From a6721b316fca0f7a460fa2e62ed679238f97a10e Mon Sep 17 00:00:00 2001
From: Wojtek Kosior <koszko@koszko.org>
Date: Wed, 17 Nov 2021 21:12:22 +0100
Subject: enable querying of patterns by url

---
 src/pydrilla/pydrilla.py  | 37 ++++++++++++++++++++++++++-----------
 src/test/test_pydrilla.py | 11 +++++++++++
 2 files changed, 37 insertions(+), 11 deletions(-)

diff --git a/src/pydrilla/pydrilla.py b/src/pydrilla/pydrilla.py
index dc472a4..af4ff1c 100644
--- a/src/pydrilla/pydrilla.py
+++ b/src/pydrilla/pydrilla.py
@@ -511,27 +511,36 @@ class Content:
                         ver=version_string(mapping['version'])
                     )
 
-    def query(self, url, max=0):
+    def query(self, url):
         '''
-        Return return registered patterns and mappings (available as
-        MappingItems) that match url. The maximum number of items yielded may be
-        limited by using the optional max argument. Its default value, 0, causes
-        no limit to be imposed.
+        Return a list of registered mappings that match url.
 
         If multiple versions of a mapping are applicable, only the most recent
         is included in the result.
         '''
         deco = DeconstructedUrl(url)
 
+        mappings = {}
+
         domain_tree = self.patterns_by_proto.get(deco.proto) \
             or PatternTreeNode()
+
+        def process_item(item):
+            if url[-1] != '/' and item.pattern[-1] == '/':
+                return
+
+            identifier = item.mapping['identifier']
+
+            if identifier not in mappings or \
+               item.mapping['version'] > mappings[identifier]['version']:
+                mappings[identifier] = item.mapping
+
         for path_tree in domain_tree.search(deco.domain):
-            for item in path_tree.search(deco.path):
-                if url[-1] == '/' or item.pattern[-1] != '/':
-                    yield item
-                    max -= 1
-                    if max == 0:
-                        return
+            for item_list in path_tree.search(deco.path):
+                for item in item_list:
+                    process_item(item)
+
+        return list(mappings.values())
 
 def load_content_from_subdir(subdir_path, source_name, content):
     index_path = subdir_path / 'index.json'
@@ -683,3 +692,9 @@ for item_type, get_dict, get_item in [
         return get_item(identifier, get_dict)
 
     bp.add_url_rule(f'/{item_type}s/<string:identifier>', item_type, _get_item)
+
+@bp.route('/query')
+def query():
+    url = request.args['url']
+
+    return json.dumps(content().query(url))
diff --git a/src/test/test_pydrilla.py b/src/test/test_pydrilla.py
index 660c8f3..db9a321 100644
--- a/src/test/test_pydrilla.py
+++ b/src/test/test_pydrilla.py
@@ -99,6 +99,17 @@ def test_api_basic(client, development_config):
     response = client.get('/sources/random-bad-identifier')
     assert response.status_code == 404
 
+    response = client.get('/query?url=https://hachettebugs.koszko.org')
+    definitions = json.loads(response.data.decode())
+    assert type(definitions) is list
+    assert all([d['type'] == 'mapping' for d in definitions])
+    assert any([p['pattern'] == 'https://hachettebugs.koszko.org/***'
+                for d in definitions for p in d['payloads']])
+
+    response = client.get('/query?url=https://random_bad_domain.org/something')
+    definitions = json.loads(response.data.decode())
+    assert definitions == []
+
 def test_normalize_version():
     assert pydrilla.normalize_version([4, 5, 3, 0, 0]) == [4, 5, 3]
     assert pydrilla.normalize_version([1, 0, 5, 0])    == [1, 0, 5]
-- 
cgit v1.2.3