Update app.py

app.py

@@ -1,78 +1,77 @@
-# app.py — CSS LexiScan MVP (Hugging Face Space)
-
+import os
+import easyocr
 import gradio as gr
-import spacy
 from fuzzywuzzy import fuzz
-from easyocr import Reader
-
-# Load SpaCy small English model
-nlp = spacy.load("en_core_web_sm")
-
-# Initialize EasyOCR reader for English
-reader = Reader(lang_list=['en'])
 
-# Seed idioms dataset
-idioms_dataset = [
-    "break the ice", "spill the beans", "once in a blue moon",
-    "at the drop of a hat", "bark up the wrong tree",
-    "come up with", "look forward to"
-]
+# -------------------------
+# SETTINGS
+# -------------------------
+# Since images are directly in root, no folder
+images_folder = "./"
 
-# Function to extract difficult vocabulary
-def extract_vocab(text, n=7):
-    doc = nlp(text)
-    candidates = [token.text for token in doc if token.is_alpha and not token.ent_type_ and len(token.text) > 5]
-    unique_words = list(dict.fromkeys(candidates))
-    return unique_words[:n]
+# Initialize EasyOCR reader
+reader = easyocr.Reader(['en'])
 
-# Function to detect idioms
-def detect_idioms(text):
-    found = []
-    doc = nlp(text)
-    sentences = [sent.text.strip() for sent in doc.sents]
-    for idiom in idioms_dataset:
-        for sent in sentences:
-            if fuzz.partial_ratio(idiom.lower(), sent.lower()) > 80:
-                found.append(idiom)
-    return list(set(found))
+# -------------------------
+# HELPER FUNCTIONS
+# -------------------------
+def extract_text_from_image(image_path):
+    """OCR the image and return full text"""
+    result = reader.readtext(image_path)
+    full_text = " ".join([res[1] for res in result])
+    return full_text
 
-# Simple CSS-style précis (first 3 sentences)
-def generate_precis(text):
-    doc = nlp(text)
-    sentences = [sent.text.strip() for sent in doc.sents]
-    return " ".join(sentences[:3])
+def get_difficult_words(text, top_n=7):
+    """Simple heuristic for difficult words"""
+    words = [w.strip('.,!?()') for w in text.split()]
+    words = [w for w in words if len(w) > 5 and w.isalpha()]
+    freq = {}
+    for w in words:
+        freq[w] = freq.get(w, 0) + 1
+    sorted_words = sorted(freq, key=freq.get, reverse=True)
+    return sorted_words[:top_n]
 
-# Main function to process uploaded image
-def process_input(image):
-    text = " ".join(reader.readtext(image, detail=0))
-    precis = generate_precis(text)
-    vocab_list = extract_vocab(text)
-    idioms_found = detect_idioms(text)
-    return (
-        precis,
-        "\n".join(f"- {w}" for w in vocab_list),
-        "\n".join(f"- {i}" for i in idioms_found)
-    )
+def get_idioms(text):
+    """Dummy idiom detection (replace with your dataset later)"""
+    seed_idioms = ["kick the bucket", "break the ice", "spill the beans"]
+    found = [i for i in seed_idioms if i in text.lower()]
+    return found
 
-# Gradio UI
-with gr.Blocks() as demo:
-    gr.Markdown("# 📰 CSS LexiScan — AI-Powered Editorial Analysis")
-    gr.Markdown("Upload a newspaper screenshot to get CSS-style précis, difficult vocabulary, and idioms detected.")
+def generate_précis(text):
+    """Extractive CSS-style précis (first 1/3 of text)"""
+    one_third_len = max(1, len(text)//3)
+    return text[:one_third_len]
 
-    with gr.Row():
-        with gr.Column():
-            gr.Markdown("### ✍️ CSS-style Précis")
-            precis_box = gr.Textbox(lines=12, placeholder="Generated CSS-style précis will appear here...")
-        with gr.Column():
-            gr.Markdown("### 📚 Difficult Vocabulary")
-            vocab_box = gr.Textbox(lines=8, placeholder="Difficult words will appear here...")
-        with gr.Column():
-            gr.Markdown("### 💡 Idioms Detected")
-            idioms_box = gr.Textbox(lines=8, placeholder="Idioms will appear here...")
+# -------------------------
+# MAIN FUNCTION
+# -------------------------
+def process_image(image):
+    text = extract_text_from_image(image)
+    vocab = get_difficult_words(text)
+    idioms = get_idioms(text)
+    precis = generate_précis(text)
+    return precis, ", ".join(vocab), ", ".join(idioms)
 
-    img_input = gr.Image(type="filepath")
-    submit_btn = gr.Button("Analyze Article 📝")
+# -------------------------
+# OPTIONAL: List of existing images in root
+# -------------------------
+existing_images = [f for f in os.listdir(images_folder) 
+                   if f.lower().endswith((".png", ".jpg", ".jpeg"))]
 
-    submit_btn.click(process_input, inputs=img_input, outputs=[precis_box, vocab_box, idioms_box])
+# -------------------------
+# GRADIO INTERFACE
+# -------------------------
+demo = gr.Interface(
+    fn=process_image,
+    inputs=gr.Image(type="filepath", label="Upload an image or select one from root"),
+    outputs=[
+        gr.Textbox(label="CSS-Style Précis"),
+        gr.Textbox(label="Difficult Words"),
+        gr.Textbox(label="Idioms / Phrasal Expressions")
+    ],
+    title="CSS LexiScan MVP",
+    description="Upload newspaper images to get grounded CSS-style précis, vocabulary, and idioms."
+)
 
-demo.launch()
+if __name__ == "__main__":
+    demo.launch()