zrb 1.9.0__py3-none-any.whl → 1.9.1__py3-none-any.whl

zrb/builtin/llm/llm_ask.py

@@ -10,7 +10,9 @@ from zrb.builtin.llm.tool.file import (
     apply_diff,
     list_files,
     read_from_file,
+    read_many_files,
     search_files,
+    write_many_files,
     write_to_file,
 )
 from zrb.builtin.llm.tool.web import (
@@ -132,8 +134,10 @@ if CFG.LLM_ALLOW_ACCESS_LOCAL_FILE:
         search_files,
         list_files,
         read_from_file,
+        read_many_files,
         apply_diff,
         write_to_file,
+        write_many_files,
     )
 
 if CFG.LLM_ALLOW_ACCESS_SHELL:

zrb/builtin/llm/tool/code.py

@@ -82,8 +82,8 @@ async def analyze_repo(
     goal: str,
     extensions: list[str] = _DEFAULT_EXTENSIONS,
     exclude_patterns: list[str] = DEFAULT_EXCLUDED_PATTERNS,
-    extraction_token_limit: int = 30000,
-    summarization_token_limit: int = 30000,
+    extraction_token_limit: int = 40000,
+    summarization_token_limit: int = 40000,
 ) -> str:
     """
     Extract and summarize information from any directory.

zrb/builtin/llm/tool/file.py

@@ -2,7 +2,7 @@ import fnmatch
 import json
 import os
 import re
-from typing import Any, Optional
+from typing import Any, Dict, List, Optional
 
 from zrb.builtin.llm.tool.sub_agent import create_sub_agent_tool
 from zrb.context.any_context import AnyContext
@@ -215,6 +215,7 @@ def read_from_file(
     end_line: Optional[int] = None,
 ) -> str:
     """Read file content (or specific lines) at a path, including line numbers.
+    This tool can read both, text and pdf file.
     Args:
         path (str): Path to read. Pass exactly as provided, including '~'.
         start_line (Optional[int]): Starting line number (1-based).
@@ -466,7 +467,7 @@ def apply_diff(
 
 
 async def analyze_file(
-    ctx: AnyContext, path: str, query: str, token_limit: int = 30000
+    ctx: AnyContext, path: str, query: str, token_limit: int = 40000
 ) -> str:
     """Analyze file using LLM capability to reduce context usage.
     Use this tool for:
@@ -498,3 +499,105 @@ async def analyze_file(
     )
     clipped_payload = llm_rate_limitter.clip_prompt(payload, token_limit)
     return await _analyze_file(ctx, clipped_payload)
+
+
+def read_many_files(paths: List[str]) -> str:
+    """
+    Read and return the content of multiple files.
+
+    This function is ideal for when you need to inspect the contents of
+    several files at once. For each file path provided in the input list,
+    it reads the entire file content. The result is a JSON string
+    containing a dictionary where keys are the file paths and values are
+    the corresponding file contents.
+
+    Use this tool when you need a comprehensive view of multiple files,
+    for example, to understand how different parts of a module interact,
+    to check configurations across various files, or to gather context
+    before making widespread changes.
+
+    Args:
+        paths (List[str]): A list of absolute or relative paths to the
+                           files you want to read. It is crucial to
+                           provide accurate paths. Use the `list_files`
+                           tool if you are unsure about the exact file
+                           locations.
+
+    Returns:
+        str: A JSON string representing a dictionary where each key is a
+             file path and the corresponding value is the content of that
+             file. If a file cannot be read, its entry in the dictionary
+             will contain an error message.
+             Example:
+             {
+                 "results": {
+                     "path/to/file1.py": "...",
+                     "path/to/file2.txt": "..."
+                 }
+             }
+    """
+    results = {}
+    for path in paths:
+        try:
+            abs_path = os.path.abspath(os.path.expanduser(path))
+            if not os.path.exists(abs_path):
+                raise FileNotFoundError(f"File not found: {path}")
+            content = read_file_with_line_numbers(abs_path)
+            results[path] = content
+        except Exception as e:
+            results[path] = f"Error reading file: {e}"
+    return json.dumps({"results": results})
+
+
+def write_many_files(files: Dict[str, str]) -> str:
+    """
+    Write content to multiple files simultaneously.
+
+    This function allows you to create, overwrite, or update multiple
+    files in a single operation. You provide a dictionary where each
+    key is a file path and the corresponding value is the content to be
+    written to that file. This is particularly useful for applying
+    changes across a project, such as refactoring code, updating
+    configuration files, or creating a set of new files from a template.
+
+    Each file is handled as a complete replacement of its content. If the
+    file does not exist, it will be created. If it already exists, its
+
+    entire content will be overwritten with the new content you provide.
+    Therefore, it is essential to provide the full, intended content for
+    each file.
+
+    Args:
+        files (Dict[str, str]): A dictionary where keys are the file paths
+                                (absolute or relative) and values are the
+                                complete contents to be written to those
+                                files.
+
+    Returns:
+        str: A JSON string summarizing the operation. It includes a list
+             of successfully written files and a list of files that
+             failed, along with the corresponding error messages.
+             Example:
+             {
+                 "success": [
+                     "path/to/file1.py",
+                     "path/to/file2.txt"
+                 ],
+                 "errors": {
+                     "path/to/problematic/file.py": "Error message"
+                 }
+             }
+    """
+    success = []
+    errors = {}
+    for path, content in files.items():
+        try:
+            abs_path = os.path.abspath(os.path.expanduser(path))
+            directory = os.path.dirname(abs_path)
+            if directory and not os.path.exists(directory):
+                os.makedirs(directory, exist_ok=True)
+            write_file(abs_path, content)
+            success.append(path)
+        except Exception as e:
+            errors[path] = f"Error writing file: {e}"
+    return json.dumps({"success": success, "errors": errors})

zrb/builtin/llm/tool/web.py

@@ -58,7 +58,7 @@ async def open_web_page(url: str) -> str:
 
 def create_search_internet_tool(serp_api_key: str) -> Callable[[str, int], str]:
     def search_internet(query: str, num_results: int = 10) -> str:
-        """Search the internet using SerpApi (Google Search) and return parsed results.
+        """Search the internet Google Search and return parsed results.
         Args:
             query (str): Search query.
             num_results (int): Search result count. Defaults to 10.

zrb/config.py

@@ -243,13 +243,19 @@ class Config:
 
     @property
     def LLM_MAX_REQUESTS_PER_MINUTE(self) -> int:
-        """Maximum number of LLM requests allowed per minute."""
-        return int(os.getenv("LLM_MAX_REQUESTS_PER_MINUTE", "60"))
+        """
+        Maximum number of LLM requests allowed per minute.
+        Default is conservative to accommodate free-tier LLM providers.
+        """
+        return int(os.getenv("LLM_MAX_REQUESTS_PER_MINUTE", "15"))
 
     @property
     def LLM_MAX_TOKENS_PER_MINUTE(self) -> int:
-        """Maximum number of LLM tokens allowed per minute."""
-        return int(os.getenv("ZRB_LLM_MAX_TOKENS_PER_MINUTE", "200000"))
+        """
+        Maximum number of LLM tokens allowed per minute.
+        Default is conservative to accommodate free-tier LLM providers.
+        """
+        return int(os.getenv("ZRB_LLM_MAX_TOKENS_PER_MINUTE", "100000"))
 
     @property
     def LLM_MAX_TOKENS_PER_REQUEST(self) -> int:
@@ -270,16 +276,16 @@ class Config:
         return to_boolean(os.getenv("ZRB_LLM_SUMMARIZE_HISTORY", "true"))
 
     @property
-    def LLM_HISTORY_SUMMARIZATION_THRESHOLD(self) -> int:
-        return int(os.getenv("ZRB_LLM_HISTORY_SUMMARIZATION_THRESHOLD", "5"))
+    def LLM_HISTORY_SUMMARIZATION_TOKEN_THRESHOLD(self) -> int:
+        return int(os.getenv("ZRB_LLM_HISTORY_SUMMARIZATION_TOKEN_THRESHOLD", "3000"))
 
     @property
     def LLM_ENRICH_CONTEXT(self) -> bool:
         return to_boolean(os.getenv("ZRB_LLM_ENRICH_CONTEXT", "true"))
 
     @property
-    def LLM_CONTEXT_ENRICHMENT_THRESHOLD(self) -> int:
-        return int(os.getenv("ZRB_LLM_CONTEXT_ENRICHMENT_THRESHOLD", "5"))
+    def LLM_CONTEXT_ENRICHMENT_TOKEN_THRESHOLD(self) -> int:
+        return int(os.getenv("ZRB_LLM_CONTEXT_ENRICHMENT_TOKEN_THRESHOLD", "3000"))
 
     @property
     def LLM_HISTORY_DIR(self) -> str:

zrb/llm_config.py

@@ -178,9 +178,9 @@ class LLMConfig:
         default_summarization_prompt: str | None = None,
         default_context_enrichment_prompt: str | None = None,
         default_summarize_history: bool | None = None,
-        default_history_summarization_threshold: int | None = None,
+        default_history_summarization_token_threshold: int | None = None,
         default_enrich_context: bool | None = None,
-        default_context_enrichment_threshold: int | None = None,
+        default_context_enrichment_token_threshold: int | None = None,
         default_model: "Model | None" = None,
         default_model_settings: "ModelSettings | None" = None,
         default_model_provider: "Provider | None" = None,
@@ -195,12 +195,12 @@ class LLMConfig:
         self._default_summarization_prompt = default_summarization_prompt
         self._default_context_enrichment_prompt = default_context_enrichment_prompt
         self._default_summarize_history = default_summarize_history
-        self._default_history_summarization_threshold = (
-            default_history_summarization_threshold
+        self._default_history_summarization_token_threshold = (
+            default_history_summarization_token_threshold
         )
         self._default_enrich_context = default_enrich_context
-        self._default_context_enrichment_threshold = (
-            default_context_enrichment_threshold
+        self._default_context_enrichment_token_threshold = (
+            default_context_enrichment_token_threshold
         )
         self._default_model_settings = default_model_settings
         self._default_model_provider = default_model_provider
@@ -316,10 +316,10 @@ class LLMConfig:
         return CFG.LLM_SUMMARIZE_HISTORY
 
     @property
-    def default_history_summarization_threshold(self) -> int:
-        if self._default_history_summarization_threshold is not None:
-            return self._default_history_summarization_threshold
-        return CFG.LLM_HISTORY_SUMMARIZATION_THRESHOLD
+    def default_history_summarization_token_threshold(self) -> int:
+        if self._default_history_summarization_token_threshold is not None:
+            return self._default_history_summarization_token_threshold
+        return CFG.LLM_HISTORY_SUMMARIZATION_TOKEN_THRESHOLD
 
     @property
     def default_enrich_context(self) -> bool:
@@ -328,10 +328,10 @@ class LLMConfig:
         return CFG.LLM_ENRICH_CONTEXT
 
     @property
-    def default_context_enrichment_threshold(self) -> int:
-        if self._default_context_enrichment_threshold is not None:
-            return self._default_context_enrichment_threshold
-        return CFG.LLM_CONTEXT_ENRICHMENT_THRESHOLD
+    def default_context_enrichment_token_threshold(self) -> int:
+        if self._default_context_enrichment_token_threshold is not None:
+            return self._default_context_enrichment_token_threshold
+        return CFG.LLM_CONTEXT_ENRICHMENT_TOKEN_THRESHOLD
 
     def set_default_persona(self, persona: str):
         self._default_persona = persona
@@ -369,18 +369,22 @@ class LLMConfig:
     def set_default_summarize_history(self, summarize_history: bool):
         self._default_summarize_history = summarize_history
 
-    def set_default_history_summarization_threshold(
-        self, history_summarization_threshold: int
+    def set_default_history_summarization_token_threshold(
+        self, history_summarization_token_threshold: int
     ):
-        self._default_history_summarization_threshold = history_summarization_threshold
+        self._default_history_summarization_token_threshold = (
+            history_summarization_token_threshold
+        )
 
     def set_default_enrich_context(self, enrich_context: bool):
         self._default_enrich_context = enrich_context
 
-    def set_default_context_enrichment_threshold(
-        self, context_enrichment_threshold: int
+    def set_default_context_enrichment_token_threshold(
+        self, context_enrichment_token_threshold: int
     ):
-        self._default_context_enrichment_threshold = context_enrichment_threshold
+        self._default_context_enrichment_token_threshold = (
+            context_enrichment_token_threshold
+        )
 
     def set_default_model_settings(self, model_settings: "ModelSettings"):
         self._default_model_settings = model_settings

zrb/task/llm/context_enrichment.py

@@ -5,7 +5,7 @@ from typing import TYPE_CHECKING
 from zrb.attr.type import BoolAttr, IntAttr
 from zrb.context.any_context import AnyContext
 from zrb.llm_config import llm_config
-from zrb.llm_rate_limitter import LLMRateLimiter
+from zrb.llm_rate_limitter import LLMRateLimiter, llm_rate_limitter
 from zrb.task.llm.agent import run_agent_iteration
 from zrb.task.llm.history import (
     count_part_in_history_list,
@@ -20,6 +20,12 @@ if TYPE_CHECKING:
     from pydantic_ai.settings import ModelSettings
 
 
+def _count_token_in_history(history_list: ListOfDict) -> int:
+    """Counts the total number of tokens in a conversation history list."""
+    text_to_count = json.dumps(history_list)
+    return llm_rate_limitter.count_token(text_to_count)
+
+
 async def enrich_context(
     ctx: AnyContext,
     model: "Model | str | None",
@@ -77,22 +83,22 @@ async def enrich_context(
     return previous_long_term_context
 
 
-def get_context_enrichment_threshold(
+def get_context_enrichment_token_threshold(
     ctx: AnyContext,
-    context_enrichment_threshold_attr: IntAttr | None,
-    render_context_enrichment_threshold: bool,
+    context_enrichment_token_threshold_attr: IntAttr | None,
+    render_context_enrichment_token_threshold: bool,
 ) -> int:
-    """Gets the context enrichment threshold, handling defaults and errors."""
+    """Gets the context enrichment token threshold, handling defaults and errors."""
     try:
         return get_int_attr(
             ctx,
-            context_enrichment_threshold_attr,
-            llm_config.default_context_enrichment_threshold,
-            auto_render=render_context_enrichment_threshold,
+            context_enrichment_token_threshold_attr,
+            llm_config.default_context_enrichment_token_threshold,
+            auto_render=render_context_enrichment_token_threshold,
         )
     except ValueError as e:
         ctx.log_warning(
-            f"Could not convert context_enrichment_threshold to int: {e}. "
+            f"Could not convert context_enrichment_token_threshold to int: {e}. "
             "Defaulting to -1 (no threshold)."
         )
         return -1
@@ -103,21 +109,25 @@ def should_enrich_context(
     history_list: ListOfDict,
     should_enrich_context_attr: BoolAttr | None,
     render_enrich_context: bool,
-    context_enrichment_threshold_attr: IntAttr | None,
-    render_context_enrichment_threshold: bool,
+    context_enrichment_token_threshold_attr: IntAttr | None,
+    render_context_enrichment_token_threshold: bool,
 ) -> bool:
     """
-    Determines if context enrichment should occur based on history, threshold, and config.
+    Determines if context enrichment should occur based on history, token threshold, and config.
     """
     history_part_count = count_part_in_history_list(history_list)
     if history_part_count == 0:
         return False
-    enrichment_threshold = get_context_enrichment_threshold(
+    enrichment_token_threshold = get_context_enrichment_token_threshold(
         ctx,
-        context_enrichment_threshold_attr,
-        render_context_enrichment_threshold,
+        context_enrichment_token_threshold_attr,
+        render_context_enrichment_token_threshold,
     )
-    if enrichment_threshold == -1 or enrichment_threshold > history_part_count:
+    history_token_count = _count_token_in_history(history_list)
+    if (
+        enrichment_token_threshold == -1
+        or enrichment_token_threshold > history_token_count
+    ):
         return False
     return get_bool_attr(
         ctx,
@@ -133,22 +143,22 @@ async def maybe_enrich_context(
     long_term_context: str,
     should_enrich_context_attr: BoolAttr | None,
     render_enrich_context: bool,
-    context_enrichment_threshold_attr: IntAttr | None,
-    render_context_enrichment_threshold: bool,
+    context_enrichment_token_threshold_attr: IntAttr | None,
+    render_context_enrichment_token_threshold: bool,
     model: "str | Model | None",
     model_settings: "ModelSettings | None",
     context_enrichment_prompt: str,
     rate_limitter: LLMRateLimiter | None = None,
 ) -> str:
-    """Enriches context based on history if enabled and threshold met."""
+    """Enriches context based on history if enabled and token threshold met."""
     shorten_history_list = replace_system_prompt_in_history_list(history_list)
     if should_enrich_context(
         ctx,
         shorten_history_list,
         should_enrich_context_attr,
         render_enrich_context,
-        context_enrichment_threshold_attr,
-        render_context_enrichment_threshold,
+        context_enrichment_token_threshold_attr,
+        render_context_enrichment_token_threshold,
     ):
         return await enrich_context(
             ctx=ctx,

zrb/task/llm/history_summarization.py

@@ -5,7 +5,7 @@ from typing import TYPE_CHECKING
 from zrb.attr.type import BoolAttr, IntAttr
 from zrb.context.any_context import AnyContext
 from zrb.llm_config import llm_config
-from zrb.llm_rate_limitter import LLMRateLimiter
+from zrb.llm_rate_limitter import LLMRateLimiter, llm_rate_limitter
 from zrb.task.llm.agent import run_agent_iteration
 from zrb.task.llm.history import (
     count_part_in_history_list,
@@ -20,22 +20,28 @@ if TYPE_CHECKING:
     from pydantic_ai.settings import ModelSettings
 
 
-def get_history_summarization_threshold(
+def _count_token_in_history(history_list: ListOfDict) -> int:
+    """Counts the total number of tokens in a conversation history list."""
+    text_to_count = json.dumps(history_list)
+    return llm_rate_limitter.count_token(text_to_count)
+
+
+def get_history_summarization_token_threshold(
     ctx: AnyContext,
-    history_summarization_threshold_attr: IntAttr | None,
-    render_history_summarization_threshold: bool,
+    history_summarization_token_threshold_attr: IntAttr | None,
+    render_history_summarization_token_threshold: bool,
 ) -> int:
-    """Gets the history summarization threshold, handling defaults and errors."""
+    """Gets the history summarization token threshold, handling defaults and errors."""
     try:
         return get_int_attr(
             ctx,
-            history_summarization_threshold_attr,
-            llm_config.default_history_summarization_threshold,
-            auto_render=render_history_summarization_threshold,
+            history_summarization_token_threshold_attr,
+            llm_config.default_history_summarization_token_threshold,
+            auto_render=render_history_summarization_token_threshold,
         )
     except ValueError as e:
         ctx.log_warning(
-            f"Could not convert history_summarization_threshold to int: {e}. "
+            f"Could not convert history_summarization_token_threshold to int: {e}. "
             "Defaulting to -1 (no threshold)."
         )
         return -1
@@ -46,19 +52,23 @@ def should_summarize_history(
     history_list: ListOfDict,
     should_summarize_history_attr: BoolAttr | None,
     render_summarize_history: bool,
-    history_summarization_threshold_attr: IntAttr | None,
-    render_history_summarization_threshold: bool,
+    history_summarization_token_threshold_attr: IntAttr | None,
+    render_history_summarization_token_threshold: bool,
 ) -> bool:
-    """Determines if history summarization should occur based on length and config."""
+    """Determines if history summarization should occur based on token length and config."""
     history_part_count = count_part_in_history_list(history_list)
     if history_part_count == 0:
         return False
-    summarization_threshold = get_history_summarization_threshold(
+    summarization_token_threshold = get_history_summarization_token_threshold(
         ctx,
-        history_summarization_threshold_attr,
-        render_history_summarization_threshold,
+        history_summarization_token_threshold_attr,
+        render_history_summarization_token_threshold,
     )
-    if summarization_threshold == -1 or summarization_threshold > history_part_count:
+    history_token_count = _count_token_in_history(history_list)
+    if (
+        summarization_token_threshold == -1
+        or summarization_token_threshold > history_token_count
+    ):
         return False
     return get_bool_attr(
         ctx,
@@ -102,7 +112,7 @@ async def summarize_history(
             history_list=[],
             rate_limitter=rate_limitter,
         )
-        if summary_run and summary_run.result.output:
+        if summary_run and summary_run.result and summary_run.result.output:
             new_summary = str(summary_run.result.output)
             usage = summary_run.result.usage()
             ctx.print(stylize_faint(f"📝 Summarization Token: {usage}"), plain=True)
@@ -126,8 +136,8 @@ async def maybe_summarize_history(
     conversation_summary: str,
     should_summarize_history_attr: BoolAttr | None,
     render_summarize_history: bool,
-    history_summarization_threshold_attr: IntAttr | None,
-    render_history_summarization_threshold: bool,
+    history_summarization_token_threshold_attr: IntAttr | None,
+    render_history_summarization_token_threshold: bool,
     model: "str | Model | None",
     model_settings: "ModelSettings | None",
     summarization_prompt: str,
@@ -140,8 +150,8 @@ async def maybe_summarize_history(
         shorten_history_list,
         should_summarize_history_attr,
         render_summarize_history,
-        history_summarization_threshold_attr,
-        render_history_summarization_threshold,
+        history_summarization_token_threshold_attr,
+        render_history_summarization_token_threshold,
     ):
         new_summary = await summarize_history(
             ctx=ctx,

zrb/task/llm_task.py

@@ -75,8 +75,8 @@ class LLMTask(BaseTask):
         render_enrich_context: bool = True,
         context_enrichment_prompt: StrAttr | None = None,
         render_context_enrichment_prompt: bool = True,
-        context_enrichment_threshold: IntAttr | None = None,
-        render_context_enrichment_threshold: bool = True,
+        context_enrichment_token_threshold: IntAttr | None = None,
+        render_context_enrichment_token_threshold: bool = True,
         tools: (
             list["ToolOrCallable"]
             | Callable[[AnySharedContext], list["ToolOrCallable"]]
@@ -102,8 +102,8 @@ class LLMTask(BaseTask):
         summarize_history: BoolAttr | None = None,
         render_summarize_history: bool = True,
         summarization_prompt: StrAttr | None = None,
-        history_summarization_threshold: IntAttr | None = None,
-        render_history_summarization_threshold: bool = True,
+        history_summarization_token_threshold: IntAttr | None = None,
+        render_history_summarization_token_threshold: bool = True,
         rate_limitter: LLMRateLimiter | None = None,
         execute_condition: bool | str | Callable[[AnySharedContext], bool] = True,
         retries: int = 2,
@@ -161,8 +161,10 @@ class LLMTask(BaseTask):
         self._render_enrich_context = render_enrich_context
         self._context_enrichment_prompt = context_enrichment_prompt
         self._render_context_enrichment_prompt = render_context_enrichment_prompt
-        self._context_enrichment_threshold = context_enrichment_threshold
-        self._render_context_enrichment_threshold = render_context_enrichment_threshold
+        self._context_enrichment_token_threshold = context_enrichment_token_threshold
+        self._render_context_enrichment_token_threshold = (
+            render_context_enrichment_token_threshold
+        )
         self._tools = tools
         self._rate_limitter = rate_limitter
         self._additional_tools: list["ToolOrCallable"] = []
@@ -175,9 +177,11 @@ class LLMTask(BaseTask):
         self._render_history_file = render_history_file
         self._should_summarize_history = summarize_history
         self._render_summarize_history = render_summarize_history
-        self._history_summarization_threshold = history_summarization_threshold
-        self._render_history_summarization_threshold = (
-            render_history_summarization_threshold
+        self._history_summarization_token_threshold = (
+            history_summarization_token_threshold
+        )
+        self._render_history_summarization_token_threshold = (
+            render_history_summarization_token_threshold
         )
         self._max_call_iteration = max_call_iteration
         self._conversation_context = conversation_context
@@ -199,14 +203,16 @@ class LLMTask(BaseTask):
     def set_should_enrich_context(self, enrich_context: bool):
         self._should_enrich_context = enrich_context
 
-    def set_context_enrichment_threshold(self, enrichment_threshold: int):
-        self._context_enrichment_threshold = enrichment_threshold
+    def set_context_enrichment_token_threshold(self, enrichment_token_threshold: int):
+        self._context_enrichment_token_threshold = enrichment_token_threshold
 
     def set_should_summarize_history(self, summarize_history: bool):
         self._should_summarize_history = summarize_history
 
-    def set_history_summarization_threshold(self, summarization_threshold: int):
-        self._history_summarization_threshold = summarization_threshold
+    def set_history_summarization_token_threshold(
+        self, summarization_token_threshold: int
+    ):
+        self._history_summarization_token_threshold = summarization_token_threshold
 
     async def _exec_action(self, ctx: AnyContext) -> Any:
         # Get dependent configurations first
@@ -255,8 +261,8 @@ class LLMTask(BaseTask):
             long_term_context=long_term_context,
             should_enrich_context_attr=self._should_enrich_context,
             render_enrich_context=self._render_enrich_context,
-            context_enrichment_threshold_attr=self._context_enrichment_threshold,
-            render_context_enrichment_threshold=self._render_context_enrichment_threshold,
+            context_enrichment_token_threshold_attr=self._context_enrichment_token_threshold,
+            render_context_enrichment_token_threshold=self._render_context_enrichment_token_threshold,
             model=model,
             model_settings=model_settings,
             context_enrichment_prompt=context_enrichment_prompt,
@@ -268,9 +274,9 @@ class LLMTask(BaseTask):
             conversation_summary=conversation_summary,
             should_summarize_history_attr=self._should_summarize_history,
             render_summarize_history=self._render_summarize_history,
-            history_summarization_threshold_attr=self._history_summarization_threshold,
-            render_history_summarization_threshold=(
-                self._render_history_summarization_threshold
+            history_summarization_token_threshold_attr=self._history_summarization_token_threshold,
+            render_history_summarization_token_threshold=(
+                self._render_history_summarization_token_threshold
             ),
             model=model,
             model_settings=model_settings,
@@ -358,3 +364,134 @@ class LLMTask(BaseTask):
         except Exception as e:
             ctx.log_error(f"Error during agent execution or history saving: {str(e)}")
             raise  # Re-raise the exception after logging
+
+
+def llm_task(
+    name: str,
+    color: int | None = None,
+    icon: str | None = None,
+    description: str | None = None,
+    cli_only: bool = False,
+    input: list[AnyInput | None] | AnyInput | None = None,
+    env: list[AnyEnv | None] | AnyEnv | None = None,
+    model: "Callable[[AnySharedContext], Model | str | fstring] | Model | None" = None,
+    render_model: bool = True,
+    model_base_url: StrAttr | None = None,
+    render_model_base_url: bool = True,
+    model_api_key: StrAttr | None = None,
+    render_model_api_key: bool = True,
+    model_settings: "ModelSettings | Callable[[AnySharedContext], ModelSettings] | None" = None,
+    agent: "Agent | Callable[[AnySharedContext], Agent] | None" = None,
+    persona: StrAttr | None = None,
+    system_prompt: StrAttr | None = None,
+    special_instruction_prompt: StrAttr | None = None,
+    message: StrAttr | None = None,
+    render_message: bool = True,
+    enrich_context: BoolAttr | None = None,
+    render_enrich_context: bool = True,
+    context_enrichment_prompt: StrAttr | None = None,
+    render_context_enrichment_prompt: bool = True,
+    context_enrichment_token_threshold: IntAttr | None = None,
+    render_context_enrichment_token_threshold: bool = True,
+    tools: (
+        list["ToolOrCallable"] | Callable[[AnySharedContext], list["ToolOrCallable"]]
+    ) = [],
+    mcp_servers: (
+        list["MCPServer"] | Callable[[AnySharedContext], list["MCPServer"]]
+    ) = [],
+    conversation_history: (
+        ConversationHistoryData
+        | Callable[[AnySharedContext], ConversationHistoryData | dict | list]
+        | dict
+        | list
+    ) = ConversationHistoryData(),
+    conversation_history_reader: (
+        Callable[[AnySharedContext], ConversationHistoryData | dict | list | None]
+        | None
+    ) = None,
+    conversation_history_writer: (
+        Callable[[AnySharedContext, ConversationHistoryData], None] | None
+    ) = None,
+    conversation_history_file: StrAttr | None = None,
+    render_history_file: bool = True,
+    summarize_history: BoolAttr | None = None,
+    render_summarize_history: bool = True,
+    summarization_prompt: StrAttr | None = None,
+    history_summarization_token_threshold: IntAttr | None = None,
+    render_history_summarization_token_threshold: bool = True,
+    rate_limitter: LLMRateLimiter | None = None,
+    execute_condition: bool | str | Callable[[AnySharedContext], bool] = True,
+    retries: int = 2,
+    retry_period: float = 0,
+    readiness_check: list[AnyTask] | AnyTask | None = None,
+    readiness_check_delay: float = 0.5,
+    readiness_check_period: float = 5,
+    readiness_failure_threshold: int = 1,
+    readiness_timeout: int = 60,
+    monitor_readiness: bool = False,
+    max_call_iteration: int = 20,
+    upstream: list[AnyTask] | AnyTask | None = None,
+    fallback: list[AnyTask] | AnyTask | None = None,
+    successor: list[AnyTask] | AnyTask | None = None,
+    conversation_context: (
+        dict[str, Any] | Callable[[AnySharedContext], dict[str, Any]] | None
+    ) = None,
+) -> LLMTask:
+    """
+    Create a new LLM task.
+    """
+    return LLMTask(
+        name=name,
+        color=color,
+        icon=icon,
+        description=description,
+        cli_only=cli_only,
+        input=input,
+        env=env,
+        model=model,
+        render_model=render_model,
+        model_base_url=model_base_url,
+        render_model_base_url=render_model_base_url,
+        model_api_key=model_api_key,
+        render_model_api_key=render_model_api_key,
+        model_settings=model_settings,
+        agent=agent,
+        persona=persona,
+        system_prompt=system_prompt,
+        special_instruction_prompt=special_instruction_prompt,
+        message=message,
+        render_message=render_message,
+        enrich_context=enrich_context,
+        render_enrich_context=render_enrich_context,
+        context_enrichment_prompt=context_enrichment_prompt,
+        render_context_enrichment_prompt=render_context_enrichment_prompt,
+        context_enrichment_token_threshold=context_enrichment_token_threshold,
+        render_context_enrichment_token_threshold=render_context_enrichment_token_threshold,
+        tools=tools,
+        mcp_servers=mcp_servers,
+        conversation_history=conversation_history,
+        conversation_history_reader=conversation_history_reader,
+        conversation_history_writer=conversation_history_writer,
+        conversation_history_file=conversation_history_file,
+        render_history_file=render_history_file,
+        summarize_history=summarize_history,
+        render_summarize_history=render_summarize_history,
+        summarization_prompt=summarization_prompt,
+        history_summarization_token_threshold=history_summarization_token_threshold,
+        render_history_summarization_token_threshold=render_history_summarization_token_threshold,
+        rate_limitter=rate_limitter,
+        execute_condition=execute_condition,
+        retries=retries,
+        retry_period=retry_period,
+        readiness_check=readiness_check,
+        readiness_check_delay=readiness_check_delay,
+        readiness_check_period=readiness_check_period,
+        readiness_failure_threshold=readiness_failure_threshold,
+        readiness_timeout=readiness_timeout,
+        monitor_readiness=monitor_readiness,
+        max_call_iteration=max_call_iteration,
+        upstream=upstream,
+        fallback=fallback,
+        successor=successor,
+        conversation_context=conversation_context,
+    )

zrb/util/file.py

@@ -31,9 +31,10 @@ def _read_text_file_content(file_path: str) -> str:
 
 
 def _read_pdf_file_content(file_path: str) -> str:
-    from pdfplumber.pdf import PDF, open
+    import pdfplumber
+    from pdfplumber.pdf import PDF
 
-    with open(file_path) as pdf:
+    with pdfplumber.open(file_path) as pdf:
         pdf: PDF
         return "\n".join(
             page.extract_text() for page in pdf.pages if page.extract_text()

{zrb-1.9.0.dist-info → zrb-1.9.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: zrb
-Version: 1.9.0
+Version: 1.9.1
 Summary: Your Automation Powerhouse
 Home-page: https://github.com/state-alchemists/zrb
 License: AGPL-3.0-or-later

{zrb-1.9.0.dist-info → zrb-1.9.1.dist-info}/RECORD

@@ -12,16 +12,16 @@ zrb/builtin/jwt.py,sha256=3M5uaQhJZbKQLjTUft1OwPz_JxtmK-xtkjxWjciOQho,2859
 zrb/builtin/llm/chat_session.py,sha256=iDNHbNX268NWl0uQmjecD3Bn686-mByqqXuMc6IKjOE,7146
 zrb/builtin/llm/history.py,sha256=cnkOyO43uiMQ9cEvmqk-pPoCk1zCAH_fwAqSgBtsjzY,3079
 zrb/builtin/llm/input.py,sha256=Nw-26uTWp2QhUgKJcP_IMHmtk-b542CCSQ_vCOjhvhM,877
-zrb/builtin/llm/llm_ask.py,sha256=QUV29gOAFKiMfJlAKbY9YfGPoxYv-4RPv6p7cWogK4U,4438
+zrb/builtin/llm/llm_ask.py,sha256=A7MmDybKSdlO0T-Wted9t1tEtw6ULjV5aG4SAwpkt-w,4532
 zrb/builtin/llm/previous-session.js,sha256=xMKZvJoAbrwiyHS0OoPrWuaKxWYLoyR5sguePIoCjTY,816
 zrb/builtin/llm/tool/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 zrb/builtin/llm/tool/api.py,sha256=yR9I0ZsI96OeQl9pgwORMASVuXsAL0a89D_iPS4C8Dc,1699
 zrb/builtin/llm/tool/cli.py,sha256=_CNEmEc6K2Z0i9ppYeM7jGpqaEdT3uxaWQatmxP3jKE,858
-zrb/builtin/llm/tool/code.py,sha256=q6YrVJkRJg4AQpnK2KHE6AEMo8nMbRN4XUZ3QtMI_Og,8090
-zrb/builtin/llm/tool/file.py,sha256=ufLCAaHB0JkEAqQS4fbM9OaTfLluqlCuSyMmnYhI0rY,18491
+zrb/builtin/llm/tool/code.py,sha256=t_0-D1ToSm3ATYGiHfkHBpvkRGMnrmwWSYk_M73AVXs,8090
+zrb/builtin/llm/tool/file.py,sha256=z0_RihenF09WtJqAoKLMeVmkqTS7Z5W04lwkKbyfESo,22683
 zrb/builtin/llm/tool/rag.py,sha256=yqx7vXXyrOCJjhQJl4s0TnLL-2uQUTuKRnkWlSQBW0M,7883
 zrb/builtin/llm/tool/sub_agent.py,sha256=Xz_nwNA8hW52gCeNdNBT7TXDUU3x7Ube-t17FYxx0-E,4671
-zrb/builtin/llm/tool/web.py,sha256=GYp6e_eaw-dj7MDpB4CP1fplUbfguuJawem9lPJM9TY,5481
+zrb/builtin/llm/tool/web.py,sha256=SeTT59yRaBJQILi4TETaLDMna2ooeYStYMkAnn0uTZc,5465
 zrb/builtin/md5.py,sha256=690RV2LbW7wQeTFxY-lmmqTSVEEZv3XZbjEUW1Q3XpE,1480
 zrb/builtin/project/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 zrb/builtin/project/add/fastapp/fastapp_input.py,sha256=MKlWR_LxWhM_DcULCtLfL_IjTxpDnDBkn9KIqNmajFs,310
@@ -217,7 +217,7 @@ zrb/callback/callback.py,sha256=PFhCqzfxdk6IAthmXcZ13DokT62xtBzJr_ciLw6I8Zg,4030
 zrb/cmd/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 zrb/cmd/cmd_result.py,sha256=L8bQJzWCpcYexIxHBNsXj2pT3BtLmWex0iJSMkvimOA,597
 zrb/cmd/cmd_val.py,sha256=7Doowyg6BK3ISSGBLt-PmlhzaEkBjWWm51cED6fAUOQ,1014
-zrb/config.py,sha256=L4nrdO8uK1MNOJiTgywkjlYn5kX1dOsq48b5mPZ2004,10364
+zrb/config.py,sha256=TRVQgUgX9yIso0Z6GioX8HASX_tITv5m_AP0Z4UIWb4,10574
 zrb/content_transformer/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 zrb/content_transformer/any_content_transformer.py,sha256=v8ZUbcix1GGeDQwB6OKX_1TjpY__ksxWVeqibwa_iZA,850
 zrb/content_transformer/content_transformer.py,sha256=STl77wW-I69QaGzCXjvkppngYFLufow8ybPLSyAvlHs,2404
@@ -246,7 +246,7 @@ zrb/input/option_input.py,sha256=TQB82ko5odgzkULEizBZi0e9TIHEbIgvdP0AR3RhA74,213
 zrb/input/password_input.py,sha256=szBojWxSP9QJecgsgA87OIYwQrY2AQ3USIKdDZY6snU,1465
 zrb/input/str_input.py,sha256=NevZHX9rf1g8eMatPyy-kUX3DglrVAQpzvVpKAzf7bA,81
 zrb/input/text_input.py,sha256=6T3MngWdUs0u0ZVs5Dl11w5KS7nN1RkgrIR_zKumzPM,3695
-zrb/llm_config.py,sha256=s-8stT_9lsH35IQuE6YBdLbi2actYgwLDFqEqgedtCE,16964
+zrb/llm_config.py,sha256=FaX_v_EeZoXoZtwFpjqpvPHkxyaeqAziwcu-zb8lGXM,17144
 zrb/llm_rate_limitter.py,sha256=uM9zmSgV10fQq1dlaDGLDrv72uLj6ldBxMoGjO2Az14,4429
 zrb/runner/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 zrb/runner/cli.py,sha256=y_n7O4kQKXpiLAsBsf1WnHNfWgKlrxE9TU0MU1ICUMg,6989
@@ -340,15 +340,15 @@ zrb/task/llm/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 zrb/task/llm/agent.py,sha256=IUGhU7vSfGVpVbd0rGLJ9kxIpt04Fm6UBjBWCVBVaZs,6695
 zrb/task/llm/config.py,sha256=WsO_kLXi0JXB6SPNL_dWm1Izu04KRrzmuX6DSrRYeBE,3483
 zrb/task/llm/context.py,sha256=LGGQ_mb1dWorfshHjGgXEW_pRweGj-6MZcIUFq3AHy4,2213
-zrb/task/llm/context_enrichment.py,sha256=NrRsKLY2Kojx2lejdEJxORgz-7uiQ1_A_IAyHJ3WVPQ,5552
+zrb/task/llm/context_enrichment.py,sha256=MBs5lkI1GFZwtW_P7fJfIc6CWcAS0UWQeBO6ggSBSk0,6033
 zrb/task/llm/error.py,sha256=s5PSK3ibGupMzOM0P81nstHWrMr3205H0FSDwfhWUDA,3662
 zrb/task/llm/history.py,sha256=cpaNqoEzsNAgzZGPPdohA8U5nTrVWmZ3P1Y5RTtXDgc,7986
-zrb/task/llm/history_summarization.py,sha256=-A7oLLHYYxlYL6RQYhaL_yu2yT1qlB9s8kAWGNqPWfU,5534
+zrb/task/llm/history_summarization.py,sha256=TejPLktJmlxvhM-UmC29m_g6U9bC5e4fxRSEwt22GN0,6032
 zrb/task/llm/print_node.py,sha256=VB_ZXD3iQPZcaNWvIMNRxOhmXIJp9__5utRrSbiIRXo,4457
 zrb/task/llm/prompt.py,sha256=obSsLHF4AHlwMIEQ8XTJmtEnkLlD5XTEkk0RJUmIz4Y,3410
 zrb/task/llm/tool_wrapper.py,sha256=8_bL8m_WpRf-pVKSrvQIVqT-m2sUA87a1RBQG13lhp4,6457
 zrb/task/llm/typing.py,sha256=c8VAuPBw_4A3DxfYdydkgedaP-LU61W9_wj3m3CAX1E,58
-zrb/task/llm_task.py,sha256=-1-QV3wN70CgGSdtAj4F5QR9wMQSLsFJcUcFkeiCoWs,15362
+zrb/task/llm_task.py,sha256=qNXS078y2s6iF50OoVH3-7UWmV5phgfY2_NiZ9KxGSU,21055
 zrb/task/make_task.py,sha256=PD3b_aYazthS8LHeJsLAhwKDEgdurQZpymJDKeN60u0,2265
 zrb/task/rsync_task.py,sha256=GSL9144bmp6F0EckT6m-2a1xG25AzrrWYzH4k3SVUKM,6370
 zrb/task/scaffolder.py,sha256=rME18w1HJUHXgi9eTYXx_T2G4JdqDYzBoNOkdOOo5-o,6806
@@ -376,7 +376,7 @@ zrb/util/codemod/modify_function_call.py,sha256=wbyoRRsM4V9fPYkT5kHN0zpBetpRDq2S
 zrb/util/codemod/modify_method.py,sha256=5fioXjqNQmrf4CV2qlTZHpViF9PMnNer4FvuKam7byo,6344
 zrb/util/codemod/modify_module.py,sha256=2mzi_NxJ-kNFo5G0U_Rqb3JoYQl1s6Izt7b_wAl10F0,715
 zrb/util/cron.py,sha256=UWqqhhM7DDTPx6wjCIdBndnVh3NRu-sdKazp8aZkXh8,3833
-zrb/util/file.py,sha256=RJosRqX63iwJ-TLx0yzJ4pJ97evKR-oxxs4r-v5chq0,2779
+zrb/util/file.py,sha256=tm_8qn0vEM8Hz46yXUcvFHfsLtQNqidQaEuB85xqhFE,2806
 zrb/util/git.py,sha256=gS_Y9sQgJbY0PfgSQiowLvV3Nf0y9C8nT3j6z6oEsG8,8186
 zrb/util/git_subtree.py,sha256=E_UB5OIgm8WkHL9beifRxpZ25_BB9p1H578OhLZTgRU,4611
 zrb/util/group.py,sha256=T82yr3qg9I5k10VPXkMyrIRIqyfzadSH813bqzwKEPI,4718
@@ -391,7 +391,7 @@ zrb/util/todo.py,sha256=r9_KYF2-hLKMNjsp6AFK9zivykMrywd-kJ4bCwfdafI,19323
 zrb/util/todo_model.py,sha256=0SJ8aLYfJAscDOk5JsH7pXP3h1rAG91VMCS20-c2Y6A,1576
 zrb/xcom/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 zrb/xcom/xcom.py,sha256=o79rxR9wphnShrcIushA0Qt71d_p3ZTxjNf7x9hJB78,1571
-zrb-1.9.0.dist-info/METADATA,sha256=2j4pOOx_l_wmHxFKn4gVX6wCeI9AOqmKszZ7PASZzSI,9321
-zrb-1.9.0.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-zrb-1.9.0.dist-info/entry_points.txt,sha256=-Pg3ElWPfnaSM-XvXqCxEAa-wfVI6BEgcs386s8C8v8,46
-zrb-1.9.0.dist-info/RECORD,,
+zrb-1.9.1.dist-info/METADATA,sha256=ohOmILh0PYQYiYzfuNy3dfuqNT7UorikVrHQO5jIE8I,9321
+zrb-1.9.1.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+zrb-1.9.1.dist-info/entry_points.txt,sha256=-Pg3ElWPfnaSM-XvXqCxEAa-wfVI6BEgcs386s8C8v8,46
+zrb-1.9.1.dist-info/RECORD,,