Create app.py

app.py

@@ -0,0 +1,195 @@
+import gradio as gr
+import sqlite3
+import re
+from unidecode import unidecode
+import arabic_reshaper
+
+# Database connection
+def get_db_connection():
+    conn = sqlite3.connect('asawal_amqran.db')
+    conn.row_factory = sqlite3.Row
+    return conn
+
+# Arabic normalization
+def normalize_arabic(text):
+    text = re.sub(r'[إأٱآا]', 'ا', text)
+    text = re.sub(r'[ًٌٍَُِْ]', '', text)
+    return text
+
+# Amazigh normalization
+def normalize_amazigh(text, additional_options=False):
+    text = text.replace('ⵕ', 'ⵔ')
+    text = text.replace('ⵯ', '')
+    if additional_options:
+        text = re.sub(r'(.)\1', r'\1', text)
+        text = text.replace('*', '%')
+    return text
+
+# French normalization
+def normalize_french(text):
+    return unidecode(text)
+
+# Search function with dictionary-style output
+def search_dictionary(query, search_type="general", language="amazigh", additional_options=False):
+    conn = get_db_connection()
+    cursor = conn.cursor()
+    
+    # Prepare query based on language
+    if language == "amazigh":
+        columns = ['word', 'latin', 'construct', 'plural', 'acc', 'accneg', 
+                 'inacc', 'variante', 'feminine', 'frem_construct', 'fem_plural_construct', 'exp_zgh']
+        normalized_query = normalize_amazigh(query, additional_options)
+    elif language == "french":
+        columns = ['french', 'exp_fra']
+        normalized_query = normalize_french(query)
+    else:  # arabic
+        columns = ['arabic', 'exp_ara', 'mean_ara']
+        normalized_query = normalize_arabic(query)
+
+    # Build SQL query based on search type
+    conditions = []
+    params = []
+    
+    if search_type == "exact":
+        conditions = [f"{col} = ?" for col in columns]
+        params = [query] * len(columns)
+    elif search_type == "word":
+        conditions = [f"{col} LIKE ?" for col in columns]
+        params = [f"% {query} %" for _ in columns]
+    elif search_type == "contains":
+        conditions = [f"{col} LIKE ?" for col in columns]
+        params = [f"%{normalized_query}%" for _ in columns]
+    elif search_type == "starts":
+        conditions = [f"{col} LIKE ?" for col in columns]
+        params = [f"{normalized_query}%" for _ in columns]
+    elif search_type == "ends":
+        conditions = [f"{col} LIKE ?" for col in columns]
+        params = [f"%{normalized_query}" for _ in columns]
+    else:  # general
+        conditions = []
+        params = []
+        conditions.extend([f"{col} = ?" for col in columns])
+        params.extend([query] * len(columns))
+        conditions.extend([f"{col} LIKE ?" for col in columns])
+        params.extend([f"% {query} %" for _ in columns])
+        conditions.extend([f"{col} LIKE ?" for col in columns])
+        params.extend([f"%{normalized_query}%" for _ in columns])
+        conditions.extend([f"{col} LIKE ?" for col in columns])
+        params.extend([f"{normalized_query}%" for _ in columns])
+        conditions.extend([f"{col} LIKE ?" for col in columns])
+        params.extend([f"%{normalized_query}" for _ in columns])
+
+    where_clause = " OR ".join(conditions)
+    query_sql = f"SELECT * FROM lexie WHERE {where_clause}"
+    
+    cursor.execute(query_sql, params)
+    results = cursor.fetchall()
+    conn.close()
+
+    # Format results as dictionary-style HTML
+    if not results:
+        return "<h3>No results found</h3>"
+    
+    html = "<div style='font-family: Arial, sans-serif; max-width: 800px;'>"
+    
+    for i, row in enumerate(results, 1):
+        html += f"<div style='margin-bottom: 20px; padding: 10px; border-bottom: 1px solid #eee;'>"
+        html += f"<h3 style='color: #2c3e50; margin-bottom: 10px;'>Entry {i}</h3>"
+        
+        # Main word
+        if row['word']:
+            html += f"<div><strong>{row['word']}</strong>"
+            if row['latin']:
+                html += f" ({row['latin']})"
+            html += "</div>"
+        
+        # Morphological forms
+        forms = []
+        for field, label in [
+            ('construct', 'Construct:'),
+            ('plural', 'Plural:'),
+            ('acc', 'Accusative:'),
+            ('accneg', 'Acc. Negative:'),
+            ('inacc', 'Inaccusative:'),
+            ('variante', 'Variant:'),
+            ('feminine', 'Feminine:'),
+            ('frem_construct', 'Fem. Construct:'),
+            ('fem_plural_construct', 'Fem. Pl. Construct:')
+        ]:
+            if row[field]:
+                forms.append(f"{label} {row[field]}")
+        if forms:
+            html += "<div style='margin-left: 20px;'>" + "<br>".join(forms) + "</div>"
+        
+        # Translations
+        translations = []
+        if row['french']:
+            translations.append(f"French: {row['french']}")
+        if row['arabic']:
+            translations.append(f"Arabic: {row['arabic']}")
+        if translations:
+            html += "<div style='margin-left: 20px; margin-top: 5px;'>" + "<br>".join(translations) + "</div>"
+        
+        # Explanations
+        explanations = []
+        if row['exp_zgh']:
+            explanations.append(f"Amazigh Exp: {row['exp_zgh']}")
+        if row['exp_fra']:
+            explanations.append(f"French Exp: {row['exp_fra']}")
+        if row['exp_ara']:
+            explanations.append(f"Arabic Exp: {row['exp_ara']}")
+        if row['mean_ara']:
+            explanations.append(f"Arabic Meaning: {row['mean_ara']}")
+        if explanations:
+            html += "<div style='margin-left: 20px; margin-top: 5px; color: #666;'>" + "<br>".join(explanations) + "</div>"
+        
+        html += "</div>"
+
+    html += "</div>"
+    return html
+
+# Gradio interface
+with gr.Blocks(title="Amazigh Dictionary") as demo:
+    gr.Markdown("# Amazigh Dictionary Search")
+    
+    with gr.Row():
+        search_box = gr.Textbox(
+            placeholder="Enter search term and press Enter or click Search",
+            label="Search"
+        )
+        search_btn = gr.Button("Search")
+    
+    with gr.Row():
+        search_type = gr.Radio(
+            choices=["general", "exact", "word", "contains", "starts", "ends"],
+            value="general",
+            label="Search Type"
+        )
+    
+    with gr.Row():
+        language = gr.Radio(
+            choices=["amazigh", "french", "arabic"],
+            value="amazigh",
+            label="Language"
+        )
+    
+    with gr.Row():
+        additional_options = gr.Checkbox(
+            label="Additional options (Amazigh only: normalize double letters, use * for missing letters)",
+            value=False
+        )
+    
+    output = gr.HTML()
+    
+    search_box.submit(
+        search_dictionary,
+        inputs=[search_box, search_type, language, additional_options],
+        outputs=output
+    )
+    search_btn.click(
+        search_dictionary,
+        inputs=[search_box, search_type, language, additional_options],
+        outputs=output
+    )
+
+demo.launch()