import gradio as gr
import pandas as pd
import re
from newspaper import Article
import requests
import io
import os
import requests
from bs4 import BeautifulSoup
from transformers import pipeline

# Sumy and NLTK imports
from nltk.tokenize import sent_tokenize
from sumy.parsers.plaintext import PlaintextParser
from sumy.nlp.tokenizers import Tokenizer
from sumy.summarizers.lsa import LsaSummarizer
from sumy.nlp.stemmers import Stemmer
from sumy.utils import get_stop_words

classifier = pipeline("zero-shot-classification", model="facebook/bart-large-mnli")

# -------- Summary Cleaning and Extraction -------- #
def preprocess_text(text):
    if not isinstance(text, str):
        return ""
    text = re.sub(r'http\S+', ' ', text)
    lines = text.splitlines()
    kept = []
    for line in lines:
        line = line.strip()
        if not line:
            continue
        if re.match(r'By\s+\S+', line): continue
        if re.search(r'\bFollow\b', line): continue
        if re.search(r'\d+\s+min\s+read', line, flags=re.IGNORECASE): continue
        if re.search(r'\b(?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)\s+\d{1,2},\s+\d{4}\b', line): continue
        if line.lower().startswith((
            "read more", "continue reading", "more from medium",
            "about the author", "related stories", "you might also like"
        )): continue
        if line.isupper() and len(line.split()) > 3:
            continue
        kept.append(line)
    text = "\n".join(kept)
    text = re.sub(r'[^\w\s.,!?;:]', ' ', text)
    text = re.sub(r'\s+', ' ', text).strip()
    sents = sent_tokenize(text)
    return ' '.join(dict.fromkeys([s for s in sents if len(s.split()) > 3]))

def summarize_with_sumy_auto(text, summary_frac=0.2, min_sentences=3, max_sentences=10):
    if not isinstance(text, str):
        return ""
    cleaned = preprocess_text(text)
    orig = sent_tokenize(cleaned)
    total = len(orig)
    if total <= min_sentences:
        return ' '.join(orig)
    n = max(min_sentences, min(max_sentences, int(total * summary_frac)))
    parser = PlaintextParser.from_string(cleaned, Tokenizer("english"))
    stemmer = Stemmer("english")
    summarizer = LsaSummarizer(stemmer)
    summarizer.stop_words = get_stop_words("english")
    sents = summarizer(parser.document, n)
    return ' '.join(str(s) for s in sents)

# -------- Utility Functions -------- #
def check_url_status(url: str, timeout: int = 5) -> str:
    try:
        resp = requests.head(url, allow_redirects=True, timeout=timeout)
        if resp.status_code == 405:
            resp = requests.get(url, allow_redirects=True, timeout=timeout)
        return 'Workable' if resp.status_code == 200 else f'Not Workable ({resp.status_code})'
    except requests.RequestException:
        return 'Not Workable'

def detect_keywords_and_score(content, url):
    keywords = []
    score = 0
    imarticus_found = False
    pga_link_found = False
    pga_link = "https://imarticus.org/postgraduate-program-in-data-science-analytics/"
    if content and re.search(r'imarticus', content, re.IGNORECASE):
        keywords.append('Imarticus')
        imarticus_found = True
        if pga_link in content or pga_link in url:
            pga_link_found = True
    if content and re.search(r'post graduate', content, re.IGNORECASE):
        keywords.append('post graduate')
    if imarticus_found:
        score = 5 if pga_link_found else 3
        return keywords, score
    else:
        return [], 0

def detect_code_snippet(content):
    if not content:
        return False
    code_markers = [
        r'```', r'<code>', r'</code>', r'\n    ', r'\t',
        r'def ', r'class ', r'\{', r'\}', r';', r'\(', r'\)', r'import ', r'from ', r'print\('
    ]
    for marker in code_markers:
        if re.search(marker, content):
            return True
    return False

# ------ Originality Check -----------#
def extract_blog_text(url):
    headers = {'User-Agent': 'Mozilla/5.0'}
    response = requests.get(url, headers=headers)
    soup = BeautifulSoup(response.text, 'html.parser')
    paragraphs = soup.find_all('p')
    return ' '.join([p.get_text() for p in paragraphs])

def get_ai_generated_score(url, classifier=classifier):
    text = extract_blog_text(url)
    #classifier = pipeline("zero-shot-classification", model="facebook/bart-large-mnli")
    labels = ["Human-written", "AI-generated"]
    result = classifier(text, candidate_labels=labels)
    scores = dict(zip(result['labels'], result['scores']))
    return scores.get("AI-generated", 0.0)

# -------- Main Summary Extraction -------- #
def extract_summary(file):
    df = pd.read_excel(file)
    total_blogs = len(df)
    imarticus_count = 0
    code_snippet_count = 0
    filtered_rows = []
    full_analysis = []

    for _, row in df.iterrows():
        url = row.get("Blog Link(Medium link)") or row.get("URL") or row.get("url")
        if pd.isna(url):
            continue

        status = check_url_status(url)

        name = row.get("Participant") or row.get("Name")
        if not name:
            continue

        centre = row.get("Centre") or row.get("Center")
        if not centre:
            continue
        #originality = get_ai_generated_score(url)

        try:
            article = Article(url)
            article.download()
            article.parse()
            title = article.title
            content = article.text

            if len(content.strip()) == 0:
                continue

            summary = summarize_with_sumy_auto(content)

            keywords, score = detect_keywords_and_score(content, url)
            code_snippet = detect_code_snippet(content)

            if score > 0:
                imarticus_count += 1
            if code_snippet:
                code_snippet_count += 1

            filtered_rows.append({
                "Participant": name,
                "Centre": centre,
                "URL": url,
                "Status": status,
                "Title": title,
                "Content": content,
                "Summary": summary,
                "Identified_Keywords": ', '.join(keywords) if keywords else "None",
                "Code_Snippet": code_snippet,
                "Score": score
               # "Originality(AI-Score)": originality
            })

            full_analysis.append({
                "Participant": name,
                "Centre": centre,
                "URL": url,
                "Title": title,
                "Identified_Keywords": ', '.join(keywords) if keywords else "None",
                "Code_Snippet": code_snippet,
                "Score": score,
                "Summary": summary,
                "Status": status
               # "Originality(AI-Score)": originality
            })

        except Exception as e:
            print(f"Error processing {url}: {e}")
            continue

    filtered_df = pd.DataFrame(filtered_rows)
    full_df = pd.DataFrame(full_analysis)

    return (
        str(total_blogs),
        str(code_snippet_count),
        str(imarticus_count),
        filtered_df,
        full_df
    )

def filter_analysis(full_df, status_filter, score_filter):
    df = full_df.copy()
    if status_filter != "All":
        df = df[df["Status"].str.contains(status_filter)]
    if score_filter != "All":
        df = df[df["Score"] == int(score_filter)]
    df = df[["Title", "Identified_Keywords", "Code_Snippet", "Score", "Summary"]]
    return df

def download_file(full_df):
    if full_df is None or full_df.empty:
        print("No data to download.")
        return None
    output_dir = "./output"
    os.makedirs(output_dir, exist_ok=True)
    file_path = os.path.join(output_dir, "Full_Analysis.xlsx")
    try:
        full_df.to_excel(file_path, index=False)
    except Exception as e:
        print(f"Error saving file: {e}")
        return None
    return file_path

def trigger_download(full_df):
    path = download_file(full_df)
    return path, gr.update(visible=True) if path else gr.update(visible=False)

# -------- Gradio UI -------- #
with gr.Blocks(css="""
    .sidebar { background-color: #00664d; color: white; padding: 20px; height: 100%; border-radius: 10px; }
    .sidebar label, .sidebar h2, .sidebar h3, .sidebar span, .sidebar p { color: black !important; }
    .main-content { padding: 20px; background-color: #ffffff; border-radius: 10px; }
    h1, h3 { color: #00664d; }
    @media (min-width: 1024px) {
        .gr-block.gr-box { max-width: 1000px; margin: auto; }
    }
""") as demo:

    with gr.Row():
        with gr.Column(scale=1, elem_classes="sidebar"):
            gr.Markdown("## 📅 Upload & Filter", elem_id="sidebar-title")
            file_input = gr.File(label="Upload Excel File (.xlsx)", file_types=[".xlsx"])
            analyze_btn = gr.Button("Run Summary")

            gr.Markdown("## 🔎 Filter")
            status_filter = gr.Dropdown(["All", "Workable", "Not Workable"], label="Status", value="All")
            score_filter = gr.Dropdown(["All", "0", "3", "5"], label="Score", value="All")

            download_btn = gr.Button("Download Full Analysis")
            download_file_output = gr.File(label="")

        with gr.Column(scale=3, elem_classes="main-content"):
            gr.Markdown("<h1>📊 Blog Evaluator </h1>")
            gr.Markdown("<h3>Analyze blog URLs for educational content, keywords, and coding examples</h3>")

            with gr.Row():
                total_blogs = gr.Textbox(label="Total Blogs", interactive=False)
                code_snippets = gr.Textbox(label="Blogs with Code Snippets", interactive=False)
                imarticus_hits = gr.Textbox(label="Blogs with 'Imarticus' Mentions", interactive=False)

            gr.Markdown("### 📋 Filtered Results Table")
            full_table = gr.Dataframe(
                headers=["Participant", "Centre","URL","Status","Title","Content","Summary","Identified_Keywords", "Code_Snippet", "Score"],
                interactive=False,
                datatype=["str", "str", "str", "str", "str","str","str","str","bool","number"],
                row_count=10,
                col_count=(10, "fixed")
            )

            gr.Markdown("### 📋 Full Analyzed Blog Data Table")
            filtered_table = gr.Dataframe(headers=["URL", "Status", "Title", "Content", "Summary"], interactive=False)


    state_full_df = gr.State()

    def analyze(file):
        total, codes, imarts, filtered_df, full_df = extract_summary(file)
        return total, codes, imarts, filtered_df, full_df.values.tolist(), full_df

    def apply_filters(full_df, status, score):
        df = filter_analysis(full_df, status, score)
        return df.values.tolist()

    analyze_btn.click(
        fn=analyze,
        inputs=file_input,
        outputs=[total_blogs, code_snippets, imarticus_hits, filtered_table, full_table, state_full_df]
    )

    status_filter.change(
        fn=apply_filters,
        inputs=[state_full_df, status_filter, score_filter],
        outputs=full_table
    )

    score_filter.change(
        fn=apply_filters,
        inputs=[state_full_df, status_filter, score_filter],
        outputs=full_table
    )

    download_btn.click(
    fn=download_file,
    inputs=state_full_df,
    outputs=download_file_output
    )


demo.launch(share=True)