barrett/cp.nvim
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

fix(usaco): fix login field, submit endpoint, form parsing, and lang match

Browse source 
Problem: Login used `user=` field (correct is `uname=`) and checked for
`"success"` key in response (USACO returns `{"code":1}`). Submit used
the wrong endpoint, wrong file field name (`sub_file[]` vs `sourcefile`),
had an off-by-one in hidden field extraction (`group(2)` vs `group(1)`),
and `_pick_lang_option` matched C++11 before C++17 due to iterating
options before keywords. Login fast path also accepted stale cookies
with wrong credentials.

Solution: Fix `uname` field and `code==1` check. Fix submit to
`submit-solution.php`, `sourcefile` field, and `group(1)`. Invert
`_pick_lang_option` loop order (keywords outer, options inner) so
specific keywords like `"c++17"` match before `"c++"`. Remove fast
path from login subcommand so credentials are always validated.
This commit is contained in:
Barrett Ruth 2026-03-06 12:29:31 -05:00
parent 01fc2f26e9
commit 57f57f931f
Signed by: barrett
GPG key ID: A6C96C9349D2FC81
1 changed files with 17 additions and 25 deletions
Download patch file Download diff file Expand all files Collapse all files

42
scrapers/usaco.py
View file

@ -29,7 +29,7 @@ CONNECTIONS = 4
_COOKIE_PATH = Path.home() / ".cache" / "cp-nvim" / "usaco-cookies.json" _COOKIE_PATH = Path.home() / ".cache" / "cp-nvim" / "usaco-cookies.json"
_LOGIN_PATH = "/current/tpcm/login-session.php" _LOGIN_PATH = "/current/tpcm/login-session.php"
_SUBMIT_PATH = "/current/tpcm/submitproblem.php" _SUBMIT_PATH = "/current/tpcm/submit-solution.php"
_LANG_KEYWORDS: dict[str, list[str]] = { _LANG_KEYWORDS: dict[str, list[str]] = {
"cpp": ["c++17", "c++ 17", "g++17", "c++", "cpp"], "cpp": ["c++17", "c++ 17", "g++17", "c++", "cpp"],
@ -141,13 +141,16 @@ def _parse_problem_page(html: str) -> dict[str, Any]:
def _pick_lang_option(select_body: str, language_id: str) -> str | None: def _pick_lang_option(select_body: str, language_id: str) -> str | None:
keywords = _LANG_KEYWORDS.get(language_id.lower(), [language_id.lower()]) keywords = _LANG_KEYWORDS.get(language_id.lower(), [language_id.lower()])
for m in re.finditer( options = [
r'<option\b[^>]*\bvalue=["\']([^"\']*)["\'][^>]*>([^<]+)', (m.group(1), m.group(2).strip().lower())
select_body, for m in re.finditer(
re.IGNORECASE, r'<option\b[^>]*\bvalue=["\']([^"\']*)["\'][^>]*>([^<]+)',
): select_body,
val, text = m.group(1), m.group(2).strip().lower() re.IGNORECASE,
for kw in keywords: )
]
for kw in keywords:
for val, text in options:
if kw in text: if kw in text:
return val return val
return None return None
@ -165,7 +168,7 @@ def _parse_submit_form(
re.DOTALL | re.IGNORECASE, re.DOTALL | re.IGNORECASE,
): ):
action, body = form_m.group(1), form_m.group(2) action, body = form_m.group(1), form_m.group(2)
if "sub_file" not in body.lower(): if "sourcefile" not in body.lower():
continue continue
if action.startswith("http"): if action.startswith("http"):
form_action = action form_action = action
@ -182,7 +185,7 @@ def _parse_submit_form(
name_m = re.search(r'\bname=["\']([^"\']+)["\']', tag, re.IGNORECASE) name_m = re.search(r'\bname=["\']([^"\']+)["\']', tag, re.IGNORECASE)
val_m = re.search(r'\bvalue=["\']([^"\']*)["\']', tag, re.IGNORECASE) val_m = re.search(r'\bvalue=["\']([^"\']*)["\']', tag, re.IGNORECASE)
if name_m and val_m: if name_m and val_m:
hidden[name_m.group(1)] = val_m.group(2) hidden[name_m.group(1)] = val_m.group(1)
for sel_m in re.finditer( for sel_m in re.finditer(
r'<select\b[^>]*\bname=["\']([^"\']+)["\'][^>]*>(.*?)</select>', r'<select\b[^>]*\bname=["\']([^"\']+)["\'][^>]*>(.*?)</select>',
body, body,
@ -231,16 +234,15 @@ async def _do_usaco_login(
) -> bool: ) -> bool:
r = await client.post( r = await client.post(
f"{_AUTH_BASE}{_LOGIN_PATH}", f"{_AUTH_BASE}{_LOGIN_PATH}",
data={"user": username, "password": password}, data={"uname": username, "password": password},
headers=HEADERS, headers=HEADERS,
timeout=HTTP_TIMEOUT, timeout=HTTP_TIMEOUT,
) )
r.raise_for_status() r.raise_for_status()
try: try:
data = r.json() return r.json().get("code") == 1
return bool(data.get("success") or data.get("status") == "success")
except Exception: except Exception:
return r.status_code == 200 and "error" not in r.text.lower() return False
class USACOScraper(BaseScraper): class USACOScraper(BaseScraper):
@ -453,7 +455,7 @@ class USACOScraper(BaseScraper):
r = await client.post( r = await client.post(
form_url, form_url,
data=data, data=data,
files={"sub_file[]": (f"solution.{ext}", source, "text/plain")}, files={"sourcefile": (f"solution.{ext}", source, "text/plain")},
headers=HEADERS, headers=HEADERS,
timeout=HTTP_TIMEOUT, timeout=HTTP_TIMEOUT,
) )
@ -477,16 +479,6 @@ class USACOScraper(BaseScraper):
return self._login_error("Missing username or password") return self._login_error("Missing username or password")
async with httpx.AsyncClient(follow_redirects=True) as client: async with httpx.AsyncClient(follow_redirects=True) as client:
await _load_usaco_cookies(client)
if client.cookies:
print(json.dumps({"status": "checking_login"}), flush=True)
if await _check_usaco_login(client, username):
return LoginResult(
success=True,
error="",
credentials={"username": username, "password": password},
)
print(json.dumps({"status": "logging_in"}), flush=True) print(json.dumps({"status": "logging_in"}), flush=True)
try: try:
ok = await _do_usaco_login(client, username, password) ok = await _do_usaco_login(client, username, password)