perf: paginate docs list, lazy thumbnails, static cache headers
This commit is contained in:
Krikorios
+94
-40
@@ -131,8 +131,7 @@ async def verify_page_correlation(
|
||||
}
|
||||
],
|
||||
)
|
||||
raw_text = _strip_code_fences(message.content[0].text)
|
||||
result = json.loads(raw_text)
|
||||
result = _parse_json_lenient(message.content[0].text)
|
||||
elif provider == "gemini":
|
||||
import asyncio
|
||||
from google import genai
|
||||
@@ -176,8 +175,7 @@ async def verify_page_correlation(
|
||||
return response.text
|
||||
|
||||
raw_text = await asyncio.get_event_loop().run_in_executor(None, _call)
|
||||
raw_text = _strip_code_fences(raw_text)
|
||||
result = json.loads(raw_text)
|
||||
result = _parse_json_lenient(raw_text)
|
||||
else:
|
||||
raise ValueError(f"Provider {provider} does not support AI correlation verification")
|
||||
|
||||
@@ -221,6 +219,36 @@ def _strip_code_fences(text: str) -> str:
|
||||
return text.strip()
|
||||
|
||||
|
||||
def _parse_json_lenient(text: str) -> dict:
|
||||
"""Parse JSON even if the model added prose, trailing commas, or truncation."""
|
||||
text = _strip_code_fences(text or "")
|
||||
if not text:
|
||||
raise ValueError("Empty response from model")
|
||||
|
||||
# Fast path
|
||||
try:
|
||||
return json.loads(text)
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
# Extract the first {...} block (greedy to last })
|
||||
start = text.find("{")
|
||||
end = text.rfind("}")
|
||||
if start != -1 and end > start:
|
||||
candidate = text[start:end + 1]
|
||||
try:
|
||||
return json.loads(candidate)
|
||||
except Exception:
|
||||
# Remove trailing commas before } or ]
|
||||
cleaned = re.sub(r",(\s*[}\]])", r"\1", candidate)
|
||||
try:
|
||||
return json.loads(cleaned)
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
raise ValueError(f"Could not parse model response as JSON: {text[:200]}")
|
||||
|
||||
|
||||
def get_available_providers() -> list[dict]:
|
||||
"""Return list of all available providers (EasyOCR is always available)."""
|
||||
providers = [
|
||||
@@ -248,41 +276,55 @@ def get_default_provider() -> str:
|
||||
|
||||
async def _extract_with_claude(image_path: str) -> dict:
|
||||
import anthropic
|
||||
import asyncio
|
||||
|
||||
client = anthropic.AsyncAnthropic(api_key=ANTHROPIC_API_KEY)
|
||||
full_path = _resolve_path(image_path)
|
||||
img_data, media_type = _encode_image(full_path)
|
||||
|
||||
message = await client.messages.create(
|
||||
model=CLAUDE_MODEL,
|
||||
max_tokens=4096,
|
||||
system=[
|
||||
{
|
||||
"type": "text",
|
||||
"text": SYSTEM_PROMPT,
|
||||
"cache_control": {"type": "ephemeral"},
|
||||
}
|
||||
],
|
||||
messages=[
|
||||
{
|
||||
"role": "user",
|
||||
"content": [
|
||||
last_error = None
|
||||
for attempt in range(3):
|
||||
try:
|
||||
message = await client.messages.create(
|
||||
model=CLAUDE_MODEL,
|
||||
max_tokens=8192,
|
||||
system=[
|
||||
{
|
||||
"type": "image",
|
||||
"source": {
|
||||
"type": "base64",
|
||||
"media_type": media_type,
|
||||
"data": img_data,
|
||||
},
|
||||
},
|
||||
{"type": "text", "text": USER_PROMPT},
|
||||
"type": "text",
|
||||
"text": SYSTEM_PROMPT,
|
||||
"cache_control": {"type": "ephemeral"},
|
||||
}
|
||||
],
|
||||
}
|
||||
],
|
||||
)
|
||||
|
||||
raw_text = _strip_code_fences(message.content[0].text)
|
||||
return json.loads(raw_text)
|
||||
messages=[
|
||||
{
|
||||
"role": "user",
|
||||
"content": [
|
||||
{
|
||||
"type": "image",
|
||||
"source": {
|
||||
"type": "base64",
|
||||
"media_type": media_type,
|
||||
"data": img_data,
|
||||
},
|
||||
},
|
||||
{"type": "text", "text": USER_PROMPT},
|
||||
],
|
||||
}
|
||||
],
|
||||
)
|
||||
return _parse_json_lenient(message.content[0].text)
|
||||
except Exception as e:
|
||||
last_error = e
|
||||
err_str = str(e)
|
||||
# Transient: overload/rate-limit/network — back off and retry
|
||||
if any(code in err_str for code in ("529", "503", "502", "500", "overload", "Overload", "timeout", "Timeout")):
|
||||
await asyncio.sleep(2 * (attempt + 1))
|
||||
continue
|
||||
if "429" in err_str and attempt < 2:
|
||||
await asyncio.sleep(5 * (attempt + 1))
|
||||
continue
|
||||
raise
|
||||
raise last_error
|
||||
|
||||
|
||||
# ─── Gemini extraction ───────────────────────────────────────────
|
||||
@@ -320,24 +362,36 @@ async def _extract_with_gemini(image_path: str) -> dict:
|
||||
max_output_tokens=8192,
|
||||
),
|
||||
)
|
||||
return response.text
|
||||
text = response.text
|
||||
if not text:
|
||||
try:
|
||||
parts = []
|
||||
for cand in (response.candidates or []):
|
||||
for part in (cand.content.parts or []):
|
||||
if getattr(part, "text", None):
|
||||
parts.append(part.text)
|
||||
text = "".join(parts)
|
||||
except Exception:
|
||||
pass
|
||||
if not text:
|
||||
raise ValueError("Gemini returned empty response")
|
||||
return text
|
||||
|
||||
last_error = None
|
||||
for model_name in models_to_try:
|
||||
for attempt in range(2):
|
||||
for attempt in range(3):
|
||||
try:
|
||||
raw_text = await asyncio.get_event_loop().run_in_executor(None, lambda m=model_name: _call(m))
|
||||
raw_text = _strip_code_fences(raw_text)
|
||||
return json.loads(raw_text)
|
||||
return _parse_json_lenient(raw_text)
|
||||
except Exception as e:
|
||||
last_error = e
|
||||
err_str = str(e)
|
||||
# On quota exhaustion, skip to next model immediately (don't waste retries)
|
||||
# Quota exhaustion — skip to next model immediately
|
||||
if "429" in err_str or "RESOURCE_EXHAUSTED" in err_str:
|
||||
break
|
||||
# On temporary overload, wait and retry same model
|
||||
if "503" in err_str or "UNAVAILABLE" in err_str:
|
||||
await asyncio.sleep(3 * (attempt + 1))
|
||||
# Transient errors — back off and retry same model
|
||||
if any(code in err_str for code in ("503", "502", "500", "UNAVAILABLE", "DEADLINE", "timeout", "Timeout", "empty response", "Could not parse")):
|
||||
await asyncio.sleep(2 * (attempt + 1))
|
||||
continue
|
||||
raise
|
||||
|
||||
|
||||
@@ -0,0 +1,61 @@
|
||||
"""On-demand thumbnail generation with disk caching.
|
||||
|
||||
Thumbnails are stored under UPLOAD_DIR/.thumbs/<size>/<original relative path>.jpg
|
||||
and generated lazily the first time they are requested. This keeps the
|
||||
documents list fast even with hundreds of full-resolution uploads.
|
||||
"""
|
||||
from __future__ import annotations
|
||||
|
||||
from pathlib import Path
|
||||
from typing import Optional
|
||||
|
||||
from PIL import Image, ImageOps
|
||||
|
||||
from config import UPLOAD_DIR
|
||||
|
||||
THUMB_ROOT = Path(UPLOAD_DIR) / ".thumbs"
|
||||
DEFAULT_SIZE = 320 # px on the longest edge
|
||||
JPEG_QUALITY = 78
|
||||
|
||||
|
||||
def _safe_rel(rel_path: str) -> Optional[Path]:
|
||||
"""Resolve rel_path against UPLOAD_DIR, rejecting traversal."""
|
||||
base = Path(UPLOAD_DIR).resolve()
|
||||
target = (base / rel_path).resolve()
|
||||
try:
|
||||
target.relative_to(base)
|
||||
except ValueError:
|
||||
return None
|
||||
return target
|
||||
|
||||
|
||||
def get_or_create_thumbnail(rel_path: str, size: int = DEFAULT_SIZE) -> Optional[Path]:
|
||||
"""Return a path to a cached JPEG thumbnail for the given upload,
|
||||
creating it if needed. Returns None if the source doesn't exist."""
|
||||
source = _safe_rel(rel_path)
|
||||
if source is None or not source.is_file():
|
||||
return None
|
||||
|
||||
thumb_dir = THUMB_ROOT / str(size)
|
||||
thumb_path = thumb_dir / (rel_path + ".jpg")
|
||||
|
||||
# Regenerate if missing or source is newer
|
||||
if thumb_path.is_file():
|
||||
try:
|
||||
if thumb_path.stat().st_mtime >= source.stat().st_mtime:
|
||||
return thumb_path
|
||||
except OSError:
|
||||
pass
|
||||
|
||||
thumb_path.parent.mkdir(parents=True, exist_ok=True)
|
||||
try:
|
||||
with Image.open(source) as im:
|
||||
im = ImageOps.exif_transpose(im)
|
||||
im.thumbnail((size, size), Image.LANCZOS)
|
||||
if im.mode not in ("RGB", "L"):
|
||||
im = im.convert("RGB")
|
||||
im.save(thumb_path, "JPEG", quality=JPEG_QUALITY, optimize=True)
|
||||
except (OSError, Image.UnidentifiedImageError):
|
||||
return None
|
||||
|
||||
return thumb_path
|
||||
Reference in New Issue
Block a user