streamlit-octostar-utils 2.11a6__tar.gz → 2.11a8__tar.gz

{streamlit_octostar_utils-2.11a6 → streamlit_octostar_utils-2.11a8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: streamlit-octostar-utils
-Version: 2.11a6
+Version: 2.11a8
 Summary: 
 License: MIT
 License-File: LICENSE

{streamlit_octostar_utils-2.11a6 → streamlit_octostar_utils-2.11a8}/pyproject.toml

@@ -5,7 +5,7 @@ include = '\.pyi?$'
 
 [tool.poetry]
 name = "streamlit-octostar-utils"
-version = "2.11a6"
+version = "2.11a8"
 description = ""
 license = "MIT"
 authors = ["Octostar"]

{streamlit_octostar_utils-2.11a6 → streamlit_octostar_utils-2.11a8}/streamlit_octostar_utils/api_crafter/fastapi.py

@@ -330,7 +330,7 @@ class DefaultErrorRoute:
         if len(message) > MAX_ERROR_MESSAGE_BYTES:
             message = message[-MAX_ERROR_MESSAGE_BYTES:]
         try:
-            tcbk = traceback.format_exception(exc)
+            tcbk = "\n".join(traceback.format_exception(exc))
             if len(tcbk) > MAX_ERROR_TRACEBACK_BYTES:
                 tcbk = tcbk[-MAX_ERROR_TRACEBACK_BYTES:]
         except:

{streamlit_octostar_utils-2.11a6 → streamlit_octostar_utils-2.11a8}/streamlit_octostar_utils/api_crafter/nifi.py

@@ -23,7 +23,7 @@ from octostar.client import make_client
 from ..core.dict import recursive_update_dict, travel_dict, jsondict_hash
 from ..core.timestamp import now, string_to_datetime
 from .fastapi import DefaultErrorRoute, Route
-from ..ontology.inheritance import is_child_concept as is_child_concept_fn
+from ..ontology.inheritance import is_child_concept as is_child_concept_fn, get_label_keys
 from ..ontology.expand_entities import expand_entities
 
 RELATIONSHIP_ENTITY_NAME = "os_relationship"
@@ -69,6 +69,7 @@ class NifiEntityModel(BaseModel):
         class OntologyInfoModel(BaseModel):
             parents: List[str]
             relationships: List[str]
+            label_keys: List[str]
 
         class ContentsPointerModel(BaseModel):
             location: NifiContentsPointerLocationModel
@@ -806,15 +807,8 @@ class NifiEntity(object):
 
     @property
     def label(self):
-        if not self.context.ontology:
-            return None
-        label_fields = self.context.ontology["concepts"][
-            self.record.get("os_concept") or self.record.get("entity_type")
-        ]["labelKeys"]
-        label_fields = [field for field in label_fields if field]
-        label = " ".join(
-            [(self.record.get(field) or "") for field in label_fields]
-        ).strip()
+        label_keys = self.request["ontology_info"]["label_keys"]
+        label = " ".join([(self.record.get(field) or "") for field in label_keys]).strip()
         if not label:
             label = None
         return label
@@ -923,14 +917,21 @@ class NifiEntity(object):
         now_time = now()
         random_id = str(uuid.uuid4())
         username = self.jwt_data["username"]
+        if entity_type == self.record["entity_type"]:
+            ont_parents = self.request["ontology_info"]["parents"]
+            ont_relationships = self.request["ontology_info"]["relationships"]
+            ont_label_keys = self.request["ontology_info"]["label_keys"]
+        else:
+            ont_parents = self.context.ontology["concepts"][entity_type]["parents"]
+            ont_relationships = self.context.ontology["concepts"][entity_type]["relationships"]
+            ont_label_keys = get_label_keys(entity_type, self.context.ontology)
         child_request = {
             "jwt": self.request["jwt"],
             "ontology_name": self.request["ontology_name"],
             "ontology_info": {
-                "parents": self.context.ontology["concepts"][entity_type]["parents"],
-                "relationships": self.context.ontology["concepts"][entity_type][
-                    "relationships"
-                ],
+                "parents": ont_parents,
+                "relationships": ont_relationships,
+                "label_keys": ont_label_keys,
             },
             "entity_timestamp": None,
             "sync_params": {},

{streamlit_octostar_utils-2.11a6 → streamlit_octostar_utils-2.11a8}/streamlit_octostar_utils/nlp/language.py

@@ -1,7 +1,9 @@
 import re
+from typing import Optional
+
 import py3langid as langid
 
-from iso639 import Lang
+from iso639 import Lang, NonExistentLanguageError
 
 FLAIR_MODELS = {
     "en": "flair/ner-english-large",
@@ -31,7 +33,7 @@ def language_to_alpha2(language_name: str) -> str:
     if not language_name:
         raise ValueError("Language name must be a non-empty string.")
 
-    name = re.sub(r'\b\w+', lambda m: m.group(0).capitalize(), name)
+    name = re.sub(r'\b\w+', lambda m: m.group(0).capitalize(), language_name)
     return Lang(name).pt1
 
 
@@ -46,19 +48,39 @@ def detect_language(text, min_confidence=None):
     return detected_lang, confidence
 
 
+def is_language_available(language: Optional[str], type: str) -> bool:
+    if not language:
+        return False
+
+    try:
+        lang_code = language_to_alpha2(language)
+
+    except NonExistentLanguageError:
+        lang_code = language
+
+    match type:
+        case "spacy":
+            return SPACY_MODELS.get(lang_code, None) is not None
+
+        case "flair":
+            return FLAIR_MODELS.get(lang_code, None) is not None
+
+
 def load_language_model(language, type):
     from flair.models import SequenceTagger
     from spacy_download import load_spacy
 
-    model = None
-
     match type:
         case "spacy":
-            model_name = SPACY_MODELS.get(language_to_alpha2(language), SPACY_MODELS["en"])
-            model = load_spacy(model_name)
+            if is_language_available(language, "spacy"):
+                model_name = SPACY_MODELS.get(language_to_alpha2(language), SPACY_MODELS["en"])
+                return load_spacy(model_name)
+
+            raise Exception(f"SpaCy model for language '{language}' is not available.")
 
         case "flair":
-            model_name = FLAIR_MODELS.get(language, "flair/ner-multi")
-            model = SequenceTagger.load(model_name)
+            if is_language_available(language, "flair"):
+                model_name = FLAIR_MODELS.get(language, "flair/ner-multi")
+                return SequenceTagger.load(model_name)
 
-    return model
+            raise Exception(f"Flair model for language '{language}' is not available.")

streamlit_octostar_utils-2.11a8/streamlit_octostar_utils/ontology/inheritance.py

@@ -0,0 +1,15 @@
+def is_child_concept(type, parent_type, ontology):
+    return type == parent_type or parent_type in ontology["concepts"][type]["parents"]
+
+def get_label_keys(type, ontology):
+    parents = set(ontology["concepts"][type]["parents"])
+    parents.add(type)
+    parents = list(parents)
+    parents.reverse()
+    label_keys = {} # for guaranteed insertion order
+    for parent in parents:
+        for label_key in ontology["concepts"][parent]["labelKeys"]:
+            if not label_key:
+                continue
+            label_keys[label_key] = None
+    return list(label_keys.keys())

streamlit_octostar_utils-2.11a6/streamlit_octostar_utils/ontology/inheritance.py

@@ -1,2 +0,0 @@
-def is_child_concept(type, parent_type, ontology):
-    return type == parent_type or parent_type in ontology["concepts"][type]["parents"]