Spaces:
Running
Running
anyonehomep1mane commited on
Commit ·
c01193a
1
Parent(s): ed86e9f
Dockerfile changes
Browse files- services/summarizer.py +37 -1
services/summarizer.py
CHANGED
|
@@ -1 +1,37 @@
|
|
| 1 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from utils.file_loader import extract_text
|
| 2 |
+
from prompts.summarize_prompt import SUMMARY_PROMPT
|
| 3 |
+
from llm.hf_llm import get_chat_model
|
| 4 |
+
|
| 5 |
+
chat_model = get_chat_model()
|
| 6 |
+
summarize_chain = SUMMARY_PROMPT | chat_model
|
| 7 |
+
|
| 8 |
+
MAX_CHARS = 18_000
|
| 9 |
+
|
| 10 |
+
def summarize_document(file):
|
| 11 |
+
if not file:
|
| 12 |
+
return "Please upload a document."
|
| 13 |
+
|
| 14 |
+
text = extract_text(file)
|
| 15 |
+
|
| 16 |
+
if text.startswith("❌"):
|
| 17 |
+
return text
|
| 18 |
+
|
| 19 |
+
if len(text.strip()) < 80:
|
| 20 |
+
return "Not enough meaningful text extracted."
|
| 21 |
+
|
| 22 |
+
warning = ""
|
| 23 |
+
if len(text) > MAX_CHARS:
|
| 24 |
+
text = text[:MAX_CHARS]
|
| 25 |
+
warning = "⚠️ Document truncated for processing.\n\n"
|
| 26 |
+
|
| 27 |
+
try:
|
| 28 |
+
response = summarize_chain.invoke({"text": text})
|
| 29 |
+
return warning + response.content.strip()
|
| 30 |
+
|
| 31 |
+
except Exception as e:
|
| 32 |
+
msg = str(e).lower()
|
| 33 |
+
if "token" in msg:
|
| 34 |
+
return "❌ Invalid or missing Hugging Face token."
|
| 35 |
+
if "rate" in msg:
|
| 36 |
+
return "❌ Rate limit exceeded. Try later."
|
| 37 |
+
return f"❌ Error: {str(e)}"
|