None