Add clean_domains

src/vcr_cleaner/cleaners/uri.py

@@ -1,3 +1,7 @@
+import json
+import re
+
+
 def clean_uri(old: str, new: str):
     """Returns a cleaner function that replaces the request URI
     string with all occurrences of substring old replaced by new.
@@ -16,3 +20,26 @@ def clean_uri(request: dict, response: dict):
     clean_uri.__doc__ = f"Replaces the request URI string with all " \
         f"occurrences of substring '{old}' replaced by '{new}'."
     return clean_uri
+
+
+def _regex_sub_dict(message: dict, rule: str, replacement: str):
+    '''Update the dictionary with rule matches replaced.'''
+    cleaned = re.sub(rule, replacement, json.dumps(message))
+
+    # Update the original dict
+    message.clear()
+    message.update(json.loads(cleaned))
+
+
+def clean_domains(domain: str, replacement: str = 'cleaned.example.edu'):
+    '''Replace anything that looks like the given domain and sub-domains.'''
+    rule = f"/[^/]*{domain.replace('.', r'\.')}"
+    rep = f"/{replacement}"
+
+    def wrapper(request: dict, response: dict):
+        _regex_sub_dict(request, rule, rep)
+        _regex_sub_dict(response, rule, rep)
+
+    wrapper.__doc__ = clean_domains.__doc__
+
+    return wrapper

tests/test_uri.py

@@ -1,4 +1,7 @@
-from vcr_cleaner.cleaners.uri import clean_uri
+from vcr_cleaner.cleaners.uri import (
+    clean_domains,
+    clean_uri,
+)
 
 
 def test_simple_clean_uri():
@@ -13,3 +16,27 @@ def test_simple_clean_uri():
     assert str(cleaner.__doc__) != 'None'
     assert 'example' in str(cleaner.__doc__)
     assert 'foo' in str(cleaner.__doc__)
+
+
+def test_clean_domain():
+    request = {
+        'sub-domain': 'https://foo.illinois.edu',
+        'sub-sub-domain': 'https://foo.bar.illinois.edu',
+        'uri': 'https://illinois.edu',
+        'insecure': 'http://illinois.edu',
+        'essay': 'Lorum ipsum https://illinois.edu, and so on...',
+    }
+    response = request.copy()
+
+    cleaner = clean_domains('illinois.edu')
+    cleaner(request, response)
+
+    assert str(cleaner.__doc__) != 'None'
+
+    for key in ['uri', 'sub-domain', 'sub-sub-domain']:
+        assert request[key] == 'https://cleaned.example.edu'
+
+    assert request['insecure'] == 'http://cleaned.example.edu'
+    assert 'illinois.edu' not in request['essay']
+
+    assert request == response