Commit @94c75432f4ac1f02aed03a31b42bd5017441065f - yjyoon/whisper_server

9f56267

94c7543

.pre-commit-config.yaml

--- .pre-commit-config.yaml

+++ .pre-commit-config.yaml


       - id: end-of-file-fixer
       - id: check-yaml
       - id: check-added-large-files
-
-  - repo: https://github.com/pre-commit/mirrors-mypy
-    rev: v1.10.0
-    hooks:
-      - id: mypy
-
-  # - repo: https://github.com/PyCQA/isort
-  #   rev: 5.13.2
+  # TODO: enable
+  # - repo: https://github.com/pre-commit/mirrors-mypy
+  #   rev: v1.10.0
   #   hooks:
-  #     - id: isort
-  #
-  # - repo: https://github.com/psf/black
-  #   rev: 24.4.2
+  #     - id: mypy
+  #       args: [--strict]
+  # TODO: enable
+  # - repo: https://github.com/RobertCraigie/pyright-python
+  #   rev: v1.1.363
   #   hooks:
-  #     - id: black
+  #   - id: pyright
+  # Disabled because it doesn't work on NixOS
+  # - repo: https://github.com/astral-sh/ruff-pre-commit
+  #   rev: v0.4.4
+  #   hooks:
+  #     - id: ruff # linter
+  #       args: [--fix]
+  #     - id: ruff-format

9f56267

94c7543

flake.nix

--- flake.nix

+++ flake.nix


               pv
               pyright
               python311
+              ruff
               websocat
             ];
             shellHook = ''

9f56267

94c7543

pyproject.toml

--- pyproject.toml

+++ pyproject.toml


 httpx = "^0.27.0"
 httpx-ws = "^0.6.0"
 
+[tool.ruff]
+target-version = "py311"
+
 [build-system]
 requires = ["poetry-core"]
 build-backend = "poetry.core.masonry.api"

9f56267

94c7543

speaches/audio.py

--- speaches/audio.py

+++ speaches/audio.py


 
 
 def audio_samples_from_file(file: BinaryIO) -> NDArray[np.float32]:
-    audio_and_sample_rate: tuple[NDArray[np.float32], Any] = sf.read(  # type: ignore
+    audio_and_sample_rate = sf.read(
         file,
         format="RAW",
         channels=1,

         endian="LITTLE",
     )
     audio = audio_and_sample_rate[0]
-    return audio
+    return audio  # type: ignore
 
 
 class Audio:

         self.modify_event = asyncio.Event()
 
     def extend(self, data: NDArray[np.float32]) -> None:
-        assert self.closed == False
+        assert not self.closed
         super().extend(data)
         self.modify_event.set()
 
     def close(self) -> None:
-        assert self.closed == False
+        assert not self.closed
         self.closed = True
         self.modify_event.set()
         logger.info("AudioStream closed")

9f56267

94c7543

speaches/core.py

--- speaches/core.py

+++ speaches/core.py


 
 
 def test_segment_is_eos():
-    assert Segment("Hello").is_eos == False
-    assert Segment("Hello...").is_eos == False
-    assert Segment("Hello.").is_eos == True
-    assert Segment("Hello!").is_eos == True
-    assert Segment("Hello?").is_eos == True
-    assert Segment("Hello. Yo").is_eos == False
-    assert Segment("Hello. Yo...").is_eos == False
-    assert Segment("Hello. Yo.").is_eos == True
+    assert not Segment("Hello").is_eos
+    assert not Segment("Hello...").is_eos
+    assert Segment("Hello.").is_eos
+    assert Segment("Hello!").is_eos
+    assert Segment("Hello?").is_eos
+    assert not Segment("Hello. Yo").is_eos
+    assert not Segment("Hello. Yo...").is_eos
+    assert Segment("Hello. Yo.").is_eos
 
 
 def to_full_sentences(words: list[Word]) -> list[Segment]:

9f56267

94c7543

speaches/main.py

--- speaches/main.py

+++ speaches/main.py


 from io import BytesIO
 from typing import Annotated
 
-from fastapi import (Depends, FastAPI, Response, UploadFile, WebSocket,
-                     WebSocketDisconnect)
+from fastapi import (
+    Depends,
+    FastAPI,
+    Response,
+    UploadFile,
+    WebSocket,
+    WebSocketDisconnect,
+)
 from fastapi.websockets import WebSocketState
 from faster_whisper import WhisperModel
 from faster_whisper.vad import VadOptions, get_speech_timestamps

 from speaches.config import SAMPLES_PER_SECOND, Language, config
 from speaches.core import Transcription
 from speaches.logger import logger
-from speaches.server_models import (ResponseFormat, TranscriptionResponse,
-                                    TranscriptionVerboseResponse)
+from speaches.server_models import (
+    ResponseFormat,
+    TranscriptionResponse,
+    TranscriptionVerboseResponse,
+)
 from speaches.transcriber import audio_transcriber
 
 whisper: WhisperModel = None  # type: ignore

...	...	@@ -92,14 +92,14 @@
92	92
93	93
94	94	def test_segment_is_eos():
95		- assert Segment("Hello").is_eos == False
96		- assert Segment("Hello...").is_eos == False
97		- assert Segment("Hello.").is_eos == True
98		- assert Segment("Hello!").is_eos == True
99		- assert Segment("Hello?").is_eos == True
100		- assert Segment("Hello. Yo").is_eos == False
101		- assert Segment("Hello. Yo...").is_eos == False
102		- assert Segment("Hello. Yo.").is_eos == True
	95	+ assert not Segment("Hello").is_eos
	96	+ assert not Segment("Hello...").is_eos
	97	+ assert Segment("Hello.").is_eos
	98	+ assert Segment("Hello!").is_eos
	99	+ assert Segment("Hello?").is_eos
	100	+ assert not Segment("Hello. Yo").is_eos
	101	+ assert not Segment("Hello. Yo...").is_eos
	102	+ assert Segment("Hello. Yo.").is_eos
103	103
104	104
105	105	def to_full_sentences(words: list[Word]) -> list[Segment]:

...	...	@@ -8,18 +8,21 @@
8	8	- id: end-of-file-fixer
9	9	- id: check-yaml
10	10	- id: check-added-large-files
11		-
12		- - repo: https://github.com/pre-commit/mirrors-mypy
13		- rev: v1.10.0
14		- hooks:
15		- - id: mypy
16		-
17		- # - repo: https://github.com/PyCQA/isort
18		- # rev: 5.13.2
	11	+ # TODO: enable
	12	+ # - repo: https://github.com/pre-commit/mirrors-mypy
	13	+ # rev: v1.10.0
19	14	# hooks:
20		- # - id: isort
21		- #
22		- # - repo: https://github.com/psf/black
23		- # rev: 24.4.2
	15	+ # - id: mypy
	16	+ # args: [--strict]
	17	+ # TODO: enable
	18	+ # - repo: https://github.com/RobertCraigie/pyright-python
	19	+ # rev: v1.1.363
24	20	# hooks:
25		- # - id: black
	21	+ # - id: pyright
	22	+ # Disabled because it doesn't work on NixOS
	23	+ # - repo: https://github.com/astral-sh/ruff-pre-commit
	24	+ # rev: v0.4.4
	25	+ # hooks:
	26	+ # - id: ruff # linter
	27	+ # args: [--fix]
	28	+ # - id: ruff-format

...	...	@@ -26,6 +26,7 @@
26	26	pv
27	27	pyright
28	28	python311
	29	+ ruff
29	30	websocat
30	31	];
31	32	shellHook = ''

...	...	@@ -27,6 +27,9 @@
27	27	httpx = "^0.27.0"
28	28	httpx-ws = "^0.6.0"
29	29
	30	+[tool.ruff]
	31	+target-version = "py311"
	32	+
30	33	[build-system]
31	34	requires = ["poetry-core"]
32	35	build-backend = "poetry.core.masonry.api"

...	...	@@ -12,7 +12,7 @@
12	12
13	13
14	14	def audio_samples_from_file(file: BinaryIO) -> NDArray[np.float32]:
15		- audio_and_sample_rate: tuple[NDArray[np.float32], Any] = sf.read( # type: ignore
	15	+ audio_and_sample_rate = sf.read(
16	16	file,
17	17	format="RAW",
18	18	channels=1,
...	...	@@ -22,7 +22,7 @@
22	22	endian="LITTLE",
23	23	)
24	24	audio = audio_and_sample_rate[0]
25		- return audio
	25	+ return audio # type: ignore
26	26
27	27
28	28	class Audio:
...	...	@@ -68,12 +68,12 @@
68	68	self.modify_event = asyncio.Event()
69	69
70	70	def extend(self, data: NDArray[np.float32]) -> None:
71		- assert self.closed == False
	71	+ assert not self.closed
72	72	super().extend(data)
73	73	self.modify_event.set()
74	74
75	75	def close(self) -> None:
76		- assert self.closed == False
	76	+ assert not self.closed
77	77	self.closed = True
78	78	self.modify_event.set()
79	79	logger.info("AudioStream closed")

...	...	@@ -7,8 +7,14 @@
7	7	from io import BytesIO
8	8	from typing import Annotated
9	9
10		-from fastapi import (Depends, FastAPI, Response, UploadFile, WebSocket,
11		- WebSocketDisconnect)
	10	+from fastapi import (
	11	+ Depends,
	12	+ FastAPI,
	13	+ Response,
	14	+ UploadFile,
	15	+ WebSocket,
	16	+ WebSocketDisconnect,
	17	+)
12	18	from fastapi.websockets import WebSocketState
13	19	from faster_whisper import WhisperModel
14	20	from faster_whisper.vad import VadOptions, get_speech_timestamps
...	...	@@ -18,8 +24,11 @@
18	24	from speaches.config import SAMPLES_PER_SECOND, Language, config
19	25	from speaches.core import Transcription
20	26	from speaches.logger import logger
21		-from speaches.server_models import (ResponseFormat, TranscriptionResponse,
22		- TranscriptionVerboseResponse)
	27	+from speaches.server_models import (
	28	+ ResponseFormat,
	29	+ TranscriptionResponse,
	30	+ TranscriptionVerboseResponse,
	31	+)
23	32	from speaches.transcriber import audio_transcriber
24	33
25	34	whisper: WhisperModel = None # type: ignore

Delete comment