motif_tagging.py

# motif_tagging.py

import re

MOTIF_PATTERNS = {
    "threat": [
        r"\b(i['’]m|i am) going to (hurt|kill|break|ruin|destroy|end|bash|beat|smash|wreck) you\b",
        r"\b(i['’]?ll|i will) (hurt|kill|ruin|wreck|end|destroy|obliterate|crush|eliminate|beat) you\b",
        r"\b(i['’]?ll|i will) make you disappear\b",
        r"\b(i['’]?ll|i will) put you in (a|the) (grave|hospital|ground|dirt|bodybag)\b",
        r"\b(i['’]?ll|i will) snap your (neck|spine|arm|leg)\b",
        r"\b(i['’]?ll|i will) break your (face|jaw|legs|arms|neck)\b",
        r"\b(i['’]?ll|i will) beat the (shit|hell|crap) out of you\b",
        r"\b(i['’]?ll|i will) f\*?ck you up\b",
        r"\b(i['’]?ll|i will) (come for|find|hunt|get) you\b",
        r"\b(i['’]?ll|i will) put you down\b",
        r"\b(i['’]?ll|i will) end you\b",
        r"\bi hope you (die|suffer|disappear)\b",
        r"\byou better (run|hide|watch your back)\b",
        r"\bsay goodbye\b.*\byou\b",
        r"\bsleep with one eye open\b",
        r"\b(i['’]?ll|i will) send you to (hell|the morgue)\b",
        r"\b(i['’]?ll|i will) make sure you’re never seen again\b",
        r"\b(i['’]?ll|i will|ill) put you in (a|the) rose garden\b"
        r"\b(i (won[’']?t|will not) let you (leave|go|talk to|see|text|spend time with) (him|her|them|anyone))\b",
        r"\b(you (can[’']?t|are not allowed to|shouldn’t)) (go out|leave|talk to|see|text|spend time with|have friends)\b",
        r"\b(i (decide|get to decide|control|own) what you (do|say|wear|eat|spend|see|talk to))\b",
        r"\b(you['’]?re|you are) not allowed to (leave|talk to anyone|have your phone)\b",
        r"\bi make the rules\b",
        r"\bit’s my way or nothing\b",
        r"\bi’ll take your (car|keys|money|phone)\b",
        r"\byou don’t need anyone else\b",
        r"\byou belong to me\b",
        r"\bwithout me, you’re (nothing|no one)\b",
        r"\byou do what i say\b",
        r"\b(you better|you have to) do what i say\b",
        r"\b(you['’]?re not going|you’re not leaving|i won’t let you go)\b"
        r"\b(i['’]?ll|i will) (kill|hurt) myself (if you leave|if you don’t come back|unless you stay)\b",
        r"\bi (don’t|do not) want to live (without you|if you leave)\b",
        r"\bi['’]?ll (die|disappear|end it) (if|unless) you (leave|go|don’t call)\b",
        r"\b(if )?i['’]?ll make you watch me (suffer|bleed)\b",
        r"\bi['’]?ll do something bad (to myself|if you leave)\b",
        r"\bi['’]?ll regret it when i’m gone\b",
        r"\bi can’t live without you\b",
        r"\bmaybe (i should just die|it’d be better if i were dead)\b",
        r"\bi’m better off dead\b",
        r"\bi['’]?ll disappear forever if you walk out\b"
    ]
}

def detect_motifs(text):
    tags = []
    matched_phrases = []
    text_lower = text.lower()

    for label, patterns in MOTIF_PATTERNS.items():
        for pattern in patterns:
            match = re.search(pattern, text_lower)
            if match:
                tags.append(label)
                matched_phrases.append((label, match.group()))
                break

    return tags, matched_phrases