janus-llm 3.5.2__tar.gz → 3.5.3__tar.gz

{janus_llm-3.5.2 → janus_llm-3.5.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: janus-llm
-Version: 3.5.2
+Version: 3.5.3
 Summary: A transcoding library using LLMs.
 Home-page: https://github.com/janus-llm/janus-llm
 License: Apache 2.0

{janus_llm-3.5.2 → janus_llm-3.5.3}/janus/__init__.py

@@ -5,7 +5,7 @@ from langchain_core._api.deprecation import LangChainDeprecationWarning
 from janus.converter.translate import Translator
 from janus.metrics import *  # noqa: F403
 
-__version__ = "3.5.2"
+__version__ = "3.5.3"
 
 # Ignoring a deprecation warning from langchain_core that I can't seem to hunt down
 warnings.filterwarnings("ignore", category=LangChainDeprecationWarning)

{janus_llm-3.5.2 → janus_llm-3.5.3}/janus/cli.py

@@ -1,6 +1,7 @@
 import json
 import logging
 import os
+import subprocess  # nosec
 from pathlib import Path
 from typing import List, Optional
 
@@ -1156,5 +1157,33 @@ app.add_typer(evaluate, name="evaluate")
 app.add_typer(embedding, name="embedding")
 
 
+@app.command()
+def render(
+    input_dir: Annotated[
+        str,
+        typer.Option(
+            "--input",
+            "-i",
+        ),
+    ],
+    output_dir: Annotated[str, typer.Option("--output", "-o")],
+):
+    input_dir = Path(input_dir)
+    output_dir = Path(output_dir)
+    for input_file in input_dir.rglob("*.json"):
+        with open(input_file, "r") as f:
+            data = json.load(f)
+        input_tail = input_file.relative_to(input_dir)
+        output_file = output_dir / input_tail
+        output_file = output_file.with_suffix(".txt")
+        if not output_file.parent.exists():
+            output_file.parent.mkdir()
+        with open(output_file, "w") as f:
+            f.write(data["output"])
+        jar_path = homedir / ".janus/lib/plantuml.jar"
+        subprocess.run(["java", "-jar", jar_path, output_file])  # nosec
+        output_file.unlink()
+
+
 if __name__ == "__main__":
     app()

{janus_llm-3.5.2 → janus_llm-3.5.3}/janus/language/alc/alc.py

@@ -63,7 +63,7 @@ class AlcSplitter(TreeSitterSplitter):
             #  instruction and containing all the subsequent nodes up until the
             #  next csect or dsect instruction
             sects: list[list[CodeBlock]] = [[]]
-            for c in block.children:
+            for c in sorted(block.children):
                 if c.node_type == "csect_instruction":
                     c.context_tags["alc_section"] = "CSECT"
                     sects.append([c])

{janus_llm-3.5.2 → janus_llm-3.5.3}/janus/language/naive/simple_ast.py

@@ -19,6 +19,7 @@ def get_flexible_ast(language: str, **kwargs) -> Splitter:
     Returns:
         A flexible AST splitter for the given language.
     """
+    kwargs.update(protected_node_types=())
     if language == "ibmhlasm":
         return AlcSplitter(**kwargs)
     elif language == "mumps":
@@ -28,7 +29,7 @@ def get_flexible_ast(language: str, **kwargs) -> Splitter:
 
 
 @register_splitter("ast-strict")
-def get_strict_ast(language: str, **kwargs) -> Splitter:
+def get_strict_ast(language: str, prune_unprotected=True, **kwargs) -> Splitter:
     """Get a strict AST splitter for the given language.
 
     The strict splitter will only return nodes that are of a functional type.
@@ -41,7 +42,7 @@ def get_strict_ast(language: str, **kwargs) -> Splitter:
     """
     kwargs.update(
         protected_node_types=LANGUAGES[language]["functional_node_types"],
-        prune_unprotected=True,
+        prune_unprotected=prune_unprotected,
     )
     if language == "ibmhlasm":
         return AlcSplitter(**kwargs)

{janus_llm-3.5.2 → janus_llm-3.5.3}/janus/language/splitter.py

@@ -387,7 +387,10 @@ class Splitter(FileManager):
             return
 
         if self._is_protected(node):
-            raise TokenLimitError(r"Irreducible node too large for context!")
+            log.error(
+                "Protected node too large for context!"
+                f" ({node.tokens} > {self.max_tokens})"
+            )
 
         if node.children:
             for child in node.children:
@@ -423,7 +426,7 @@ class Splitter(FileManager):
             name = f"{node.name}-L#{node_line}"
             tokens = self._count_tokens(line)
             if tokens > self.max_tokens:
-                raise TokenLimitError(
+                log.error(
                     "Irreducible node too large for context!"
                     f" ({tokens} > {self.max_tokens})"
                 )

{janus_llm-3.5.2 → janus_llm-3.5.3}/janus/llm/models_info.py

@@ -1,5 +1,6 @@
 import json
 import os
+import time
 from pathlib import Path
 from typing import Any, Callable
 
@@ -247,6 +248,12 @@ def load_model(
     model_args = model_config["model_args"]
     if model_config["model_type"] == "OpenAI":
         model_args.update(_open_ai_defaults)
+        log.warning("Do NOT use this model in sensitive environments!")
+        log.warning("If you would like to cancel, please press Ctrl+C.")
+        log.warning("Waiting 10 seconds...")
+        # Give enough time for the user to read the warnings and cancel
+        time.sleep(10)
+
     model = model_constructor(**model_args)
     return (
         model,

{janus_llm-3.5.2 → janus_llm-3.5.3}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "janus-llm"
-version = "3.5.2"
+version = "3.5.3"
 description = "A transcoding library using LLMs."
 authors = ["Michael Doyle <mdoyle@mitre.org>", "Chris Glasz <cglasz@mitre.org>",
            "Chris Tohline <ctohline@mitre.org>", "William Macke <wmacke@mitre.org>",