|
| 1 | +"""PII masking for the analyze-files tool. |
| 2 | +
|
| 3 | +Encapsulates the policy evaluation, PII detection request, and rehydration of |
| 4 | +masked LLM output behind a single :class:`PiiMasker` class. |
| 5 | +""" |
| 6 | + |
| 7 | +import logging |
| 8 | +from typing import Any |
| 9 | + |
| 10 | +from uipath.core.feature_flags import FeatureFlags |
| 11 | +from uipath.platform import UiPath |
| 12 | +from uipath.platform.semantic_proxy import ( |
| 13 | + PiiDetectionRequest, |
| 14 | + PiiDetectionResponse, |
| 15 | + PiiDocument, |
| 16 | + PiiEntityThreshold, |
| 17 | + PiiFile, |
| 18 | + rehydrate_from_pii_response, |
| 19 | +) |
| 20 | + |
| 21 | +from uipath_langchain.agent.multimodal import FileInfo |
| 22 | + |
| 23 | +logger = logging.getLogger("uipath") |
| 24 | + |
| 25 | +_FEATURE_FLAG = "FilePiiMaskingEnabled" |
| 26 | + |
| 27 | + |
| 28 | +class PiiMasker: |
| 29 | + """Runs PII detection against prompts/files and rehydrates masked LLM output. |
| 30 | +
|
| 31 | + Two gates (both must allow) control whether masking runs: |
| 32 | +
|
| 33 | + 1. Opt-in kill-switch — the ``FilePiiMaskingEnabled`` feature flag |
| 34 | + (defaults to ``False``; enable via ``FeatureFlags.configure_flags`` |
| 35 | + or the ``UIPATH_FEATURE_FilePiiMaskingEnabled`` env var). |
| 36 | + 2. Platform policy — ``data.container.pii-in-flight-agents`` from the |
| 37 | + AutomationOps deployed-policy response. |
| 38 | + """ |
| 39 | + |
| 40 | + def __init__(self, client: UiPath, policy: dict[str, Any] | None) -> None: |
| 41 | + self._client = client |
| 42 | + self._policy = policy |
| 43 | + self._result: PiiDetectionResponse | None = None |
| 44 | + |
| 45 | + @staticmethod |
| 46 | + def is_policy_enabled(policy: dict[str, Any] | None) -> bool: |
| 47 | + """Return True when both the feature flag and platform policy allow masking.""" |
| 48 | + flag_enabled = FeatureFlags.is_flag_enabled(_FEATURE_FLAG, default=False) |
| 49 | + logger.info("PII masking feature flag %s=%s", _FEATURE_FLAG, flag_enabled) |
| 50 | + if not flag_enabled: |
| 51 | + return False |
| 52 | + if not policy: |
| 53 | + return False |
| 54 | + container = policy.get("data", {}).get("container", {}) |
| 55 | + return bool(container.get("pii-in-flight-agents", False)) |
| 56 | + |
| 57 | + async def apply( |
| 58 | + self, analysis_task: str, files: list[FileInfo] |
| 59 | + ) -> tuple[str, list[FileInfo]]: |
| 60 | + """Run PII detection and return the masked prompt and redacted files. |
| 61 | +
|
| 62 | + The underlying detection response is retained so the LLM output can be |
| 63 | + rehydrated later via :meth:`rehydrate`. |
| 64 | + """ |
| 65 | + request = PiiDetectionRequest( |
| 66 | + documents=[ |
| 67 | + PiiDocument(id="user-prompt", role="user", document=analysis_task) |
| 68 | + ], |
| 69 | + files=[ |
| 70 | + PiiFile( |
| 71 | + file_name=f.name, |
| 72 | + file_url=f.url, |
| 73 | + file_type=f.name.rsplit(".", 1)[-1].lower() |
| 74 | + if "." in f.name |
| 75 | + else "", |
| 76 | + ) |
| 77 | + for f in files |
| 78 | + ], |
| 79 | + entity_thresholds=self._entity_thresholds_from_policy() or None, |
| 80 | + ) |
| 81 | + self._result = await self._client.semantic_proxy.detect_pii_async(request) |
| 82 | + logger.info( |
| 83 | + "PII detection completed: %d document entities, %d file entities", |
| 84 | + sum(len(d.pii_entities) for d in self._result.response), |
| 85 | + sum(len(f.pii_entities) for f in self._result.files), |
| 86 | + ) |
| 87 | + |
| 88 | + masked_prompt = analysis_task |
| 89 | + for doc in self._result.response: |
| 90 | + if doc.id == "user-prompt": |
| 91 | + if doc.masked_document != analysis_task: |
| 92 | + logger.info( |
| 93 | + "User prompt masked (%d entities replaced)", |
| 94 | + len(doc.pii_entities), |
| 95 | + ) |
| 96 | + masked_prompt = doc.masked_document |
| 97 | + break |
| 98 | + |
| 99 | + redacted_by_name = {f.file_name: f.file_url for f in self._result.files} |
| 100 | + if redacted_by_name: |
| 101 | + masked_files = [ |
| 102 | + self._rename_for_masking(f, redacted_by_name.get(f.name, f.url)) |
| 103 | + for f in files |
| 104 | + ] |
| 105 | + logger.info("Renamed %d file(s) with pii_masked_ prefix", len(masked_files)) |
| 106 | + else: |
| 107 | + masked_files = files |
| 108 | + |
| 109 | + return masked_prompt, masked_files |
| 110 | + |
| 111 | + def rehydrate(self, text: str) -> str: |
| 112 | + """Replace masked placeholders in ``text`` with the original PII values. |
| 113 | +
|
| 114 | + Returns ``text`` unchanged if :meth:`apply` hasn't been called. |
| 115 | + """ |
| 116 | + if self._result is None: |
| 117 | + return text |
| 118 | + rehydrated = rehydrate_from_pii_response(text, self._result) |
| 119 | + if rehydrated != text: |
| 120 | + logger.info("Rehydrated LLM response with PII entities") |
| 121 | + return rehydrated |
| 122 | + |
| 123 | + def _entity_thresholds_from_policy(self) -> list[PiiEntityThreshold]: |
| 124 | + """Extract enabled entity thresholds from the policy's ``pii-entity-table``.""" |
| 125 | + if not self._policy: |
| 126 | + return [] |
| 127 | + table = self._policy.get("data", {}).get("pii-entity-table", []) |
| 128 | + thresholds: list[PiiEntityThreshold] = [] |
| 129 | + for entry in table: |
| 130 | + if not entry.get("pii-entity-is-enabled", False): |
| 131 | + continue |
| 132 | + category = entry.get("pii-entity-category") |
| 133 | + confidence = entry.get("pii-entity-confidence-threshold") |
| 134 | + if category is None or confidence is None: |
| 135 | + continue |
| 136 | + thresholds.append( |
| 137 | + PiiEntityThreshold( |
| 138 | + category=category, |
| 139 | + confidence_threshold=confidence, |
| 140 | + ) |
| 141 | + ) |
| 142 | + return thresholds |
| 143 | + |
| 144 | + @staticmethod |
| 145 | + def _rename_for_masking(file: FileInfo, redacted_url: str) -> FileInfo: |
| 146 | + """Return a FileInfo pointing at ``redacted_url`` with a ``pii_masked_`` prefix.""" |
| 147 | + if "." in file.name: |
| 148 | + base, ext = file.name.rsplit(".", 1) |
| 149 | + new_name = f"pii_masked_{base}.{ext}" |
| 150 | + else: |
| 151 | + new_name = f"pii_masked_{file.name}" |
| 152 | + return FileInfo(url=redacted_url, name=new_name, mime_type=file.mime_type) |
0 commit comments