Ensuring that we change type hints from list to List.

list is only supported from Python 3.10 onwards, and we'd like to support 3.9 onwards.

docs/examples/unstructured-element-splitter.ipynb

@@ -103,9 +103,10 @@
    "source": [
     "from unstructured.documents.elements import Element\n",
     "from colorama import Fore, Style\n",
+    "from typing import List\n",
     "\n",
     "\n",
-    "def group_elements_by_title(elements: list[Element]) -> dict:\n",
+    "def group_elements_by_title(elements: List[Element]) -> dict:\n",
     "    grouped_elements = {}\n",
     "    current_title = \"Untitled\"  # Default title for initial text without a title\n",
     "\n",

semantic_router/encoders/tfidf.py

 import string
 from collections import Counter
-from typing import Dict
+from typing import Dict, List
 
 import numpy as np
 from numpy import ndarray
@@ -20,7 +20,7 @@ class TfidfEncoder(BaseEncoder):
         self.word_index = {}
         self.idf = np.array([])
 
-    def __call__(self, docs: list[str]) -> list[list[float]]:
+    def __call__(self, docs: List[str]) -> List[List[float]]:
         if len(self.word_index) == 0 or self.idf.size == 0:
             raise ValueError("Vectorizer is not initialized.")
         if len(docs) == 0:
@@ -31,7 +31,7 @@ class TfidfEncoder(BaseEncoder):
         tfidf = tf * self.idf
         return tfidf.tolist()
 
-    def fit(self, routes: list[Route]):
+    def fit(self, routes: List[Route]):
         docs = []
         for route in routes:
             for doc in route.utterances:
@@ -39,7 +39,7 @@ class TfidfEncoder(BaseEncoder):
         self.word_index = self._build_word_index(docs)
         self.idf = self._compute_idf(docs)
 
-    def _build_word_index(self, docs: list[str]) -> dict:
+    def _build_word_index(self, docs: List[str]) -> dict:
         words = set()
         for doc in docs:
             for word in doc.split():
@@ -47,7 +47,7 @@ class TfidfEncoder(BaseEncoder):
         word_index = {word: i for i, word in enumerate(words)}
         return word_index
 
-    def _compute_tf(self, docs: list[str]) -> np.ndarray:
+    def _compute_tf(self, docs: List[str]) -> np.ndarray:
         if len(self.word_index) == 0:
             raise ValueError("Word index is not initialized.")
         tf = np.zeros((len(docs), len(self.word_index)))
@@ -60,7 +60,7 @@ class TfidfEncoder(BaseEncoder):
         tf = tf / norm(tf, axis=1, keepdims=True)
         return tf
 
-    def _compute_idf(self, docs: list[str]) -> np.ndarray:
+    def _compute_idf(self, docs: List[str]) -> np.ndarray:
         if len(self.word_index) == 0:
             raise ValueError("Word index is not initialized.")
         idf = np.zeros(len(self.word_index))

semantic_router/layer.py

@@ -196,7 +196,7 @@ class RouteLayer:
         else:
             self.encoder = encoder
         self.llm = llm
-        self.routes: list[Route] = routes if routes is not None else []
+        self.routes: List[Route] = routes if routes is not None else []
         self.score_threshold = self.encoder.score_threshold
         self.top_k = top_k
         if self.top_k < 1:

semantic_router/llms/base.py

@@ -20,7 +20,7 @@ class BaseLLM(BaseModel):
         raise NotImplementedError("Subclasses must implement this method")
     
     def _is_valid_inputs(
-        self, inputs: list[dict[str, Any]], function_schemas: list[dict[str, Any]]
+        self, inputs: List[dict[str, Any]], function_schemas: List[dict[str, Any]]
     ) -> bool:
         """Determine if the functions chosen by the LLM exist within the function_schemas, 
         and if the input arguments are valid for those functions."""
@@ -68,7 +68,7 @@ class BaseLLM(BaseModel):
             logger.error(f"Single input validation error: {str(e)}")
             return False
 
-    def _extract_parameter_info(self, signature: str) -> tuple[list[str], list[str]]:
+    def _extract_parameter_info(self, signature: str) -> tuple[List[str], List[str]]:
         """Extract parameter names and types from the function signature."""
         param_info = [param.strip() for param in signature[1:-1].split(",")]
         param_names = [info.split(":")[0].strip() for info in param_info]
@@ -78,7 +78,7 @@ class BaseLLM(BaseModel):
         return param_names, param_types
 
     def extract_function_inputs(
-        self, query: str, function_schemas: list[dict[str, Any]]
+        self, query: str, function_schemas: List[dict[str, Any]]
     ) -> dict:
         logger.info("Extracting function input...")
 

semantic_router/llms/llamacpp.py

 from contextlib import contextmanager
 from pathlib import Path
-from typing import Any, Optional
+from typing import Any, Optional, List
 
 from pydantic.v1 import PrivateAttr
 
@@ -48,7 +48,7 @@ class LlamaCppLLM(BaseLLM):
 
     def __call__(
         self,
-        messages: list[Message],
+        messages: List[Message],
     ) -> str:
         try:
             completion = self.llm.create_chat_completion(

semantic_router/llms/openai.py

@@ -41,7 +41,7 @@ class OpenAILLM(BaseLLM):
         self.temperature = temperature
         self.max_tokens = max_tokens
 
-    def _extract_tool_calls_info(self, tool_calls: list[dict[str, Any]]) -> list[dict[str, Any]]:
+    def _extract_tool_calls_info(self, tool_calls: List[dict[str, Any]]) -> List[dict[str, Any]]:
         tool_calls_info = []
         for tool_call in tool_calls:
             if tool_call.function.arguments is None:
@@ -57,7 +57,7 @@ class OpenAILLM(BaseLLM):
     def __call__(
         self,
         messages: List[Message],
-        function_schemas: Optional[list[dict[str, Any]]] = None,
+        function_schemas: Optional[List[dict[str, Any]]] = None,
     ) -> str:
         if self.client is None:
             raise ValueError("OpenAI client is not initialized.")
@@ -99,7 +99,7 @@ class OpenAILLM(BaseLLM):
             raise Exception(f"LLM error: {e}") from e
 
     def extract_function_inputs(
-        self, query: str, function_schemas: list[dict[str, Any]]
+        self, query: str, function_schemas: List[dict[str, Any]]
     ) -> dict:
         messages = []
         system_prompt = "You are an intelligent AI. Given a command or request from the user, call the function to complete the request."

semantic_router/route.py

@@ -47,7 +47,7 @@ class Route(BaseModel):
     name: str
     utterances: Union[List[str], List[Union[Any, "Image"]]]
     description: Optional[str] = None
-    function_schemas: Optional[list[Dict[str, Any]]] = None
+    function_schemas: Optional[List[Dict[str, Any]]] = None
     llm: Optional[BaseLLM] = None
     score_threshold: Optional[float] = None
 

semantic_router/schema.py

@@ -24,7 +24,7 @@ class EncoderInfo(BaseModel):
 
 class RouteChoice(BaseModel):
     name: Optional[str] = None
-    function_call: Optional[list[dict]] = None
+    function_call: Optional[List[dict]] = None
     similarity_score: Optional[float] = None
 
 

semantic_router/splitters/rolling_window.py

@@ -340,7 +340,7 @@ class RollingWindowSplitter(BaseSplitter):
         self,
         similarities: List[float],
         split_indices: List[int],
-        splits: list[DocumentSplit],
+        splits: List[DocumentSplit],
     ):
         try:
             from matplotlib import pyplot as plt

semantic_router/splitters/utils.py

 import regex
 import tiktoken
+from typing import List
 
-
-def split_to_sentences(text: str) -> list[str]:
+def split_to_sentences(text: str) -> List[str]:
     """
     Enhanced regex pattern to split a given text into sentences more accurately.
 

semantic_router/text.py

@@ -157,7 +157,7 @@ class Conversation(BaseModel):
         :raises ValueError: If the splitter is not configured before calling this method.
 
         :return: A tuple containing the updated list of topics and the list of new topics generated in this call.
-        :rtype: tuple[list[tuple[int, str]], list[DocumentSplit]]
+        :rtype: tuple[List[tuple[int, str]], List[DocumentSplit]]
         """
 
         if self.splitter is None: