optimizations

packages/stand-off-data-py/stand_off_data/models.py

@@ -2,12 +2,26 @@ from xml.sax.saxutils import escape
 
 
 class StandOffData:
-    def __init__(self, attrs):
+    def __init__(self, attrs={}):
         self.meta = attrs.get('meta', {})
-        self.lookup = {tag_definition.id: tag_definition for tag_definition in
-                       [TagDefinition(x) for x in attrs.get('tags', [])]}
-        self.annotations = [TagAnnotation(x, self.lookup) for x in
-                            attrs.get('annotations', [])]
+        self.lookup = {}
+        for x in attrs.get('tags', []):
+            self.add_tag_definition(x)
+        self.annotations = [TagAnnotation(x, self.lookup)
+                            for x in attrs.get('annotations', [])]
+
+    def add_tag_definition(self, attrs):
+        tag_definition = TagDefinition(attrs)
+        if tag_definition.id in self.lookup:
+            raise Exception('Tag id already in use: {}'.format(self.to_dict()))
+        self.lookup[tag_definition.id] = tag_definition
+
+    def to_dict(self):
+        return {
+            'meta': self.meta,
+            'lookup': {k: v.to_dict() for k, v in self.lookup.items()},
+            'annotations': [x.to_dict() for x in self.annotations]
+        }
 
     def to_vrt(self, text):
         # Devide annotations into CWB's verticalized text format (.vrt) logic
@@ -88,21 +102,23 @@ class TagAnnotation:
     def __init__(self, attrs, lookup):
         self.lookup = lookup
         self.tag_id = attrs['tag_id']
-        if self.tag_id not in self.lookup:
-            raise Exception('Unknown tag id: {}'.format(self.tag_id))
         self.start = attrs['start']
         self.end = attrs['end']
-        if self.start >= self.end:
-            raise Exception('start must be lower then end')
         self.properties = [
-            PropertyAnnotation({**x, 'tag_id': self.tag_id}, self.lookup)
+            PropertyAnnotation(x, self.lookup[self.tag_id].properties)
             for x in attrs.get('properties', [])
         ]
+        ''' Sanity checks '''
+        if self.tag_id not in self.lookup:
+            raise Exception('Unknown tag: {}'.format(self.to_dict()))
+        if self.end < self.start:
+            raise Exception('Annotation end less then start: '
+                            '{}'.format(self.to_dict()))
         property_ids = [x.property_id for x in self.properties]
-        for required_property_id in self.lookup[self.tag_id].required_properties:
+        for required_property_id, required_property in self.lookup[self.tag_id].required_properties.items():  # noqa
             if required_property_id not in property_ids:
-                raise Exception(
-                    'Missing required property: {}'.format(required_property_id))
+                raise Exception('Missing required property: '
+                                '{}'.format(required_property.to_dict()))
 
     @property
     def name(self):
@@ -151,15 +167,15 @@ class PropertyAnnotation:
     def __init__(self, attrs, lookup):
         self.lookup = lookup
         self.property_id = attrs['property_id']
-        self.tag_id = attrs['tag_id']
-        if self.property_id not in self.lookup[self.tag_id].properties:
-            raise Exception('Unknown property id: {}'.format(self.property_id))
         self.value = attrs['value']
         # TODO: Process attrs['possibleValues'] as self.labels (no id?)
+        ''' Sanity checks '''
+        if self.property_id not in self.lookup:
+            raise Exception('Unknown property: {}'.format(self.to_dict()))
 
     @property
     def name(self):
-        return self.lookup[self.tag_id].properties[self.property_id].name
+        return self.lookup[self.property_id].name
 
     def to_dict(self):
         return {
@@ -174,18 +190,30 @@ class TagDefinition:
         self.id = attrs['id']
         self.name = attrs['name']
         self.description = attrs.get('description', '')
-        self.properties = {
-            property_definition.id: property_definition
-            for property_definition in [
-                PropertyDefinition(x) for x in attrs.get('properties', [])
-            ]
-        }
+        self.properties = {}
+        for x in attrs.get('properties', []):
+            self.add_property_definition(x)
+
+    def add_property_definition(self, attrs):
+        property_definition = PropertyDefinition(attrs)
+        if property_definition.id in self.properties:
+            raise Exception('Property id already in use: '
+                            '{}'.format(property_definition.to_dict()))
+        self.properties[property_definition.id] = property_definition
 
     @property
     def required_properties(self):
         return {property.id: property for property in self.properties.values()
                 if property.is_required}
 
+    def to_dict(self):
+        return {
+            'id': self.id,
+            'name': self.name,
+            'description': self.description,
+            'properties': {k: v.to_dict() for k, v in self.properties.items()}
+        }
+
 
 class PropertyDefinition:
     def __init__(self, attrs):
@@ -202,3 +230,12 @@ class PropertyDefinition:
     @property
     def has_multiple_values(self):
         return 'multiple' in self.flags
+
+    def to_dict(self):
+        return {
+            'id': self.id,
+            'name': self.name,
+            'description': self.description,
+            'flags': self.flags,
+            'labels': self.labels
+        }