Merge pull request #106 from simulacrum6/master

Add new job type "text_label_multiple_span_select".

basemodels/manifest/data/groundtruth.py

@@ -1,7 +1,7 @@
 from typing import List, Optional, Union
 
 import requests
-from pydantic import BaseModel, HttpUrl, ValidationError
+from pydantic import BaseModel, HttpUrl, ValidationError, conlist, validator, root_validator, Field
 from requests import RequestException
 from typing_extensions import Literal
 
@@ -78,13 +78,36 @@ class ILASGroundtruthEntry(BaseModel):
 ilas_groundtruth_entry_type = List[List[ILASGroundtruthEntry]]
 ILASGroundtruthEntryModel = create_wrapper_model(ilas_groundtruth_entry_type)
 
+class TLMSSGroundTruthEntry(BaseModel):
+    start: int
+    end: int
+    label: str
+
+
+"""
+Groundtruth file format for `text_label_multiple_span_select` job type
+
+{
+  "https://domain.com/file1.txt": [
+    {
+      "start": 0,
+      "end": 4,
+      "label": "0"
+    }
+  ]
+}
+"""
+tlmss_groundtruth_entry_type = List[TLMSSGroundTruthEntry]
+TLMSSGroundTruthEntryModel = create_wrapper_model(tlmss_groundtruth_entry_type)
+
+
 groundtruth_entry_models_map = {
     "image_label_binary": ILBGroundtruthEntryModel,
     "image_label_multiple_choice": ILMCGroundtruthEntryModel,
     "image_label_area_select": ILASGroundtruthEntryModel,
+    "text_label_multiple_span_select": TLMSSGroundTruthEntryModel,
 }
 
-
 def validate_content_type(uri: str) -> None:
     """Validate uri content type"""
     try:

basemodels/manifest/manifest.py

@@ -39,6 +39,7 @@ class BaseJobTypesEnum(str, Enum):
     image_label_binary = "image_label_binary"
     image_label_multiple_choice = "image_label_multiple_choice"
     text_free_entry = "text_free_entry"
+    text_label_multiple_span_select = "text_label_multiple_span_select"
     text_multiple_choice_one_option = "text_multiple_choice_one_option"
     text_multiple_choice_multiple_options = "text_multiple_choice_multiple_options"
     image_label_area_adjust = "image_label_area_adjust"

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "hmt-basemodels"
-version = "0.2.5"
+version = "0.2.6"
 description = ""
 authors = ["Intuition Machines, Inc <[email protected]>"]
 packages = [

setup.py

@@ -2,7 +2,7 @@
 
 setuptools.setup(
     name="hmt-basemodels",
-    version="0.2.5",
+    version="0.2.6",
     author="HUMAN Protocol",
     description="Common data models shared by various components of the Human Protocol stack",
     url="https://github.com/hCaptcha/hmt-basemodels",

tests/test_manifest_validation.py

@@ -801,6 +801,43 @@ def test_groundtruth_uri_ilas_invalid_key(self):
         with self.assertRaises(ValidationError):
             self.validate_groundtruth_response("image_label_area_select", body)
 
+    def test_groundtruth_uri_tlmss_valid(self):
+        groundtruth_uri = "https://domain.com/file1.txt"
+        body = {
+            groundtruth_uri: [
+              {"start": 0, "end": 4, "label": "0"},
+              {"start": 17, "end": 89, "label": "1"},
+            ]
+        }
+        self.register_http_response(groundtruth_uri, method=httpretty.HEAD, headers={"Content-Type": "text/plain"})
+        self.validate_groundtruth_response("text_label_multiple_span_select", body)
+
+
+    def test_groundtruth_uri_tlmss_invalid_key(self):
+        body = {
+            "not_uri": [
+              {"start": 0, "end": 4, "label": "0"},
+              {"start": 17, "end": 89, "label": "1"},
+            ]
+        }
+
+        with self.assertRaises(ValidationError):
+            self.validate_groundtruth_response("text_label_multiple_span_select", body)
+
+
+    def test_groundtruth_uri_tlmss_invalid_value(self):
+        body = {
+            "https://www.domain.com/file1.txt": [
+              {"span": [0, 4]},
+              {"span": [17, 89], "label": "1"},
+            ]
+        }
+
+        with self.assertRaises(ValidationError):
+            self.validate_groundtruth_response("text_label_multiple_span_select", body)
+
+
+
     def test_groundtruth_uri_ilas_invalid_value(self):
         body = {"https://domain.com/file1.jpeg": [[True]]}