"""
Advance an oTree participant through all remaining pages.
"""

import re
import sys
import requests

START_URL = "http://localhost:8000/p/g4qy7hsr/welfare_study/Application/29"
MAX_PAGES = 200

Q_HIDDEN = re.compile(r'<input[^>]+type=["\']hidden["\'][^>]*>', re.IGNORECASE)
Q_NAME = re.compile(r'name=["\']([^"\']+)["\']')
Q_VALUE = re.compile(r'value=["\']([^"\']*)["\']')
Q_INPUT = re.compile(r'<input[^>]+>', re.IGNORECASE)
Q_TYPE = re.compile(r'type=["\']([^"\']+)["\']')
Q_SELECT = re.compile(r'<select[^>]+name=["\']([^"\']+)["\']', re.IGNORECASE)
Q_TEXTAREA = re.compile(r'<textarea[^>]+name=["\']([^"\']+)["\']', re.IGNORECASE)


def extract_hidden(html):
    fields = {}
    for match in Q_HIDDEN.finditer(html):
        tag = match.group(0)
        nm = Q_NAME.search(tag)
        vm = Q_VALUE.search(tag)
        if nm:
            fields[nm.group(1)] = vm.group(1) if vm else ""
    return fields


def extract_visible(html):
    fields = {}
    for match in Q_INPUT.finditer(html):
        tag = match.group(0)
        if "hidden" in tag.lower():
            continue
        nm = Q_NAME.search(tag)
        tm = Q_TYPE.search(tag)
        if nm:
            fields[nm.group(1)] = tm.group(1).lower() if tm else "text"
    for match in Q_SELECT.finditer(html):
        fields[match.group(1)] = "select"
    for match in Q_TEXTAREA.finditer(html):
        fields[match.group(1)] = "textarea"
    return fields


def default_val(ft):
    return {"number": "1", "text": "test", "textarea": "test", "radio": "1",
            "select": "1", "checkbox": "on", "range": "50"}.get(ft, "1")


def page_name(url):
    parts = url.rstrip("/").split("/")
    if len(parts) >= 2:
        return parts[-2] if parts[-1].isdigit() else parts[-1]
    return url


def main():
    s = requests.Session()
    url = START_URL
    count = 0
    print(f"Starting at: {url}")
    print("=" * 70)

    while count < MAX_PAGES:
        count += 1
        try:
            resp = s.get(url, allow_redirects=True)
        except requests.ConnectionError as e:
            print(f"ERROR: Cannot connect: {e}")
            sys.exit(1)

        cur = resp.url
        name = page_name(cur)
        print(f"\n[{count}] {name} | {cur} | {resp.status_code}")

        if "Redirect" in cur or "OutOfRangeNotification" in cur:
            print(f"\n{'=' * 70}\nDONE after {count} pages. Final: {cur}")
            return

        if resp.status_code == 404:
            print("404 - stopping.")
            sys.exit(1)

        html = resp.text
        if "otree-wait-page" in html or "wait-page-body" in html:
            print("Wait page - stopping.")
            sys.exit(1)

        hidden = extract_hidden(html)
        csrf = hidden.get("csrfmiddlewaretoken", "")
        if not csrf and "<form" not in html.lower():
            print(f"No form. DONE after {count} pages.")
            return

        post_data = dict(hidden)
        for fn, ft in extract_visible(html).items():
            if fn not in post_data:
                post_data[fn] = default_val(ft)

        print(f"  Submitting {len(post_data)} fields...")
        try:
            pr = s.post(cur, data=post_data, allow_redirects=True,
                        headers={"Referer": cur})
        except requests.ConnectionError as e:
            print(f"POST error: {e}")
            sys.exit(1)

        nxt = pr.url
        if nxt == cur and pr.status_code == 200:
            visible2 = extract_visible(pr.text)
            hidden2 = extract_hidden(pr.text)
            post_data.update(hidden2)
            for fn, ft in visible2.items():
                if fn not in post_data or not post_data[fn]:
                    post_data[fn] = default_val(ft)
                    print(f"    Filling {fn} = {post_data[fn]}")
            pr = s.post(cur, data=post_data, allow_redirects=True,
                        headers={"Referer": cur})
            nxt = pr.url
            if nxt == cur:
                print("  STUCK on this page!")
                errs = re.findall(r'error[^>]*>([^<]+)', pr.text)
                for e in errs[:5]:
                    print(f"    {e.strip()}")

        url = nxt
        print(f"  -> {nxt}")

    print(f"Safety limit ({MAX_PAGES}) reached.")
    sys.exit(1)


if __name__ == "__main__":
    main()