Add more session tests and small error improvements

fbchat/_session.py

@@ -4,7 +4,10 @@ import requests
 import random
 import re
 import json
-import urllib.parse
+
+# TODO: Only import when required
+# Or maybe just replace usage with `html.parser`?
+import bs4
 
 from ._common import log, kw_only
 from . import _graphql, _util, _exception
@@ -27,6 +30,10 @@ def parse_server_js_define(html: str) -> Mapping[str, Any]:
     rtn = []
     if not define_splits:
         raise _exception.ParseError("Could not find any ServerJSDefine", data=html)
+    if len(define_splits) < 2:
+        raise _exception.ParseError("Could not find enough ServerJSDefine", data=html)
+    if len(define_splits) > 2:
+        raise _exception.ParseError("Found too many ServerJSDefine", data=define_splits)
     # Parse entries (should be two)
     for entry in define_splits:
         try:
@@ -94,16 +101,7 @@ def client_id_factory() -> str:
     return hex(int(random.random() * 2 ** 31))[2:]
 
 
-def get_next_url(url: str) -> Optional[str]:
-    parsed_url = urllib.parse.urlparse(url)
-    query = urllib.parse.parse_qs(parsed_url.query)
-    return query.get("next", [None])[0]
-
-
 def find_form_request(html: str):
-    # Only import when required
-    import bs4
-
     soup = bs4.BeautifulSoup(html, "html.parser", parse_only=bs4.SoupStrainer("form"))
 
     form = soup.form
@@ -114,6 +112,7 @@ def find_form_request(html: str):
     if not url:
         raise _exception.ParseError("Could not find url to submit to", data=form)
 
+    # From what I've seen, it'll always do this!
     if url.startswith("/"):
         url = "https://www.facebook.com" + url
 
@@ -168,15 +167,12 @@ def two_factor_helper(session: requests.Session, r, on_2fa_callback):
 
 def get_error_data(html: str) -> Optional[str]:
     """Get error message from a request."""
-    # Only import when required
-    import bs4
-
     soup = bs4.BeautifulSoup(
         html, "html.parser", parse_only=bs4.SoupStrainer("form", id="login_form")
     )
     # Attempt to extract and format the error string
     # The error message is in the user's own language!
-    return ". ".join(list(soup.stripped_strings)[:2]) or None
+    return " ".join(list(soup.stripped_strings)[1:3]) or None
 
 
 def get_fb_dtsg(define) -> Optional[str]:
@@ -298,7 +294,7 @@ class Session:
                 )
             # Get a facebook.com url that handles the 2FA flow
             # This probably works differently for Messenger-only accounts
-            url = get_next_url(url)
+            url = _util.get_url_parameter(url, "next")
             # Explicitly allow redirects
             r = session.get(url, allow_redirects=True)
             url = two_factor_helper(session, r, on_2fa_callback)

tests/test_session.py

@@ -1,15 +1,47 @@
 import datetime
 import pytest
+from fbchat import ParseError
 from fbchat._session import (
+    parse_server_js_define,
     base36encode,
     prefix_url,
     generate_message_id,
+    session_factory,
     client_id_factory,
-    is_home,
+    find_form_request,
     get_error_data,
 )
 
 
+def test_parse_server_js_define():
+    html = """
+    some data;require("TimeSliceImpl").guard(function(){(require("ServerJSDefine")).handleDefines([["DTSGInitialData",[],{"token":"123"},100]])
+
+    <script>require("TimeSliceImpl").guard(function() {require("ServerJSDefine").handleDefines([["DTSGInitData",[],{"token":"123","async_get_token":"12345"},3333]])
+
+    </script>
+    other irrelevant data
+    """
+    define = parse_server_js_define(html)
+    assert define == {
+        "DTSGInitialData": {"token": "123"},
+        "DTSGInitData": {"async_get_token": "12345", "token": "123"},
+    }
+
+
+def test_parse_server_js_define_error():
+    with pytest.raises(ParseError, match="Could not find any"):
+        parse_server_js_define("")
+
+    html = 'function(){(require("ServerJSDefine")).handleDefines([{"a": function(){}}])'
+    with pytest.raises(ParseError, match="Invalid"):
+        parse_server_js_define(html + html)
+
+    html = 'function(){require("ServerJSDefine").handleDefines({"a": "b"})'
+    with pytest.raises(ParseError, match="Invalid"):
+        parse_server_js_define(html + html)
+
+
 @pytest.mark.parametrize(
     "number,expected",
     [(1, "1"), (10, "a"), (123, "3f"), (1000, "rs"), (123456789, "21i3v9")],
@@ -19,8 +51,10 @@ def test_base36encode(number, expected):
 
 
 def test_prefix_url():
-    assert prefix_url("/") == "https://www.facebook.com/"
+    static_url = "https://upload.messenger.com/"
-    assert prefix_url("/abc") == "https://www.facebook.com/abc"
+    assert prefix_url(static_url) == static_url
+    assert prefix_url("/") == "https://www.messenger.com/"
+    assert prefix_url("/abc") == "https://www.messenger.com/abc"
 
 
 def test_generate_message_id():
@@ -28,46 +62,106 @@ def test_generate_message_id():
     assert generate_message_id(datetime.datetime.utcnow(), "def")
 
 
+def test_session_factory():
+    session = session_factory()
+    assert session.headers
+
+
 def test_client_id_factory():
     # Returns random output, so hard to test more thoroughly
     assert client_id_factory()
 
 
-def test_is_home():
+def test_find_form_request():
-    assert not is_home("https://m.facebook.com/login/?...")
+    html = """
-    assert is_home("https://m.facebook.com/home.php?refsrc=...")
+    <div>
+    <form action="/checkpoint/?next=https%3A%2F%2Fwww.messenger.com%2F" class="checkpoint" id="u_0_c" method="post" onsubmit="">
+        <input autocomplete="off" name="jazoest" type="hidden" value="some-number" />
+        <input autocomplete="off" name="fb_dtsg" type="hidden" value="some-base64" />
+        <input class="hidden_elem" data-default-submit="true" name="submit[Continue]" type="submit" />
+        <input autocomplete="off" name="nh" type="hidden" value="some-hex" />
+        <div class="_4-u2 _5x_7 _p0k _5x_9 _4-u8">
+            <div class="_2e9n" id="u_0_d">
+                <strong id="u_0_e">Two factor authentication required</strong>
+                <div id="u_0_f"></div>
+            </div>
+            <div class="_2ph_">
+                <input autocomplete="off" name="no_fido" type="hidden" value="true" />
+                <div class="_50f4">You've asked us to require a 6-digit login code when anyone tries to access your account from a new device or browser.</div>
+                <div class="_3-8y _50f4">Enter the 6-digit code from your Code Generator or 3rd party app below.</div>
+                <div class="_2pie _2pio">
+                    <span>
+                        <input aria-label="Login code" autocomplete="off" class="inputtext" id="approvals_code" name="approvals_code" placeholder="Login code" tabindex="1" type="text" />
+                    </span>
+                </div>
+            </div>
+            <div class="_5hzs" id="checkpointBottomBar">
+                <div class="_2s5p">
+                    <button class="_42ft _4jy0 _2kak _4jy4 _4jy1 selected _51sy" id="checkpointSubmitButton" name="submit[Continue]" type="submit" value="Continue">Continue</button>
+                </div>
+                <div class="_2s5q">
+                    <div class="_25b6" id="u_0_g">
+                        <a href="#" id="u_0_h" role="button">Need another way to authenticate?</a>
+                    </div>
+                </div>
+            </div>
+        </div>
+    </form>
+    </div>
+    """
+    url, data = find_form_request(html)
+
+
+def test_find_form_request_error():
+    with pytest.raises(ParseError, match="Could not find form to submit"):
+        assert find_form_request("")
+    with pytest.raises(ParseError, match="Could not find url to submit to"):
+        assert find_form_request("<form></form>")
 
 
-@pytest.mark.skip
 def test_get_error_data():
-    html = """<?xml version="1.0" encoding="utf-8"?>
+    html = """<!DOCTYPE html>
-    <!DOCTYPE html PUBLIC "-//WAPFORUM//DTD XHTML Mobile 1.0//EN" "http://www.wapforum.org/DTD/xhtml-mobile10.dtd">
+    <html lang="da" id="facebook" class="no_js">
-    <html xmlns="http://www.w3.org/1999/xhtml">
 
     <head>
-        <title>Log in to Facebook | Facebook</title>
+        <meta charset="utf-8" />
-        <meta name="referrer" content="origin-when-crossorigin" id="meta_referrer" />
+        <title id="pageTitle">Messenger</title>
-        <style type="text/css">...</style>
+        <meta name="referrer" content="default" id="meta_referrer" />
-        <meta name="description" content="..." />
-        <link rel="canonical" href="https://www.facebook.com/login/" />
     </head>
 
-    <body tabindex="0" class="b c d e f g">
+    <body class="_605a x1 Locale_da_DK" dir="ltr">
-    <div class="h"><div id="viewport">...<div id="objects_container"><div class="g" id="root" role="main">
+    <div class="_3v_o" id="XMessengerDotComLoginViewPlaceholder">
-    <table class="x" role="presentation"><tbody><tr><td class="y">
+    <form id="login_form" action="/login/password/" method="post" onsubmit="">
-    <div class="z ba bb" style="" id="login_error">
+        <input type="hidden" name="jazoest" value="2222" autocomplete="off" />
-        <div class="bc">
+        <input type="hidden" name="lsd" value="xyz-abc" autocomplete="off" />
-            <span>The password you entered is incorrect. <a href="/recover/initiate/?ars=facebook_login_pw_error&amp;email=abc@mail.com&amp;__ccr=XXX" class="bd" aria-label="Have you forgotten your password?">Did you forget your password?</a></span>
+        <div class="_3403 _3404">
+            <div>Type your password again</div>
+            <div>The password you entered is incorrect. <a href="https://www.facebook.com/recover/initiate?ars=facebook_login_pw_error">Did you forget your password?</a></div>
         </div>
+        <div id="loginform">
+            <input type="hidden" autocomplete="off" id="initial_request_id" name="initial_request_id" value="xxx" />
+            <input type="hidden" autocomplete="off" name="timezone" value="" id="u_0_1" />
+            <input type="hidden" autocomplete="off" name="lgndim" value="" id="u_0_2" />
+            <input type="hidden" name="lgnrnd" value="aaa" />
+            <input type="hidden" id="lgnjs" name="lgnjs" value="n" />
+            <input type="text" class="inputtext _55r1 _43di" id="email" name="email" placeholder="E-mail or phone number" value="some@email.com" tabindex="0" aria-label="E-mail or phone number" />
+            <input type="password" class="inputtext _55r1 _43di" name="pass" id="pass" tabindex="0" placeholder="Password" aria-label="Password" />
+            <button value="1" class="_42ft _4jy0 _2m_r _43dh _4jy4 _517h _51sy" id="loginbutton" name="login" tabindex="0" type="submit">Continue</button>
+            <div class="_43dj">
+                <div class="uiInputLabel clearfix">
+                    <label class="uiInputLabelInput">
+                        <input type="checkbox" value="1" name="persistent" tabindex="0" class="" id="u_0_0" />
+                        <span class=""></span>
+                    </label>
+                    <label for="u_0_0" class="uiInputLabelLabel">Stay logged in</label>
+                </div>
+                <input type="hidden" autocomplete="off" id="default_persistent" name="default_persistent" value="0" />
+            </div>
+    </form>
     </div>
-    ...
-    </td></tr></tbody></table>
-    <div style="display:none"></div><span><img src="https://facebook.com/security/hsts-pixel.gif" width="0" height="0" style="display:none" /></span>
-    </div></div><div></div></div></div>
     </body>
 
     </html>
     """
-    url = "https://m.facebook.com/login/?email=abc@mail.com&li=XXX&e=1348092"
     msg = "The password you entered is incorrect. Did you forget your password?"
-    assert (1348092, msg) == get_error_data(html)
+    assert msg == get_error_data(html)