Internalize settings in validate function

tests/tests.py

@@ -49,8 +49,7 @@ def validate_csv_str(**options):
 def test_valid_delimiter():
     source = """A,B,C
 a,b,c"""
-    inspector = Inspector()
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 0
     assert len(report['tables'][0]['errors']) == 0
 
@@ -58,8 +57,7 @@ a,b,c"""
 def test_invalid_delimiter_semicolon():
     source = """A;B;C
 a;b;c"""
-    inspector = Inspector()
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 1
     assert len(report['tables'][0]['errors']) == 1
     assert report['tables'][0]['errors'][0]['code'] == 'invalid-column-delimiter'
@@ -70,8 +68,7 @@ a;b;c"""
 def test_invalid_delimiter_percent():
     source = """A%B%C
 a%b%c"""
-    inspector = Inspector()
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 1
     assert len(report['tables'][0]['errors']) == 1
     assert report['tables'][0]['errors'][0]['code'] == 'invalid-column-delimiter'
@@ -82,9 +79,7 @@ a%b%c"""
 def test_missing_header_start():
     source = """B,C
 b,c"""
-    inspector = Inspector(checks=['structure', 'schema', {'extra-or-missing-header': {}}],
-                          skip_checks=['non-matching-header', 'missing-header'])
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 1
     assert len(report['tables'][0]['errors']) == 1
     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
@@ -94,9 +89,7 @@ b,c"""
 def test_missing_header_middle():
     source = """A,C
 a,c"""
-    inspector = Inspector(checks=['structure', 'schema', {'extra-or-missing-header': {}}],
-                          skip_checks=['non-matching-header', 'missing-header'])
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 1
     assert len(report['tables'][0]['errors']) == 1
     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
@@ -106,9 +99,7 @@ a,c"""
 def test_missing_header_end():
     source = """A,B
 a,b"""
-    inspector = Inspector(checks=['structure', 'schema', {'extra-or-missing-header': {}}],
-                          skip_checks=['non-matching-header', 'missing-header'])
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 1
     assert len(report['tables'][0]['errors']) == 1
     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
@@ -118,9 +109,7 @@ a,b"""
 def test_missing_and_extra_header_end():
     source = """A,B,Z
 a,b,z"""
-    inspector = Inspector(checks=['structure', 'schema', {'extra-or-missing-header': {}}],
-                          skip_checks=['non-matching-header', 'missing-header'])
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 2
     assert len(report['tables'][0]['errors']) == 2
     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
@@ -132,9 +121,7 @@ a,b,z"""
 def test_missing_and_extra_header_middle():
     source = """A,Z,B
 a,z,b"""
-    inspector = Inspector(checks=['structure', 'schema', {'extra-or-missing-header': {}}],
-                          skip_checks=['non-matching-header', 'missing-header'])
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 2
     assert len(report['tables'][0]['errors']) == 2
     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
@@ -146,9 +133,7 @@ a,z,b"""
 def test_missing_and_extra_header_multiple():
     source = """A,Z
 a,z"""
-    inspector = Inspector(checks=['structure', 'schema', {'extra-or-missing-header': {}}],
-                          skip_checks=['non-matching-header', 'missing-header'])
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 2
     assert len(report['tables'][0]['errors']) == 2
     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
@@ -160,9 +145,7 @@ a,z"""
 def test_extra_header():
     source = """A,B,C,D
 a,b,c,d"""
-    inspector = Inspector(checks=['structure', 'schema', {'extra-or-missing-header': {}}],
-                          skip_checks=['non-matching-header', 'missing-header'])
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 1
     assert len(report['tables'][0]['errors']) == 1
     assert report['tables'][0]['errors'][0]['code'] == 'extra-header'
@@ -172,9 +155,7 @@ a,b,c,d"""
 def test_missing_and_extra_headers_multiple():
     source = """A,Z,D
 a,z,d"""
-    inspector = Inspector(checks=['structure', 'schema', {'extra-or-missing-header': {}}],
-                          skip_checks=['non-matching-header', 'missing-header'])
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 3
     assert len(report['tables'][0]['errors']) == 3
     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
@@ -188,9 +169,7 @@ a,z,d"""
 def test_header_order():
     source = """A,C,B
 a,c,b"""
-    inspector = Inspector(checks=['structure', 'schema', {'extra-or-missing-header': {}}],
-                          skip_checks=['non-matching-header', 'missing-header'])
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 1
     assert len(report['tables'][0]['errors']) == 1
     assert report['tables'][0]['errors'][0]['code'] == 'wrong-headers-order'
@@ -199,9 +178,7 @@ a,c,b"""
 def test_missing_and_duplicate_headers():
     source = """A,A,B
 a,a,b"""
-    inspector = Inspector(checks=['structure', 'schema', {'extra-or-missing-header': {}}],
-                          skip_checks=['non-matching-header', 'missing-header'])
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 2
     assert len(report['tables'][0]['errors']) == 2
     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
@@ -214,9 +191,7 @@ a,a,b"""
 def test_invalid_delimiter_and_missing_header():
     source = """A;C
 a;c"""
-    inspector = Inspector(checks=['structure', 'schema', {'extra-or-missing-header': {}}],
-                          skip_checks=['non-matching-header', 'missing-header'])
-    report = validate_csv_str(source=source, schema=schema1, inspector=inspector, pre_checks_conf=pre_checks_conf)
+    report = validate_csv_str(source=source, schema=schema1, pre_checks_conf=pre_checks_conf)
     assert report['tables'][0]['error-count'] == 2
     assert len(report['tables'][0]['errors']) == 2
     assert report['tables'][0]['errors'][0]['code'] == 'invalid-column-delimiter'

validata_validate/__init__.py

@@ -4,10 +4,15 @@ from pathlib import Path
 
 from toolz import thread_first, update_in
 
+import goodtables
+
 from . import pre_checks
 
 log = logging.getLogger(__name__)
 
+VALIDATA_MAX_ROWS = 100000
+
+
 # Import all custom checks classes
 cs_dir = Path(__file__).parent / 'custom_checks'
 for check_file in cs_dir.glob('*.py'):
@@ -36,10 +41,18 @@ def add_error(report, table_index, error):
         ))))
 
 
-def validate(source, schema, inspector, pre_checks_conf=None, **options):
-    """Validate a `source` with a given `inspector`, applying pre-checks from `pre_checks_conf` if given."""
+def validate(source, schema, checks=None, pre_checks_conf=None, **options):
+    """Validate a `source` applying pre-checks and checks."""
     report = None
 
+    inspector = goodtables.Inspector(
+        checks=(['structure', 'schema'] if checks is None else checks) + [{'extra-or-missing-header': {}}],
+        skip_checks=['non-matching-header', 'missing-header'],
+        row_limit=VALIDATA_MAX_ROWS)
+
+    # Validata design choices
+    options = {**options, "force_strings": True}
+
     for pre_check_conf in pre_checks_conf or []:
         pre_check_name = pre_check_conf['name']
         pre_check_options = {**options, **pre_check_conf.get('params', {})}

validate.py

@@ -10,11 +10,8 @@ import logging
 import sys
 from pathlib import Path
 
-from goodtables import Inspector
 from validata_validate import validate
 
-VALIDATA_MAX_ROWS = 100000
-
 
 def main():
     parser = argparse.ArgumentParser(description=__doc__,
@@ -53,12 +50,10 @@ def main():
                 for custom_check_conf in custom_checks_conf:
                     checks.append({custom_check_conf['name']: custom_check_conf['params']})
 
-    inspector = Inspector(checks=checks, row_limit=VALIDATA_MAX_ROWS, order_fields=True)
-
     report = validate(
         source=args.source,
+        checks=checks,
         pre_checks_conf=pre_checks_conf,
-        inspector=inspector,
         schema=args.schema,
     )