from __future__ import annotations

OPENAI_TAXONOMY = [
    "harassment",
    "harassment_threatening",
    "hate",
    "hate_threatening",
    "self_harm",
    "self_harm_intent",
    "self_harm_instructions",
    "sexual",
    "sexual_minors",
    "violence",
    "violence_graphic",
]
EXTENDED_TAXONOMY = [
    "identity_attack",
    "insult",
    "profanity_obscene",
    "threat",
    "sexual_explicit",
    "severe_toxicity",
    "dehumanize",
    "incitement_violence",
]
TAXONOMY = OPENAI_TAXONOMY + EXTENDED_TAXONOMY
SYNTHETIC_ONLY = {
    "self_harm",
    "self_harm_intent",
    "self_harm_instructions",
    "sexual_minors",
    "violence_graphic",
}
JIGSAW_LABELS = [
    "toxic",
    "severe_toxic",
    "obscene",
    "threat",
    "insult",
    "identity_hate",
]
SPAM_LABELS = ["spam", "scam", "phishing"]
JAILBREAK_LABELS = [
    "jailbreak",
    "instruction_override",
    "role_hijack",
    "data_exfiltration",
]
NSFW_LABELS = ["sexual_suggestive", "sexual_explicit"]
IDENTITY_LABELS = [
    "race",
    "religion",
    "gender",
    "sexuality",
    "disability",
    "origin",
    "age",
]
V2_HEADS = {
    "spam": SPAM_LABELS,
    "jailbreak": JAILBREAK_LABELS,
    "nsfw": NSFW_LABELS,
    "identity": IDENTITY_LABELS,
}
PII_ENTITIES = [
    "PERSON",
    "FIRSTNAME",
    "LASTNAME",
    "USERNAME",
    "EMAIL",
    "PHONE",
    "FAX",
    "STREET_ADDRESS",
    "CITY",
    "STATE",
    "ZIPCODE",
    "COUNTRY",
    "BUILDINGNUMBER",
    "SECONDARYADDRESS",
    "NEARBYGPSCOORDINATE",
    "DOB",
    "AGE",
    "SSN",
    "PASSPORT",
    "DRIVERLICENSE",
    "IDCARD",
    "CREDITCARD",
    "CREDITCARDCVV",
    "CREDITCARDEXP",
    "IBAN",
    "BIC_SWIFT",
    "ACCOUNTNUMBER",
    "ROUTINGNUMBER",
    "BITCOINADDRESS",
    "ETHEREUMADDRESS",
    "IPV4",
    "IPV6",
    "MACADDRESS",
    "URL",
    "USERAGENT",
    "COMPANYNAME",
    "JOBTITLE",
    "MRN",
    "MEDICALLICENSE",
    "HEALTHINSURANCEID",
    "VEHICLEVIN",
    "VEHICLEPLATE",
    "TIME",
    "DATE",
    "PIN",
    "PASSWORD",
    "APIKEY",
    "ORDINALDIRECTION",
    "CURRENCY",
    "AMOUNT",
    "GENDER",
    "ETHNICITY",
    "RELIGION",
    "SEXUALITY",
    "POLITICAL",
    "MARITALSTATUS",
]
PII_TAGS = ["O"] + [f"{p}-{e}" for e in PII_ENTITIES for p in ("B", "I")]
N_PII_TAGS = len(PII_TAGS)
TAG_TO_ID = {t: i for i, t in enumerate(PII_TAGS)}


def n_tax() -> int:
    return len(TAXONOMY)