check email before login; update logs

twscrape/accounts_pool.py

@@ -42,6 +42,7 @@ class AccountsPool:
         if not required.issubset(tokens):
             raise ValueError(f"Invalid line format: {line_format}")
 
+        accounts = []
         with open(filepath, "r") as f:
             lines = f.read().split("\n")
             lines = [x.strip() for x in lines if x.strip()]
@@ -49,12 +50,14 @@ class AccountsPool:
             for line in lines:
                 data = [x.strip() for x in line.split(line_delim)]
                 if len(data) < len(tokens):
-                    logger.warning(f"Invalid line format: {line}")
-                    continue
+                    raise ValueError(f"Invalid line: {line}")
 
                 data = data[: len(tokens)]
                 vals = {k: v for k, v in zip(tokens, data) if k != "_"}
-                await self.add_account(**vals)
+                accounts.append(vals)
+
+        for x in accounts:
+            await self.add_account(**x)
 
     async def add_account(
         self,

twscrape/imap.py

@@ -50,7 +50,7 @@ def search_email_code(imap: imaplib.IMAP4_SSL, count: int, min_t: datetime | Non
                 msg_time = datetime.strptime(msg.get("Date", ""), "%a, %d %b %Y %H:%M:%S %z")
                 msg_from = str(msg.get("From", "")).lower()
                 msg_subj = str(msg.get("Subject", "")).lower()
-                logger.debug(f"({i} of {count}) {msg_from} - {msg_time} - {msg_subj}")
+                logger.info(f"({i} of {count}) {msg_from} - {msg_time} - {msg_subj}")
 
                 if min_t is not None and msg_time < min_t:
                     return None
@@ -62,17 +62,11 @@ def search_email_code(imap: imaplib.IMAP4_SSL, count: int, min_t: datetime | Non
     return None
 
 
-async def get_email_code(email: str, password: str, min_t: datetime | None = None) -> str:
-    domain = get_imap_domain(email)
-    start_time = time.time()
-    with imaplib.IMAP4_SSL(domain) as imap:
-        try:
-            imap.login(email, password)
-        except imaplib.IMAP4.error as e:
-            logger.error(f"Error logging into {email}: {e}")
-            raise EmailLoginError() from e
-
-        was_count = 0
+async def imap_get_email_code(
+    imap: imaplib.IMAP4_SSL, email: str, min_t: datetime | None = None
+) -> str:
+    try:
+        start_time, was_count = time.time(), 0
         while True:
             _, rep = imap.select("INBOX")
             now_count = int(rep[0].decode("utf-8")) if len(rep) > 0 and rep[0] is not None else 0
@@ -81,8 +75,28 @@ async def get_email_code(email: str, password: str, min_t: datetime | None = Non
                 if code is not None:
                     return code
 
-            logger.debug(f"Waiting for confirmation code for {email}, msg_count: {now_count}")
+            logger.info(f"Waiting for confirmation code for {email}, msg_count: {now_count}")
             if MAX_WAIT_SEC < time.time() - start_time:
-                logger.error(f"Timeout waiting for confirmation code for {email}")
+                logger.info(f"Timeout waiting for confirmation code for {email}")
                 raise EmailCodeTimeoutError()
             await asyncio.sleep(5)
+    except Exception as e:
+        imap.select("INBOX")
+        imap.close()
+        logger.error(f"Error getting confirmation code for {email}: {e}")
+        raise e
+
+
+async def imap_try_login(email: str, password: str):
+    domain = get_imap_domain(email)
+    imap = imaplib.IMAP4_SSL(domain)
+
+    try:
+        imap.login(email, password)
+    except imaplib.IMAP4.error as e:
+        logger.error(f"Error logging into {email} on {domain}: {e}")
+        imap.select("INBOX")
+        imap.close()
+        raise EmailLoginError() from e
+
+    return imap

twscrape/login.py

@@ -4,7 +4,7 @@ from httpx import AsyncClient, HTTPStatusError, Response
 
 from .account import Account
 from .constants import LOGIN_URL
-from .imap import get_email_code
+from .imap import imap_get_email_code, imap_try_login
 from .logger import logger
 from .utils import raise_for_status
 
@@ -100,7 +100,7 @@ async def login_duplication_check(client: AsyncClient, acc: Account, prev: dict)
     return rep
 
 
-async def login_confirm_email(client: AsyncClient, acc: Account, prev: dict) -> Response:
+async def login_confirm_email(client: AsyncClient, acc: Account, prev: dict, imap) -> Response:
     payload = {
         "flow_token": prev["flow_token"],
         "subtask_inputs": [
@@ -116,9 +116,9 @@ async def login_confirm_email(client: AsyncClient, acc: Account, prev: dict) ->
     return rep
 
 
-async def login_confirm_email_code(client: AsyncClient, acc: Account, prev: dict):
+async def login_confirm_email_code(client: AsyncClient, acc: Account, prev: dict, imap):
     now_time = datetime.now(timezone.utc) - timedelta(seconds=30)
-    value = await get_email_code(acc.email, acc.email_password, now_time)
+    value = await imap_get_email_code(imap, acc.email, now_time)
 
     payload = {
         "flow_token": prev["flow_token"],
@@ -146,7 +146,7 @@ async def login_success(client: AsyncClient, acc: Account, prev: dict) -> Respon
     return rep
 
 
-async def next_login_task(client: AsyncClient, acc: Account, rep: Response):
+async def next_login_task(client: AsyncClient, acc: Account, rep: Response, imap):
     ct0 = client.cookies.get("ct0", None)
     if ct0:
         client.headers["x-csrf-token"] = ct0
@@ -164,9 +164,8 @@ async def next_login_task(client: AsyncClient, acc: Account, rep: Response):
                 return await login_success(client, acc, prev)
             if task_id == "LoginAcid":
                 is_code = x["enter_text"]["hint_text"].lower() == "confirmation code"
-                # logger.debug(f"is login code: {is_code}")
                 fn = login_confirm_email_code if is_code else login_confirm_email
-                return await fn(client, acc, prev)
+                return await fn(client, acc, prev, imap)
             if task_id == "AccountDuplicationCheck":
                 return await login_duplication_check(client, acc, prev)
             if task_id == "LoginEnterPassword":
@@ -186,10 +185,12 @@ async def next_login_task(client: AsyncClient, acc: Account, rep: Response):
 async def login(acc: Account, fresh=False) -> Account:
     log_id = f"{acc.username} - {acc.email}"
     if acc.active and not fresh:
-        logger.debug(f"account already active {log_id}")
+        logger.info(f"account already active {log_id}")
         return acc
 
-    logger.debug(f"logging in {log_id}")
+    # check if email is valid first
+    imap = await imap_try_login(acc.email, acc.email_password)
+
     client = acc.make_client()
     guest_token = await get_guest_token(client)
     client.headers["x-guest-token"] = guest_token
@@ -200,7 +201,7 @@ async def login(acc: Account, fresh=False) -> Account:
             break
 
         try:
-            rep = await next_login_task(client, acc, rep)
+            rep = await next_login_task(client, acc, rep, imap)
         except HTTPStatusError as e:
             if e.response.status_code == 403:
                 logger.error(f"403 error {log_id}")

twscrape/queue_client.py

@@ -105,7 +105,8 @@ class QueueClient:
         return self.ctx
 
     async def _check_rep(self, rep: httpx.Response):
-        dump_rep(rep)
+        if self.debug:
+            dump_rep(rep)
 
         try:
             res = rep.json()
@@ -116,8 +117,9 @@ class QueueClient:
         if "errors" in res:
             msg = "; ".join([f'({x.get("code", -1)}) {x["message"]}' for x in res["errors"]])
 
-        fn = logger.info if rep.status_code == 200 else logger.warning
-        fn(f"{rep.status_code:3d} - {req_id(rep)} - {msg}")
+        if self.debug:
+            fn = logger.info if rep.status_code == 200 else logger.warning
+            fn(f"{rep.status_code:3d} - {req_id(rep)} - {msg}")
 
         if msg.startswith("The following features cannot be null"):
             logger.error(f"Invalid request: {msg}")

twscrape/utils.py

@@ -22,7 +22,7 @@ def raise_for_status(rep: Response, label: str):
     try:
         rep.raise_for_status()
     except HTTPStatusError as e:
-        logger.debug(f"{label} - {rep.status_code} - {rep.text}")
+        logger.info(f"{label} - {rep.status_code} - {rep.text}")
         raise e