Merge pull request #1 from acdh-oeaw/dev

new function added

.gitignore

@@ -3,3 +3,6 @@
 /dist
 /composer.lock
 /vendor
+env
+.tox
+*.pyc

AcdhUriNormRules/AcdhUriNormRules.py

@@ -1,9 +1,25 @@
 import json
 import pkg_resources
+import re
 
+def get_rules():
+  """ returns a list of regex pattern mateches
 
-def getRules():
-  #with open(pkg_resources.resource_string(__name__, 'rules.json'), 'r') as f:
+  :return: a list of regex patterns to normalize Authority IDs
+  :rtype: list
+  
+  """
   return json.loads(pkg_resources.resource_string(__name__, 'rules.json'))
-  #return rules
 
+
+def get_normalized_uri(uri):
+  """ takes a normdata uri and returns a normlalized version
+  :param uri: A normdata uri
+  :param type: str
+  
+  :return: The normalized URI
+  :rtype: str
+  """
+  for x in get_rules():
+    uri = re.sub(x['match'], x['replace'], uri)
+  return uri

README.md

@@ -20,8 +20,14 @@ Rules are stored as a JSON in the `UriNormRules/rules.json` file.
   ```
 * Use with
   ```Python
-  from AcdhUriNormRules import *
-  print(AcdhUriNormRules.getRules())
+  from AcdhUriNormRules import get_rules, get_normalized_uri
+  print(AcdhUriNormRules.get_rules())
+
+  wrong_id = "http://sws.geonames.org/1232324343/linz.html"
+
+  good_id = get_normalized_uri(wrong_id)
+  print(good_id)
+  # "https://www.geonames.org/1232324343"
   ```
 
 ## PHP

tests/test_normalizer.py

@@ -0,0 +1,27 @@
+import unittest
+from AcdhUriNormRules.AcdhUriNormRules import get_rules, get_normalized_uri
+
+SAMPLES = [
+    [
+        "http://sws.geonames.org/1232324343/linz.html",
+        "https://www.geonames.org/1232324343"
+    ],
+    [
+        "http://d-nb.info/gnd/4074255-6/",
+        "https://d-nb.info/gnd/4074255-6"
+    ],
+    [
+        "https://d-nb.info/gnd/4074255-6",
+        "https://d-nb.info/gnd/4074255-6"
+    ]
+]
+
+class TestNormalizer(unittest.TestCase):
+    def test__001_load_list(self):
+        rules = get_rules()
+        self.assertEqual(type(rules), list, "should be type 'list' ")
+
+    def test__002_test_patterns(self):
+        for x in SAMPLES:
+            new_uri = get_normalized_uri(x[0])
+            self.assertEqual(x[1], new_uri)

tox.ini

@@ -0,0 +1,8 @@
+[tox]
+envlist = py35, py36, py37, py38
+
+[testenv]
+setenv =
+    PYTHONPATH = {toxinidir}
+
+commands = python setup.py test