Fetch students who passed first stage for analysing.

analyse.py

import json

with open("cac.json", "r", encoding="utf-8") as f:
    cac_db = json.load(f)

with open("ntut.json", "r", encoding="utf-8") as f:
    ntut = json.load(f)

with open("ntust.json", "r", encoding="utf-8") as f:
    ntust = json.load(f)

ncu = set(cac_db["016222"]["number"])
ntut = set(ntut)
ntust = set(ntust)

ntust_inter = ncu.intersection(ntust)
ntut_inter = ncu.intersection(ntut)
jctv_inter = ntut.intersection(ntust)

ncu_inter = jctv_inter.intersection(ncu)


print(len(ntust_inter))
print(len(ntut_inter))
print(len(jctv_inter))

print(len(ncu_inter))

with open("dup.json", "w", encoding="utf-8") as f:
    data = dict()
    data["ntust_inter"] = list(ntust_inter)
    data["ntut_inter"] = list(ntut_inter)
    data["jctv_inter"] = list(jctv_inter)
    data["ncu_inter"] = list(ncu_inter)
    json.dump(data, f, indent=2)

cac.py

import json
import re
import requests as r
from tqdm import tqdm
from bs4 import BeautifulSoup
from time import sleep

# list api
api = "https://www.cac.edu.tw/CacLink/apply109/109apply_6pSieveXw_ju9sd98/html_sieve_109pgx3/ColPost/common/apply/{}.htm"
department_name = "資訊工程"

# getting departments list
try:
    with open("departments.json", "r", encoding="utf-8") as f:
        departments = json.load(f)
        print("Read local departments.json.")
except Exception as err:
    print(err)
    departments = dict()
    print("Fetching departments...")

    res = r.post(
        "https://www.cac.edu.tw/apply109/system/109ColQrytk4p_forapply_os92k5w/ShowGsd.php", 
        {
            "TxtGsd": department_name,
            "SubTxtGsd": "依學系名稱查詢",
            "action": "SubTxtGsd"
        }
    )
    res.encoding = "utf-8"

    doc = BeautifulSoup(res.text, "html.parser")
    l = len(doc.select("table tr"))
    for i, row in enumerate(doc.select("table tr")):
        if i == 0 or i == (l - 1):
            continue
        ele = row.findChildren("td")[0] # name
        string = re.sub(r"\s+", " ", ele.get_text())
        code = string[-7:-1]
        departments[code] = string[0:-9]
    with open("departments.json", "w", encoding="utf-8") as f:
        json.dump(departments, f, ensure_ascii=False, indent=2)

print("Total: {} departments.".format(len(departments.keys())))

# database
data = dict()
bar = tqdm(total=len(departments.keys()), ascii=True, bar_format="{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}]", position=0, leave=True)
for k in departments.keys():
    bar.set_description_str(k)
    tmp = {
        "name": departments.get(k),
        "number": []
    }
    res = r.get(api.format(k))
    res.encoding = "utf-8"
    doc = BeautifulSoup(res.text, "html.parser")
    l = len(doc.select("div > span"))
    for i, ele in enumerate(doc.select("div > span")):
        if i == 0 or i == (l - 1):
            continue
        for num in ele.findChildren("span"):
            tmp["number"].append(num.string)
    data[k] = tmp
    bar.update()
    sleep(0.01)

bar.close()

# save data
with open("cac.json", "w", encoding="utf-8") as f:
    json.dump(data, f, ensure_ascii=False, indent=2)
    print("data saved.")

jctv.js

// https://ent01.jctv.ntut.edu.tw/applys1result/college.html
var data = [];
Array.from(document.querySelector(".enterTable tbody").children).forEach((e) => {
	if (e.children[0].textContent == "101009") { // 101009 department code
		data.push(e.children[2].textContent.replace(/\s/g, "").slice(-9, -1));
	}
});
console.log(JSON.stringify(data, null, 2));