implement dynamic length adjustments for summarization and add translation endpoint

main.py

@@ -1,6 +1,11 @@
 from fastapi import FastAPI
 from pydantic import BaseModel
-from transformers import pipeline
+from transformers import pipeline, AutoTokenizer,T5ForConditionalGeneration
+
+
+tokenizer = AutoTokenizer.from_pretrained("google-t5/t5-base")
+model = T5ForConditionalGeneration.from_pretrained("google-t5/t5-base")
+
 
 app = FastAPI()
 summarizer = pipeline("summarization", model="facebook/bart-large-cnn")
@@ -16,5 +21,48 @@ async def root():
 
 @app.post("/summarize")
 async def summarize_text(input: TextInput):
-    summary = (summarizer(input.text, max_length=130, min_length=30, do_sample=False ))
-    return {"summary": summary[0]['summary_text']}
+    # Count approximate number of words (could be improved with tokenizer count)
+    word_count = len(input.text.split())
+    
+    # Set dynamic parameters based on input length
+    if word_count < 50:
+        max_length = max(10, word_count // 2)  # Half the original length, minimum 10
+        min_length = max(3, word_count // 4)   # Quarter the original length, minimum 3
+    elif word_count < 200:
+        max_length = max(50, word_count // 3)
+        min_length = max(15, word_count // 6)
+    else:
+        max_length = max(100, word_count // 4)
+        min_length = max(30, word_count // 8)
+    
+    # Prevent max_length from being too large (BART has token limits)
+    max_length = min(max_length, 1024)
+    
+    # Generate summary with dynamic parameters
+    summary = summarizer(
+        input.text, 
+        max_length=max_length, 
+        min_length=min_length, 
+        do_sample=True,
+        temperature=0.7,
+        num_beams=4
+    )
+    
+    return {
+        "summary": summary[0]["summary_text"],
+        "parameters_used": {
+            "input_word_count": word_count,
+            "max_length": max_length,
+            "min_length": min_length
+        }
+    }
+    
+
+
+@app.post("/translateFrench")
+async def translate(input: TextInput):
+    input.text = "translate English to French: " + input.text
+    input_ids = tokenizer(input.text, return_tensors="pt").input_ids
+    output = model.generate(input_ids, max_length=50, num_beams=4, early_stopping=True)
+    translated_text = tokenizer.decode(output[0], skip_special_tokens=True)
+    return {"translated_text": translated_text}