Integrating v4 schema support

Author: glenrobson

fixtures/4/bad/dup_id.json

@@ -0,0 +1,42 @@
+{
+  "@context": "http://iiif.io/api/presentation/4/context.json",
+  "id": "https://iiif.io/api/presentation/4.0/example/03_canvas.json",
+  "type": "Manifest",
+  "label": {
+    "en": [
+      "Canvas and first annotation page have same id"
+    ]
+  },
+  "items": [
+    {
+      "id": "https://iiif.io/api/presentation/4.0/example/03_canvas/canvas/p1",
+      "type": "Canvas",
+      "height": 1800,
+      "width": 1200,
+      "items": [
+        {
+          "id": "https://iiif.io/api/presentation/4.0/example/03_canvas/canvas/p1",
+          "type": "AnnotationPage",
+          "items": [
+            {
+              "id": "https://iiif.io/api/presentation/4.0/example/03_canvas/annotation/p0001-image",
+              "type": "Annotation",
+              "motivation": [ "painting" ],
+              "body": {
+                "id": "http://iiif.io/api/presentation/2.1/example/fixtures/resources/page1-full.png",
+                "type": "Image",
+                "format": "image/png",
+                "height": 1800,
+                "width": 1200
+              },
+              "target": {
+                "id": "https://iiif.io/api/presentation/4.0/example/03_canvas/canvas/p1",
+                "type": "Canvas"
+              }
+            }
+          ]
+        }
+      ]
+    }
+  ]
+}

fixtures/4/ok/02_timeline.json

@@ -0,0 +1,40 @@
+{
+  "@context": "http://iiif.io/api/presentation/4/context.json",
+  "id": "https://iiif.io/api/presentation/4.0/example/02_timeline.json",
+  "type": "Manifest",
+  "label": {
+    "en": [
+      "Simplest Audio Example (IIIF Presentation v4)"
+    ]
+  },
+  "items": [
+    {
+      "id": "https://iiif.io/api/presentation/4.0/example/02",
+      "type": "Timeline",
+      "duration": 1985.024,
+      "items": [
+        {
+          "id": "https://iiif.io/api/presentation/4.0/example/02/page",
+          "type": "AnnotationPage",
+          "items": [
+            {
+              "id": "https://iiif.io/api/presentation/4.0/example/02/page/anno",
+              "type": "Annotation",
+              "motivation": ["painting"],
+              "body": {
+                "id": "https://fixtures.iiif.io/audio/indiana/mahler-symphony-3/CD1/medium/128Kbps.mp4",
+                "type": "Audio",
+                "format": "audio/mp4",
+                "duration": 1985.024
+              },
+              "target": {
+                "id": "https://iiif.io/api/presentation/4.0/example/02",
+                "type": "Timeline"
+              }
+            }
+          ]
+        }
+      ]
+    }
+  ]
+}

presentation_validator/v3/schemavalidator.py

@@ -24,6 +24,39 @@ def printPath(pathObj, fields):
     path += '/[{}]'.format(fields)
     return path    
 
+def create_snippet(data):
+    # Take possibly a large JSON document and only show the fields at the current level
+    for key in data:
+        if isinstance(data[key], list): 
+            data[key] = '[ ... ]'
+        elif isinstance(data[key], dict):
+            data[key] = '{ ... }'    
+
+    return data        
+
+def convertValidationError(err, errorCount, total):
+    detail = ''
+    if 'title' in err.schema:
+        detail = err.schema['title']
+    description = ''    
+    if 'description' in err.schema:
+        detail += ' ' + err.schema['description']
+    context = err.instance
+    if isinstance(context, dict):
+        for key in context:
+            if isinstance(context[key], list): 
+                context[key] = '[ ... ]'
+            elif isinstance(context[key], dict):
+                context[key] = '{ ... }'
+
+    return ErrorDetail(
+            f"Error {errorCount} of {total}.\n Message: {err.message}",
+            detail, 
+            description, 
+            printPath(err.path, err.message), 
+            context,
+            err)    
+
 def validate(data, version, url):
     if version == IIIFVersion.V3_0:
         with open(f'{SCHEMA_DIR}/iiif_3_0.json') as json_file:
@@ -90,31 +123,12 @@ def validate(data, version, url):
             if errorPath not in seen_titles:
                 errors.append(errorDup)
                 seen_titles.add(errorPath)
+
         errorCount = 1
         # Now create some useful messsages to pass on
         for err in errors:
-            detail = ''
-            if 'title' in err.schema:
-                detail = err.schema['title']
-            description = ''    
-            if 'description' in err.schema:
-                detail += ' ' + err.schema['description']
-            context = err.instance
-            if isinstance(context, dict):
-                for key in context:
-                    if isinstance(context[key], list): 
-                        context[key] = '[ ... ]'
-                    elif isinstance(context[key], dict):
-                        context[key] = '{ ... }'
-
-            result.errorList.append(ErrorDetail(
-                    'Error {} of {}.\n Message: {}'.format(errorCount, len(errors), err.message),
-                    detail, 
-                    description, 
-                    printPath(err.path, err.message), 
-                    context,
-                    err))
-            #print (json.dumps(err.instance, indent=4))
+            result.errorList.append(convertValidationError(err, errorCount, len(errors)))
+            
             errorCount += 1
 
         # Return:

presentation_validator/v4/unique_ids.py

@@ -1,5 +1,7 @@
 import sys
 import json
+from presentation_validator.model import ErrorDetail
+from presentation_validator.v3.schemavalidator import create_snippet
 
 ignore = ["target", "lookAt", "range","structures","first","last","start","source"]
 # create a method where you pass in a manifest and it checks to see if the id is unique
@@ -8,13 +10,14 @@ def check(manifest):
 
     duplicates = []
     ids = []
-    print ("Looking at manifest")
     checkNode(manifest, ids, duplicates) 
 
     if len(duplicates) > 0:
-        raise ValueError(f"Duplicate ids: {duplicates}")
+        return duplicates
+    else:
+        return None    
 
-def checkNode(node, ids=[], duplicates=[]):
+def checkNode(node, ids=[], duplicates=[], path = ""):
     if type(node) != dict:
         return 
 
@@ -23,19 +26,28 @@ def checkNode(node, ids=[], duplicates=[]):
             if type(value) != str:
                 raise ValueError(f"Id must be a string: {value}")
             if value in ids:
-                duplicates.append(value)
+                duplicates.append(ErrorDetail(
+                    f"Duplicate id found",
+                    "The id field must be unique",
+                    f"Duplicate id: {value}",
+                    path + "/" + key,
+                    create_snippet(node),
+                    None
+                ))
             ids.append(value)
         else:
             # Don't look further in fields that point to other resources
             if key in ignore:
                 continue
 
             if type(value) == list:
+                count = 0
                 for item in value:
-                    checkNode(item, ids, duplicates)
+                    checkNode(item, ids, duplicates, path + "/" + key + "[" + str(count) + "]")
+                    count += 1
 
             elif type(value) != str:
-                checkNode(value, ids, duplicates)    
+                checkNode(value, ids, duplicates, path + "/" + key)    
 
 def main():
     # pass in manifest by command line argument

presentation_validator/v4/validation4.py

@@ -2,9 +2,12 @@
 from pathlib import Path
 import json
 import sys
-from unique_ids import check
+from presentation_validator.model import ValidationResult
+from presentation_validator.v3.schemavalidator import convertValidationError
+from presentation_validator.v4.unique_ids import check
 
 from jsonschema import Draft202012Validator
+from jsonschema.exceptions import relevance
 from referencing import Registry, Resource
 from referencing.jsonschema import DRAFT202012
 
@@ -37,11 +40,30 @@ def validate(instance):
     main = schemas[f"{BASE_URI}/main.json"]
     validator = Draft202012Validator(main, registry=registry)
 
-    validator.validate(instance)
+    result = ValidationResult()
 
-    # check ids
-    check(instance)
-    print ("Validation successful")
+    results = validator.iter_errors(instance)
+    errors = sorted(results, key=relevance)
+
+    if errors:
+        result.passed = False
+        errorCount = 1
+        # Now create some useful messsages to pass on
+        for err in errors:
+            result.errorList.append(convertValidationError(err, errorCount, len(errors)))
+            
+            errorCount += 1
+    else:
+        result.passed = True
+
+    duplicate_ids = check(instance)
+    if duplicate_ids:
+        result.passed = False
+
+        # Add all of the examples of duplicated ids
+        result.errorList.extend(duplicate_ids)
+    
+    return result
 
 def main():
     # Check if command line argument is provided

presentation_validator/validator.py

@@ -2,6 +2,7 @@
 from jsonschema.exceptions import ValidationError
 from presentation_validator.model import ValidationResult, ErrorDetail
 from presentation_validator.v3 import schemavalidator
+from presentation_validator.v4 import validation4
 from presentation_validator.enum import IIIFVersion
 
 import requests
@@ -67,6 +68,9 @@ def check_manifest(
             traceback.print_exc()
             result.passed = False
             result.error = f'Presentation Validator bug: "{e}". Please create a <a href="https://github.com/IIIF/presentation-validator/issues">Validator Issue</a>, including a link to the manifest.'
+
+    elif version == IIIFVersion.V4_0:
+        result = validation4.validate(manifest)
     else:
         if isinstance(data, dict):
             data = json.dumps(data, indent=3)

tests/test_validator.py

@@ -2,6 +2,7 @@
 import unittest
 from unittest.mock import Mock
 import importlib
+from pathlib import Path
 
 from bottle import Response, request, LocalRequest
 
@@ -108,6 +109,40 @@ def test_bad_manifests_v3(self):
 
                 self.assertFalse(result.passed)
 
+    def test_good_manifests_v4(self):
+        base = Path("fixtures/4/ok")
+        data = []
+        for path in base.rglob("*.json"):
+            with path.open("r", encoding="utf-8") as f:
+                print ('Testing: {}'.format(path))
+                data = json.load(f)
+        
+                result = check_manifest(data, '4.0')
+                if not result.passed:
+                    if 'errorList' in result.errorList:
+                        self.printValidationerror(path, result.errorList)
+                    else:
+                        print ('Failed to find errors but manifest {} failed validation'.format(path))
+                        print (json.dumps(result.json(), indent=2))
+
+                self.assertTrue(result.passed, 'Expected manifest {} to pass validation but it failed'.format(path))            
+                
+    def test_bad_manifests_v4(self):
+        base = Path("fixtures/4/bad")
+        data = []
+        for path in base.rglob("*.json"):
+            with path.open("r", encoding="utf-8") as f:
+                print ('Testing: {}'.format(path))
+                data = json.load(f)
+
+                result = check_manifest(data, '3.0')
+
+                if result.passed:
+                    print(f"Expected {path} to fail validation but it passed....")
+                
+                self.assertFalse(result.passed)
+    
+
     def printValidationerror(self, filename, errors):
         print('Failed to validate: {}'.format(filename))