update: isolate presidio dependencies

guardrails_genie/guardrails/entity_recognition/__init__.py

@@ -1,5 +1,14 @@
+import warnings
+
 from .llm_judge_entity_recognition_guardrail import RestrictedTermsJudge
-from .presidio_entity_recognition_guardrail import PresidioEntityRecognitionGuardrail
+
+try:
+    from .presidio_entity_recognition_guardrail import PresidioEntityRecognitionGuardrail
+except ImportError:
+    warnings.warn(
+        "Presidio is not installed. You can install it using `pip install -e .[presidio]`"
+    )
+
 from .regex_entity_recognition_guardrail import RegexEntityRecognitionGuardrail
 from .transformers_entity_recognition_guardrail import (
     TransformersEntityRecognitionGuardrail,

guardrails_genie/utils.py

@@ -1,18 +1,5 @@
-import os
-
 import pandas as pd
-import pymupdf4llm
 import weave
-import weave.trace
-from firerequests import FireRequests
-
-
-@weave.op()
-def get_markdown_from_pdf_url(url: str) -> str:
-    FireRequests().download(url, "temp.pdf", show_progress=False)
-    markdown = pymupdf4llm.to_markdown("temp.pdf", show_progress=False)
-    os.remove("temp.pdf")
-    return markdown
 
 
 class EvaluationCallManager:

pyproject.toml

@@ -9,25 +9,27 @@ dependencies = [
     "evaluate>=0.4.3",
     "google-generativeai>=0.8.3",
     "openai>=1.52.2",
-    "isort>=5.13.2",
-    "black>=24.10.0",
-    "ruff>=0.6.9",
-    "pip>=24.2",
-    "uv>=0.4.20",
     "weave @ git+https://github.com/wandb/weave@feat/eval-progressbar",
     "streamlit>=1.40.1",
     "python-dotenv>=1.0.1",
     "watchdog>=6.0.0",
-    "firerequests>=0.1.1",
-    "pymupdf4llm>=0.0.17",
     "transformers>=4.46.3",
     "torch>=2.5.1",
-    "presidio-analyzer>=2.2.355",
-    "presidio-anonymizer>=2.2.355",
     "instructor>=1.7.0",
 ]
 
 [project.optional-dependencies]
+presidio = [
+    "presidio-analyzer>=2.2.355",
+    "presidio-anonymizer>=2.2.355",
+]
+dev = [
+    "isort>=5.13.2",
+    "black>=24.10.0",
+    "ruff>=0.6.9",
+    "pip>=24.2",
+    "uv>=0.4.20",
+]
 docs = [
     "mkdocs>=1.6.1",
     "mkdocstrings>=0.26.1",