Improve errors handling (wip)

tests/test_core.py

@@ -388,57 +388,56 @@ def test_missing_header_start(schema_abc):
     source = """B,C
 b,c"""
     report = validate_csv_text(source=source, schema=schema_abc)
-    import ipdb; ipdb.set_trace()
     assert report['tables'][0]['error-count'] == 1
     assert len(report['tables'][0]['errors']) == 1
     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
-    assert report['tables'][0]['errors'][0]['message-data']['headers'] == ['A']
+    assert report['tables'][0]['errors'][0]['cells'] == ['A']
 
 
-# def test_missing_header_middle(schema_abc):
-#     source = """A,C
-# a,c"""
-#     report = validate_csv_text(source=source, schema=schema_abc)
-#     assert report['tables'][0]['error-count'] == 1
-#     assert len(report['tables'][0]['errors']) == 1
-#     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
-#     assert report['tables'][0]['errors'][0]['message-data']['headers'] == ['B']
+def test_missing_header_middle(schema_abc):
+    source = """A,C
+a,c"""
+    report = validate_csv_text(source=source, schema=schema_abc)
+    assert report['tables'][0]['error-count'] == 1
+    assert len(report['tables'][0]['errors']) == 1
+    assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
+    assert report['tables'][0]['errors'][0]['cells'] == ['B']
 
 
-# def test_missing_header_end(schema_abc):
-#     source = """A,B
-# a,b"""
-#     report = validate_csv_text(source=source, schema=schema_abc)
-#     assert report['tables'][0]['error-count'] == 1
-#     assert len(report['tables'][0]['errors']) == 1
-#     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
-#     assert report['tables'][0]['errors'][0]['message-data']['headers'] == ['C']
+def test_missing_header_end(schema_abc):
+    source = """A,B
+a,b"""
+    report = validate_csv_text(source=source, schema=schema_abc)
+    assert report['tables'][0]['error-count'] == 1
+    assert len(report['tables'][0]['errors']) == 1
+    assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
+    assert report['tables'][0]['errors'][0]['cells'] == ['C']
 
 
-# def test_missing_and_extra_header_end(schema_abc):
-#     source = """A,B,Z
-# a,b,z"""
-#     report = validate_csv_text(source=source, schema=schema_abc)
-#     assert report['tables'][0]['error-count'] == 2
-#     assert len(report['tables'][0]['errors']) == 2
-#     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
-#     assert report['tables'][0]['errors'][0]['message-data']['headers'] == ['C']
-#     assert report['tables'][0]['errors'][1]['code'] == 'extra-headers'
-#     assert report['tables'][0]['errors'][1]['message-data']['headers'] == ['Z']
-
+def test_missing_and_extra_header_end(schema_abc):
+    source = """A,B,Z
+a,b,z"""
+    report = validate_csv_text(source=source, schema=schema_abc)
+    assert report['tables'][0]['error-count'] == 2
+    assert len(report['tables'][0]['errors']) == 2
+    assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
+    assert report['tables'][0]['errors'][0]['cells'] == ['C']
+    assert report['tables'][0]['errors'][1]['code'] == 'extra-headers'
+    assert report['tables'][0]['errors'][1]['cells'] == ['Z']
 
-# def test_missing_and_extra_header_middle(schema_abc):
-#     source = """A,Z,B
-# a,z,b"""
-#     report = validate_csv_text(source=source, schema=schema_abc)
-#     assert report['tables'][0]['error-count'] == 2
-#     assert len(report['tables'][0]['errors']) == 2
-#     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
-#     assert report['tables'][0]['errors'][0]['message-data']['headers'] == ['C']
-#     assert report['tables'][0]['errors'][1]['code'] == 'extra-headers'
-#     assert report['tables'][0]['errors'][1]['message-data']['headers'] == ['Z']
 
+def test_missing_and_extra_header_middle(schema_abc):
+    source = """A,Z,B
+a,z,b"""
+    report = validate_csv_text(source=source, schema=schema_abc)
+    assert report['tables'][0]['error-count'] == 2
+    assert len(report['tables'][0]['errors']) == 2
+    assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
+    assert report['tables'][0]['errors'][0]['cells'] == ['C']
+    assert report['tables'][0]['errors'][1]['code'] == 'extra-headers'
+    assert report['tables'][0]['errors'][1]['cells'] == ['Z']
 
+# waiting for https://github.com/frictionlessdata/frictionless-py/issues/551
 # def test_missing_and_extra_header_multiple(schema_abc):
 #     source = """A,Z
 # a,z"""
@@ -446,11 +445,21 @@ b,c"""
 #     assert report['tables'][0]['error-count'] == 2
 #     assert len(report['tables'][0]['errors']) == 2
 #     assert report['tables'][0]['errors'][0]['code'] == 'missing-headers'
-#     assert report['tables'][0]['errors'][0]['message-data']['headers'] == ['B', 'C']
+#     assert report['tables'][0]['errors'][0]['cells'] == ['B', 'C']
 #     assert report['tables'][0]['errors'][1]['code'] == 'extra-headers'
-#     assert report['tables'][0]['errors'][1]['message-data']['headers'] == ['Z']
+#     assert report['tables'][0]['errors'][1]['cells'] == ['Z']
 
+# waiting for https://github.com/frictionlessdata/frictionless-py/issues/551
+# def test_extra_header_start(schema_abc):
+#     source = """X,A,B,C
+# x,a,b,c"""
+#     report = validate_csv_text(source=source, schema=schema_abc)
+#     assert report['tables'][0]['error-count'] == 1
+#     assert len(report['tables'][0]['errors']) == 1
+#     assert report['tables'][0]['errors'][1]['code'] == 'extra-headers'
+#     assert report['tables'][0]['errors'][1]['cells'] == ['X']
 
+# waiting for https://github.com/frictionlessdata/frictionless-py/issues/551
 # def test_extra_header(schema_abc):
 #     source = """A,B,C,D
 # a,b,c,d"""
@@ -458,7 +467,7 @@ b,c"""
 #     assert report['tables'][0]['error-count'] == 1
 #     assert len(report['tables'][0]['errors']) == 1
 #     assert report['tables'][0]['errors'][0]['code'] == 'extra-headers'
-#     assert report['tables'][0]['errors'][0]['message-data']['headers'] == ['D']
+#     assert report['tables'][0]['errors'][0]['cells'] == ['D']
 
 
 # def test_extra_multiple(schema_abc):
@@ -469,7 +478,7 @@ b,c"""
 #     assert len(report['tables'][0]['errors']) == 1
 #     assert report['tables'][0]['error-stats']['count'] == 2
 #     assert report['tables'][0]['errors'][0]['code'] == 'extra-headers'
-#     assert report['tables'][0]['errors'][0]['message-data']['headers'] == ['X', 'Y']
+#     assert report['tables'][0]['errors'][0]['cells'] == ['X', 'Y']
 
 
 # def test_missing_and_extra_headers_multiple(schema_abc):

validata_core/__init__.py

@@ -90,7 +90,7 @@ def compute_error_statistics(errors, columns):
     # Fill in error stats
     for err in errors:
         err_tag = err['tag']
-        errors_nb = len(err['message-data']['headers']) \
+        errors_nb = len(err['cells']) \
             if err['code'] in ('extra-headers', 'missing-headers') else 1
         errors_nb_dict[err_tag] += errors_nb
         errors_dist_dict[err_tag][err['code']] += errors_nb
@@ -370,9 +370,15 @@ def validate(source, schema, with_repair=True, **options):
     # TODO: merge options
     inspector_options = {
         **{
-            # TODO: Fix `pick_errors` content
-            #'pick_errors': checks,
-            'skip_errors': ['non-matching-header', 'extra-header', 'missing-header', "missing-cell"],
+            # TODO: We hide "extra-cell" or "missing-cell" to make validata header errors
+            # work but at the risk that extra cells and missing cells are no more detected :(
+            'skip_errors': [
+                'non-matching-header',
+                'extra-header',
+                'missing-header',
+                "missing-cell",
+                "extra-cell",
+            ],
             'query': frictionless.Query(limit_rows=VALIDATA_MAX_ROWS),
             'extra_checks': extra_checks,
         },

validata_core/custom_checks/extra_or_missing_header.py

@@ -40,7 +40,6 @@ class ExtraOrMissingHeader(Check):
         header_set = set(header)
 
         default_error_params = {
-            "cells": [],
             "cell": "",
             "field_name": "",
             "field_number":0,
@@ -54,8 +53,7 @@ class ExtraOrMissingHeader(Check):
             else:
                 cols = ", ".join(f"'{col}'" for col in missing_headers)
                 note = f"les colonnes {cols} n'ont pas été trouvées dans le fichier"
-            yield MissingHeadersError(note=note, **default_error_params)
-            return
+            yield MissingHeadersError(note=note, cells=list(missing_headers), **default_error_params)
 
         extra_headers = header_set - field_names_set
         if extra_headers:
@@ -64,9 +62,7 @@ class ExtraOrMissingHeader(Check):
             else:
                 cols = ", ".join(f"'{col}'" for col in extra_headers)
                 note = f"les colonnes {cols} ne sont pas déclarées dans le schéma"
-                yield ExtraHeadersError(note=note, **default_error_params)
-                return
+            yield ExtraHeadersError(note=note, cells=list(extra_headers), **default_error_params)
 
         if field_names_set == header_set and field_names != header:
-            yield WrongHeadersOrderError(note="", **default_error_params)
-            return
+            yield WrongHeadersOrderError(note="", cells=[], **default_error_params)