[feat] utils.py: add altcha captcha solver

searx/utils.py

@@ -1,6 +1,7 @@
 # SPDX-License-Identifier: AGPL-3.0-or-later
 """Utility functions for the engines"""
 
+from hashlib import pbkdf2_hmac
 import time
 
 import re
@@ -811,3 +812,34 @@ def format_duration(duration: str | int) -> str:
     if length.tm_hour:
         return time.strftime("%H:%M:%S", length)
     return time.strftime("%M:%S", length)
+
+
+def _array_startswith(arr: bytes, prefix: bytes) -> bool:
+    return arr[: len(prefix)] == prefix
+
+
+def solve_altcha(parameters: dict[str, t.Any], maxCounter: int = 1000) -> tuple[str, int] | None:
+    """Solves Altcha CAPTCHAs. It derives keys using PBKDF2 until the derived
+    key starts with the ``keyPrefix``` from the challenge. If the solver does not
+    finish after ``maxCounter`` iterations, returns ``None``."""
+
+    nonce = bytes.fromhex(parameters["nonce"])
+    salt = bytes.fromhex(parameters["salt"])
+    keyPrefix = bytes.fromhex(parameters["keyPrefix"])
+    cost = parameters["cost"]
+    keyLength = parameters["keyLength"]
+    # e.g. "PBKDF2/SHA-256" -> "sha256"
+    hashAlgorithm = parameters["algorithm"].split("/")[-1].replace("-", "").lower()
+
+    counter = 0
+    while counter < maxCounter:
+        # Appends the counter to the nonce, i.e. writes the counter as a big-endian 32-bit integer.
+        secret = nonce + counter.to_bytes(length=4)
+
+        key = pbkdf2_hmac(hash_name=hashAlgorithm, password=secret, salt=salt, iterations=cost, dklen=keyLength)
+        if _array_startswith(key, keyPrefix):
+            return key.hex(), counter
+
+        counter += 1
+
+    return None

tests/unit/test_utils.py

@@ -17,7 +17,6 @@ def random_string(length, choices=string.ascii_letters):
 
 
 class TestUtils(SearxTestCase):
-
     def test_gen_useragent(self):
         self.assertIsInstance(utils.gen_useragent(), str)
         self.assertIsNotNone(utils.gen_useragent())
@@ -116,7 +115,6 @@ class TestUtils(SearxTestCase):
 
 
 class TestXPathUtils(SearxTestCase):  # pylint: disable=missing-class-docstring
-
     TEST_DOC = """<ul>
         <li>Text in <b>bold</b> and <i>italic</i> </li>
         <li>Another <b>text</b> <img src="data:image/gif;base64,R0lGODlhAQABAIAAAAUEBAAAACwAAAAAAQABAAACAkQBADs="></li>
@@ -194,3 +192,25 @@ class TestXPathUtils(SearxTestCase):  # pylint: disable=missing-class-docstring
         with self.assertRaises(SearxEngineXPathException) as context:
             utils.eval_xpath_getindex(doc, 'count(//i)', 1)
         self.assertEqual(context.exception.message, 'the result is not a list')
+
+    def test_altcha_solver(self):
+        # copied from a real Mojeek challenge
+        challenge = {
+            "parameters": {
+                "algorithm": "PBKDF2/SHA-256",
+                "cost": 8000,
+                "keyLength": 32,
+                "keyPrefix": "71087e0b10d819181fc5463588a230b5",
+                "nonce": "1449f02d6089fc29dac33c44aa0630cd",
+                "salt": "860b25b9ee8c123e9ad20f24d9752c9e",
+                "keySignature": "9945d6f881232806edf82a953e1742843b30cf9fe9d56054e1e17a9f47435aa6",
+                "expiresAt": 1789121497,
+            },
+            "signature": "aecf8725f019643766920b67c7e21bc383d8ff166a8e1137c6b5f4d6d19d00c1",
+        }
+        solution = utils.solve_altcha(challenge["parameters"])  # pyright: ignore[reportArgumentType]
+        self.assertIsNotNone(solution)
+
+        key, counter = solution  # pyright: ignore[reportGeneralTypeIssues]
+        self.assertEqual(key, "71087e0b10d819181fc5463588a230b5744e6b9f19b86667c71ff73a9c7567aa")
+        self.assertEqual(counter, 260)