fix: fix data source to get data from biography

data-sources/instagram.py

@@ -1,8 +1,8 @@
 #!/usr/bin/env python3
 import json
 import sys
-import urllib.parse
 import requests
+import re
 from typing import Optional
 import cryptography.hazmat.primitives.asymmetric.utils as crypto
 from cryptography.hazmat.primitives.asymmetric import ec
@@ -34,15 +34,26 @@ def __init__(self, address: str, pub_key: str, value: str, signature: str):
         self.value = value
 
 
-def get_user_data(data: CallData) -> Optional[VerificationData]:
+def get_urls_from_biography(user: str) -> [str]:
     """
-    Tries getting the verification data for the user having the given Instagram username.
-    :param data: Data used to get the VerificationData
-    :return: An OptionalData object if the call was successful, or None if it errored somehow.
+    Returns all the URLs that are found inside the biography of the user having the given user username.
+    :param user: Username of the Instagram user.
+    :return: List of URLs that are found inside the biography
+    """
+    url = f"{ENDPOINT}/users/{user}"
+    result = requests.request("GET", url, headers=HEADERS).json()
+    return re.findall(r'(https?://[^\s]+)', result['biography'])
+
+
+def get_signature_from_url(url: str) -> Optional[VerificationData]:
+    """
+    Tries getting the signature object linked to the given URL.
+    :param url: URL that should contain the signature object.
+    :return: A dictionary containing 'valid' to tell whether the search was valid, and an optional 'data' containing
+    the signature object.
     """
     try:
-        url_encoded_username = urllib.parse.quote(data.username)
-        result = requests.request("GET", f"{ENDPOINT}/{url_encoded_username}", headers=HEADERS).json()
+        result = requests.request("GET", url, headers=HEADERS).json()
         if validate_json(result):
             return VerificationData(
                 result['address'],
@@ -142,12 +153,24 @@ def main(args: str):
     json_obj = json.loads(decoded)
     call_data = check_values(json_obj)
 
-    result = get_user_data(call_data)
-    if result is None:
-        raise Exception(f"No valid signature data found for user with username {call_data.username}")
-
+    # Get the URLs to check from the user biography
+    urls = get_urls_from_biography(call_data.username)
+    if len(urls) == 0:
+        raise Exception(f"No URL found inside {call_data.username} biography")
+
+    # Find the signature following the URLs
+    data = None
+    for url in urls:
+        result = get_signature_from_url(url)
+        if result is not None:
+            data = result
+            break
+
+    if data is None:
+        raise Exception(f"No valid signature data found inside {call_data.username} biography")
+
     # Verify the signature
-    signature_valid = verify_signature(result)
+    signature_valid = verify_signature(data)
     if not signature_valid:
         raise Exception("Invalid signature")
 

data-sources/instagram_test.py

@@ -1,31 +1,49 @@
 import unittest
-import instagram
+
 import httpretty
 
+import instagram
 
-class instagramTest(unittest.TestCase):
+
+class TestInstagram(unittest.TestCase):
 
     @httpretty.activate(verbose=True, allow_net_connect=False)
-    def test_get_user_data(self):
+    def test_get_urls_from_biography(self):
         # Register fake HTTP call
         httpretty.register_uri(
             httpretty.GET,
-            "https://themis.mainnet.desmos.network/instagram/test_user_desmos",
+            "https://themis.mainnet.desmos.network/instagram/users/riccardomontagnin",
             status=200,
-            body='{"address":"71b0310267b49279116835ed35791c24c110012f","pub_key":"0203233fabd69a1b7a90bb968a0ab66e3af61989f65cf0bc1f8e9518740a302f1f","value":"746573745f757365725f6465736d6f73","signature":"c12605456b8652df655bb43d0166586dfc0c5d758b03f127ca6b027d0ec140ca29b9569a20c9b78b72e13d15c1a7fa0b142dc0e624f3f51ef76bd94e55345d2a"}',
+            body='{"biography":"https://pastebin.com/raw/D95NyR91"}',
+        )
+
+        url = instagram.get_urls_from_biography('riccardomontagnin')
+        self.assertEqual(['https://pastebin.com/raw/TgSpUCz6'], url)
+
+    @httpretty.activate(verbose=True, allow_net_connect=False)
+    def test_get_signature_from_url(self):
+        # Register fake HTTP call
+        httpretty.register_uri(
+            httpretty.GET,
+            "https://pastebin.com/raw/xz4S8WrW",
+            status=200,
+            body='{"address":"desmos13yp2fq3tslq6mmtq4628q38xzj75ethzela9uu","pub_key":"033024e9e0ad4f93045ef5a60bb92171e6418cd13b082e7a7bc3ed05312a0b417d","signature":"a00a7d5bd45e42615645fcaeb4d800af22704e54937ab235e5e50bebd38e88b765fdb696c22712c0cab1176756b6346cbc11481c544d1f7828cb233620c06173","value":"ricmontagnin"}',
         )
 
         # Valid signature
-        data = instagram.get_user_data(instagram.CallData('test_user_desmos'))
+        data = instagram.get_signature_from_url('https://pastebin.com/raw/xz4S8WrW')
         self.assertIsNotNone(data)
 
-        # Invalid signature
+        # Register fake HTTP call
         httpretty.register_uri(
             httpretty.GET,
-            "https://themis.mainnet.desmos.network/instagram/test_user_desmos",
-            status=404,
+            "https://bitcoin.org",
+            status=200,
+            body='Bitcoin website',
         )
-        data = instagram.get_user_data(instagram.CallData('test_user_desmos'))
+
+        # Invalid signature
+        data = instagram.get_signature_from_url('https://bitcoin.org')
         self.assertIsNone(data)
 
     def test_validate_json(self):