File "debug_ibx.py"

Full path: /home/algopkco/public_html/scraper/debug_ibx.py
File size: 819 B (819 B bytes)
MIME-type: text/plain
Charset: utf-8

Download   Open   Edit   Advanced Editor &nnbsp; Back

import importlib.util
import re

spec = importlib.util.spec_from_file_location("fibx", r"scraper\fetch_indiabix.py")
mod = importlib.util.module_from_spec(spec)
spec.loader.exec_module(mod)

html = open("https_www_indiabix_comaptitudeprobability.html", encoding="utf-8").read()
blocks = mod.split_blocks(html)
print("blocks:", len(blocks))
for i, b in enumerate(blocks):
    print("block", i, "len:", len(b), "img:", "<img" in b, "qtxt:", 'bix-td-qtxt' in b)
    if "bix-td-qtxt" in b and "<img" not in b:
        qm = re.search(r'class="bix-td-qtxt[^"]*">(.*?)</div>', b, re.S)
        print("  q:", mod.clean_html(qm.group(1))[:100] if qm else None)
        opts = re.findall(r'id="tdOptionDt_[A-Z]_[0-9]+">(.*?)\s*</div>\s*</div>', b, re.S)
        print("  opts:", len(opts), [mod.clean_html(o)[:40] for o in opts])