1037 lines
38 KiB
Python
1037 lines
38 KiB
Python
"""Local privacy gateway. Personal generative egress is not allowed to skip this layer."""
|
|
from __future__ import annotations
|
|
|
|
from dataclasses import dataclass, field
|
|
from datetime import datetime, timezone
|
|
from typing import Any
|
|
import contextvars
|
|
import re
|
|
import time
|
|
|
|
from entity_detect import DetectError, detect_personal_egress, reset_detect_test_hooks, user_detect_message
|
|
from identity_store import KINSHIP, is_maskable_label
|
|
from journal_reconstruct import claim_texts, fake_reconstruction, is_dialogue_role_line
|
|
from prompt_budget import (
|
|
ERROR_PROVIDER_CONTEXT_LENGTH,
|
|
JOURNAL_PURPOSES,
|
|
JournalBudgetError,
|
|
USER_MESSAGES,
|
|
assert_input_fits,
|
|
merge_usage,
|
|
sum_usages,
|
|
)
|
|
from pronoun_bind import bind_user_lines
|
|
from providers import ChatResult, ProviderError, complete_chat, generate_provider
|
|
|
|
MAX_EGRESS_CHARS = 24000
|
|
debug_calls = 0
|
|
last_compact: dict[str, Any] | None = None
|
|
_test_recorder: contextvars.ContextVar[list | None] = contextvars.ContextVar(
|
|
"kansho_gateway_recorder",
|
|
default=None,
|
|
)
|
|
COMPACT_DIAGNOSTIC_KEYS = (
|
|
"model",
|
|
"actual_model",
|
|
"provider",
|
|
"purpose",
|
|
"prompt_slug",
|
|
"effective_context_window",
|
|
"estimated_input_tokens",
|
|
"prompt_tokens",
|
|
"completion_tokens",
|
|
"total_tokens",
|
|
"reserved_output_tokens",
|
|
"max_tokens",
|
|
"safety_margin",
|
|
"context_compression",
|
|
"cost",
|
|
"budget_ok",
|
|
"abort_reason",
|
|
"status",
|
|
"available_mapping_count",
|
|
"active_mapping_count",
|
|
"masked_occurrence_count",
|
|
"active_tokens",
|
|
"active_entity_types",
|
|
"leak_tokens",
|
|
"leak_entity_types",
|
|
"pre_egress_validation",
|
|
"response_validation",
|
|
"response_validation_retry",
|
|
"full_detection_coverage",
|
|
"detect_provider",
|
|
"detect_model",
|
|
"detect_note",
|
|
"source_chars",
|
|
"chunk_count",
|
|
"chunks_ok",
|
|
"entity_counts",
|
|
"confirmed_registry_hits",
|
|
"request_local_hits",
|
|
"detect_calls",
|
|
"detect_prompt_tokens",
|
|
"detect_completion_tokens",
|
|
"detect_total_tokens",
|
|
"detect_cost",
|
|
"detect_ms",
|
|
"generate_called",
|
|
)
|
|
|
|
ERROR_EGRESS_VALIDATION = "egress_validation_failed"
|
|
|
|
|
|
class PrivacyGatewayError(Exception):
|
|
def __init__(self, code: str, message: str, status_code: int = 503, diagnostics: dict | None = None):
|
|
super().__init__(message)
|
|
self.code = code
|
|
self.message = message
|
|
self.status_code = status_code
|
|
self.diagnostics = diagnostics or {}
|
|
|
|
|
|
@dataclass
|
|
class GatewayRequest:
|
|
prompt_id: str | None
|
|
purpose: str
|
|
data_class: str
|
|
payload: dict[str, Any] = field(default_factory=dict)
|
|
profile_id: str | None = None
|
|
|
|
|
|
@dataclass
|
|
class GatewayResult:
|
|
allowed: bool
|
|
reason: str
|
|
provider: str | None = None
|
|
content: str | None = None
|
|
diagnostics: dict[str, Any] = field(default_factory=dict)
|
|
trace: dict[str, Any] | None = None
|
|
local_identities: list[dict[str, Any]] = field(default_factory=list)
|
|
checked_at: str = field(default_factory=lambda: datetime.now(timezone.utc).isoformat())
|
|
|
|
|
|
ALLOWED_CLASSES = {"A", "B", "C"}
|
|
LOCAL_ONLY_CLASS = "A"
|
|
|
|
|
|
@dataclass(frozen=True)
|
|
class ActiveReplacement:
|
|
"""One mapping that was actually replaced as identity in this request."""
|
|
|
|
token: str
|
|
entity_type: str
|
|
occurrence_count: int
|
|
local_label: str
|
|
demask_label: str = ""
|
|
|
|
def restore_label(self) -> str:
|
|
return self.demask_label or self.local_label
|
|
|
|
|
|
@dataclass
|
|
class MaskingManifest:
|
|
"""Request-scoped mask result. local_label stays on this object only."""
|
|
|
|
masked_text: str
|
|
available_mapping_count: int
|
|
replacements: tuple[ActiveReplacement, ...] = ()
|
|
restore_by_token: dict[str, str] = field(default_factory=dict)
|
|
|
|
def public_summary(self) -> dict[str, Any]:
|
|
return {
|
|
"available_mapping_count": self.available_mapping_count,
|
|
"active_mapping_count": len(self.replacements),
|
|
"masked_occurrence_count": sum(item.occurrence_count for item in self.replacements),
|
|
"active_tokens": [item.token for item in self.replacements],
|
|
"active_entity_types": sorted({item.entity_type for item in self.replacements}),
|
|
}
|
|
|
|
|
|
def compact_diagnostics(data: dict[str, Any] | None) -> dict[str, Any]:
|
|
payload = {}
|
|
for key in COMPACT_DIAGNOSTIC_KEYS:
|
|
if data and key in data and data[key] is not None:
|
|
payload[key] = data[key]
|
|
return payload
|
|
|
|
|
|
_LOG_FORBIDDEN_KEYS = frozenset(
|
|
{
|
|
"local_label",
|
|
"label",
|
|
"masked_text",
|
|
"rendered",
|
|
"intern",
|
|
"egress",
|
|
"raw",
|
|
"reply",
|
|
"content",
|
|
"messages",
|
|
"prompt",
|
|
"mapping_table",
|
|
"replacements",
|
|
}
|
|
)
|
|
|
|
|
|
def _log_event(events: list[dict[str, Any]], started: float, kind: str, **fields: Any) -> None:
|
|
"""Compact run log. No prompts, no answers, no mapping labels."""
|
|
row: dict[str, Any] = {
|
|
"kind": kind,
|
|
"t_ms": int((time.perf_counter() - started) * 1000),
|
|
}
|
|
for key, value in fields.items():
|
|
if value is None or value == "" or key in _LOG_FORBIDDEN_KEYS:
|
|
continue
|
|
row[key] = value
|
|
events.append(row)
|
|
|
|
|
|
def _usage_bits(usage: dict[str, Any] | None) -> dict[str, Any]:
|
|
data = usage or {}
|
|
bits: dict[str, Any] = {}
|
|
for key in ("prompt_tokens", "completion_tokens", "total_tokens", "cost"):
|
|
if data.get(key) is not None:
|
|
bits[key] = data[key]
|
|
return bits
|
|
|
|
|
|
def _raise_with_log(
|
|
exc: PrivacyGatewayError,
|
|
events: list[dict[str, Any]],
|
|
*,
|
|
purpose: str | None = None,
|
|
extra: dict[str, Any] | None = None,
|
|
) -> None:
|
|
diagnostics = dict(exc.diagnostics or {})
|
|
diagnostics["log"] = events
|
|
if purpose:
|
|
diagnostics["purpose"] = purpose
|
|
if extra:
|
|
diagnostics.update({key: value for key, value in extra.items() if value is not None})
|
|
raise PrivacyGatewayError(exc.code, exc.message, exc.status_code, diagnostics) from exc
|
|
|
|
|
|
def install_test_recorder() -> list[dict[str, Any]]:
|
|
"""Test-local observer. Production never installs this."""
|
|
records: list[dict[str, Any]] = []
|
|
_test_recorder.set(records)
|
|
return records
|
|
|
|
|
|
def _record_test(event: dict[str, Any]) -> None:
|
|
records = _test_recorder.get()
|
|
if records is not None:
|
|
records.append(event)
|
|
|
|
|
|
def reset_debug() -> None:
|
|
global last_compact, debug_calls
|
|
last_compact = None
|
|
debug_calls = 0
|
|
_test_recorder.set(None)
|
|
reset_detect_test_hooks()
|
|
|
|
|
|
def inspect(request: GatewayRequest) -> GatewayResult:
|
|
"""Fail closed unless a policy-complete generative provider is configured."""
|
|
data_class = (request.data_class or "").upper()
|
|
if data_class not in ALLOWED_CLASSES:
|
|
return GatewayResult(allowed=False, reason="unknown_data_class")
|
|
if data_class == LOCAL_ONLY_CLASS:
|
|
return GatewayResult(allowed=False, reason="class_a_never_leaves_local_zone")
|
|
config = generate_provider()
|
|
if not config:
|
|
return GatewayResult(allowed=False, reason="no_egress_provider_configured")
|
|
return GatewayResult(allowed=True, reason="policy_ok", provider=config.name)
|
|
|
|
|
|
def _minimize(text: str, purpose: str = "") -> str:
|
|
cleaned = (text or "").strip()
|
|
if purpose in JOURNAL_PURPOSES:
|
|
return cleaned
|
|
if len(cleaned) <= MAX_EGRESS_CHARS:
|
|
return cleaned
|
|
return cleaned[:MAX_EGRESS_CHARS]
|
|
|
|
|
|
_LETTER = r"A-Za-zÄÖÜäöüß"
|
|
_WORD = re.compile(rf"[{_LETTER}]+")
|
|
THING_GOVERNORS = {
|
|
"esse", "essen", "isst", "aß", "aßest", "gegessen",
|
|
"koche", "kochen", "kochte", "gekocht",
|
|
"trinke", "trinken", "trank", "getrunken",
|
|
"kaufe", "kaufen", "kaufte", "gekauft",
|
|
"bestelle", "bestellen", "bestellte",
|
|
"hole", "holen", "holte",
|
|
}
|
|
THING_OBJECT_AFTER = {
|
|
"essen", "kochen", "trinken", "kaufen", "bestellen", "holen",
|
|
"gegessen", "gekocht", "getrunken", "gekauft",
|
|
}
|
|
|
|
|
|
def _words_before(text: str, index: int, n: int = 4) -> list[str]:
|
|
return [word.lower() for word in _WORD.findall(text[:index])[-n:]]
|
|
|
|
|
|
def _words_after(text: str, index: int, n: int = 3) -> list[str]:
|
|
return [word.lower() for word in _WORD.findall(text[index:])[:n]]
|
|
|
|
|
|
def _is_person_token(token: str) -> bool:
|
|
raw = (token or "").strip().upper()
|
|
raw = raw[2:-2] if raw.startswith("[[") and raw.endswith("]]") else raw
|
|
return raw.startswith("PERSON:")
|
|
|
|
|
|
def _mask_person_hit(text: str, start: int, end: int) -> bool:
|
|
prev = _words_before(text, start)
|
|
nxt = _words_after(text, end)
|
|
if prev and prev[-1] in KINSHIP:
|
|
return True
|
|
if any(word in THING_GOVERNORS for word in prev):
|
|
return False
|
|
if nxt and nxt[0] in THING_OBJECT_AFTER:
|
|
return False
|
|
return True
|
|
|
|
|
|
def canonical_token(token: str) -> str:
|
|
"""Public, intent-neutral: placeholder token without surrounding [[ ]]."""
|
|
raw = (token or "").strip()
|
|
if raw.startswith("[[") and raw.endswith("]]"):
|
|
return raw[2:-2]
|
|
return raw
|
|
|
|
|
|
def identity_label_pattern(label: str) -> re.Pattern[str]:
|
|
"""Public, intent-neutral: word-boundary pattern for a local identity label."""
|
|
return re.compile(
|
|
rf"(?<![{_LETTER}]){re.escape(label)}(?![{_LETTER}])",
|
|
re.IGNORECASE,
|
|
)
|
|
|
|
|
|
def is_identity_mention(text: str, start: int, end: int, token: str) -> bool:
|
|
"""Public, intent-neutral: same rule as masking. Food/thing homonyms are not identity."""
|
|
if not _is_person_token(token):
|
|
return True
|
|
return _mask_person_hit(text, start, end)
|
|
|
|
|
|
def identity_occurrence_count(text: str, label: str, token: str) -> int:
|
|
"""Public, intent-neutral count of identity mentions for one mapping."""
|
|
if not label or not is_maskable_label(label):
|
|
return 0
|
|
return sum(
|
|
1
|
|
for match in identity_label_pattern(label).finditer(text or "")
|
|
if is_identity_mention(text, match.start(), match.end(), token)
|
|
)
|
|
|
|
|
|
def _is_identity_mention(text: str, start: int, end: int, token: str) -> bool:
|
|
return is_identity_mention(text, start, end, token)
|
|
|
|
|
|
def _label_pattern(label: str) -> re.Pattern[str]:
|
|
return identity_label_pattern(label)
|
|
|
|
|
|
def _canonical_token(token: str) -> str:
|
|
return canonical_token(token)
|
|
|
|
|
|
def _entity_type_of(item: dict) -> str:
|
|
explicit = (item.get("entity_type") or "").strip().upper()
|
|
if explicit:
|
|
return explicit
|
|
token = _canonical_token(item.get("token") or "").upper()
|
|
if ":" in token:
|
|
return token.split(":", 1)[0]
|
|
return token or "UNKNOWN"
|
|
|
|
|
|
def _identity_occurrence_count(text: str, label: str, token: str) -> int:
|
|
return identity_occurrence_count(text, label, token)
|
|
|
|
|
|
def _active_replacements(text: str, mappings: list[dict]) -> tuple[ActiveReplacement, ...]:
|
|
found: list[ActiveReplacement] = []
|
|
for item in mappings or []:
|
|
label = (item.get("local_label") or "").strip()
|
|
token = (item.get("token") or "").strip()
|
|
if not label or not token or not is_maskable_label(label):
|
|
continue
|
|
count = _identity_occurrence_count(text, label, token)
|
|
if count <= 0:
|
|
continue
|
|
found.append(
|
|
ActiveReplacement(
|
|
token=_canonical_token(token),
|
|
entity_type=_entity_type_of(item),
|
|
occurrence_count=count,
|
|
local_label=label,
|
|
demask_label=(item.get("demask_label") or item.get("canonical_label") or label),
|
|
)
|
|
)
|
|
return tuple(found)
|
|
|
|
|
|
def _mask_body(text: str, mappings: list[dict]) -> str:
|
|
masked = text
|
|
for item in sorted(mappings, key=lambda row: len(row.get("local_label") or ""), reverse=True):
|
|
label = (item.get("local_label") or "").strip()
|
|
token = (item.get("token") or "").strip()
|
|
if not label or not token or not is_maskable_label(label):
|
|
continue
|
|
placeholder = token if token.startswith("[[") else f"[[{token}]]"
|
|
pattern = _label_pattern(label)
|
|
|
|
def repl(match: re.Match, *, _token=token, _ph=placeholder) -> str:
|
|
if not _is_identity_mention(masked, match.start(), match.end(), _token):
|
|
return match.group(0)
|
|
return _ph
|
|
|
|
masked = pattern.sub(repl, masked)
|
|
return masked
|
|
|
|
|
|
def _mask(text: str, mappings: list[dict], *, personal_lines_only: bool = False) -> str:
|
|
if not personal_lines_only:
|
|
return _mask_body(text, mappings)
|
|
parts: list[str] = []
|
|
for line in (text or "").splitlines(keepends=True):
|
|
raw = line[:-1] if line.endswith("\n") else line
|
|
newline = "\n" if line.endswith("\n") else ""
|
|
if is_dialogue_role_line(raw):
|
|
parts.append(_mask_body(raw, mappings) + newline)
|
|
else:
|
|
parts.append(line)
|
|
return "".join(parts)
|
|
|
|
|
|
IDENTITY_LEAK_RETRY = (
|
|
"Korrektur: Keine Klartext-Identität. Kopiere die im Auftragstext bereits "
|
|
"vorhandenen Platzhalter zeichengetreu. Keine Klarnamen, keine Klarorte, "
|
|
"keine neuen Platzhalter, keine Auslassungspunkte."
|
|
)
|
|
|
|
PLACEHOLDER_RE = re.compile(r"\[\[\s*([^\[\]]+?)\s*\]\]")
|
|
GENERIC_PLACEHOLDER_INNER = frozenset({"…", "...", "..", "...."})
|
|
|
|
|
|
def _restore_table(manifest: MaskingManifest | None) -> dict[str, str]:
|
|
table: dict[str, str] = {}
|
|
if not manifest:
|
|
return table
|
|
for token, label in (manifest.restore_by_token or {}).items():
|
|
key = canonical_token(str(token)).upper().replace(" ", "")
|
|
if key and label:
|
|
table[key] = label
|
|
for item in manifest.replacements:
|
|
key = canonical_token(item.token).upper().replace(" ", "")
|
|
label = item.restore_label()
|
|
if key and label:
|
|
table.setdefault(key, label)
|
|
return table
|
|
|
|
|
|
def _lookup_restore(table: dict[str, str], inner: str) -> str | None:
|
|
raw = canonical_token(inner or "").upper().replace(" ", "")
|
|
if not raw or raw in GENERIC_PLACEHOLDER_INNER:
|
|
return None
|
|
if raw in table:
|
|
return table[raw]
|
|
if ":" in raw:
|
|
prefix, suffix = raw.split(":", 1)
|
|
if suffix.isdigit():
|
|
padded = f"{prefix}:{int(suffix):02d}"
|
|
if padded in table:
|
|
return table[padded]
|
|
unpadded = f"{prefix}:{int(suffix)}"
|
|
if unpadded in table:
|
|
return table[unpadded]
|
|
return None
|
|
|
|
|
|
def _restore_map_from_mappings(mappings: list[dict] | None) -> dict[str, str]:
|
|
table: dict[str, str] = {}
|
|
for item in mappings or []:
|
|
token = canonical_token(item.get("token") or "").upper().replace(" ", "")
|
|
label = (item.get("demask_label") or item.get("canonical_label") or item.get("local_label") or "").strip()
|
|
if token and label:
|
|
table[token] = label
|
|
return table
|
|
|
|
|
|
def mask_prompt(rendered: str, mappings: list[dict], purpose: str) -> MaskingManifest:
|
|
"""Request-scoped mask result. Active mappings are those actually replaced."""
|
|
masked = rendered or ""
|
|
replacements: list[ActiveReplacement] = []
|
|
for item in sorted(mappings or [], key=lambda row: len(row.get("local_label") or ""), reverse=True):
|
|
label = (item.get("local_label") or "").strip()
|
|
token = (item.get("token") or "").strip()
|
|
if not label or not token or not is_maskable_label(label):
|
|
continue
|
|
placeholder = token if token.startswith("[[") else f"[[{_canonical_token(token)}]]"
|
|
pattern = _label_pattern(label)
|
|
current = masked
|
|
count = 0
|
|
|
|
def repl(match: re.Match, *, _token=token, _ph=placeholder, _text=current) -> str:
|
|
nonlocal count
|
|
if not _is_identity_mention(_text, match.start(), match.end(), _token):
|
|
return match.group(0)
|
|
count += 1
|
|
return _ph
|
|
|
|
masked = pattern.sub(repl, current)
|
|
if count <= 0:
|
|
continue
|
|
replacements.append(
|
|
ActiveReplacement(
|
|
token=_canonical_token(token),
|
|
entity_type=_entity_type_of(item),
|
|
occurrence_count=count,
|
|
local_label=label,
|
|
demask_label=(item.get("demask_label") or item.get("canonical_label") or label),
|
|
)
|
|
)
|
|
if purpose == "dialogue_turn":
|
|
masked = bind_user_lines(masked)
|
|
return MaskingManifest(
|
|
masked_text=masked,
|
|
available_mapping_count=len(mappings or []),
|
|
replacements=tuple(replacements),
|
|
restore_by_token=_restore_map_from_mappings(mappings),
|
|
)
|
|
|
|
|
|
def mask_for_egress(rendered: str, mappings: list[dict], purpose: str) -> str:
|
|
"""String wrapper for callers that only need the masked egress text."""
|
|
return mask_prompt(rendered, mappings, purpose).masked_text
|
|
|
|
|
|
def validate_pre_egress(masked_text: str, manifest: MaskingManifest) -> None:
|
|
"""Fail closed before the provider if an active identity remains as plaintext."""
|
|
leaked_tokens: list[str] = []
|
|
leaked_types: list[str] = []
|
|
for item in manifest.replacements:
|
|
for match in _label_pattern(item.local_label).finditer(masked_text or ""):
|
|
if _is_identity_mention(masked_text, match.start(), match.end(), item.token):
|
|
leaked_tokens.append(item.token)
|
|
leaked_types.append(item.entity_type)
|
|
break
|
|
if leaked_tokens:
|
|
raise PrivacyGatewayError(
|
|
ERROR_EGRESS_VALIDATION,
|
|
"Aktive Identität war nach der Maskierung noch als Klartext im Egress.",
|
|
diagnostics={
|
|
"pre_egress_validation": "failed",
|
|
"leak_tokens": leaked_tokens,
|
|
"leak_entity_types": leaked_types,
|
|
**manifest.public_summary(),
|
|
},
|
|
)
|
|
|
|
|
|
def _demask(text: str, manifest: MaskingManifest | None) -> str:
|
|
"""Rehydrate placeholders that belong to this request. Unknown tokens stay placeholders."""
|
|
result = text or ""
|
|
table = _restore_table(manifest)
|
|
if not table:
|
|
return result
|
|
|
|
def repl(match: re.Match) -> str:
|
|
inner = match.group(1) or ""
|
|
restored = _lookup_restore(table, inner)
|
|
return restored if restored else match.group(0)
|
|
|
|
return PLACEHOLDER_RE.sub(repl, result)
|
|
|
|
|
|
def _validate_response(content: str, manifest: MaskingManifest | None = None) -> str:
|
|
text = (content or "").strip()
|
|
if not text:
|
|
raise PrivacyGatewayError("empty_provider_response", "Der Provider lieferte keine Antwort.")
|
|
leaked_tokens: list[str] = []
|
|
leaked_types: list[str] = []
|
|
for item in manifest.replacements if manifest else ():
|
|
for match in _label_pattern(item.local_label).finditer(text):
|
|
if _is_identity_mention(text, match.start(), match.end(), item.token):
|
|
leaked_tokens.append(item.token)
|
|
leaked_types.append(item.entity_type)
|
|
break
|
|
if leaked_tokens:
|
|
raise PrivacyGatewayError(
|
|
"response_validation_failed",
|
|
"Antwort enthielt Klartext-Identität vor der Demaskierung.",
|
|
diagnostics={
|
|
"response_validation": "failed",
|
|
"leak_tokens": leaked_tokens,
|
|
"leak_entity_types": leaked_types,
|
|
},
|
|
)
|
|
return text
|
|
|
|
|
|
def _fake_journal_from_sources(rendered: str) -> str:
|
|
"""Deterministic fake body from labeled CURRENT_DAY_SOURCES. Not a quality claim."""
|
|
text = rendered or ""
|
|
marker = "CURRENT_DAY_SOURCES"
|
|
if marker not in text:
|
|
return ""
|
|
after = text.split(marker, 1)[1]
|
|
for stop in ("WRITING_PROFILE", "STYLE_EXAMPLES", "EXISTING_TEXT", "EDITORIAL_MODE"):
|
|
if f"\n{stop}" in after:
|
|
after = after.split(f"\n{stop}", 1)[0]
|
|
parts: list[str] = []
|
|
current: list[str] = []
|
|
for line in after.splitlines():
|
|
stripped = line.strip()
|
|
if stripped.startswith("[u") and stripped.endswith("]"):
|
|
if current:
|
|
parts.append(" ".join(current).strip())
|
|
current = []
|
|
continue
|
|
if stripped and not stripped.startswith("(") and "einzige Tatsachen" not in stripped:
|
|
current.append(stripped)
|
|
if current:
|
|
parts.append(" ".join(current).strip())
|
|
return " ".join(part for part in parts if part).strip()
|
|
|
|
|
|
def _fake_complete(purpose: str, rendered: str) -> str:
|
|
if purpose == "profile_review":
|
|
return (
|
|
'{"kind":"kansho.profile_analysis_result","format_version":1,"target":"writing",'
|
|
'"evidence_basis":["kansho_sources"],'
|
|
'"changes":[{"layer":"trait","key":"dry_humor","slug":"dry_humor","facet_key":"autobiographical_journal",'
|
|
'"action":"add","label":"Trockener Humor",'
|
|
'"proposed_value":"gelegentlich trocken, nie aufgesetzt",'
|
|
'"evidence_basis":["kansho_sources"],'
|
|
'"rationale":"Fake-Review: Journal-Evidenz trägt Humor in autobiographical_journal, nicht als globalen Core.",'
|
|
'"evidence_ids":[],"exemplars":[{"excerpt":"haha das war irgendwie lustig","role":"exemplar","evidence_basis":["kansho_sources"]}]}]}'
|
|
)
|
|
if purpose == "journal_reconstruct":
|
|
return fake_reconstruction(rendered)
|
|
if purpose == "journal_generate":
|
|
body = _fake_journal_from_sources(rendered)
|
|
if body:
|
|
return f"Ein Tag\n\n{body}"
|
|
match = re.search(r"\{.*\}", rendered, re.DOTALL)
|
|
if match:
|
|
try:
|
|
import json
|
|
|
|
data = json.loads(match.group(0))
|
|
parts = claim_texts(data)
|
|
body = " ".join(str(part) for part in parts if part)
|
|
if body:
|
|
return f"Ein Tag\n\n{body}"
|
|
except (ValueError, TypeError):
|
|
pass
|
|
lines = []
|
|
for line in rendered.splitlines():
|
|
if line.startswith("user:"):
|
|
lines.append(line[5:].strip())
|
|
body = " ".join(part for part in lines if part) or "Ein stiller Tag."
|
|
return f"Ein Tag\n\n{body}"
|
|
if "Keine Interviewfrage" in rendered:
|
|
return '{"operation":"fortfuehren","impulse":"Der letzte Faden bleibt offen."}'
|
|
impulse = (
|
|
"Was davon möchtest du vertiefen, ohne Unerwähntes als nicht geschehen zu behandeln?"
|
|
if ("nicht geschehen" in rendered or "nicht erwähnt" in rendered)
|
|
else "Was davon möchtest du festhalten?"
|
|
)
|
|
return '{"operation":"erleben_vertiefen","impulse":"' + impulse + '"}'
|
|
|
|
|
|
def complete_model(messages: list[dict], policy: dict[str, Any]) -> ChatResult:
|
|
config = generate_provider()
|
|
if not config:
|
|
raise PrivacyGatewayError("no_egress_provider_configured", "Es ist kein Egress-Provider konfiguriert.")
|
|
if config.mode == "fake":
|
|
rendered = "\n".join(item.get("content") or "" for item in messages)
|
|
return ChatResult(
|
|
content=_fake_complete(policy.get("purpose") or "", rendered),
|
|
model=config.model,
|
|
usage={},
|
|
context_compression="disabled" if policy.get("disable_context_compression") else "not_applicable",
|
|
)
|
|
try:
|
|
return complete_chat(
|
|
config,
|
|
messages,
|
|
timeout=90 if (policy.get("purpose") in JOURNAL_PURPOSES) else 60,
|
|
max_tokens=policy.get("max_tokens"),
|
|
disable_context_compression=bool(policy.get("disable_context_compression")),
|
|
)
|
|
except ProviderError as exc:
|
|
if exc.code == "provider_context_length_rejected":
|
|
raise PrivacyGatewayError(
|
|
ERROR_PROVIDER_CONTEXT_LENGTH,
|
|
USER_MESSAGES[ERROR_PROVIDER_CONTEXT_LENGTH],
|
|
exc.status_code,
|
|
diagnostics=exc.diagnostics,
|
|
) from exc
|
|
raise PrivacyGatewayError(exc.code, exc.message, exc.status_code, getattr(exc, "diagnostics", None)) from exc
|
|
|
|
|
|
def complete(request: GatewayRequest) -> GatewayResult:
|
|
global last_compact, debug_calls
|
|
started = time.perf_counter()
|
|
events: list[dict[str, Any]] = []
|
|
result = inspect(request)
|
|
if not result.allowed:
|
|
_log_event(events, started, "blocked", purpose=request.purpose, code=result.reason)
|
|
raise PrivacyGatewayError(
|
|
result.reason,
|
|
"Persönlicher KI-Aufruf wurde vom Privacy Gateway blockiert. "
|
|
"Es ist kein Egress-Provider konfiguriert."
|
|
if result.reason == "no_egress_provider_configured"
|
|
else "Persönlicher KI-Aufruf wurde vom Privacy Gateway blockiert.",
|
|
diagnostics={"log": events, "purpose": request.purpose},
|
|
)
|
|
rendered = _minimize(str((request.payload or {}).get("rendered") or ""), request.purpose)
|
|
detect_name = None
|
|
detect_note = None
|
|
detect_stats: dict[str, Any] = {}
|
|
local_identities: list[dict[str, Any]] = []
|
|
try:
|
|
outcome = detect_personal_egress(request.profile_id, rendered)
|
|
mappings = outcome.mappings
|
|
detect_stats = outcome.stats.public()
|
|
detect_name = outcome.stats.detect_provider
|
|
detect_note = outcome.stats.detect_note
|
|
local_identities = outcome.local_identities
|
|
except DetectError as exc:
|
|
detect_stats = dict(exc.diagnostics or {})
|
|
_log_event(
|
|
events,
|
|
started,
|
|
"detect",
|
|
status="error",
|
|
code=exc.code,
|
|
full_detection_coverage=False,
|
|
generate_called=False,
|
|
chunk_count=detect_stats.get("chunk_count"),
|
|
chunks_ok=detect_stats.get("chunks_ok"),
|
|
detect_calls=detect_stats.get("detect_calls"),
|
|
)
|
|
raise PrivacyGatewayError(
|
|
exc.code,
|
|
user_detect_message(exc.code, exc.message),
|
|
exc.status_code,
|
|
{"log": events, "purpose": request.purpose, "generate_called": False, **detect_stats},
|
|
) from exc
|
|
_log_event(
|
|
events,
|
|
started,
|
|
"detect",
|
|
status="ok",
|
|
detect_provider=detect_name,
|
|
detect_note=detect_note,
|
|
detect_model=detect_stats.get("detect_model"),
|
|
full_detection_coverage=detect_stats.get("full_detection_coverage"),
|
|
chunk_count=detect_stats.get("chunk_count"),
|
|
chunks_ok=detect_stats.get("chunks_ok"),
|
|
detect_calls=detect_stats.get("detect_calls"),
|
|
request_local_hits=detect_stats.get("request_local_hits"),
|
|
confirmed_registry_hits=detect_stats.get("confirmed_registry_hits"),
|
|
entity_counts=detect_stats.get("entity_counts"),
|
|
detect_ms=detect_stats.get("detect_ms"),
|
|
detect_prompt_tokens=detect_stats.get("detect_prompt_tokens"),
|
|
detect_completion_tokens=detect_stats.get("detect_completion_tokens"),
|
|
detect_total_tokens=detect_stats.get("detect_total_tokens"),
|
|
detect_cost=detect_stats.get("detect_cost"),
|
|
mapping_count=len(mappings),
|
|
)
|
|
manifest = mask_prompt(rendered, mappings, request.purpose)
|
|
masked = manifest.masked_text
|
|
summary = manifest.public_summary()
|
|
budget = (request.payload or {}).get("budget")
|
|
diagnostics = dict((request.payload or {}).get("diagnostics") or {})
|
|
diagnostics.update(summary)
|
|
diagnostics.update({key: value for key, value in detect_stats.items() if value is not None})
|
|
diagnostics["generate_called"] = False
|
|
if request.purpose in JOURNAL_PURPOSES:
|
|
if budget is None:
|
|
_log_event(events, started, "budget", status="error", code="model_metadata_unknown")
|
|
raise PrivacyGatewayError(
|
|
"model_metadata_unknown",
|
|
USER_MESSAGES["model_metadata_unknown"],
|
|
503,
|
|
diagnostics={"reason": "journal_budget_missing", "log": events},
|
|
)
|
|
try:
|
|
assert_input_fits(budget, masked)
|
|
except JournalBudgetError as exc:
|
|
_log_event(events, started, "budget", status="error", code=exc.code)
|
|
raise PrivacyGatewayError(
|
|
exc.code,
|
|
exc.message,
|
|
exc.status_code,
|
|
{**(exc.diagnostics or {}), "log": events},
|
|
) from exc
|
|
diagnostics.update(budget.as_diagnostics())
|
|
diagnostics["estimated_input_tokens"] = budget.estimated_input_tokens
|
|
_log_event(events, started, "budget", status="ok")
|
|
debug_calls += 1
|
|
config = generate_provider()
|
|
layer = {
|
|
"journal_generate": "journalentwurf",
|
|
"journal_reconstruct": "journalrekonstruktion",
|
|
"profile_review": "profilreview",
|
|
}.get(request.purpose, "dialogzug")
|
|
max_tokens = (request.payload or {}).get("max_tokens")
|
|
if max_tokens is not None:
|
|
diagnostics["max_tokens"] = max_tokens
|
|
diagnostics["provider"] = result.provider
|
|
diagnostics["purpose"] = request.purpose
|
|
diagnostics["prompt_slug"] = (request.payload or {}).get("prompt_slug")
|
|
diagnostics["model"] = config.model if config else None
|
|
request_trace = {
|
|
"purpose": request.purpose,
|
|
"layer": layer,
|
|
"data_class": request.data_class,
|
|
"prompt_slug": (request.payload or {}).get("prompt_slug"),
|
|
"rendered": rendered,
|
|
"masked": masked,
|
|
"mask_input": rendered,
|
|
"provider": result.provider,
|
|
"model": config.model if config else None,
|
|
"detect_provider": detect_name,
|
|
"detect_note": detect_note,
|
|
"detect_model": detect_stats.get("detect_model"),
|
|
"full_detection_coverage": detect_stats.get("full_detection_coverage"),
|
|
"chunk_count": detect_stats.get("chunk_count"),
|
|
"chunks_ok": detect_stats.get("chunks_ok"),
|
|
"entity_counts": detect_stats.get("entity_counts"),
|
|
"confirmed_registry_hits": detect_stats.get("confirmed_registry_hits"),
|
|
"request_local_hits": detect_stats.get("request_local_hits"),
|
|
"detect_calls": detect_stats.get("detect_calls"),
|
|
"detect_ms": detect_stats.get("detect_ms"),
|
|
"generate_called": False,
|
|
"mapping_count": len(mappings),
|
|
"available_mapping_count": summary["available_mapping_count"],
|
|
"active_mapping_count": summary["active_mapping_count"],
|
|
"masked_occurrence_count": summary["masked_occurrence_count"],
|
|
"active_tokens": summary["active_tokens"],
|
|
"active_entity_types": summary["active_entity_types"],
|
|
"budget": compact_diagnostics(diagnostics),
|
|
"log": events,
|
|
}
|
|
model_policy = {
|
|
"purpose": request.purpose,
|
|
"zdr": True,
|
|
"no_train": True,
|
|
"max_tokens": max_tokens,
|
|
"disable_context_compression": bool(
|
|
(request.payload or {}).get("disable_context_compression")
|
|
),
|
|
}
|
|
try:
|
|
validate_pre_egress(masked, manifest)
|
|
diagnostics["pre_egress_validation"] = "ok"
|
|
request_trace["pre_egress_validation"] = "ok"
|
|
_log_event(
|
|
events,
|
|
started,
|
|
"pre_egress_validation",
|
|
status="ok",
|
|
active_mapping_count=summary["active_mapping_count"],
|
|
masked_occurrence_count=summary["masked_occurrence_count"],
|
|
)
|
|
except PrivacyGatewayError as exc:
|
|
leak = dict(exc.diagnostics or {})
|
|
diagnostics["pre_egress_validation"] = "failed"
|
|
request_trace["pre_egress_validation"] = "failed"
|
|
_log_event(
|
|
events,
|
|
started,
|
|
"pre_egress_validation",
|
|
status="failed",
|
|
code=exc.code,
|
|
leak_tokens=leak.get("leak_tokens"),
|
|
leak_entity_types=leak.get("leak_entity_types"),
|
|
)
|
|
failed = merge_usage(
|
|
{**diagnostics, "budget_ok": False, "abort_reason": exc.code, "status": "error"},
|
|
None,
|
|
config.model if config else None,
|
|
)
|
|
last_compact = compact_diagnostics(failed)
|
|
_record_test({"purpose": request.purpose, "ok": False, "code": exc.code})
|
|
_raise_with_log(exc, events, purpose=request.purpose)
|
|
usages: list[dict] = []
|
|
try:
|
|
_log_event(
|
|
events,
|
|
started,
|
|
"model_call",
|
|
attempt=1,
|
|
purpose=request.purpose,
|
|
model=config.model if config else None,
|
|
)
|
|
chat = complete_model([{"role": "user", "content": masked}], model_policy)
|
|
usages.append(chat.usage or {})
|
|
diagnostics["generate_called"] = True
|
|
request_trace["generate_called"] = True
|
|
_log_event(events, started, "model_result", attempt=1, **_usage_bits(chat.usage))
|
|
raw = chat.content
|
|
try:
|
|
validated = _validate_response(raw, manifest)
|
|
diagnostics["response_validation"] = "ok"
|
|
_log_event(events, started, "response_validation", attempt=1, status="ok")
|
|
except PrivacyGatewayError as exc:
|
|
if exc.code != "response_validation_failed":
|
|
raise
|
|
leak = dict(exc.diagnostics or {})
|
|
diagnostics["response_validation"] = "failed"
|
|
_log_event(
|
|
events,
|
|
started,
|
|
"response_validation",
|
|
attempt=1,
|
|
status="failed",
|
|
code=exc.code,
|
|
leak_tokens=leak.get("leak_tokens"),
|
|
leak_entity_types=leak.get("leak_entity_types"),
|
|
)
|
|
_log_event(events, started, "retry", attempt=2, reason="response_validation_failed")
|
|
diagnostics["response_validation_retry"] = 1
|
|
request_trace["response_validation_retry"] = 1
|
|
_log_event(
|
|
events,
|
|
started,
|
|
"model_call",
|
|
attempt=2,
|
|
purpose=request.purpose,
|
|
model=config.model if config else None,
|
|
)
|
|
chat = complete_model(
|
|
[{"role": "user", "content": masked + "\n\n" + IDENTITY_LEAK_RETRY}],
|
|
model_policy,
|
|
)
|
|
usages.append(chat.usage or {})
|
|
_log_event(events, started, "model_result", attempt=2, **_usage_bits(chat.usage))
|
|
raw = chat.content
|
|
validated = _validate_response(raw, manifest)
|
|
diagnostics["response_validation"] = "ok"
|
|
_log_event(events, started, "response_validation", attempt=2, status="ok")
|
|
except PrivacyGatewayError as exc:
|
|
if exc.code == "response_validation_failed":
|
|
leak = dict(exc.diagnostics or {})
|
|
diagnostics["response_validation"] = "failed"
|
|
_log_event(
|
|
events,
|
|
started,
|
|
"response_validation",
|
|
attempt=2 if diagnostics.get("response_validation_retry") else 1,
|
|
status="failed",
|
|
code=exc.code,
|
|
leak_tokens=leak.get("leak_tokens"),
|
|
leak_entity_types=leak.get("leak_entity_types"),
|
|
)
|
|
failed = merge_usage(
|
|
{**diagnostics, "budget_ok": False, "abort_reason": exc.code, "status": "error"},
|
|
sum_usages(usages) or None,
|
|
config.model if config else None,
|
|
)
|
|
last_compact = compact_diagnostics(failed)
|
|
_record_test({"purpose": request.purpose, "ok": False, "code": exc.code})
|
|
_raise_with_log(
|
|
exc,
|
|
events,
|
|
purpose=request.purpose,
|
|
extra={"response_validation_retry": diagnostics.get("response_validation_retry")},
|
|
)
|
|
diagnostics = merge_usage(
|
|
{
|
|
**diagnostics,
|
|
"context_compression": chat.context_compression,
|
|
"budget_ok": True,
|
|
"status": "ok",
|
|
"response_validation": "ok",
|
|
},
|
|
sum_usages(usages) or chat.usage,
|
|
chat.model or (config.model if config else None),
|
|
)
|
|
result.content = _demask(validated, manifest)
|
|
result.allowed = True
|
|
result.local_identities = local_identities
|
|
result.diagnostics = compact_diagnostics(diagnostics)
|
|
request_trace["raw"] = validated
|
|
request_trace["reply"] = result.content
|
|
request_trace["model"] = chat.model or request_trace.get("model")
|
|
request_trace["response_validation"] = "ok"
|
|
request_trace["budget"] = compact_diagnostics(diagnostics)
|
|
request_trace["log"] = events
|
|
result.trace = public_trace(request_trace)
|
|
last_compact = compact_diagnostics(diagnostics)
|
|
_record_test({"purpose": request.purpose, "ok": True, "prompt_slug": request_trace.get("prompt_slug")})
|
|
return result
|
|
|
|
|
|
def last_trace() -> dict | None:
|
|
"""Compact diagnostics only. Full prompts live on the current GatewayResult."""
|
|
return last_compact
|
|
|
|
|
|
def public_trace(trace: dict | None) -> dict | None:
|
|
"""Admin test view for the current response. No mapping table, no secrets."""
|
|
if not trace:
|
|
return None
|
|
budget = dict(trace.get("budget") or {})
|
|
for key in ("rendered", "masked", "intern", "raw", "reply", "prompt", "messages"):
|
|
budget.pop(key, None)
|
|
return {
|
|
"purpose": trace.get("purpose"),
|
|
"layer": trace.get("layer"),
|
|
"data_class": trace.get("data_class"),
|
|
"prompt_slug": trace.get("prompt_slug"),
|
|
"provider": trace.get("provider"),
|
|
"model": trace.get("model"),
|
|
"detect_provider": trace.get("detect_provider"),
|
|
"detect_note": trace.get("detect_note"),
|
|
"detect_model": trace.get("detect_model"),
|
|
"full_detection_coverage": trace.get("full_detection_coverage"),
|
|
"chunk_count": trace.get("chunk_count"),
|
|
"chunks_ok": trace.get("chunks_ok"),
|
|
"entity_counts": trace.get("entity_counts"),
|
|
"confirmed_registry_hits": trace.get("confirmed_registry_hits"),
|
|
"request_local_hits": trace.get("request_local_hits"),
|
|
"detect_calls": trace.get("detect_calls"),
|
|
"detect_ms": trace.get("detect_ms"),
|
|
"generate_called": trace.get("generate_called"),
|
|
"mapping_count": trace.get("mapping_count"),
|
|
"available_mapping_count": trace.get("available_mapping_count"),
|
|
"active_mapping_count": trace.get("active_mapping_count"),
|
|
"masked_occurrence_count": trace.get("masked_occurrence_count"),
|
|
"active_tokens": trace.get("active_tokens"),
|
|
"active_entity_types": trace.get("active_entity_types"),
|
|
"pre_egress_validation": trace.get("pre_egress_validation"),
|
|
"response_validation": trace.get("response_validation"),
|
|
"leak_tokens": trace.get("leak_tokens"),
|
|
"leak_entity_types": trace.get("leak_entity_types"),
|
|
"intern": trace.get("rendered") or trace.get("intern"),
|
|
"egress": trace.get("masked") or trace.get("egress"),
|
|
"mask_input": trace.get("mask_input"),
|
|
"raw": trace.get("raw"),
|
|
"reply": trace.get("reply"),
|
|
"budget": budget or None,
|
|
"log": trace.get("log") or None,
|
|
"response_validation_retry": trace.get("response_validation_retry"),
|
|
"stage1": trace.get("stage1"),
|
|
"reason": trace.get("reason"),
|
|
"guard": trace.get("guard"),
|
|
"stages": trace.get("stages"),
|
|
"status": trace.get("status"),
|
|
"coverage": trace.get("coverage"),
|
|
"source_count": trace.get("source_count"),
|
|
}
|