import os
import re
import json
import uuid
import urllib.request
import urllib.parse

PROJECT_ID = "gdvj8e5j"
DATASET = "production"
TOKEN = "skdQbMqjLbrlUWE69MCg8uUwYlwz0yDv3T6hlkmk8TnwvpexPoOF7xVK7VhVimpnSVMbzn5lXjSqlSkdPwJYA6eF86SZAdHguvAtVrJeAXUQjAOjLgifGjhafmoRdK86wwLDcDBSIP7IdMju1JABNsD3PwfElzoiGcZDYL83ZUK8VQia0Cp7"
API_VERSION = "2024-10-31"

BASE_MUTATE_URL = f"https://{PROJECT_ID}.api.sanity.io/v{API_VERSION}/data/mutate/{DATASET}"
BASE_QUERY_URL = f"https://{PROJECT_ID}.api.sanity.io/v{API_VERSION}/data/query/{DATASET}"

IMAGES_DIR = "/opt/ai-os/products/ceo/projects/blog/vwg-content/tai-sao-chon-viet-nam/images"
DRAFT_PATH = "/opt/ai-os/products/ceo/projects/blog/vwg-content/tai-sao-chon-viet-nam/draft_vi_v1.md"

def get_existing_assets():
    # Query assets already uploaded
    query = '*[_type == "sanity.imageAsset"]{_id, originalFilename, url}'
    url = f"{BASE_QUERY_URL}?" + urllib.parse.urlencode({"query": query})
    req = urllib.request.Request(url, headers={"Authorization": f"Bearer {TOKEN}"})
    with urllib.request.urlopen(req) as resp:
        res = json.loads(resp.read().decode())["result"]
    assets = {}
    for a in res:
        fn = a.get("originalFilename")
        if fn:
            assets[fn] = {"id": a["_id"], "url": a["url"]}
    return assets

def make_span(text, marks=None):
    return {
        "_key": uuid.uuid4().hex[:8],
        "_type": "span",
        "marks": marks or [],
        "text": text
    }

def make_block(text, style="normal", list_item=None, level=None):
    return {
        "_key": uuid.uuid4().hex[:8],
        "_type": "block",
        "style": style,
        "markDefs": [],
        "children": [make_span(text)],
        **({"listItem": list_item, "level": level or 1} if list_item else {})
    }

def parse_draft(md_path, uploaded_assets):
    with open(md_path, "r", encoding="utf-8") as f:
        content = f.read()

    lines = content.split("\n")
    
    # Defaults
    seo_title = "Tại sao chọn Việt Nam tổ chức đám cưới điểm đến?"
    meta_desc = "Tìm hiểu lý do nhiều cặp đôi chọn Việt Nam làm nơi tổ chức đám cưới điểm đến cùng các bối cảnh đa dạng và những điều cần lưu ý."

    # Parse metadata header strictly
    body_start_idx = 0
    for idx, line in enumerate(lines):
        stripped = line.strip()
        if stripped.startswith("Meta title:"):
            seo_title = stripped.replace("Meta title:", "").strip()
        elif stripped.startswith("Meta description:"):
            meta_desc = stripped.replace("Meta description:", "").strip()
        elif stripped.startswith("# "):
            body_start_idx = idx
            break

    body_lines = lines[body_start_idx:]
    blocks = []
    
    idx = 0
    in_faq_section = False
    faq_items = []
    cur_q = ""
    cur_a = []

    while idx < len(body_lines):
        line = body_lines[idx]
        stripped = line.strip()
        
        if not stripped:
            idx += 1
            continue

        # Stop and omit any code blocks (like ```json ... ``` at the bottom)
        if stripped.startswith("```"):
            while idx < len(body_lines) and not body_lines[idx].strip().endswith("```") and idx + 1 < len(body_lines):
                idx += 1
            idx += 1
            continue

        # Handle Table
        if stripped.startswith("|") and stripped.endswith("|"):
            table_rows = []
            while idx < len(body_lines):
                t_line = body_lines[idx].strip()
                if t_line.startswith("|") and t_line.endswith("|"):
                    cells = [c.strip() for c in t_line.split("|")[1:-1]]
                    if not all(c.startswith("-") or not c for c in cells):
                        table_rows.append(cells)
                    idx += 1
                else:
                    break
            if table_rows:
                blocks.append({
                    "_key": uuid.uuid4().hex[:8],
                    "_type": "tableBlock",
                    "rows": [
                        {"_key": uuid.uuid4().hex[:8], "cells": r} for r in table_rows
                    ]
                })
            continue

        # Detect FAQ Section
        if stripped.startswith("## Các câu hỏi thường gặp") or stripped.startswith("## FAQ"):
            in_faq_section = True
            blocks.append(make_block(stripped[3:].strip(), "h2"))
            idx += 1
            continue

        if in_faq_section:
            # Check if another H2 starts (end of FAQ)
            if stripped.startswith("## ") and not stripped.startswith("### "):
                if cur_q and cur_a:
                    faq_items.append({"question": cur_q, "answer": " ".join(cur_a)})
                    cur_q, cur_a = "", []
                if faq_items:
                    blocks.append({
                        "_key": uuid.uuid4().hex[:8],
                        "_type": "faqAccordion",
                        "items": faq_items
                    })
                    faq_items = []
                in_faq_section = False
                # Do not increment, let normal processing take this H2
                continue
            
            if stripped.startswith("### "):
                if cur_q and cur_a:
                    faq_items.append({"question": cur_q, "answer": " ".join(cur_a)})
                    cur_q, cur_a = "", []
                cur_q = stripped[4:].strip()
                idx += 1
                continue
            elif cur_q:
                cur_a.append(stripped)
                idx += 1
                continue

        # Markdown image
        img_match = re.match(r'!\[(.*?)\]\((.*?)\)', stripped)
        if img_match:
            alt = img_match.group(1).strip()
            src = img_match.group(2).strip()
            filename = os.path.basename(src)
            asset_info = uploaded_assets.get(filename)
            if asset_info:
                blocks.append({
                    "_key": uuid.uuid4().hex[:8],
                    "_type": "image",
                    "asset": {
                        "_type": "reference",
                        "_ref": asset_info["id"]
                    },
                    "alt": alt
                })
            idx += 1
            continue

        # Skip comment metadata or caption lines
        if stripped.startswith("<!-- Metadata:") or (stripped.startswith("*") and stripped.endswith("*") and len(stripped) < 140):
            idx += 1
            continue

        # Headings
        if stripped.startswith("# "):
            blocks.append(make_block(stripped[2:].strip(), "h1"))
        elif stripped.startswith("## "):
            blocks.append(make_block(stripped[3:].strip(), "h2"))
        elif stripped.startswith("### "):
            blocks.append(make_block(stripped[4:].strip(), "h3"))
        elif stripped.startswith("#### "):
            blocks.append(make_block(stripped[5:].strip(), "h4"))
        elif stripped.startswith("* ") or stripped.startswith("- "):
            blocks.append(make_block(stripped[2:].strip(), "normal", list_item="bullet"))
        elif re.match(r'^\d+\.\s', stripped):
            item_text = re.sub(r'^\d+\.\s', '', stripped)
            blocks.append(make_block(item_text, "normal", list_item="number"))
        elif stripped.startswith("> "):
            blocks.append(make_block(stripped[2:].strip(), "blockquote"))
        else:
            blocks.append(make_block(stripped, "normal"))

        idx += 1

    # If FAQ section was at the end of document
    if in_faq_section and cur_q and cur_a:
        faq_items.append({"question": cur_q, "answer": " ".join(cur_a)})
    if faq_items:
        blocks.append({
            "_key": uuid.uuid4().hex[:8],
            "_type": "faqAccordion",
            "items": faq_items
        })

    # Extract excerpt
    excerpt = meta_desc
    for b in blocks:
        if b.get("_type") == "block" and b.get("style") == "normal" and not b.get("listItem"):
            text = b["children"][0]["text"]
            if len(text) > 60:
                excerpt = text[:220].strip() + "..."
                break

    return seo_title, meta_desc, excerpt, blocks

def main():
    assets = get_existing_assets()
    seo_title, meta_desc, excerpt, blocks = parse_draft(DRAFT_PATH, assets)
    
    cover_filename = "dam-cuoi-diem-den-viet-nam-01.webp"
    cover_asset = assets.get(cover_filename, {})
    
    print(f"Tổng số block sau khi chuẩn hóa: {len(blocks)}")
    types = [b.get("_type") for b in blocks]
    print(f"Các loại block: {set(types)}")
    print(f"Số lượng tableBlock: {types.count('tableBlock')}")
    print(f"Số lượng faqAccordion: {types.count('faqAccordion')}")
    print(f"Tiêu đề bài viết: {seo_title}")
    
    post_doc = {
        "_id": "post-tai-sao-chon-viet-nam",
        "_type": "post",
        "title": seo_title,
        "slug": {
            "_type": "slug",
            "current": "tai-sao-chon-viet-nam"
        },
        "excerpt": excerpt,
        "publishedAt": "2026-10-07T08:00:00.000Z",
        "featured": True,
        "author": {
            "_type": "reference",
            "_ref": "author-nhi-vo"
        },
        "categories": [
            {
                "_type": "reference",
                "_ref": "category-destination"
            }
        ],
        "image": {
            "_type": "image",
            "asset": {
                "_type": "reference",
                "_ref": cover_asset.get("id")
            },
            "alt": "Toàn cảnh lễ cưới bên bờ biển Việt Nam sang trọng và lãng mạn"
        },
        "meta": {
            "_type": "meta",
            "title": seo_title,
            "description": meta_desc,
            "noindex": False
        },
        "body": blocks
    }
    
    req = urllib.request.Request(
        BASE_MUTATE_URL,
        data=json.dumps({"mutations": [{"createOrReplace": post_doc}]}).encode("utf-8"),
        headers={
            "Authorization": f"Bearer {TOKEN}",
            "Content-Type": "application/json"
        },
        method="POST"
    )
    
    with urllib.request.urlopen(req) as resp:
        res = json.loads(resp.read().decode())
        print("✅ KẾT QUẢ MUTATION:", json.dumps(res, ensure_ascii=False))

if __name__ == "__main__":
    main()