app.py

import gradio as gr
import google.generativeai as genai
from bs4 import BeautifulSoup, NavigableString
import re
import json
import random
import os

# --- Constants & Config ---
BLACKLIST_WORDS = [
    "landscape", "realm", "navigate", "unveil", "explore", "transformative",
    "encompass", "examine", "crucial", "discover", "dive", "delve",
    "uncover", "unlock", "elevate", "unleash", "harness"
]

BRITISH_MAPPINGS = {
    "color": "colour", "flavor": "flavour", "humor": "humour", "labor": "labour",
    "neighbor": "neighbour", "favor": "favour", "honor": "honour", "behavior": "behaviour",
    "center": "centre", "fiber": "fibre", "liter": "litre", "theater": "theatre",
    "meter": "metre", "analyze": "analyse", "breathalyze": "breathalyse", "paralyze": "paralyse",
    "catalyze": "catalyse", "organization": "organisation", "realize": "realise",
    "recognize": "recognise", "standardize": "standardise", "appetizer": "appetiser",
    "leukemia": "leukaemia", "maneuver": "manoeuvre", "estrogen": "oestrogen",
    "pediatric": "paediatric", "defense": "defence", "license": "licence",
    "offense": "offence", "pretense": "pretence", "traveler": "traveller", "modeling": "modelling",
    "cancelled": "cancelled",
    "program": "programme",
}

SOCIAL_PROOF_TEMPLATES = [
    "We recently hired {KEYWORD} for our project, and the results were outstanding. The team was professional, efficient, and delivered exactly what we needed. I highly recommend their services to anyone looking for reliable {KEYWORD_LOWER}.",
    "I was struggling to find trustworthy {KEYWORD_LOWER} until I found this company. They exceeded my expectations with their attention to detail and timely completion. It was a refreshing experience to work with such dedicated professionals.",
    "If you need {KEYWORD_LOWER}, look no further. Their expertise is evident in the quality of their work, and the customer service is top-notch. I am completely satisfied with the outcome and will definitely use them again.",
    "Finding a dependable {KEYWORD} can be difficult, but this team made it easy. They communicated clearly throughout the process and finished the job to a high standard. I'm very impressed with their workmanship."
]

# --- Logic Ports ---

def capitalize(s):
    if not s: return ""
    return s[0].upper() + s[1:]

def parse_growmatic_data(text):
    term_map = {}
    if not text: return term_map
    # Regex to match: "term": number% OR term: number%
    regex = r'["\']?([\w\s]+)["\']?\s*[:=]\s*(\d+)%?'
    matches = re.findall(regex, text)
    for term, score in matches:
        term_lower = term.strip().lower()
        if term_lower:
            term_map[term_lower] = int(score)
    return term_map

def generate_titles(main_keyword, term_map):
    titles = []
    # Templates
    templates = [
        "{KEYWORD} in [location] - {TERM_A} [zip]",
        "{KEYWORD} in [location] - {TERM_B} Services [zip]",
        "Expert {KEYWORD} in [location] - {TERM_C} [zip]",
        "{KEYWORD} Services in [location] - {TERM_A} [zip]",
        "Leading {KEYWORD} in [location] - {TERM_B} [zip]",
        "{KEYWORD} Specialists in [location] - {TERM_C} [zip]",
        "Best {KEYWORD} in [location] - {TERM_A} Solutions [zip]"
    ]

    # Sort terms by score descending
    sorted_terms = sorted(term_map.keys(), key=lambda k: term_map[k], reverse=True)
    
    term_a = sorted_terms[0] if len(sorted_terms) > 0 else "Projects"
    term_b = sorted_terms[1] if len(sorted_terms) > 1 else "Installations"
    term_c = sorted_terms[2] if len(sorted_terms) > 2 else "Solutions"

    for tmpl in templates:
        t = tmpl.replace("{KEYWORD}", main_keyword)
        t = t.replace("{TERM_A}", capitalize(term_a))
        t = t.replace("{TERM_B}", capitalize(term_b))
        t = t.replace("{TERM_C}", capitalize(term_c))
        titles.append(t)

    # Variations
    variations = [
        f"{main_keyword} {capitalize(term_a)}",
        f"{main_keyword} {capitalize(term_b)} Services",
        f"{capitalize(term_a)} & {main_keyword}"
    ]
    return titles + variations

def calculate_score(title, term_map):
    title_lower = title.lower()
    
    # Blacklist check
    for bad_word in BLACKLIST_WORDS:
        if bad_word in title_lower:
            return {"title": title, "score": 0, "terms": "BLACKLISTED"}

    total_score = 0
    matched_terms = []
    
    for term, weight in term_map.items():
        if term in title_lower:
            total_score += weight
            matched_terms.append(f"{term} ({weight}%)")
    
    # Scale score (approx 0-10)
    final_score = round(total_score / 30, 1)
    if final_score > 10: final_score = 10
    
    return {
        "title": title,
        "score": final_score,
        "terms": ", ".join(matched_terms)
    }

def process_text_nodes(html_content, callback):
    if not html_content: return ""
    soup = BeautifulSoup(html_content, 'html.parser')
    
    # Recursive function specifically for NavigableStrings
    def walk(node):
        if isinstance(node, NavigableString):
            if node.parent.name not in ['script', 'style']: # Skip script/style tags
                new_text = callback(str(node))
                if new_text != str(node):
                    node.replace_with(new_text)
        elif hasattr(node, 'children'):
             for child in node.children:
                 walk(child)

    walk(soup)
    return str(soup)

def convert_to_british(html_content):
    if not html_content: return ""
    
    def replacer(text):
        processed = text
        for us, uk in BRITISH_MAPPINGS.items():
            # Regex for whole word match, case insensitive
            pattern = re.compile(r'\b' + re.escape(us) + r'\b', re.IGNORECASE)
            
            def match_handler(m):
                # Preserve case
                word = m.group(0)
                if word[0].isupper():
                    return capitalize(uk)
                return uk
                
            processed = pattern.sub(match_handler, processed)
        return processed

    return process_text_nodes(html_content, replacer)

def clean_homepage_content(html_content):
    if not html_content: return ""
    
    def replacer(text):
        clean = text
        
        # 1. Remove phrases
        phrases_to_remove = [
            r'\s+in\s+\[location\]', r'in\s+\[location\]',
            r'\s+across\s+the\s+\[location\]', r'across\s+the\s+\[location\]',
            r'\s+across\s+\[location\]', r'across\s+\[location\]',
            r'\s+around\s+the\s+\[location\]', r'around\s+the\s+\[location\]',
            r'\s+nearby\s+\[location\]', r'nearby\s+\[location\]',
            r'\s+throughout\s+\[location\]', r'throughout\s+\[location\]'
        ]
        for phrase in phrases_to_remove:
            clean = re.sub(phrase, '', clean, flags=re.IGNORECASE)

        # 2. Remove tags
        tags_to_remove = [
            r'\[location\]', r'\[county\]', r'\[region\]', r'\[zip\]'
        ]
        for tag in tags_to_remove:
            clean = re.sub(tag, '', clean, flags=re.IGNORECASE)

        # 3. Footer text
        footer_regex = r'in\s*\[region\]\.?\s*Here\s*are\s*some\s*towns\s*we\s*cover\s*near\s*\[location\]\s*\[zip\]\s*\[cities[^\]]*\]'
        clean = re.sub(footer_regex, '', clean, flags=re.IGNORECASE | re.DOTALL)

        # 4. Whitespace cleanup
        clean = re.sub(r'\s{2,}', ' ', clean)
        clean = re.sub(r'\s+\.', '.', clean)
        clean = re.sub(r'\s+\?', '?', clean)
        clean = re.sub(r'\s+\,', ',', clean)
        
        return clean.strip()

    return process_text_nodes(html_content, replacer)


# --- Gemini Integration ---

def call_gemini(prompt, api_key, model_name="gemini-1.5-flash"):
    if not api_key: return None
    try:
        genai.configure(api_key=api_key)
        model = genai.GenerativeModel(model_name)
        response = model.generate_content(prompt)
        return response.text
    except Exception as e:
        return f"Error: {str(e)}"

# --- Main Automation Logic ---

def run_automation(main_keyword, site_link, growmatic_data, api_key, article_content, model_selection):
    if not main_keyword:
        return "Error: Main Keyword is required.", ""

    term_map = parse_growmatic_data(growmatic_data)
    
    # 1. Magic Page Logic
    magic_output_html = ""
    
    # SEO Titles
    if api_key:
        # LLM Title Gen
        terms_str = ", ".join([f"{k} ({v}%)" for k, v in term_map.items()])
        prompt = f"""Act as an SEO expert. 
        Main Keyword: "{main_keyword}"
        Semantic Terms (Growmatic Data): {terms_str}

        Task:
        1. Generate 3 highly optimized Meta Titles for a page targeting "{main_keyword}". Use the semantic terms to increase relevance.
        2. Generate a list of 5-8 Meta Keywords (comma separated).
        3. Select the "Best" Title from the 3 options based on SEO scoring principles.

        Output JSON format ONLY (no markdown):
        {{
            "metaTitles": ["Title 1", "Title 2", "Title 3"],
            "bestTitle": "The Best Title",
            "metaKeywords": "keyword1, keyword2, keyword3"
        }}"""
        
        llm_resp = call_gemini(prompt, api_key, model_selection)
        
        try:
            # Clean json block if present
            clean_json = llm_resp.replace('```json', '').replace('```', '').strip()
            data = json.loads(clean_json)
            
            magic_output_html += "<h3>--- GENERATED SEO TITLES (LLM) ---</h3>"
            for t in data.get("metaTitles", []):
                is_best = t == data.get("bestTitle")
                style = "color: blue; font-weight: bold;" if is_best else ""
                suffix = "(Best Match)" if is_best else ""
                magic_output_html += f'<p style="{style}">• {t} {suffix}</p>'
            
            magic_output_html += f"<p><strong>Meta Keywords:</strong> {data.get('metaKeywords', '')}</p><br>"
            
        except:
             magic_output_html += f"<p style='color:red'>Error parsing LLM response: {llm_resp}</p>"

    else:
        # Template Gen
        titles = generate_titles(main_keyword, term_map)
        scored = [calculate_score(t, term_map) for t in titles]
        scored.sort(key=lambda x: x['score'], reverse=True)
        
        magic_output_html += "<h3>--- GENERATED SEO TITLES (Template) ---</h3>"
        for item in scored[:5]:
             magic_output_html += f"<p>• [Score: {item['score']}] {item['title']}</p>"
        magic_output_html += "<br>"

    # Social Proof
    social_proof_text = ""
    if api_key:
        sp_prompt = f"""Write 2 positive testimonials for a service provider offering "{main_keyword}". 
        Create two very non-generic names including last names. 
        Each testimonial should be max 3-4 sentences. 
        Focus on professionalism, result quality, and ease of working with them."""
        social_proof_text = call_gemini(sp_prompt, api_key, model_selection)
    else:
         tmpl = random.choice(SOCIAL_PROOF_TEMPLATES)
         social_proof_text = tmpl.replace("{KEYWORD}", main_keyword).replace("{KEYWORD_LOWER}", main_keyword.lower())

    magic_output_html += f"<h3>--- MAGIC PAGE METADATA ---</h3>"
    magic_output_html += f"<p><strong>Target Keyword:</strong> {main_keyword}</p>"
    magic_output_html += f"<p><strong>Site URL:</strong> {site_link}</p><br>"
    
    magic_output_html += f"<h3>--- SOCIAL PROOF ---</h3>"
    magic_output_html += f"<p>{social_proof_text.replace(chr(10), '<br>')}</p>"

    # 2. Homepage Logic
    clean_html = clean_homepage_content(article_content)
    british_html = convert_to_british(clean_html)

    return magic_output_html, british_html


# --- Gradio UI ---

with gr.Blocks(title="Content Automation Tool") as app:
    gr.Markdown("# Content Automation Tool (Gradio Edition)")
    gr.Markdown("Generate Magic Page & Optimized Homepage Content Instantly")
    
    with gr.Row():
        with gr.Column():
            main_keyword = gr.Textbox(label="Main Keyword", placeholder="e.g. Suspended Ceiling Contractors")
            site_link = gr.Textbox(label="Site Link", placeholder="e.g. https://example.com")
            growmatic_data = gr.TextArea(label="Growmatic Data", placeholder='"suspended": 100%, "ceiling": 73%')
            
            with gr.Row():
                api_key = gr.Textbox(label="Gemini API Key", type="password", placeholder="AIza...")
                model_selection = gr.Dropdown(
                    choices=["gemini-1.5-flash", "gemini-1.5-pro", "gemini-1.0-pro"], 
                    value="gemini-1.5-flash", 
                    label="Gemini Model"
                )
            
        with gr.Column():
            article_content = gr.Textbox(label="Article Content (HTML/Text)", lines=15, placeholder="Paste content with [tags] here...")

    generate_btn = gr.Button("Generate Output ✨", variant="primary")

    with gr.Row():
        with gr.Column():
            gr.Markdown("### Magic Page Output")
            magic_output = gr.HTML(label="Magic Page Result")
        
        with gr.Column():
            gr.Markdown("### Homepage Output")
            home_output = gr.HTML(label="Homepage Result")

    generate_btn.click(
        fn=run_automation,
        inputs=[main_keyword, site_link, growmatic_data, api_key, article_content, model_selection],
        outputs=[magic_output, home_output]
    )

if __name__ == "__main__":
    app.launch()