Update schema generation and bump version for v6 deploy

Co-authored-by: Bryant Gray <bryant.gray@qlik.com>
Co-authored-by: Andres Pineda <andres.pineda@qlik.com>

Author: 3 people

CHANGELOG.md

@@ -1,5 +1,11 @@
 # Changelog
 
+## 6.2.2
+  * Updates json schema generation to not emit dates
+  * Handle multiple schemas with anyOf and emit them in a specific order
+  * Do not emit error messages when checking multiple schemas and a subsequent schema passes
+  * [#179](https://github.com/singer-io/singer-python/pull/179)
+
 ## 6.2.1
   * Fixes json schema generation to not treat numbers as dates
   * Fixes json schema generation to handle empty arrays

setup.py

@@ -4,7 +4,7 @@
 import subprocess
 
 setup(name="singer-python",
-      version='6.2.1',
+      version='6.2.2',
       description="Singer.io utility library",
       author="Stitch",
       classifiers=['Programming Language :: Python :: 3 :: Only'],

singer/schema_generation.py

@@ -37,13 +37,6 @@ def add_observations(acc, path, data):
             return acc
         except (ValueError, TypeError):
             pass
-        try:
-            # If the string parses as a date, add an observation that it's a date
-            dateutil.parser.parse(data)
-            add_observation(acc, path + ["date"])
-            return acc
-        except (dateutil.parser.ParserError, OverflowError):
-            pass
         add_observation(acc, path + ["string"])
     elif isinstance(data, bool):
         add_observation(acc, path + ["boolean"])
@@ -60,7 +53,10 @@ def add_observations(acc, path, data):
 
 def to_json_schema(obs):
     types = []
-    for key in obs:
+    # add schema types in a specific order to anyOf list
+    for key in ['array', 'object', 'number', 'integer', 'boolean', 'string', 'null']:
+        if key not in obs:
+            continue
 
         result = {'type': ['null']}
 
@@ -75,10 +71,6 @@ def to_json_schema(obs):
             result['type'] += ['array']
             result['items'] = to_json_schema(obs['array'])
 
-        elif key == 'date':
-            result['type'] += ['string']
-            result['format'] = 'date-time'
-
         elif key == 'string':
             result['type'] += ['string']
 

tests/test_schema_generation.py

@@ -10,7 +10,7 @@ def test_simple_schema(self):
                 'a': {'type': ['null', 'integer']},
                 'b': {'type': ['null', 'string']},
                 'c': {'type': ['null', 'boolean']},
-                'dt': {'type': ['null', 'string'], 'format': 'date-time'}
+                'dt': {'type': ['null', 'string']}
             }
         }
         self.assertEqual(expected_schema, generate_schema(records))
@@ -24,11 +24,11 @@ def test_mix_n_match_records_schema(self):
         expected_schema = {
             'type': ['null', 'object'],
             'properties': {'a': {'anyOf': [{'type': ['null', 'integer']},
-                                           {'type': ['null', 'string']},
-                                           {'type': ['null', 'boolean']}]},
+                                           {'type': ['null', 'boolean']},
+                                           {'type': ['null', 'string']}]},
                            'b': {'type': ['null', 'string']},
-                           'c': {'anyOf': [{'type': ['null', 'integer']},
-                                           {'type': ['null', 'string'], 'format': 'singer.decimal'}]},
+                           'c': {'anyOf': [{'type': ['null', 'string'], 'format': 'singer.decimal'},
+                                           {'type': ['null', 'integer']}]},
                 'd': {'anyOf': [{'type': ['null', 'array'],
                                  'items': {'anyOf': [{'type': ['null', 'integer']},
                                                      {'type': ['null', 'string']}]}},