Update app.py
Browse files
app.py
CHANGED
|
@@ -83,11 +83,9 @@ def verify_api_key(req: Request):
|
|
| 83 |
|
| 84 |
token = None
|
| 85 |
|
| 86 |
-
# دعم x-api-key
|
| 87 |
if x_api_key:
|
| 88 |
token = x_api_key.strip()
|
| 89 |
|
| 90 |
-
# دعم Authorization Bearer
|
| 91 |
elif auth:
|
| 92 |
|
| 93 |
parts = auth.split(" ", 1)
|
|
@@ -95,11 +93,9 @@ def verify_api_key(req: Request):
|
|
| 95 |
if len(parts) == 2:
|
| 96 |
token = parts[1].strip()
|
| 97 |
|
| 98 |
-
# لو لم يتم استخراج توكن صالح
|
| 99 |
if not token:
|
| 100 |
return True
|
| 101 |
|
| 102 |
-
# تحقق اختياري فقط
|
| 103 |
if API_KEY and token != API_KEY:
|
| 104 |
raise HTTPException(
|
| 105 |
status_code=403,
|
|
@@ -125,7 +121,69 @@ async def root():
|
|
| 125 |
|
| 126 |
@app.head("/")
|
| 127 |
async def root_head():
|
| 128 |
-
return JSONResponse(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 129 |
|
| 130 |
|
| 131 |
# =====================================================
|
|
@@ -137,24 +195,37 @@ async def get_models():
|
|
| 137 |
|
| 138 |
models_data = []
|
| 139 |
|
| 140 |
-
|
| 141 |
|
| 142 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 143 |
|
| 144 |
-
|
|
|
|
|
|
|
| 145 |
|
| 146 |
-
|
|
|
|
| 147 |
|
| 148 |
-
|
| 149 |
-
|
| 150 |
-
"object": "model",
|
| 151 |
-
"created": int(time.time()),
|
| 152 |
-
"owned_by": "g4f"
|
| 153 |
-
})
|
| 154 |
|
| 155 |
-
|
|
|
|
|
|
|
|
|
|
| 156 |
|
| 157 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 158 |
|
| 159 |
return {
|
| 160 |
"object": "list",
|
|
@@ -162,46 +233,6 @@ async def get_models():
|
|
| 162 |
}
|
| 163 |
|
| 164 |
|
| 165 |
-
# =====================================================
|
| 166 |
-
# MODEL NORMALIZER
|
| 167 |
-
# =====================================================
|
| 168 |
-
|
| 169 |
-
def normalize_model(model: str):
|
| 170 |
-
|
| 171 |
-
if not model:
|
| 172 |
-
return model
|
| 173 |
-
|
| 174 |
-
model_map = {
|
| 175 |
-
|
| 176 |
-
# Claude
|
| 177 |
-
"claude-sonnet-4-6": "gpt-4o",
|
| 178 |
-
"claude-sonnet-4": "gpt-4o",
|
| 179 |
-
"claude-4-sonnet": "gpt-4o",
|
| 180 |
-
"claude-3-7-sonnet": "gpt-4o",
|
| 181 |
-
"claude-3.7-sonnet": "gpt-4o",
|
| 182 |
-
"claude-opus-4": "gpt-4o",
|
| 183 |
-
"claude-opus-4-6": "gpt-4o",
|
| 184 |
-
"claude-opus-4-7": "gpt-4o",
|
| 185 |
-
"claude-haiku-4": "gpt-4o-mini",
|
| 186 |
-
|
| 187 |
-
# GPT
|
| 188 |
-
"gpt-5": "gpt-4o",
|
| 189 |
-
"gpt-5-mini": "gpt-4o-mini",
|
| 190 |
-
"gpt-4.1": "gpt-4o",
|
| 191 |
-
"gpt-4.1-mini": "gpt-4o-mini",
|
| 192 |
-
|
| 193 |
-
# Gemini
|
| 194 |
-
"gemini-2.5-pro": "gemini",
|
| 195 |
-
"gemini-2.5-flash": "gemini",
|
| 196 |
-
|
| 197 |
-
# Grok
|
| 198 |
-
"grok-4": "gpt-4o",
|
| 199 |
-
"grok-4-fast": "gpt-4o-mini",
|
| 200 |
-
}
|
| 201 |
-
|
| 202 |
-
return model_map.get(model, model)
|
| 203 |
-
|
| 204 |
-
|
| 205 |
# =====================================================
|
| 206 |
# CHAT COMPLETIONS
|
| 207 |
# =====================================================
|
|
@@ -214,7 +245,9 @@ async def chat_completions(
|
|
| 214 |
|
| 215 |
verify_api_key(req)
|
| 216 |
|
| 217 |
-
model = normalize_model(
|
|
|
|
|
|
|
| 218 |
|
| 219 |
messages = [
|
| 220 |
{
|
|
@@ -240,7 +273,7 @@ async def chat_completions(
|
|
| 240 |
|
| 241 |
try:
|
| 242 |
|
| 243 |
-
client =
|
| 244 |
|
| 245 |
response = client.chat.completions.create(
|
| 246 |
model=model,
|
|
@@ -248,8 +281,6 @@ async def chat_completions(
|
|
| 248 |
stream=True
|
| 249 |
)
|
| 250 |
|
| 251 |
-
has_content = False
|
| 252 |
-
|
| 253 |
for chunk in response:
|
| 254 |
|
| 255 |
try:
|
|
@@ -264,13 +295,15 @@ async def chat_completions(
|
|
| 264 |
and chunk.choices[0].delta
|
| 265 |
and chunk.choices[0].delta.content
|
| 266 |
):
|
| 267 |
-
content =
|
|
|
|
|
|
|
|
|
|
|
|
|
| 268 |
|
| 269 |
if not content:
|
| 270 |
continue
|
| 271 |
|
| 272 |
-
has_content = True
|
| 273 |
-
|
| 274 |
payload = {
|
| 275 |
"id": chunk_id,
|
| 276 |
"object": "chat.completion.chunk",
|
|
@@ -302,21 +335,6 @@ async def chat_completions(
|
|
| 302 |
f"Chunk error: {chunk_error}"
|
| 303 |
)
|
| 304 |
|
| 305 |
-
if not has_content:
|
| 306 |
-
|
| 307 |
-
error_payload = {
|
| 308 |
-
"error": {
|
| 309 |
-
"message": "Provider returned empty stream",
|
| 310 |
-
"type": "empty_stream"
|
| 311 |
-
}
|
| 312 |
-
}
|
| 313 |
-
|
| 314 |
-
yield (
|
| 315 |
-
"data: "
|
| 316 |
-
+ json.dumps(error_payload)
|
| 317 |
-
+ "\n\n"
|
| 318 |
-
)
|
| 319 |
-
|
| 320 |
final_payload = {
|
| 321 |
"id": chunk_id,
|
| 322 |
"object": "chat.completion.chunk",
|
|
@@ -345,16 +363,13 @@ async def chat_completions(
|
|
| 345 |
f"Streaming error: {e}"
|
| 346 |
)
|
| 347 |
|
| 348 |
-
error_payload = {
|
| 349 |
-
"error": {
|
| 350 |
-
"message": str(e),
|
| 351 |
-
"type": "server_error"
|
| 352 |
-
}
|
| 353 |
-
}
|
| 354 |
-
|
| 355 |
yield (
|
| 356 |
"data: "
|
| 357 |
-
+ json.dumps(
|
|
|
|
|
|
|
|
|
|
|
|
|
| 358 |
+ "\n\n"
|
| 359 |
)
|
| 360 |
|
|
@@ -371,12 +386,12 @@ async def chat_completions(
|
|
| 371 |
)
|
| 372 |
|
| 373 |
# =================================================
|
| 374 |
-
# NORMAL
|
| 375 |
# =================================================
|
| 376 |
|
| 377 |
try:
|
| 378 |
|
| 379 |
-
client =
|
| 380 |
|
| 381 |
response = client.chat.completions.create(
|
| 382 |
model=model,
|
|
@@ -452,16 +467,16 @@ async def anthropic_messages(
|
|
| 452 |
|
| 453 |
verify_api_key(req)
|
| 454 |
|
|
|
|
|
|
|
| 455 |
model = normalize_model(
|
| 456 |
-
|
| 457 |
)
|
| 458 |
|
| 459 |
messages = body.get("messages", [])
|
| 460 |
|
| 461 |
-
stream = True
|
| 462 |
-
|
| 463 |
logger.info(
|
| 464 |
-
f"Anthropic request model={model} stream=
|
| 465 |
)
|
| 466 |
|
| 467 |
converted_messages = []
|
|
@@ -493,153 +508,187 @@ async def anthropic_messages(
|
|
| 493 |
"content": content
|
| 494 |
})
|
| 495 |
|
| 496 |
-
|
| 497 |
-
# STREAM
|
| 498 |
-
# =============================================
|
| 499 |
-
|
| 500 |
-
if stream:
|
| 501 |
-
|
| 502 |
-
def generate_stream():
|
| 503 |
-
|
| 504 |
-
try:
|
| 505 |
-
|
| 506 |
-
client = Client()
|
| 507 |
|
| 508 |
-
|
| 509 |
-
model=model,
|
| 510 |
-
messages=converted_messages,
|
| 511 |
-
stream=True
|
| 512 |
-
)
|
| 513 |
|
| 514 |
-
|
| 515 |
-
|
| 516 |
-
try:
|
| 517 |
|
| 518 |
-
|
| 519 |
|
| 520 |
-
|
| 521 |
-
|
| 522 |
-
|
| 523 |
-
|
| 524 |
-
|
| 525 |
-
and chunk.choices[0].delta
|
| 526 |
-
and chunk.choices[0].delta.content
|
| 527 |
-
):
|
| 528 |
-
content = chunk.choices[0].delta.content
|
| 529 |
|
| 530 |
-
|
| 531 |
-
|
|
|
|
| 532 |
|
| 533 |
-
|
| 534 |
-
|
| 535 |
-
|
| 536 |
-
|
| 537 |
-
|
| 538 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 539 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
| 540 |
|
| 541 |
-
|
| 542 |
-
|
| 543 |
-
|
| 544 |
-
payload,
|
| 545 |
-
ensure_ascii=False
|
| 546 |
-
)
|
| 547 |
-
+ "\n\n"
|
| 548 |
-
)
|
| 549 |
-
|
| 550 |
-
except Exception as e:
|
| 551 |
-
|
| 552 |
-
logger.error(
|
| 553 |
-
f"Anthropic chunk error: {e}"
|
| 554 |
-
)
|
| 555 |
|
| 556 |
-
|
| 557 |
-
|
| 558 |
-
|
| 559 |
-
|
| 560 |
-
|
| 561 |
-
|
| 562 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 563 |
|
| 564 |
-
|
| 565 |
|
| 566 |
-
|
| 567 |
-
f"Anthropic stream error: {e}"
|
| 568 |
-
)
|
| 569 |
|
| 570 |
-
|
| 571 |
-
"data: "
|
| 572 |
-
+ json.dumps({
|
| 573 |
-
"error": str(e)
|
| 574 |
-
})
|
| 575 |
-
+ "\n\n"
|
| 576 |
-
)
|
| 577 |
|
| 578 |
-
|
| 579 |
-
|
| 580 |
-
|
| 581 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 582 |
|
| 583 |
-
|
| 584 |
-
|
| 585 |
-
# =============================================
|
| 586 |
|
| 587 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 588 |
|
| 589 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 590 |
|
| 591 |
-
|
| 592 |
-
model=model,
|
| 593 |
-
messages=converted_messages,
|
| 594 |
-
stream=False
|
| 595 |
-
)
|
| 596 |
|
| 597 |
-
|
|
|
|
|
|
|
| 598 |
|
| 599 |
-
|
|
|
|
|
|
|
| 600 |
|
| 601 |
-
|
| 602 |
-
|
| 603 |
-
|
| 604 |
-
.
|
| 605 |
-
|
|
|
|
|
|
|
|
|
|
| 606 |
)
|
| 607 |
|
| 608 |
-
|
| 609 |
-
|
| 610 |
-
|
| 611 |
-
|
| 612 |
-
|
| 613 |
-
|
| 614 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 615 |
|
| 616 |
-
|
|
|
|
|
|
|
| 617 |
|
| 618 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 619 |
|
| 620 |
-
|
| 621 |
|
| 622 |
-
|
| 623 |
-
{
|
| 624 |
-
|
| 625 |
-
"text": text
|
| 626 |
-
}
|
| 627 |
-
],
|
| 628 |
|
| 629 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 630 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 631 |
}
|
| 632 |
-
|
| 633 |
-
except Exception as e:
|
| 634 |
-
|
| 635 |
-
logger.error(
|
| 636 |
-
f"Anthropic error: {e}"
|
| 637 |
-
)
|
| 638 |
-
|
| 639 |
-
raise HTTPException(
|
| 640 |
-
status_code=500,
|
| 641 |
-
detail=str(e)
|
| 642 |
-
)
|
| 643 |
|
| 644 |
|
| 645 |
# =====================================================
|
|
|
|
| 83 |
|
| 84 |
token = None
|
| 85 |
|
|
|
|
| 86 |
if x_api_key:
|
| 87 |
token = x_api_key.strip()
|
| 88 |
|
|
|
|
| 89 |
elif auth:
|
| 90 |
|
| 91 |
parts = auth.split(" ", 1)
|
|
|
|
| 93 |
if len(parts) == 2:
|
| 94 |
token = parts[1].strip()
|
| 95 |
|
|
|
|
| 96 |
if not token:
|
| 97 |
return True
|
| 98 |
|
|
|
|
| 99 |
if API_KEY and token != API_KEY:
|
| 100 |
raise HTTPException(
|
| 101 |
status_code=403,
|
|
|
|
| 121 |
|
| 122 |
@app.head("/")
|
| 123 |
async def root_head():
|
| 124 |
+
return JSONResponse(
|
| 125 |
+
status_code=200,
|
| 126 |
+
content={}
|
| 127 |
+
)
|
| 128 |
+
|
| 129 |
+
|
| 130 |
+
# =====================================================
|
| 131 |
+
# MODEL NORMALIZER
|
| 132 |
+
# =====================================================
|
| 133 |
+
|
| 134 |
+
def normalize_model(model: str):
|
| 135 |
+
|
| 136 |
+
if not model:
|
| 137 |
+
return "gpt-4o-mini"
|
| 138 |
+
|
| 139 |
+
model_lower = model.lower()
|
| 140 |
+
|
| 141 |
+
# GPT
|
| 142 |
+
if "gpt-4" in model_lower:
|
| 143 |
+
return "gpt-4"
|
| 144 |
+
|
| 145 |
+
if "gpt-3.5" in model_lower:
|
| 146 |
+
return "gpt-3.5-turbo"
|
| 147 |
+
|
| 148 |
+
# Claude
|
| 149 |
+
if "claude" in model_lower:
|
| 150 |
+
return "gpt-4"
|
| 151 |
+
|
| 152 |
+
# Gemini
|
| 153 |
+
if "gemini" in model_lower:
|
| 154 |
+
return "gemini"
|
| 155 |
+
|
| 156 |
+
# Cohere / Command
|
| 157 |
+
if "command-r-plus" in model_lower:
|
| 158 |
+
return "command-r-plus"
|
| 159 |
+
|
| 160 |
+
if "command-r" in model_lower:
|
| 161 |
+
return "command-r"
|
| 162 |
+
|
| 163 |
+
# Grok
|
| 164 |
+
if "grok" in model_lower:
|
| 165 |
+
return "gpt-4"
|
| 166 |
+
|
| 167 |
+
# Llama
|
| 168 |
+
if "llama" in model_lower:
|
| 169 |
+
return "llama-3-70b"
|
| 170 |
+
|
| 171 |
+
# Mistral
|
| 172 |
+
if "mistral" in model_lower:
|
| 173 |
+
return "mistral-7b"
|
| 174 |
+
|
| 175 |
+
return model
|
| 176 |
+
|
| 177 |
+
|
| 178 |
+
# =====================================================
|
| 179 |
+
# SAFE CLIENT
|
| 180 |
+
# =====================================================
|
| 181 |
+
|
| 182 |
+
def create_client():
|
| 183 |
+
|
| 184 |
+
return Client(
|
| 185 |
+
provider=g4f.Provider.Blackbox
|
| 186 |
+
)
|
| 187 |
|
| 188 |
|
| 189 |
# =====================================================
|
|
|
|
| 195 |
|
| 196 |
models_data = []
|
| 197 |
|
| 198 |
+
visible_models = [
|
| 199 |
|
| 200 |
+
"gpt-4",
|
| 201 |
+
"gpt-4-turbo",
|
| 202 |
+
"gpt-4o",
|
| 203 |
+
"gpt-4o-mini",
|
| 204 |
+
"gpt-3.5-turbo",
|
| 205 |
|
| 206 |
+
"claude-sonnet-4-6",
|
| 207 |
+
"claude-opus-4",
|
| 208 |
+
"claude-3-7-sonnet",
|
| 209 |
|
| 210 |
+
"gemini-2.5-pro",
|
| 211 |
+
"gemini-2.5-flash",
|
| 212 |
|
| 213 |
+
"command-r",
|
| 214 |
+
"command-r-plus",
|
|
|
|
|
|
|
|
|
|
|
|
|
| 215 |
|
| 216 |
+
"grok-4",
|
| 217 |
+
"llama-3-70b",
|
| 218 |
+
"mistral-7b"
|
| 219 |
+
]
|
| 220 |
|
| 221 |
+
for model in visible_models:
|
| 222 |
+
|
| 223 |
+
models_data.append({
|
| 224 |
+
"id": model,
|
| 225 |
+
"object": "model",
|
| 226 |
+
"created": int(time.time()),
|
| 227 |
+
"owned_by": "g4f"
|
| 228 |
+
})
|
| 229 |
|
| 230 |
return {
|
| 231 |
"object": "list",
|
|
|
|
| 233 |
}
|
| 234 |
|
| 235 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 236 |
# =====================================================
|
| 237 |
# CHAT COMPLETIONS
|
| 238 |
# =====================================================
|
|
|
|
| 245 |
|
| 246 |
verify_api_key(req)
|
| 247 |
|
| 248 |
+
model = normalize_model(
|
| 249 |
+
body.model
|
| 250 |
+
)
|
| 251 |
|
| 252 |
messages = [
|
| 253 |
{
|
|
|
|
| 273 |
|
| 274 |
try:
|
| 275 |
|
| 276 |
+
client = create_client()
|
| 277 |
|
| 278 |
response = client.chat.completions.create(
|
| 279 |
model=model,
|
|
|
|
| 281 |
stream=True
|
| 282 |
)
|
| 283 |
|
|
|
|
|
|
|
| 284 |
for chunk in response:
|
| 285 |
|
| 286 |
try:
|
|
|
|
| 295 |
and chunk.choices[0].delta
|
| 296 |
and chunk.choices[0].delta.content
|
| 297 |
):
|
| 298 |
+
content = (
|
| 299 |
+
chunk
|
| 300 |
+
.choices[0]
|
| 301 |
+
.delta.content
|
| 302 |
+
)
|
| 303 |
|
| 304 |
if not content:
|
| 305 |
continue
|
| 306 |
|
|
|
|
|
|
|
| 307 |
payload = {
|
| 308 |
"id": chunk_id,
|
| 309 |
"object": "chat.completion.chunk",
|
|
|
|
| 335 |
f"Chunk error: {chunk_error}"
|
| 336 |
)
|
| 337 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 338 |
final_payload = {
|
| 339 |
"id": chunk_id,
|
| 340 |
"object": "chat.completion.chunk",
|
|
|
|
| 363 |
f"Streaming error: {e}"
|
| 364 |
)
|
| 365 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 366 |
yield (
|
| 367 |
"data: "
|
| 368 |
+
+ json.dumps({
|
| 369 |
+
"error": {
|
| 370 |
+
"message": str(e)
|
| 371 |
+
}
|
| 372 |
+
})
|
| 373 |
+ "\n\n"
|
| 374 |
)
|
| 375 |
|
|
|
|
| 386 |
)
|
| 387 |
|
| 388 |
# =================================================
|
| 389 |
+
# NORMAL
|
| 390 |
# =================================================
|
| 391 |
|
| 392 |
try:
|
| 393 |
|
| 394 |
+
client = create_client()
|
| 395 |
|
| 396 |
response = client.chat.completions.create(
|
| 397 |
model=model,
|
|
|
|
| 467 |
|
| 468 |
verify_api_key(req)
|
| 469 |
|
| 470 |
+
requested_model = body.get("model")
|
| 471 |
+
|
| 472 |
model = normalize_model(
|
| 473 |
+
requested_model
|
| 474 |
)
|
| 475 |
|
| 476 |
messages = body.get("messages", [])
|
| 477 |
|
|
|
|
|
|
|
| 478 |
logger.info(
|
| 479 |
+
f"Anthropic request model={model} stream=True"
|
| 480 |
)
|
| 481 |
|
| 482 |
converted_messages = []
|
|
|
|
| 508 |
"content": content
|
| 509 |
})
|
| 510 |
|
| 511 |
+
def generate_stream():
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 512 |
|
| 513 |
+
message_id = f"msg_{uuid.uuid4().hex}"
|
|
|
|
|
|
|
|
|
|
|
|
|
| 514 |
|
| 515 |
+
try:
|
|
|
|
|
|
|
| 516 |
|
| 517 |
+
client = create_client()
|
| 518 |
|
| 519 |
+
response = client.chat.completions.create(
|
| 520 |
+
model=model,
|
| 521 |
+
messages=converted_messages,
|
| 522 |
+
stream=True
|
| 523 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 524 |
|
| 525 |
+
# =========================================
|
| 526 |
+
# message_start
|
| 527 |
+
# =========================================
|
| 528 |
|
| 529 |
+
yield (
|
| 530 |
+
"event: message_start\n"
|
| 531 |
+
"data: "
|
| 532 |
+
+ json.dumps({
|
| 533 |
+
"type": "message_start",
|
| 534 |
+
"message": {
|
| 535 |
+
"id": message_id,
|
| 536 |
+
"type": "message",
|
| 537 |
+
"role": "assistant",
|
| 538 |
+
"model": requested_model,
|
| 539 |
+
"content": [],
|
| 540 |
+
"stop_reason": None,
|
| 541 |
+
"stop_sequence": None,
|
| 542 |
+
"usage": {
|
| 543 |
+
"input_tokens": 0,
|
| 544 |
+
"output_tokens": 0
|
| 545 |
}
|
| 546 |
+
}
|
| 547 |
+
})
|
| 548 |
+
+ "\n\n"
|
| 549 |
+
)
|
| 550 |
|
| 551 |
+
# =========================================
|
| 552 |
+
# content_block_start
|
| 553 |
+
# =========================================
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 554 |
|
| 555 |
+
yield (
|
| 556 |
+
"event: content_block_start\n"
|
| 557 |
+
"data: "
|
| 558 |
+
+ json.dumps({
|
| 559 |
+
"type": "content_block_start",
|
| 560 |
+
"index": 0,
|
| 561 |
+
"content_block": {
|
| 562 |
+
"type": "text",
|
| 563 |
+
"text": ""
|
| 564 |
+
}
|
| 565 |
+
})
|
| 566 |
+
+ "\n\n"
|
| 567 |
+
)
|
| 568 |
|
| 569 |
+
for chunk in response:
|
| 570 |
|
| 571 |
+
try:
|
|
|
|
|
|
|
| 572 |
|
| 573 |
+
content = ""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 574 |
|
| 575 |
+
if (
|
| 576 |
+
hasattr(chunk, "choices")
|
| 577 |
+
and chunk.choices
|
| 578 |
+
and len(chunk.choices) > 0
|
| 579 |
+
and hasattr(chunk.choices[0], "delta")
|
| 580 |
+
and chunk.choices[0].delta
|
| 581 |
+
and chunk.choices[0].delta.content
|
| 582 |
+
):
|
| 583 |
+
content = (
|
| 584 |
+
chunk
|
| 585 |
+
.choices[0]
|
| 586 |
+
.delta.content
|
| 587 |
+
)
|
| 588 |
|
| 589 |
+
if not content:
|
| 590 |
+
continue
|
|
|
|
| 591 |
|
| 592 |
+
payload = {
|
| 593 |
+
"type": "content_block_delta",
|
| 594 |
+
"index": 0,
|
| 595 |
+
"delta": {
|
| 596 |
+
"type": "text_delta",
|
| 597 |
+
"text": content
|
| 598 |
+
}
|
| 599 |
+
}
|
| 600 |
|
| 601 |
+
yield (
|
| 602 |
+
"event: content_block_delta\n"
|
| 603 |
+
"data: "
|
| 604 |
+
+ json.dumps(
|
| 605 |
+
payload,
|
| 606 |
+
ensure_ascii=False
|
| 607 |
+
)
|
| 608 |
+
+ "\n\n"
|
| 609 |
+
)
|
| 610 |
|
| 611 |
+
except Exception as e:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 612 |
|
| 613 |
+
logger.error(
|
| 614 |
+
f"Anthropic chunk error: {e}"
|
| 615 |
+
)
|
| 616 |
|
| 617 |
+
# =========================================
|
| 618 |
+
# content_block_stop
|
| 619 |
+
# =========================================
|
| 620 |
|
| 621 |
+
yield (
|
| 622 |
+
"event: content_block_stop\n"
|
| 623 |
+
"data: "
|
| 624 |
+
+ json.dumps({
|
| 625 |
+
"type": "content_block_stop",
|
| 626 |
+
"index": 0
|
| 627 |
+
})
|
| 628 |
+
+ "\n\n"
|
| 629 |
)
|
| 630 |
|
| 631 |
+
# =========================================
|
| 632 |
+
# message_delta
|
| 633 |
+
# =========================================
|
| 634 |
+
|
| 635 |
+
yield (
|
| 636 |
+
"event: message_delta\n"
|
| 637 |
+
"data: "
|
| 638 |
+
+ json.dumps({
|
| 639 |
+
"type": "message_delta",
|
| 640 |
+
"delta": {
|
| 641 |
+
"stop_reason": "end_turn",
|
| 642 |
+
"stop_sequence": None
|
| 643 |
+
},
|
| 644 |
+
"usage": {
|
| 645 |
+
"output_tokens": 0
|
| 646 |
+
}
|
| 647 |
+
})
|
| 648 |
+
+ "\n\n"
|
| 649 |
+
)
|
| 650 |
|
| 651 |
+
# =========================================
|
| 652 |
+
# message_stop
|
| 653 |
+
# =========================================
|
| 654 |
|
| 655 |
+
yield (
|
| 656 |
+
"event: message_stop\n"
|
| 657 |
+
"data: "
|
| 658 |
+
+ json.dumps({
|
| 659 |
+
"type": "message_stop"
|
| 660 |
+
})
|
| 661 |
+
+ "\n\n"
|
| 662 |
+
)
|
| 663 |
|
| 664 |
+
except Exception as e:
|
| 665 |
|
| 666 |
+
logger.error(
|
| 667 |
+
f"Anthropic stream error: {e}"
|
| 668 |
+
)
|
|
|
|
|
|
|
|
|
|
| 669 |
|
| 670 |
+
yield (
|
| 671 |
+
"event: error\n"
|
| 672 |
+
"data: "
|
| 673 |
+
+ json.dumps({
|
| 674 |
+
"type": "error",
|
| 675 |
+
"error": {
|
| 676 |
+
"type": "api_error",
|
| 677 |
+
"message": str(e)
|
| 678 |
+
}
|
| 679 |
+
})
|
| 680 |
+
+ "\n\n"
|
| 681 |
+
)
|
| 682 |
|
| 683 |
+
return StreamingResponse(
|
| 684 |
+
generate_stream(),
|
| 685 |
+
media_type="text/event-stream",
|
| 686 |
+
headers={
|
| 687 |
+
"Cache-Control": "no-cache",
|
| 688 |
+
"Connection": "keep-alive",
|
| 689 |
+
"X-Accel-Buffering": "no"
|
| 690 |
}
|
| 691 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 692 |
|
| 693 |
|
| 694 |
# =====================================================
|