label-studio-sdk 1.0.7__py3-none-any.whl → 1.0.10__py3-none-any.whl

label_studio_sdk/label_interface/control_tags.py

@@ -93,6 +93,15 @@ class ControlTag(LabelStudioTag):
             and tag.tag not in _NOT_CONTROL_TAGS
         )
 
+    def to_json_schema(self):
+        """
+        Converts the current ControlTag instance into a JSON Schema.
+
+        Returns:
+            dict: A dictionary representing the JSON Schema.
+        """
+        return {"type": "string"}
+
     @classmethod
     def parse_node(cls, tag: xml.etree.ElementTree.Element, tags_mapping=None) -> "ControlTag":
         """
@@ -261,11 +270,11 @@ class ControlTag(LabelStudioTag):
 
     def _validate_labels(self, labels):
         """Check that labels is a subset of self.labels, used for
-        example when you're validate the annotaion or prediction to
+        example when you're validate the annotation or prediction to
         make sure there no undefined labels used.
 
         """
-        if not self.labels:
+        if not self.labels or not labels:
             return True
 
         return set(labels).issubset(set(self.labels))
@@ -335,10 +344,11 @@ class ControlTag(LabelStudioTag):
 
             return to_name
         else:
-            if len(self.to_name) > 1:
-                raise Exception(
-                    "Multiple to_name in control tag, specify to_name in function"
-                )
+            # TODO: "Pairwise" tag has multiple to_name
+            # if len(self.to_name) > 1:
+            #     raise Exception(
+            #         "Multiple to_name in control tag, specify to_name in function"
+            #     )
 
             return self.to_name[0]
 
@@ -415,7 +425,7 @@ class ControlTag(LabelStudioTag):
             )
 
         kwargs[self._label_attr_name] = label
-
+        
         return self._label_simple(to_name=to_name, **kwargs)
 
     def label(
@@ -446,12 +456,29 @@ class ControlTag(LabelStudioTag):
         Region
             A new Region object with the specified label applied.
         """
-        if hasattr(self, "_label_attr_name"):
+        if hasattr(self, "_label_attr_name") and label is not None:
             return self._label_with_labels(
                 label=label, to_name=to_name, *args, **kwargs
             )
         else:
             return self._label_simple(to_name=to_name, *args, **kwargs)
+        
+    def get_labels(self, regions: List[Dict]):
+        """
+        Returns the simplified representation of the label. Sort of a reverse to label() method to retrieve an input `label` from an output regions
+        """
+        values = [region.get('value') for region in regions if region.get('from_name') == self.name]
+        values = list(filter(lambda x: x is not None, values))
+        if not hasattr(self, "_label_attr_name"):
+            return values
+        labels = []
+        for value in values:
+            if len(value) == 1 and self._label_attr_name in value:
+                v = value[self._label_attr_name]
+                labels.append(v[0] if type(v) == list and len(v) == 1 else v)
+            else:
+                labels.append(value)
+        return labels[0] if len(labels) == 1 else labels
 
     def as_tuple(self):
         """ """
@@ -485,6 +512,34 @@ class ChoicesTag(ControlTag):
     _label_attr_name: str = "choices"
     _value_class: Type[ChoicesValue] = ChoicesValue
 
+    @property
+    def is_multiple_choice(self):
+        return self.attr.get("choice") == "multiple"
+
+    def to_json_schema(self):
+        """
+        Converts the current ChoicesTag instance into a JSON Schema.
+
+        Returns:
+            dict: A dictionary representing the JSON Schema compatible with OpenAPI 3.
+        """
+        if self.is_multiple_choice:
+            return {
+                "type": "array",
+                "items": {
+                    "type": "string",
+                    "enum": self.labels,
+                },
+                "uniqueItems": True,
+                "description": f"Choices for {self.to_name[0]}",
+            }
+
+        return {
+            "type": "string",
+            "enum": self.labels,
+            "description": f"Choices for {self.to_name[0]}"
+        }
+
 
 class LabelsValue(SpanSelection):
     labels: List[str]
@@ -496,6 +551,41 @@ class LabelsTag(ControlTag):
     _label_attr_name: str = "labels"
     _value_class: Type[LabelsValue] = LabelsValue
 
+    def to_json_schema(self):
+        """
+        Converts the current LabelsTag instance into a JSON Schema.
+
+        Returns:
+            dict: A dictionary representing the JSON Schema compatible with OpenAPI 3.
+        """
+        return {
+            "type": "array",
+            "items": {
+                "type": "object",
+                "required": ["start", "end", "labels"],
+                "properties": {
+                    "start": {
+                        "type": "integer",
+                        "minimum": 0
+                    },
+                    "end": {
+                        "type": "integer",
+                        "minimum": 0
+                    },
+                    "labels": {
+                        "type": "array",
+                        "items": {
+                            "type": "string",
+                            "enum": self.labels
+                        }
+                    },
+                    "text": {
+                        "type": "string"
+                    }
+                }
+            },
+            "description": f"Labels and span indices for {self.to_name[0]}"
+        }
 
 ## Image tags
 
@@ -684,6 +774,26 @@ class NumberTag(ControlTag):
     """ """
     tag: str = "Number"
     _value_class: Type[NumberValue] = NumberValue
+    _label_attr_name: str = "number"
+
+    def to_json_schema(self):
+        """
+        Converts the current NumberTag instance into a JSON Schema.
+
+        Returns:
+            dict: A dictionary representing the JSON Schema compatible with OpenAPI 3.
+        """
+        schema = {
+            "type": "number",
+            "description": f"Number for {self.to_name[0]}"
+        }
+        
+        if 'min' in self.attr:
+            schema["minimum"] = float(self.attr['min'])
+        if 'max' in self.attr:
+            schema["maximum"] = float(self.attr['max'])
+        
+        return schema
     
 
 class DateTimeValue(BaseModel):
@@ -694,6 +804,25 @@ class DateTimeTag(ControlTag):
     """ """
     tag: str = "DateTime"
     _value_class: Type[DateTimeValue] = DateTimeValue
+    _label_attr_name: str = "datetime"
+
+    def _label_simple(self, to_name: Optional[str] = None, *args, **kwargs) -> Region:
+        # TODO: temporary fix to force datetime to be a string
+        kwargs['datetime'] = kwargs['datetime'][0]
+        return super()._label_simple(to_name, *args, **kwargs)
+
+    def to_json_schema(self):
+        """
+        Converts the current DateTimeTag instance into a JSON Schema.
+
+        Returns:
+            dict: A dictionary representing the JSON Schema compatible with OpenAPI 3.
+        """
+        return {
+            "type": "string",
+            "format": "date-time",
+            "description": f"Date and time for {self.to_name[0]}"
+        }
 
 
 class HyperTextLabelsValue(SpanSelectionOffsets):
@@ -715,12 +844,27 @@ class PairwiseTag(ControlTag):
     """ """
     tag: str = "Pairwise"
     _value_class: Type[PairwiseValue] = PairwiseValue
+    _label_attr_name: str = "selected"
 
-    def label(self, side):
+    def label(self, label):
         """ """
-        value = PairwiseValue(selected=side)
+        value = PairwiseValue(selected=label)
+        # <Pairwise> tag has equal from_name and to_name, and string label that's not a list of strings
         return Region(from_tag=self, to_tag=self, value=value)
 
+    def to_json_schema(self):
+        """
+        Converts the current PairwiseTag instance into a JSON Schema.
+
+        Returns:
+            dict: A dictionary representing the JSON Schema compatible with OpenAPI 3.
+        """
+        return {
+            "type": "string",
+            "enum": ["left", "right"],
+            "description": f"Pairwise selection between {self.to_name[0]} (left) and {self.to_name[1]} (right)"
+        }
+
 
 class ParagraphLabelsValue(SpanSelectionOffsets):
     paragraphlabels: List[str]
@@ -759,6 +903,22 @@ class RatingTag(ControlTag):
     """ """
     tag: str = "Rating"
     _value_class: Type[RatingValue] = RatingValue
+    _label_attr_name: str = "rating"
+
+    def to_json_schema(self):
+        """
+        Converts the current RatingTag instance into a JSON Schema.
+
+        Returns:
+            dict: A dictionary representing the JSON Schema compatible with OpenAPI 3.
+        """
+        max_rating = int(self.attr.get('maxRating', 5))  # Default to 5 if not specified
+        return {
+            "type": "integer",
+            "minimum": 0,
+            "maximum": max_rating,
+            "description": f"Rating for {self.to_name[0]} (0 to {max_rating})"
+        }
 
 
 class RelationsTag(ControlTag):
@@ -784,6 +944,27 @@ class TaxonomyTag(ControlTag):
     """ """
     tag: str = "Taxonomy"
     _value_class: Type[TaxonomyValue] = TaxonomyValue
+    _label_attr_name: str = "taxonomy"
+
+    def to_json_schema(self):
+        """
+        Converts the current TaxonomyTag instance into a JSON Schema.
+
+        Returns:
+            dict: A dictionary representing the JSON Schema compatible with OpenAPI 3.
+        """
+        return {
+            "type": "array",
+            "items": {
+                "type": "array",
+                "items": {
+                    "type": "string",
+                    # TODO: enforce the order of the enums according to the taxonomy tree
+                    "enum": self.labels
+                }
+            },
+            "description": f"Taxonomy for {self.to_name[0]}. Each item is a path from root to selected node."
+        }
 
 
 class TextAreaValue(BaseModel):
@@ -794,6 +975,20 @@ class TextAreaTag(ControlTag):
     """ """
     tag: str = "TextArea"
     _value_class: Type[TextAreaValue] = TextAreaValue
+    _label_attr_name: str = "text"
+
+    def to_json_schema(self):
+        """
+        Converts the current TextAreaTag instance into a JSON Schema.
+
+        Returns:
+            dict: A dictionary representing the JSON Schema compatible with OpenAPI 3.
+        """
+        return {
+            "type": "string",
+            "description": f"Text for {self.to_name[0]}"
+        }
+
 
 
 class TimeSeriesValue(SpanSelection):

label_studio_sdk/label_interface/interface.py

@@ -15,6 +15,7 @@ from pydantic import BaseModel
 from collections import defaultdict, OrderedDict
 from lxml import etree
 import xmljson
+from jsf import JSF
 
 from label_studio_sdk._legacy.exceptions import (
     LSConfigParseException,
@@ -320,7 +321,7 @@ class LabelInterface:
             # 1. we should allow control.label to process custom payload outside of those strictly containing "label"
             # 2. we should be less open regarding the payload type and defining the strict typing elsewhere,
             # but likely that requires rewriting of how ControlTag.label() is working now
-            if isinstance(payload, str):
+            if isinstance(payload, (str, int, float)):
                 payload = {'label': payload}
             elif isinstance(payload, list):
                 if len(payload) > 0:
@@ -331,6 +332,7 @@ class LabelInterface:
 
             if isinstance(payload, Dict):
                 payload = [payload]
+
             for item in payload:
                 regions.append(control.label(**item))
 
@@ -531,12 +533,26 @@ class LabelInterface:
         tree.task_loaded = True
         
         for obj in tree.objects:
-            print(obj.value_is_variable, obj.value_name)
             if obj.value_is_variable and obj.value_name in task:
                 obj.value = task.get(obj.value_name)
 
         return tree
     
+    def to_json_schema(self):
+        """
+        Converts the current LabelInterface instance into a JSON Schema.
+
+        Returns:
+            dict: A dictionary representing the JSON Schema.
+        """
+        return {
+            "type": "object",
+            "properties": {
+                name: control.to_json_schema() for name, control in self._controls.items()
+            },
+            "required": list(self._controls.keys())
+        }
+    
     def parse(self, config_string: str) -> Tuple[Dict, Dict, Dict, etree._Element]:
         """Parses the received configuration string into dictionaries
         of ControlTags, ObjectTags, and Labels, along with an XML tree
@@ -755,7 +771,7 @@ class LabelInterface:
             return False
 
         # type of the region should match the tag name        
-        if control.tag.lower() != region["type"]:
+        if control.tag.lower() != region["type"].lower():
             return False
         
         # make sure that in config it connects to the same tag as
@@ -824,9 +840,67 @@ class LabelInterface:
 
         return task
 
-    def generate_sample_annotation(self):
-        """ """
-        raise NotImplemented()
+    def _generate_sample_regions(self):
+        """ Generate an example of each control tag's JSON schema and validate it as a region"""
+        return self.create_regions({
+            control.name: JSF(control.to_json_schema()).generate()
+            for control in self.controls
+        })
+
+    def generate_sample_prediction(self) -> Optional[dict]:
+        """Generates a sample prediction that is valid for this label config.
+
+        Example:
+            {'model_version': 'sample model version',
+             'score': 0.0,
+             'result': [{'id': 'e7bd76e6-4e88-4eb3-b433-55e03661bf5d',
+               'from_name': 'sentiment',
+               'to_name': 'text',
+               'type': 'choices',
+               'value': {'choices': ['Neutral']}}]}
+
+        NOTE: `id` field in result is not required when importing predictions; it will be generated automatically.
+        NOTE: for each control tag, depends on tag.to_json_schema() being implemented correctly
+        """
+        prediction = PredictionValue(
+            model_version='sample model version',
+            result=self._generate_sample_regions()
+        )
+        prediction_dct = prediction.model_dump()
+        if self.validate_prediction(prediction_dct):
+            return prediction_dct
+        else:
+            logger.debug(f'Sample prediction {prediction_dct} failed validation for label config {self.config}')
+            return None
+
+    def generate_sample_annotation(self) -> Optional[dict]:
+        """Generates a sample annotation that is valid for this label config.
+
+        Example:
+            {'was_cancelled': False,
+             'ground_truth': False,
+             'lead_time': 0.0,
+             'result_count': 0,
+             'completed_by': -1,
+             'result': [{'id': 'b05da11d-3ffc-4657-8b8d-f5bc37cd59ac',
+               'from_name': 'sentiment',
+               'to_name': 'text',
+               'type': 'choices',
+               'value': {'choices': ['Negative']}}]}
+
+        NOTE: `id` field in result is not required when importing predictions; it will be generated automatically.
+        NOTE: for each control tag, depends on tag.to_json_schema() being implemented correctly
+        """
+        annotation = AnnotationValue(
+            completed_by=-1,  # annotator's user id
+            result=self._generate_sample_regions()
+        )
+        annotation_dct = annotation.model_dump()
+        if self.validate_annotation(annotation_dct):
+            return annotation_dct
+        else:
+            logger.debug(f'Sample annotation {annotation_dct} failed validation for label config {self.config}')
+            return None
 
     #####
     ##### COMPATIBILITY LAYER

label_studio_sdk/label_interface/region.py

@@ -10,16 +10,7 @@ from pydantic import BaseModel, Field
 
 class Region(BaseModel):
     """
-    Class for Region Tag
-
-    Attributes:
-    -----------
-    id: str
-        The unique identifier of the region
-    x: int
-        The x coordinate of the region
-    y: int
-
+    A Region is an item in the `result` list of a PredictionValue or AnnotationValue.
     """
 
     id: str = Field(default_factory=lambda: str(uuid4()))