improve captcha solving mechanisms (#19)

* add solver callback mechanism
* add PIL viewer for displaying captcha in CLI
* remove rsbbs captcha viewer
* bump version and fix dependencies

---------

Co-authored-by: Pairman Guo <[email protected]>

libxduauth/sites/ids.py

@@ -1,21 +1,17 @@
 import time
-from io import BytesIO
 
-from PIL import Image
 from bs4 import BeautifulSoup
 
 from ..AuthSession import AuthSession
 from ..utils.page import parse_form_hidden_inputs
 from ..utils.aes import encrypt
+from ..utils.vcode import get_solver
 
 
 class IDSSession(AuthSession):
     cookie_name = 'ids'
 
-    def __init__(
-            self, target, username, password,
-            *args, **kwargs
-    ):
+    def __init__(self, target, username, password):
         super().__init__(f'{self.cookie_name}_{username}')
         if self.is_logged_in():
             return
@@ -25,34 +21,23 @@ def __init__(
             'http://ids.xidian.edu.cn/authserver/login',
             params={'service': target}
         ).text
-        is_need_captcha = self.get(
+        while self.get(
             'https://ids.xidian.edu.cn/authserver/checkNeedCaptcha.htl',
             params={'username': username, '_': str(int(time.time() * 1000))}
-        ).json()['isNeed']
-        if is_need_captcha:
-            captcha = self.get(
-                'https://ids.xidian.edu.cn/authserver/common/openSliderCaptcha.htl',
-                params={'_': str(int(time.time() * 1000))}
-            )
-            # 返回: {
-            #     'bigImage': ..., # 背景图(base64)
-            #     'smallImage': ..., # 滑块图(base64)
-            #     'tagWidth": 93, # 无用, 恒93
-            #     'yHeight': 0 # 无用, 恒0
-            # }
-            img = Image.open(BytesIO(captcha.json()['bigImage']))
-            img.show()
-            # move_len: 背景图左侧到滑块目标位置左侧的宽度
-            move_len = input('滑块位移:')
-            # canvasLength: canvas宽度, 硬编码280
-            # moveLength: 按比例缩放后的滑块位移, 有容错
-            verify = self.post(
+        ).json()['isNeed']:
+            if self.post(
                 'https://ids.xidian.edu.cn/authserver/common/verifySliderCaptcha.htl',
                 data={
-                      'canvasLength': '280',
-                      'moveLength': str(move_len * 280 // img.width)
+                    # canvasLength: canvas宽度, 硬编码280
+                    'canvasLength': '280',
+                    # moveLength: 按比例缩放后的滑块位移, 有容错
+                    'moveLength': str(get_solver('ids.xidian.edu.cn')(self.get(
+                        'https://ids.xidian.edu.cn/authserver/common/openSliderCaptcha.htl',
+                        params={'_': str(int(time.time() * 1000))}
+                    ).json()))
                 }
-            )
+            ).json()['errorMsg'] == 'success':
+                break
             # 返回: {
             #     'errorCode': ..., # 验证通过时为1
             #     'errorMsg': ... # 验证通过时为'success'

libxduauth/sites/rsbbs.py

@@ -5,7 +5,7 @@
 
 from ..AuthSession import AuthSession
 from ..utils.page import parse_form_hidden_inputs
-from ..utils.vcode import _process_vcode
+from ..utils.vcode import get_solver
 
 
 class RSBBSSession(AuthSession):
@@ -26,13 +26,9 @@ def login(self, username, password):
         soup = BeautifulSoup(login.text, 'lxml')
 
         img = soup.find('img', {'class': 'seccodeimg'}).get('src')
-        img = _process_vcode(Image.open(
-            BytesIO(self.get(f'http://{self.HOST}/{img}', headers={
-                'Referer': login.url
-            }).content)
-        ))
-        img.show()
-        vcode = input('验证码：')
+        vcode = get_solver('rsbbs.xidian.edu.cn')(self.get(f'http://{self.HOST}/{img}', headers={
+            'Referer': login.url
+        }).content)
         page = self.post(
             f'http://{self.HOST}/' +
             soup.find('form', id='loginform').get('action'), data=dict(

libxduauth/sites/xk.py

@@ -1,13 +1,12 @@
 import json
 from base64 import b64encode, b64decode
-from io import BytesIO
 from re import search
 
 import requests
 from Crypto.Cipher import AES
 from Crypto.Util.Padding import pad
-from PIL import Image
 
+from ..utils.vcode import get_solver
 from ..AuthSession import AuthSession
 
 
@@ -70,10 +69,7 @@ def login(self, username, password):
 
         # show captcha
         captcha_img = captcha['captcha'][22:]  # data:image/png;base64,
-        Image.open(BytesIO(b64decode(captcha_img))).show()
-
-        # TODO:input function shouldn't exist here
-        captcha_code = input('验证码：')
+        captcha_code = get_solver('xk.xidian.edu.cn')(b64decode(captcha_img))
 
         login_resp = self.post(f'{self.BASE}/auth/login', data={
             'loginname': username,

libxduauth/utils/cli_viewer.py

@@ -0,0 +1,28 @@
+from PIL.ImageShow import Viewer, register as register_pil_imshow
+from PIL import Image
+
+
+def _image_to_ascii(img, size=(80, 16), invert_pallete=True):
+    # convert image to ascii art for cli view
+    # given char dimension 24*16, optimal sizes are
+    # (80, 16) for xk, (80, 21) for rsbbs
+    if invert_pallete:
+        chs = '@&%QWNM0gB$#DR8mHXKAUbGOpV4d9h6Pkqwaxoenut1ivsz/*cr!+<>;=^:\'-.` '
+    else:
+        chs = ' `.-\':^=;><+!rc*/zsvi1tuneoxawqkP6h9d4VpOGbUAKXHm8RD#$Bg0MNWQ%&@'
+    w, h = size
+    img.load()
+    im = img.im.convert('L').resize((w, h))
+    return '\n'.join(''.join(
+        chs[im[i] // 4] for i in range(y, y + w)
+    ) for y in range(0, w * h, w))
+
+
+class _CliViewer(Viewer):
+    def show(self, image: Image.Image, **options: Image.Any) -> int:
+        print(_image_to_ascii(image))
+        # always attempt other viewers
+        return False
+
+def register():
+    register_pil_imshow(_CliViewer, 0)

libxduauth/utils/vcode.py

@@ -1,35 +1,32 @@
-class Processor:
-    def __init__(self, img):
-        self.img = img.convert('L')
-        self.img_arr = self.img.load()
-        self.paint()
-
-    DX = [1, 0, -1, 0]
-    DY = [0, 1, 0, -1]
-
-    def paint(self):
-        w, h = self.img.size
-        visited = set()
-        q = []
-        q.append((0, 0, 255))
-        while q:
-            x, y, value = q.pop()
-            if x < 0 or y < 0 or x >= w or y >= h or \
-                    (x, y) in visited:
-                continue
-            visited.add((x, y))
-            for i in range(4):
-                try:
-                    pixel = self.img_arr[x + self.DX[i], y + self.DY[i]]
-                except IndexError:
-                    continue
-                if abs(pixel - self.img_arr[x, y]) > 5:
-                    q.append((x + self.DX[i], y + self.DY[i], 255 - value))
-                else:
-                    q.append((x + self.DX[i], y + self.DY[i], value))
-            self.img_arr[x, y] = value
-
-
-def _process_vcode(img):
-    p = Processor(img)
-    return p.img
+from base64 import b64decode
+from typing import Callable
+from io import BytesIO
+from PIL import Image
+
+
+def _default_solver(data):
+    Image.open(BytesIO(data)).show()
+    return input('验证码: ')
+
+
+def _ids_solver(data):
+    # data is {
+    #     'bigImage': ..., # 背景图(base64)
+    #     'smallImage': ..., # 滑块图(base64)
+    #     'tagWidth": 93, # 无用, 恒93
+    #     'yHeight': 0 # 无用, 恒0
+    # }
+    img = Image.open(BytesIO(b64decode(data['bigImage'])))
+    img.show()
+
+    # 输入背景图左侧到滑块目标位置左侧的宽度
+    return int(input('滑块位移: ')) * 280 // img.width
+
+
+_solvers = {
+    'ids.xidian.edu.cn': _ids_solver,
+}
+
+
+def get_solver(key) -> Callable:
+    return _solvers.get(key, _default_solver)

setup.py

@@ -5,7 +5,7 @@
 
 setuptools.setup(
     name="libxduauth",
-    version="1.8.2",
+    version="1.9.0",
     author="Frank",
     author_email="[email protected]",
     description="login utilities for XDU",
@@ -19,7 +19,7 @@
     ],
     python_requires='>=3.7',
     install_requires=[
-        "requests", "bs4", "pycryptodome",
+        "requests", "bs4", "pycryptodome", "lxml",
         "importlib-resources", "Pillow",
     ],
 )