Fix login; Add suppport for extract CAUTH from browser cookie; Add su…

…pport for login by browser

README.md

@@ -14,6 +14,7 @@
     + [Installation (recommended)](#installation-recommended)
     + [Manual Installation](#manual-installation)
     + [Docker container](#docker-container)
+  * [Before the start](#before-the-start)
   * [Quick Start](#quick-start)
     + [Examples](#examples)
   * [Troubleshooting](#troubleshooting)
@@ -55,6 +56,32 @@ docker run --rm -it -v \
     coursera-helper/coursera_helper -u <USER> -p <PASSWORD>
 ```
 
+## Before the start
+
+`coursera-helper` supports four authentication methods:
+
+1. **CAUTH (recommended)**
+
+   Just use the `--cauth 'CAUTH-value-from-browser'` option when running the program.
+
+   [How to get the cauth value?](#CAUTH)
+
+2. Browser cookies
+
+   Just use the `--browser-cookie` option when running the program.
+
+   Automatically extract CAUTH value from the browser cookie. If this method fails, please use other authentication methods.
+
+3. Username and Password
+
+   Just use the `-u <user> -p <pass>` options when running the program.
+
+   Please note that this method will open the browser, you may have to click on the reCAPTCHA.
+
+4. netrc File
+
+   Just use the `--netrc` options when running the program.
+
 ## Quick Start
 
 Run the following command to query the usage and options:
@@ -133,6 +160,42 @@ coursera-helper data-analysis-with-python
 
 ## Troubleshooting
 
+### CAUTH
+
+Find your coursera CAUTH:
+
+* Open and login to https://www.coursera.org/
+* Right-click on the touchpad or mouse until you find *inspect*.
+* Go to Applications > Cookies (and click dropdown) > click https://www.coursera.org/ > find and click CAUTH > Copy value CAUTH.
+
+**Chrome**:
+
+1. Open the browser and login to https://www.coursera.org/
+
+2. Open the last DevTools panel
+
+   Windows or Linux: Press **F12** on the keyboard. Or press the **Ctrl** + **Shift** + **I** keys.
+
+   Mac: Press **Fn** + **F12** on the keyboard. Or press the **Cmd** + **Option** + **I** keys.
+
+3. Open **Application** > **Storage** > **Cookies** and select https://www.coursera.org/.
+
+4. find and click CAUTH > Copy value CAUTH
+
+**Firefox **:
+
+1. Open the browser and login to https://www.coursera.org/
+
+2. Open the last DevTools panel
+
+   Windows or Linux: Press **F12** on the keyboard. Or press the **Ctrl** + **Shift** + **I** keys.
+
+   Mac: Press **Fn** + **F12** on the keyboard. Or press the **Cmd** + **Option** + **I** keys.
+
+3. Open **Storage** > **Cookies** and select https://www.coursera.org/.
+
+4. find and click CAUTH > Copy value CAUTH
+
 ### china-issues
 
 If you are from China and you're having problems downloading videos, adding 

coursera_helper/__init__.py

@@ -1 +1 @@
-__version__ = '0.11.93'
+__version__ = '0.12.1'

coursera_helper/cauth.py

@@ -0,0 +1,66 @@
+import logging
+import sys
+import random
+import platform
+import browser_cookie3
+from playwright.sync_api import sync_playwright
+
+
+def cauth_by_login(username, password, headless=True):
+    with sync_playwright() as playwright:
+        try:
+            if platform.system() == 'Windows':
+                browser = playwright.chromium.launch(channel="msedge", headless=headless)
+            elif platform.system() == 'Linux':
+                browser = playwright.firefox.launch(headless=headless)
+            elif platform.system() == 'Darwin':
+                browser = playwright.webkit.launch(headless=headless)
+            else:
+                browser = playwright.chromium.launch(channel="chrome", headless=headless)
+        except AttributeError:
+            logging.info('Please run "playwright install" in terminal and then try again!')
+            sys.exit()
+
+        context = browser.new_context(
+            user_agent='Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) '
+                       'Chrome/117.0.0.0 Safari/537.36',
+        )
+        context.add_init_script("Object.defineProperty(navigator, 'webdriver', {get: () => undefined})")
+
+        page = context.new_page()
+
+        page.goto("https://www.coursera.org/", timeout=0)
+        page.click('a[href*="/?authMode=login"]')
+
+        email_elem = page.locator('#email')
+        password_elem = page.locator('#password')
+
+        email_elem.press_sequentially(username, delay=random.random() * 1000)
+        password_elem.press_sequentially(password, delay=random.random() * 1000)
+
+        page.locator('button[type*="submit"]').click()
+
+        page.wait_for_selector('p[data-e2e="UserPortraitFullName"]', timeout=0)
+
+        cookies = context.cookies('https://www.coursera.org/')
+        cauth = ''
+        cookie_dict = {cookie['name']: cookie['value'] for cookie in cookies}
+        if 'CAUTH' in cookie_dict:
+            cauth = cookie_dict['CAUTH']
+        browser.close()
+        return cauth
+
+
+def cauth_by_cookie():
+    domain = 'coursera.org'
+    browsers = ['chrome', 'chromium', 'edge', 'firefox', 'safari']
+    for browser in browsers:
+        try:
+            statement = 'browser_cookie3.{}(domain_name="{}")'.format(browser, domain)
+            cookies = eval(statement)
+            cookie_dict = {cookie.name: cookie.value for cookie in cookies}
+            if 'CAUTH' in cookie_dict:
+                cauth = cookie_dict['CAUTH']
+                return cauth
+        except browser_cookie3.BrowserCookieError:
+            pass

coursera_helper/commandline.py

@@ -448,6 +448,22 @@ def parse_args(args=None):
         help='uses or creates local cached version of syllabus'
         ' page')
 
+    parser.add_argument(
+        '--browser-cookie',
+        dest='browser_cookie',
+        action='store_true',
+        default=False,
+        help='Extract CAUTH value from cookies of browser'
+        ' (default: False)')
+
+    parser.add_argument(
+        '--headless',
+        dest='headless',
+        action='store_true',
+        default=False,
+        help='Whether to run browser in headless mode'
+        ' (default: False)')
+
     # Final parsing of the options
     args = parser.parse_args(args)
 
@@ -500,7 +516,7 @@ def parse_args(args=None):
         logging.error('Cookies file not found: %s', args.cookies_file)
         sys.exit(1)
 
-    if not args.cookies_file and not args.cookies_cauth:
+    if not args.cookies_file and not args.cookies_cauth and not args.browser_cookie:
         try:
             args.username, args.password = get_credentials(
                 username=args.username, password=args.password,

coursera_helper/coursera_dl.py

@@ -74,6 +74,7 @@
 from .network import get_page, get_page_and_url
 from .commandline import parse_args
 from .extractors import CourseraExtractor
+from coursera_helper.cauth import cauth_by_login, cauth_by_cookie
 
 from coursera_helper import __version__
 
@@ -97,15 +98,16 @@ def get_session():
     return session
 
 
-def list_courses(args):
+def list_courses(session, args):
     """
     List enrolled courses.
 
+    @param session: session.
+    @type session: session
+
     @param args: Command-line arguments.
     @type args: namedtuple
     """
-    session = get_session()
-    login(session, args.username, args.password)
     extractor = CourseraExtractor(session)
     courses = extractor.list_courses()
     logging.info('Found %d courses', len(courses))
@@ -227,16 +229,23 @@ def main():
     mkdir_p(PATH_CACHE, 0o700)
     if args.clear_cache:
         shutil.rmtree(PATH_CACHE)
-    if args.list_courses:
-        logging.info('Listing enrolled courses')
-        list_courses(args)
-        return
 
     session = get_session()
     if args.cookies_cauth:
         session.cookies.set('CAUTH', args.cookies_cauth)
+    elif args.browser_cookie:
+        cauth = cauth_by_cookie()
+        session.cookies.set('CAUTH', cauth)
     else:
-        login(session, args.username, args.password)
+        cauth = cauth_by_login(args.username, args.password, headless=args.headless)
+        session.cookies.set('CAUTH', cauth)
+        # login(session, args.username, args.password)
+
+    if args.list_courses:
+        logging.info('Listing enrolled courses')
+        list_courses(session, args)
+        return
+
     if args.specialization:
         args.class_names = expand_specializations(session, args.class_names)
 

pyproject.toml

@@ -44,6 +44,8 @@ dependencies = [
     "keyring>=4.0",
     "configargparse>=0.12.0",
     "attrs==18.1.0",
+    "playwright>=1.38.0",
+    "browser_cookie3>=0.19.1",
 ]
 
 [project.optional-dependencies]

requirements.txt

@@ -6,3 +6,5 @@ pyasn1>=0.1.7
 keyring>=4.0
 configargparse>=0.12.0
 attrs==18.1.0
+playwright>=1.38.0
+browser_cookie3>=0.19.1