hamtaa-texttools 1.1.13__py3-none-any.whl → 1.1.15__py3-none-any.whl

texttools/tools/async_tools.py

@@ -1,9 +1,11 @@
-from typing import Literal, Any, Callable
+from datetime import datetime
+from typing import Literal, Any
+from collections.abc import Callable
 
 from openai import AsyncOpenAI
 
 from texttools.tools.internals.async_operator import AsyncOperator
-import texttools.tools.internals.output_models as OM
+import texttools.tools.internals.models as Models
 
 
 class AsyncTheTool:
@@ -29,19 +31,23 @@ class AsyncTheTool:
     async def categorize(
         self,
         text: str,
+        categories: list[str] | Models.CategoryTree,
         with_analysis: bool = False,
         user_prompt: str | None = None,
         temperature: float | None = 0.0,
         logprobs: bool = False,
         top_logprobs: int | None = None,
+        mode: Literal["category_list", "category_tree"] = "category_list",
         validator: Callable[[Any], bool] | None = None,
         max_validation_retries: int | None = None,
-    ) -> OM.ToolOutput:
+        priority: int | None = 0,
+    ) -> Models.ToolOutput:
         """
-        Categorize a text into a single Islamic studies domain category.
+        Categorize a text into a category / category tree.
 
         Arguments:
             text: The input text to categorize
+            categories: The category / category_tree to give to LLM
             with_analysis: Whether to include detailed reasoning analysis
             user_prompt: Additional instructions for the categorization
             temperature: Controls randomness (0.0 = deterministic, 1.0 = creative)
@@ -49,30 +55,118 @@ class AsyncTheTool:
             top_logprobs: Number of top token alternatives to return if logprobs enabled
             validator: Custom validation function to validate the output
             max_validation_retries: Maximum number of retry attempts if validation fails
+            priority: Task execution priority (if enabled by vLLM and model)
 
         Returns:
             ToolOutput: Object containing:
-                - result (str): The assigned Islamic studies category
+                - result (str): The assigned category
                 - logprobs (list | None): Probability data if logprobs enabled
                 - analysis (str | None): Detailed reasoning if with_analysis enabled
                 - errors (list(str) | None): Errors occured during tool call
         """
-        return await self._operator.run(
-            # User parameters
-            text=text,
-            with_analysis=with_analysis,
-            user_prompt=user_prompt,
-            temperature=temperature,
-            logprobs=logprobs,
-            top_logprobs=top_logprobs,
-            validator=validator,
-            max_validation_retries=max_validation_retries,
-            # Internal parameters
-            prompt_file="categorizer.yaml",
-            output_model=OM.CategorizerOutput,
-            mode=None,
-            output_lang=None,
-        )
+        start = datetime.now()
+
+        if mode == "category_tree":
+            # Initializations
+            output = Models.ToolOutput()
+            levels = categories.get_level_count()
+            parent_id = 0
+            final_output = []
+
+            for _ in range(levels):
+                # Get child nodes for current parent
+                parent_node = categories.find_node(parent_id)
+                children = categories.find_children(parent_node)
+
+                # Check if child nodes exist
+                if not children:
+                    output.errors.append(
+                        f"No categories found for parent_id {parent_id} in the tree"
+                    )
+                    end = datetime.now()
+                    output.execution_time = (end - start).total_seconds()
+                    return output
+
+                # Extract category names and descriptions
+                category_list = [
+                    f"Category Name: {node.name}, Description: {node.description}"
+                    for node in children
+                ]
+                category_names = [node.name for node in children]
+
+                # Run categorization for this level
+                level_output = await self._operator.run(
+                    # User parameters
+                    text=text,
+                    category_list=category_list,
+                    with_analysis=with_analysis,
+                    user_prompt=user_prompt,
+                    temperature=temperature,
+                    logprobs=logprobs,
+                    top_logprobs=top_logprobs,
+                    mode=mode,
+                    validator=validator,
+                    max_validation_retries=max_validation_retries,
+                    # Internal parameters
+                    prompt_file="categorize.yaml",
+                    output_model=Models.create_dynamic_model(category_names),
+                    output_lang=None,
+                )
+
+                # Check for errors from operator
+                if level_output.errors:
+                    output.errors.extend(level_output.errors)
+                    end = datetime.now()
+                    output.execution_time = (end - start).total_seconds()
+                    return output
+
+                # Get the chosen category
+                chosen_category = level_output.result
+
+                # Find the corresponding node
+                parent_node = categories.find_node(chosen_category)
+                if parent_node is None:
+                    output.errors.append(
+                        f"Category '{chosen_category}' not found in tree after selection"
+                    )
+                    end = datetime.now()
+                    output.execution_time = (end - start).total_seconds()
+                    return output
+
+                parent_id = parent_node.node_id
+                final_output.append(parent_node.name)
+
+                # Copy analysis/logprobs/process from the last level's output
+                output.analysis = level_output.analysis
+                output.logprobs = level_output.logprobs
+                output.process = level_output.process
+
+            output.result = final_output
+            end = datetime.now()
+            output.execution_time = (end - start).total_seconds()
+            return output
+
+        else:
+            output = await self._operator.run(
+                # User parameters
+                text=text,
+                category_list=categories,
+                with_analysis=with_analysis,
+                user_prompt=user_prompt,
+                temperature=temperature,
+                logprobs=logprobs,
+                top_logprobs=top_logprobs,
+                mode=mode,
+                validator=validator,
+                max_validation_retries=max_validation_retries,
+                # Internal parameters
+                prompt_file="categorize.yaml",
+                output_model=Models.create_dynamic_model(categories),
+                output_lang=None,
+            )
+            end = datetime.now()
+            output.execution_time = (end - start).total_seconds()
+            return output
 
     async def extract_keywords(
         self,
@@ -83,9 +177,12 @@ class AsyncTheTool:
         temperature: float | None = 0.0,
         logprobs: bool = False,
         top_logprobs: int | None = None,
+        mode: Literal["auto", "threshold", "count"] = "auto",
+        number_of_keywords: int | None = None,
         validator: Callable[[Any], bool] | None = None,
         max_validation_retries: int | None = None,
-    ) -> OM.ToolOutput:
+        priority: int | None = 0,
+    ) -> Models.ToolOutput:
         """
         Extract salient keywords from text.
 
@@ -99,6 +196,7 @@ class AsyncTheTool:
             top_logprobs: Number of top token alternatives to return if logprobs enabled
             validator: Custom validation function to validate the output
             max_validation_retries: Maximum number of retry attempts if validation fails
+            priority: Task execution priority (if enabled by vLLM and model)
 
         Returns:
             ToolOutput: Object containing:
@@ -107,7 +205,8 @@ class AsyncTheTool:
                 - analysis (str | None): Detailed reasoning if with_analysis enabled
                 - errors (list(str) | None): Errors occured during tool call
         """
-        return await self._operator.run(
+        start = datetime.now()
+        output = await self._operator.run(
             # User parameters
             text=text,
             with_analysis=with_analysis,
@@ -116,13 +215,18 @@ class AsyncTheTool:
             temperature=temperature,
             logprobs=logprobs,
             top_logprobs=top_logprobs,
+            mode=mode,
+            number_of_keywords=number_of_keywords,
             validator=validator,
             max_validation_retries=max_validation_retries,
+            priority=priority,
             # Internal parameters
             prompt_file="extract_keywords.yaml",
-            output_model=OM.ListStrOutput,
-            mode=None,
+            output_model=Models.ListStrOutput,
         )
+        end = datetime.now()
+        output.execution_time = (end - start).total_seconds()
+        return output
 
     async def extract_entities(
         self,
@@ -135,7 +239,8 @@ class AsyncTheTool:
         top_logprobs: int | None = None,
         validator: Callable[[Any], bool] | None = None,
         max_validation_retries: int | None = None,
-    ) -> OM.ToolOutput:
+        priority: int | None = 0,
+    ) -> Models.ToolOutput:
         """
         Perform Named Entity Recognition (NER) over the input text.
 
@@ -149,6 +254,7 @@ class AsyncTheTool:
             top_logprobs: Number of top token alternatives to return if logprobs enabled
             validator: Custom validation function to validate the output
             max_validation_retries: Maximum number of retry attempts if validation fails
+            priority: Task execution priority (if enabled by vLLM and model)
 
         Returns:
             ToolOutput: Object containing:
@@ -157,7 +263,8 @@ class AsyncTheTool:
                 - analysis (str | None): Detailed reasoning if with_analysis enabled
                 - errors (list(str) | None): Errors occured during tool call
         """
-        return await self._operator.run(
+        start = datetime.now()
+        output = await self._operator.run(
             # User parameters
             text=text,
             with_analysis=with_analysis,
@@ -168,11 +275,15 @@ class AsyncTheTool:
             top_logprobs=top_logprobs,
             validator=validator,
             max_validation_retries=max_validation_retries,
+            priority=priority,
             # Internal parameters
             prompt_file="extract_entities.yaml",
-            output_model=OM.ListDictStrStrOutput,
+            output_model=Models.ListDictStrStrOutput,
             mode=None,
         )
+        end = datetime.now()
+        output.execution_time = (end - start).total_seconds()
+        return output
 
     async def is_question(
         self,
@@ -184,7 +295,8 @@ class AsyncTheTool:
         top_logprobs: int | None = None,
         validator: Callable[[Any], bool] | None = None,
         max_validation_retries: int | None = None,
-    ) -> OM.ToolOutput:
+        priority: int | None = 0,
+    ) -> Models.ToolOutput:
         """
         Detect if the input is phrased as a question.
 
@@ -197,6 +309,7 @@ class AsyncTheTool:
             top_logprobs: Number of top token alternatives to return if logprobs enabled
             validator: Custom validation function to validate the output
             max_validation_retries: Maximum number of retry attempts if validation fails
+            priority: Task execution priority (if enabled by vLLM and model)
 
         Returns:
             ToolOutput: Object containing:
@@ -205,7 +318,8 @@ class AsyncTheTool:
                 - analysis (str | None): Detailed reasoning if with_analysis enabled
                 - errors (list(str) | None): Errors occured during tool call
         """
-        return await self._operator.run(
+        start = datetime.now()
+        output = await self._operator.run(
             # User parameters
             text=text,
             with_analysis=with_analysis,
@@ -215,12 +329,16 @@ class AsyncTheTool:
             top_logprobs=top_logprobs,
             validator=validator,
             max_validation_retries=max_validation_retries,
+            priority=priority,
             # Internal parameters
             prompt_file="is_question.yaml",
-            output_model=OM.BoolOutput,
+            output_model=Models.BoolOutput,
             mode=None,
             output_lang=None,
         )
+        end = datetime.now()
+        output.execution_time = (end - start).total_seconds()
+        return output
 
     async def text_to_question(
         self,
@@ -233,7 +351,8 @@ class AsyncTheTool:
         top_logprobs: int | None = None,
         validator: Callable[[Any], bool] | None = None,
         max_validation_retries: int | None = None,
-    ) -> OM.ToolOutput:
+        priority: int | None = 0,
+    ) -> Models.ToolOutput:
         """
         Generate a single question from the given text.
 
@@ -247,6 +366,7 @@ class AsyncTheTool:
             top_logprobs: Number of top token alternatives to return if logprobs enabled
             validator: Custom validation function to validate the output
             max_validation_retries: Maximum number of retry attempts if validation fails
+            priority: Task execution priority (if enabled by vLLM and model)
 
         Returns:
             ToolOutput: Object containing:
@@ -255,7 +375,8 @@ class AsyncTheTool:
                 - analysis (str | None): Detailed reasoning if with_analysis enabled
                 - errors (list(str) | None): Errors occured during tool call
         """
-        return await self._operator.run(
+        start = datetime.now()
+        output = await self._operator.run(
             # User parameters
             text=text,
             with_analysis=with_analysis,
@@ -266,11 +387,15 @@ class AsyncTheTool:
             top_logprobs=top_logprobs,
             validator=validator,
             max_validation_retries=max_validation_retries,
+            priority=priority,
             # Internal parameters
             prompt_file="text_to_question.yaml",
-            output_model=OM.StrOutput,
+            output_model=Models.StrOutput,
             mode=None,
         )
+        end = datetime.now()
+        output.execution_time = (end - start).total_seconds()
+        return output
 
     async def merge_questions(
         self,
@@ -284,7 +409,8 @@ class AsyncTheTool:
         mode: Literal["default", "reason"] = "default",
         validator: Callable[[Any], bool] | None = None,
         max_validation_retries: int | None = None,
-    ) -> OM.ToolOutput:
+        priority: int | None = 0,
+    ) -> Models.ToolOutput:
         """
         Merge multiple questions into a single unified question.
 
@@ -299,6 +425,7 @@ class AsyncTheTool:
             mode: Merging strategy - 'default' for direct merge, 'reason' for reasoned merge
             validator: Custom validation function to validate the output
             max_validation_retries: Maximum number of retry attempts if validation fails
+            priority: Task execution priority (if enabled by vLLM and model)
 
         Returns:
             ToolOutput: Object containing:
@@ -307,10 +434,11 @@ class AsyncTheTool:
                 - analysis (str | None): Detailed reasoning if with_analysis enabled
                 - errors (list(str) | None): Errors occured during tool call
         """
-        text = ", ".join(text)
-        return await self._operator.run(
+        start = datetime.now()
+        text_combined = ", ".join(text)
+        output = await self._operator.run(
             # User parameters
-            text=text,
+            text=text_combined,
             with_analysis=with_analysis,
             output_lang=output_lang,
             user_prompt=user_prompt,
@@ -319,11 +447,15 @@ class AsyncTheTool:
             top_logprobs=top_logprobs,
             validator=validator,
             max_validation_retries=max_validation_retries,
+            priority=priority,
             # Internal parameters
             prompt_file="merge_questions.yaml",
-            output_model=OM.StrOutput,
+            output_model=Models.StrOutput,
             mode=mode,
         )
+        end = datetime.now()
+        output.execution_time = (end - start).total_seconds()
+        return output
 
     async def rewrite(
         self,
@@ -337,7 +469,8 @@ class AsyncTheTool:
         mode: Literal["positive", "negative", "hard_negative"] = "positive",
         validator: Callable[[Any], bool] | None = None,
         max_validation_retries: int | None = None,
-    ) -> OM.ToolOutput:
+        priority: int | None = 0,
+    ) -> Models.ToolOutput:
         """
         Rewrite a text with different modes.
 
@@ -352,6 +485,7 @@ class AsyncTheTool:
             mode: Rewriting mode - 'positive', 'negative', or 'hard_negative'
             validator: Custom validation function to validate the output
             max_validation_retries: Maximum number of retry attempts if validation fails
+            priority: Task execution priority (if enabled by vLLM and model)
 
         Returns:
             ToolOutput: Object containing:
@@ -360,7 +494,8 @@ class AsyncTheTool:
                 - analysis (str | None): Detailed reasoning if with_analysis enabled
                 - errors (list(str) | None): Errors occured during tool call
         """
-        return await self._operator.run(
+        start = datetime.now()
+        output = await self._operator.run(
             # User parameters
             text=text,
             with_analysis=with_analysis,
@@ -371,11 +506,15 @@ class AsyncTheTool:
             top_logprobs=top_logprobs,
             validator=validator,
             max_validation_retries=max_validation_retries,
+            priority=priority,
             # Internal parameters
             prompt_file="rewrite.yaml",
-            output_model=OM.StrOutput,
+            output_model=Models.StrOutput,
             mode=mode,
         )
+        end = datetime.now()
+        output.execution_time = (end - start).total_seconds()
+        return output
 
     async def subject_to_question(
         self,
@@ -389,7 +528,8 @@ class AsyncTheTool:
         top_logprobs: int | None = None,
         validator: Callable[[Any], bool] | None = None,
         max_validation_retries: int | None = None,
-    ) -> OM.ToolOutput:
+        priority: int | None = 0,
+    ) -> Models.ToolOutput:
         """
         Generate a list of questions about a subject.
 
@@ -404,6 +544,7 @@ class AsyncTheTool:
             top_logprobs: Number of top token alternatives to return if logprobs enabled
             validator: Custom validation function to validate the output
             max_validation_retries: Maximum number of retry attempts if validation fails
+            priority: Task execution priority (if enabled by vLLM and model)
 
         Returns:
             ToolOutput: Object containing:
@@ -412,7 +553,8 @@ class AsyncTheTool:
                 - analysis (str | None): Detailed reasoning if with_analysis enabled
                 - errors (list(str) | None): Errors occured during tool call
         """
-        return await self._operator.run(
+        start = datetime.now()
+        output = await self._operator.run(
             # User parameters
             text=text,
             number_of_questions=number_of_questions,
@@ -424,11 +566,15 @@ class AsyncTheTool:
             top_logprobs=top_logprobs,
             validator=validator,
             max_validation_retries=max_validation_retries,
+            priority=priority,
             # Internal parameters
             prompt_file="subject_to_question.yaml",
-            output_model=OM.ReasonListStrOutput,
+            output_model=Models.ReasonListStrOutput,
             mode=None,
         )
+        end = datetime.now()
+        output.execution_time = (end - start).total_seconds()
+        return output
 
     async def summarize(
         self,
@@ -441,7 +587,8 @@ class AsyncTheTool:
         top_logprobs: int | None = None,
         validator: Callable[[Any], bool] | None = None,
         max_validation_retries: int | None = None,
-    ) -> OM.ToolOutput:
+        priority: int | None = 0,
+    ) -> Models.ToolOutput:
         """
         Summarize the given subject text.
 
@@ -455,6 +602,7 @@ class AsyncTheTool:
             top_logprobs: Number of top token alternatives to return if logprobs enabled
             validator: Custom validation function to validate the output
             max_validation_retries: Maximum number of retry attempts if validation fails
+            priority: Task execution priority (if enabled by vLLM and model)
 
         Returns:
             ToolOutput: Object containing:
@@ -463,7 +611,8 @@ class AsyncTheTool:
                 - analysis (str | None): Detailed reasoning if with_analysis enabled
                 - errors (list(str) | None): Errors occured during tool call
         """
-        return await self._operator.run(
+        start = datetime.now()
+        output = await self._operator.run(
             # User parameters
             text=text,
             with_analysis=with_analysis,
@@ -474,11 +623,15 @@ class AsyncTheTool:
             top_logprobs=top_logprobs,
             validator=validator,
             max_validation_retries=max_validation_retries,
+            priority=priority,
             # Internal parameters
             prompt_file="summarize.yaml",
-            output_model=OM.StrOutput,
+            output_model=Models.StrOutput,
             mode=None,
         )
+        end = datetime.now()
+        output.execution_time = (end - start).total_seconds()
+        return output
 
     async def translate(
         self,
@@ -491,7 +644,8 @@ class AsyncTheTool:
         top_logprobs: int | None = None,
         validator: Callable[[Any], bool] | None = None,
         max_validation_retries: int | None = None,
-    ) -> OM.ToolOutput:
+        priority: int | None = 0,
+    ) -> Models.ToolOutput:
         """
         Translate text between languages.
 
@@ -505,6 +659,7 @@ class AsyncTheTool:
             top_logprobs: Number of top token alternatives to return if logprobs enabled
             validator: Custom validation function to validate the output
             max_validation_retries: Maximum number of retry attempts if validation fails
+            priority: Task execution priority (if enabled by vLLM and model)
 
         Returns:
             ToolOutput: Object containing:
@@ -513,7 +668,8 @@ class AsyncTheTool:
                 - analysis (str | None): Detailed reasoning if with_analysis enabled
                 - errors (list(str) | None): Errors occured during tool call
         """
-        return await self._operator.run(
+        start = datetime.now()
+        output = await self._operator.run(
             # User parameters
             text=text,
             target_language=target_language,
@@ -524,12 +680,73 @@ class AsyncTheTool:
             top_logprobs=top_logprobs,
             validator=validator,
             max_validation_retries=max_validation_retries,
+            priority=priority,
             # Internal parameters
             prompt_file="translate.yaml",
-            output_model=OM.StrOutput,
+            output_model=Models.StrOutput,
             mode=None,
             output_lang=None,
         )
+        end = datetime.now()
+        output.execution_time = (end - start).total_seconds()
+        return output
+
+    async def detect_entity(
+        self,
+        text: str,
+        with_analysis: bool = False,
+        output_lang: str | None = None,
+        user_prompt: str | None = None,
+        temperature: float | None = 0.0,
+        logprobs: bool = False,
+        top_logprobs: int | None = None,
+        validator: Callable[[Any], bool] | None = None,
+        max_validation_retries: int | None = None,
+        priority: int | None = 0,
+    ) -> Models.ToolOutput:
+        """
+        Detects entities in a given text based on the entity_detector.yaml prompt.
+
+        Arguments:
+            text: The input text
+            with_analysis: Whether to include detailed reasoning analysis
+            output_lang: Language for the output summary
+            user_prompt: Additional instructions for summarization
+            temperature: Controls randomness (0.0 = deterministic, 1.0 = creative)
+            logprobs: Whether to return token probability information
+            top_logprobs: Number of top token alternatives to return if logprobs enabled
+            validator: Custom validation function to validate the output
+            max_validation_retries: Maximum number of retry attempts if validation fails
+            priority: Task execution priority (if enabled by vLLM and model)
+
+        Returns:
+            ToolOutput: Object containing:
+                - result (list[Entity]): The entities
+                - logprobs (list | None): Probability data if logprobs enabled
+                - analysis (str | None): Detailed reasoning if with_analysis enabled
+                - errors (list(str) | None): Errors occured during tool call
+        """
+        start = datetime.now()
+        output = await self._operator.run(
+            # User parameters
+            text=text,
+            with_analysis=with_analysis,
+            output_lang=output_lang,
+            user_prompt=user_prompt,
+            temperature=temperature,
+            logprobs=logprobs,
+            top_logprobs=top_logprobs,
+            validator=validator,
+            max_validation_retries=max_validation_retries,
+            priority=priority,
+            # Internal parameters
+            prompt_file="detect_entity.yaml",
+            output_model=Models.EntityDetectorOutput,
+            mode=None,
+        )
+        end = datetime.now()
+        output.execution_time = (end - start).total_seconds()
+        return output
 
     async def run_custom(
         self,
@@ -541,7 +758,8 @@ class AsyncTheTool:
         top_logprobs: int | None = None,
         validator: Callable[[Any], bool] | None = None,
         max_validation_retries: int | None = None,
-    ) -> OM.ToolOutput:
+        priority: int | None = 0,
+    ) -> Models.ToolOutput:
         """
         Custom tool that can do almost anything!
 
@@ -553,6 +771,7 @@ class AsyncTheTool:
             top_logprobs: Number of top token alternatives to return if logprobs enabled
             validator: Custom validation function to validate the output
             max_validation_retries: Maximum number of retry attempts if validation fails
+            priority: Task execution priority (if enabled by vLLM and model)
 
         Returns:
             ToolOutput: Object containing:
@@ -561,7 +780,8 @@ class AsyncTheTool:
                 - analysis (str | None): Detailed reasoning if with_analysis enabled
                 - errors (list(str) | None): Errors occured during tool call
         """
-        return await self._operator.run(
+        start = datetime.now()
+        output = await self._operator.run(
             # User paramaeters
             text=prompt,
             output_model=output_model,
@@ -572,9 +792,13 @@ class AsyncTheTool:
             top_logprobs=top_logprobs,
             validator=validator,
             max_validation_retries=max_validation_retries,
+            priority=priority,
             # Internal parameters
             prompt_file="run_custom.yaml",
             user_prompt=None,
             with_analysis=False,
             mode=None,
         )
+        end = datetime.now()
+        output.execution_time = (end - start).total_seconds()
+        return output