import os
import html
import streamlit as st
import pandas as pd
from atlassian import Jira
import requests
from openai import OpenAI
from datetime import date, timedelta

# -------------------------
# Environment-based secrets
# -------------------------
JIRA_URL = os.getenv("JIRA_URL")
JIRA_USERNAME = os.getenv("JIRA_USERNAME")
JIRA_API_TOKEN = os.getenv("JIRA_API_TOKEN")

ZENDESK_EMAIL = os.getenv("ZENDESK_EMAIL")
ZENDESK_SUBDOMAIN = os.getenv("ZENDESK_SUBDOMAIN")
ZENDESK_API_KEY = os.getenv("ZENDESK_API_KEY")

OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
client = OpenAI(api_key=OPENAI_API_KEY)

# -------------------------
# JIRA Client
# -------------------------
jira = Jira(url=JIRA_URL, username=JIRA_USERNAME, password=JIRA_API_TOKEN)


# -------------------------
# OpenAI Summarization
# -------------------------
@st.cache_data(show_spinner=False)
def summarize_ticket(text: str) -> str:
    if not text:
        return "No description"
    prompt = (
        "Summarize this Zendesk ticket in 1–3 sentences:\n\n" + text + "\n\nSummary:"
    )
    resp = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[{"role": "user", "content": prompt}],
        temperature=0.3,
        max_tokens=150,
    )
    return resp.choices[0].message.content.strip()


# -------------------------
# Zendesk Search Function
# -------------------------
def search_zendesk_tickets(site_name: str, keyword: str) -> pd.DataFrame:
    terms = []
    if site_name:
        terms.append(f'"{site_name}"')
    if keyword:
        terms.append(f'"{keyword}"')
    query_str = " ".join(terms)

    url = f"https://{ZENDESK_SUBDOMAIN}.zendesk.com/api/v2/search.json"
    params = {"query": f"type:ticket {query_str}", "include": "users"}
    auth = (f"{ZENDESK_EMAIL}/token", ZENDESK_API_KEY)
    resp = requests.get(url, auth=auth, params=params)
    if not resp.ok:
        st.error(f"Zendesk error {resp.status_code}")
        return pd.DataFrame()

    tickets = resp.json().get("results", [])
    rows = []
    for t in tickets:
        rows.append(
            {
                "ID": t["id"],
                "Subject": html.escape(t.get("subject", "")),
                "Status": html.escape(t.get("status", "")),
                "Created At": t.get("created_at", ""),
                "Updated At": t.get("updated_at", ""),
                "Description": t.get("description", ""),  # keep for summary
            }
        )
    df = pd.DataFrame(rows)

    # generate summaries and attach as new column
    df["OpenAI Ticket Summary"] = df["Description"].apply(summarize_ticket)
    return df


# -------------------------
# Jira Search Function
# -------------------------


@st.cache_data(show_spinner=False)
def search_jira_issues(
    site_name: str, keyword: str, start_date: date, end_date: date
) -> pd.DataFrame:
    # Build JQL clauses
    clauses = []
    if site_name:
        clauses.append(f'text ~ "{site_name}"')
    if keyword:
        clauses.append(f'text ~ "{keyword}"')
    clauses.append(f'created >= "{start_date.isoformat()}"')
    clauses.append(f'created <= "{end_date.isoformat()}"')
    jql = " AND ".join(clauses)

    # Execute the JQL query, limiting to 100 issues
    resp = jira.jql(jql, limit=100)
    issues = resp.get("issues", [])

    rows = []
    for issue in issues:
        f = issue["fields"]
        rows.append(
            {
                "Key": issue["key"],
                "Summary": html.escape(f.get("summary", "")),
                "Status": html.escape(f.get("status", {}).get("name", "")),
                "Created At": f.get("created", ""),
                "Updated At": f.get("updated", ""),
            }
        )

    return pd.DataFrame(rows)


# -------------------------
# App Config
# -------------------------
st.set_page_config(layout="wide")
st.title("Unified Support Dashboard")

if "zendesk_df" not in st.session_state:
    st.session_state.zendesk_df = pd.DataFrame()
if "jira_df" not in st.session_state:
    st.session_state.jira_df = pd.DataFrame()

# -------------------------
# Main: Tabs
# -------------------------
tabs = st.tabs(["Zendesk Lookup", "Jira Lookup"])

# ---- Tab 1: Zendesk ----
with tabs[0]:
    st.header("Zendesk Lookup")
    site_input = st.text_input(
        "Site Name", placeholder="example.com", key="zendesk_site"
    )
    keyword_input = st.text_input(
        "Keyword", placeholder="timeout", key="zendesk_keyword"
    )
    start_input = st.date_input(
        "Created After", value=date.today() - timedelta(days=7), key="zendesk_start"
    )
    end_input = st.date_input("Created Before", value=date.today(), key="zendesk_end")
    if st.button("Search Zendesk Tickets", key="zendesk_search"):
        st.session_state.zendesk_df = search_zendesk_tickets(site_input, keyword_input)

    df_z = st.session_state.zendesk_df.copy()
    if not df_z.empty:
        # parse & filter dates
        df_z["Created At"] = pd.to_datetime(df_z["Created At"])
        df_z["Updated At"] = pd.to_datetime(df_z["Updated At"])
        mask = (df_z["Created At"].dt.date >= start_input) & (
            df_z["Created At"].dt.date <= end_input
        )
        df_z = df_z.loc[mask]

        # sort by Created At descending
        df_z = df_z.sort_values("Created At", ascending=False)

        # format timestamps 12-hour
        df_z["Created At"] = (
            df_z["Created At"].dt.strftime("%Y-%m-%d %I:%M %p").str.lower()
        )
        df_z["Updated At"] = (
            df_z["Updated At"].dt.strftime("%Y-%m-%d %I:%M %p").str.lower()
        )

        # hyperlink ID
        base_url = f"https://{ZENDESK_SUBDOMAIN}.zendesk.com/agent/tickets"
        df_z["ID"] = df_z["ID"].apply(
            lambda x: f'<a href="{base_url}/{x}" target="_blank">{x}</a>'
        )

        # render fixed-height table including the new summary column
        html_tbl = df_z.to_html(
            index=False,
            escape=False,
            columns=[
                "ID",
                "Subject",
                "Status",
                "Created At",
                "Updated At",
                "OpenAI Ticket Summary",
            ],
        )
        scrollable = f"""
        <div style="height: 400px; overflow-y: auto; border: 1px solid #ddd; padding: 4px;">
          {html_tbl}
        </div>
        """
        st.markdown(scrollable, unsafe_allow_html=True)

# ---- Tab 2: Jira ----
with tabs[1]:
    st.header("Jira Lookup")
    site_input = st.text_input("Site Name", placeholder="example.com", key="jira_site")
    keyword_input = st.text_input("Keyword", placeholder="timeout", key="jira_keyword")
    start_input = st.date_input(
        "Created After", value=date.today() - timedelta(days=7), key="jira_start"
    )
    end_input = st.date_input("Created Before", value=date.today(), key="jira_end")
    if st.button("Search Jira Issues", key="jira_search"):
        st.session_state.jira_df = search_jira_issues(
            site_input, keyword_input, start_input, end_input
        )

    df_j = st.session_state.jira_df.copy()

    if not df_j.empty:
        # parse & sort by Created At descending
        df_j["Created At"] = pd.to_datetime(df_j["Created At"])
        df_j["Updated At"] = pd.to_datetime(df_j["Updated At"])
        df_j = df_j.sort_values("Created At", ascending=False)

        # 12-hour fmt with am/pm
        df_j["Created At"] = (
            df_j["Created At"].dt.strftime("%Y-%m-%d %I:%M %p").str.lower()
        )
        df_j["Updated At"] = (
            df_j["Updated At"].dt.strftime("%Y-%m-%d %I:%M %p").str.lower()
        )

        # hyperlink the key to the JIRA issue
        base_jira = JIRA_URL.rstrip("/")
        df_j["Key"] = df_j["Key"].apply(
            lambda k: f'<a href="{base_jira}/browse/{k}" target="_blank">{k}</a>'
        )

        # render as fixed-height, scrollable HTML table
        html_tbl = df_j.to_html(
            index=False,
            escape=False,
            columns=["Key", "Summary", "Status", "Created At", "Updated At"],
        )
        scrollable = f"""
        <div style="height: 400px; overflow-y: auto; border: 1px solid #ddd; padding: 4px;">
          {html_tbl}
        </div>
        """
        st.markdown(scrollable, unsafe_allow_html=True)