Conditionally report summary or BGS metadata

The AGS `summary` data and BGS `additional_metadata` fields contain overlapping information, with the BGS response including additional fields. This commit avoids repetition by ensuring that if BGS data are available, only they are included in the response.
BritishGeologicalSurvey · Mar 8, 2024 · a532eb5 · a532eb5
1 parent e14f714
commit a532eb5
Show file tree

Hide file tree

Showing 3 changed files with 70 additions and 9 deletions.
diff --git a/app/borehole_map.py b/app/borehole_map.py
@@ -22,7 +22,7 @@ def extract_geojson(filepath: Path) -> dict:
     Read an AGS4 file and extract geojson represenation of LOCA table and
     metadata.
     """
-    logger.info("Extracting geojson from  %s", filepath.name)
+    logger.info("Extracting geojson from %s", filepath.name)
 
     # Read data file
     tables, load_error, _ = load_tables_reporting_errors(filepath)

diff --git a/app/validation.py b/app/validation.py
@@ -47,6 +47,8 @@ def validate(filename: Path,
 
     all_errors = {}
     all_checkers = []
+    bgs_additional_metadata = {}
+    ags_summary = []
     # Don't process if file is not .ags format
     if filename.suffix.lower() != '.ags':
         all_errors.update(
@@ -57,19 +59,32 @@ def validate(filename: Path,
         # Run checkers to extract errors and other metadata
         for checker in checkers:
             # result is a dictionary with 'errors', 'checker' and other keys
-            result = checker(filename, standard_AGS4_dictionary=dictionary_file)
+            result: dict = checker(filename, standard_AGS4_dictionary=dictionary_file)
+
             # Pull 'errors' out to add to running total
             all_errors.update(result.pop('errors'))
             all_checkers.append(result.pop('checker'))
-            # Handle additional metadata
+
+            # Extract checker-dependent additional metadata
             try:
-                response['additional_metadata'].update(result.pop('additional_metadata'))
+                bgs_additional_metadata = result.pop('additional_metadata')
             except KeyError:
-                # No additional metadata
                 pass
+
+            try:
+                ags_summary = result.pop('summary')
+            except KeyError:
+                pass
+
             # Add remaining keys to response
             response.update(result)
 
+    # We only want one checker-dependent metadata; use BGS if available.
+    if bgs_additional_metadata:
+        response['additional_metadata'] = bgs_additional_metadata
+    else:
+        response['summary'] = ags_summary
+
     error_count = len(reduce(lambda total, current: total + current, all_errors.values(), []))
     if error_count > 0:
         message = f'{error_count} error(s) found in file!'

diff --git a/test/unit/test_validation.py b/test/unit/test_validation.py
@@ -15,12 +15,32 @@
 
 def mock_check_ags(filename, standard_AGS4_dictionary=None):
     return dict(checker='ags', dictionary='some_dict',
-                errors={})
+                errors={},
+                summary=[{'desc': '7 groups identified in file: PROJ ABBR TRAN TYPE UNIT '
+                          'LOCA SAMP',
+                          'group': '',
+                          'line': ''},
+                         {'desc': '1 data row(s) in LOCA group', 'group': '', 'line': ''},
+                         {'desc': 'Optional DICT group present? False',
+                          'group': '',
+                          'line': ''},
+                         {'desc': 'Optional FILE group present? False',
+                          'group': '',
+                          'line': ''}]
+                )
 
 
 def mock_check_bgs(filename, **kwargs):
     return dict(checker='bgs',
-                errors={'BGS': [{}]})
+                errors={'BGS': [{}]},
+                additional_metadata={
+                    'bgs_all_groups': '7 groups identified in file: '
+                    'PROJ ABBR TRAN TYPE UNIT LOCA SAMP',
+                    'bgs_dict': 'Optional DICT group present: False',
+                    'bgs_file': 'Optional FILE group present: False',
+                    'bgs_loca_rows': '1 data rows in LOCA group',
+                    'bgs_projects': '1 projects found: 121415 (ACME Gas Works Redevelopment)'
+                })
 
 
 @freeze_time(FROZEN_TIME)
@@ -36,6 +56,17 @@ def test_validate_default_checker():
         'filesize': 0,
         'message': 'All checks passed!',
         'time': dt.datetime(2021, 8, 23, 14, 25, 43, tzinfo=dt.timezone.utc),
+        'summary': [{'desc': '7 groups identified in file: PROJ ABBR TRAN TYPE UNIT '
+                     'LOCA SAMP',
+                     'group': '',
+                     'line': ''},
+                    {'desc': '1 data row(s) in LOCA group', 'group': '', 'line': ''},
+                    {'desc': 'Optional DICT group present? False',
+                     'group': '',
+                     'line': ''},
+                    {'desc': 'Optional FILE group present? False',
+                     'group': '',
+                     'line': ''}],
         'valid': True,
         'additional_metadata': {}}
 
@@ -60,7 +91,14 @@ def test_validate_bgs_checker():
         'message': '1 error(s) found in file!',
         'time': dt.datetime(2021, 8, 23, 14, 25, 43, tzinfo=dt.timezone.utc),
         'valid': False,
-        'additional_metadata': {}}
+        'additional_metadata': {'bgs_all_groups': '7 groups identified in file: PROJ '
+                                'ABBR TRAN TYPE UNIT LOCA SAMP',
+                                'bgs_dict': 'Optional DICT group present: False',
+                                'bgs_file': 'Optional FILE group present: False',
+                                'bgs_loca_rows': '1 data rows in LOCA group',
+                                'bgs_projects': '1 projects found: 121415 (ACME Gas '
+                                'Works Redevelopment)'}
+    }
 
     # Act
     response = validation.validate(filename, checkers=[mock_check_bgs])
@@ -83,7 +121,14 @@ def test_validate_both_checkers():
         'message': '1 error(s) found in file!',
         'time': dt.datetime(2021, 8, 23, 14, 25, 43, tzinfo=dt.timezone.utc),
         'valid': False,
-        'additional_metadata': {}}
+        'additional_metadata': {
+            'bgs_all_groups': '7 groups identified in file: '
+            'PROJ ABBR TRAN TYPE UNIT LOCA SAMP',
+            'bgs_dict': 'Optional DICT group present: False',
+            'bgs_file': 'Optional FILE group present: False',
+            'bgs_loca_rows': '1 data rows in LOCA group',
+            'bgs_projects': '1 projects found: 121415 (ACME Gas Works Redevelopment)'
+        }}
 
     # Act
     response = validation.validate(filename, checkers=[mock_check_bgs, mock_check_ags])
@@ -106,6 +151,7 @@ def test_validate_non_ags():
         'message': '1 error(s) found in file!',
         'time': dt.datetime(2021, 8, 23, 14, 25, 43, tzinfo=dt.timezone.utc),
         'valid': False,
+        'summary': [],
         'additional_metadata': {}}
 
     # Act