Spaces:
Sleeping
Sleeping
File size: 10,623 Bytes
ff7b988 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 | import json
import multiprocessing
import pathlib
import traceback
from enum import Enum
from flask import Flask, abort, jsonify, request, send_file
from ...infer import Status as SheetSageStatus
from ...infer import sheetsage
from ...utils import compute_checksum, retrieve_audio_bytes
APP = Flask(__name__)
class JobStatus(Enum):
QUEUED = 0
FETCHING = 1
RUNNING = 2
FINALIZED = 3
class JobError(Exception):
pass
class FetchAudioError(JobError):
pass
class BulkyAudioError(JobError):
pass
_MANAGER = multiprocessing.Manager()
_JOB_QUEUE = _MANAGER.Queue()
_JOB_INPUTS = _MANAGER.dict()
_JOB_STATUS = _MANAGER.dict()
_JOB_OUTPUTS = _MANAGER.dict()
def _work(wid):
while True:
print(f"(WID {wid}) Waiting for job")
jid = _JOB_QUEUE.get()
job_def = _JOB_INPUTS[jid]
print(f"(WID {wid}) Working on {jid}:\n{job_def}")
def status_change_callback(s):
print(f"(WID {wid}) Status update for {jid}: {s.name}")
assert isinstance(s, JobStatus) or isinstance(s, SheetSageStatus)
_JOB_STATUS[jid] = s
output = None
stack_trace = None
# Fetch audio
if isinstance(job_def["audio_path_bytes_or_url"], str):
status_change_callback(JobStatus.FETCHING)
try:
audio_bytes = retrieve_audio_bytes(
job_def["audio_path_bytes_or_url"],
max_filesize_mb=ARGS["fetch_max_filesize_mb"],
max_duration_seconds=ARGS["fetch_max_duration_seconds"],
timeout=ARGS["fetch_timeout_seconds"],
)
job_def["audio_path_bytes_or_url"] = audio_bytes
except ValueError:
output = BulkyAudioError()
stack_trace = traceback.format_exc()
except Exception:
output = FetchAudioError()
stack_trace = traceback.format_exc()
# Run
if stack_trace is None:
status_change_callback(JobStatus.RUNNING)
try:
lead_sheet, segment_beats, segment_beats_times = sheetsage(
**job_def, status_change_callback=status_change_callback
)
output_path = pathlib.Path(ARGS["tmp_dir"], f"{jid}.json")
with open(output_path, "w") as f:
f.write(
json.dumps(
{
"lead_sheet": lead_sheet,
"segment_beats": segment_beats,
"segment_beats_times": segment_beats_times,
}
)
)
output = output_path
except Exception as e:
output = JobError()
stack_trace = traceback.format_exc()
# Finalize
print(f"(WID {wid}) Finalizing {jid}")
assert isinstance(output, pathlib.Path) or isinstance(output, JobError)
_JOB_OUTPUTS[jid] = output
if isinstance(output, pathlib.Path):
status_change_callback(JobStatus.FINALIZED)
else:
assert stack_trace is not None
print(f"(WID {wid}) Exception during {jid}:\n{stack_trace.strip()}")
@APP.errorhandler(400)
@APP.errorhandler(500)
def _api_error(e):
return jsonify(e.description), e.code
@APP.route("/ping", methods=["GET"])
def ping():
return "Pong", 200
@APP.route("/submit", methods=["POST"])
def submit():
# Check payload size
if ARGS["max_payload_size_mb"] is not None and request.content_length > (
ARGS["max_payload_size_mb"] * 1024 * 1024
):
abort(413, description="Too large")
# Define arguments
arg_to_sanitize_fn = {
"audio_url": str,
"audio_file": None,
"segment_start_hint": float,
"segment_end_hint": float,
"legacy_behavior": lambda i: bool(int(i)),
"melody_threshold": float,
"harmony_threshold": float,
}
# Check arguments
if request.json is not None:
r = dict(request.json)
elif request.form is not None:
r = dict(request.form)
else:
abort(400, description="Unknown request format")
for k in r.keys():
if k not in arg_to_sanitize_fn:
abort(400, description=f"Unknown argument: {k}")
# Sanitize arguments
for k, fn in arg_to_sanitize_fn.items():
if k in r and fn is not None:
try:
r[k] = fn(r[k])
except:
abort(400, description=f"Bad '{k}'")
# Create job definition
job_def = {
"audio_path_bytes_or_url": None,
"segment_start_hint": None,
"segment_end_hint": None,
"use_jukebox": ARGS["jukebox"],
"legacy_behavior": False,
"melody_threshold": None,
"harmony_threshold": None,
}
# Parse audio_url and audio_file
audio_file = request.files.get("audio_file")
if audio_file is not None:
# Audio was uploaded
try:
audio_mimetype = audio_file.content_type
audio_file_bytes = BytesIO()
audio_file.save(audio_file_bytes)
audio_file_bytes.seek(0)
audio_file_bytes = audio_file_bytes.read()
audio_file_checksum = compute_checksum(audio_file_bytes, algorithm="sha256")
except:
abort(400, description="Bad 'audio_file'")
try:
audio_path = pathlib.Path(ARGS["tmp_dir"], "audio", audio_file_checksum)
audio_path.parent.mkdir(parents=True, exist_ok=True)
if not audio_path.is_file():
with open(audio_path, "wb") as f:
f.write(audio_file_bytes)
except:
abort(500)
job_def["audio_path_bytes_or_url"] = audio_path
elif "audio_url" in r:
# Media needs to be retrieved from URL
try:
audio_url = r["audio_url"].strip()
assert len(audio_url) > 0
except:
abort(400, description="Bad 'audio_url'")
job_def["audio_path_bytes_or_url"] = audio_url
else:
abort(400, description="No audio specified")
# Parse float args
for k in [
"segment_start_hint",
"segment_end_hint",
"legacy_behavior",
"melody_threshold",
"harmony_threshold",
]:
if k in r:
job_def[k] = r[k]
# Compute job ID
jid = compute_checksum(
json.dumps(job_def, sort_keys=True, indent=2).encode("utf-8"),
algorithm="sha1",
)
# Submit to queue
position = None
status = _JOB_STATUS.get(jid)
output = _JOB_OUTPUTS.get(jid)
actively_processing = output is None and status is not None
already_cached = isinstance(output, pathlib.Path) and output.is_file()
if not (actively_processing or already_cached):
position = _JOB_QUEUE.qsize()
_JOB_INPUTS[jid] = job_def
_JOB_STATUS[jid] = JobStatus.QUEUED
if output is not None:
del _JOB_OUTPUTS[jid]
_JOB_QUEUE.put(jid)
return {"jid": jid, "cached": already_cached, "position": position}
@APP.route("/heartbeat/<jid>", methods=["GET"])
def heartbeat(jid):
status = _JOB_STATUS.get(jid)
if status is None:
abort(404, description="INVALID_ID")
output = _JOB_OUTPUTS.get(jid)
if isinstance(output, BulkyAudioError):
abort(400, description="AUDIO_TOO_LONG_OR_TOO_BIG")
elif isinstance(output, JobError):
abort(500, description=status.name)
return jsonify(status.name)
@APP.route("/lead-sheet/<jid>", methods=["GET"])
def download(jid):
if isinstance(jid, str) and jid.endswith(".json"):
jid = jid[:-5]
output = _JOB_OUTPUTS.get(jid)
if output is None:
abort(404, description="INVALID_ID")
if not isinstance(output, pathlib.Path):
abort(500)
return send_file(output, download_name=f"{jid}.json", max_age=7 * 24 * 60 * 60)
def __init():
import os
from argparse import ArgumentParser
from flask_cors import CORS
parser = ArgumentParser()
parser.add_argument("--port", type=int)
parser.add_argument("--cors", action="store_true")
parser.add_argument("--cors_allow", type=str)
parser.add_argument("--ssl_crt_path", type=str)
parser.add_argument("--ssl_key_path", type=str)
parser.add_argument("--jukebox", action="store_true")
parser.add_argument("--num_workers", type=int)
parser.add_argument("--max_payload_size_mb", type=int)
parser.add_argument("--fetch_max_filesize_mb", type=int)
parser.add_argument("--fetch_max_duration_seconds", type=float)
parser.add_argument("--fetch_timeout_seconds", type=int)
parser.add_argument("--tmp_dir", type=str)
parser.set_defaults(
port=8000,
cors=False,
cors_allow=None,
ssl_crt_path=None,
ssl_key_path=None,
jukebox=False,
num_workers=1,
max_payload_size_mb=32,
fetch_max_filesize_mb=128,
fetch_max_duration_seconds=660,
fetch_timeout_seconds=60,
tmp_dir="/tmp/sheetsage",
)
global ARGS
ARGS = vars(parser.parse_args())
print(ARGS)
# Indicate that Jukebox support is forthcoming
if ARGS["jukebox"] and ARGS["num_workers"] > 1:
raise NotImplementedError()
# Enable CORS
if ARGS["cors"] or ARGS["cors_allow"] is not None:
kwargs = {}
if ARGS["cors_allow"] is not None:
kwargs["origins"] = [o.strip() for o in ARGS["cors_allow"].split(",")]
CORS(APP, **kwargs)
# Create tmp dir
ARGS["tmp_dir"] = pathlib.Path(ARGS["tmp_dir"])
ARGS["tmp_dir"].mkdir(parents=True, exist_ok=True)
# Worker processes
if ARGS["num_workers"] <= 0:
raise ValueError()
processes = [
multiprocessing.Process(target=_work, args=(wid,))
for wid in range(ARGS["num_workers"])
]
[p.start() for p in processes]
# Start HTTP server
gunicorn = "gunicorn" in os.environ.get("SERVER_SOFTWARE", "")
if not gunicorn:
kwargs = {
"debug": True,
"use_reloader": True,
"host": "0.0.0.0",
"port": ARGS["port"],
}
if ARGS["ssl_crt_path"] is not None and ARGS["ssl_key_path"] is not None:
kwargs["ssl_context"] = (ARGS["ssl_crt_path"], ARGS["ssl_key_path"])
APP.run(**kwargs)
# Join workers
[p.join() for p in processes]
if __name__ == "__main__":
__init()
|