ecoloop/api/services/matching.py

from __future__ import annotations

from collections import Counter
from dataclasses import dataclass
from typing import Iterable, List, Sequence, Tuple

from django.contrib.auth.models import User
from django.db.models import Q
from django.utils import timezone

from api.models import IntroductionRequest, Opportunity, Profile


def _now():
    return timezone.now()


def tokenize_interests_text(text: str | None) -> set[str]:
    if not text:
        return set()
    raw = [t.strip().lower() for part in text.split(",") for t in part.split()]  # type: ignore
    return {t for t in raw if t}


def profile_interests(profile: Profile) -> set[str]:
    # Prefer taggit tags when available, fall back to comma-separated string
    try:
        tag_names = set(t.lower() for t in profile.tags.names())
    except Exception:
        tag_names = set()
    if tag_names:
        return tag_names
    return tokenize_interests_text(profile.interests)


def tokenize_text(text: str | None) -> List[str]:
    if not text:
        return []
    clean = "".join(ch.lower() if ch.isalnum() or ch.isspace() else " " for ch in text)
    return [t for t in clean.split() if t]


def jaccard(a: Iterable[str], b: Iterable[str]) -> float:
    sa, sb = set(a), set(b)
    if not sa and not sb:
        return 0.0
    inter = len(sa & sb)
    union = len(sa | sb)
    return inter / union if union else 0.0


def cosine(a: Sequence[str], b: Sequence[str]) -> float:
    if not a or not b:
        return 0.0
    ca, cb = Counter(a), Counter(b)
    # dot product
    dot = sum(ca[t] * cb.get(t, 0) for t in ca)
    if dot == 0:
        return 0.0
    import math

    na = math.sqrt(sum(v * v for v in ca.values()))
    nb = math.sqrt(sum(v * v for v in cb.values()))
    denom = na * nb
    return (dot / denom) if denom else 0.0


def time_decay(ts, half_life_days: float = 7.0) -> float:
    if not ts:
        return 0.0
    import math

    dt = (_now() - ts).total_seconds()
    half_life = half_life_days * 24 * 3600.0
    if dt <= 0:
        return 1.0
    return math.pow(0.5, dt / half_life)


def neighbor_set(u: User) -> set[int]:
    # Users this user has interacted with via introduction requests
    ids = set(
        IntroductionRequest.objects.filter(Q(from_user=u) | Q(to_user=u))
        .values_list("from_user_id", "to_user_id")
    )
    # Flatten and drop self id
    flat = {i for pair in ids for i in pair if i and i != u.id}
    return flat


def graph_similarity(u: User, v: User) -> float:
    nu, nv = neighbor_set(u), neighbor_set(v)
    return jaccard(nu, nv)


def has_existing_request(u: User, v: User) -> bool:
    return IntroductionRequest.objects.filter(
        (Q(from_user=u, to_user=v) | Q(from_user=v, to_user=u))
    ).exists()


@dataclass
class RankedProfile:
    profile: Profile
    score: float


def rank_users_for(user: User, k: int = 20) -> List[RankedProfile]:
    try:
        me_profile = Profile.objects.select_related("user").get(user=user)
    except Profile.DoesNotExist:
        me_profile = Profile(user=user)  # empty defaults

    my_interests = profile_interests(me_profile)
    ranked: List[Tuple[float, Profile]] = []

    qs = (
        Profile.objects.select_related("user")
        .exclude(user=user)
    )

    for p in qs.iterator():
        if has_existing_request(user, p.user):
            continue

        jac = jaccard(my_interests, profile_interests(p))
        # Industry match: exact string match and non-empty
        industry_match = 1.0 if (me_profile.industry and me_profile.industry == p.industry) else 0.0
        ver = 1.0 if p.is_verified else 0.0
        gsim = graph_similarity(user, p.user)

        # Candidate activity recency: last accepted intro involving candidate
        last_acc = (
            IntroductionRequest.objects.filter(
                Q(from_user=p.user) | Q(to_user=p.user), is_accepted=True
            )
            .order_by("-created_at")
            .values_list("created_at", flat=True)
            .first()
        )
        rec = time_decay(last_acc) if last_acc else 0.0

        score = 0.45 * jac + 0.15 * industry_match + 0.20 * gsim + 0.10 * ver + 0.10 * rec
        if score > 0:
            ranked.append((score, p))

    ranked.sort(key=lambda t: t[0], reverse=True)
    top = ranked[:k]
    return [RankedProfile(profile=p, score=s) for s, p in top]


@dataclass
class RankedOpportunity:
    opportunity: Opportunity
    score: float


def rank_opportunities_for(user: User, k: int = 20) -> List[RankedOpportunity]:
    try:
        me_profile = Profile.objects.select_related("user").get(user=user)
    except Profile.DoesNotExist:
        me_profile = Profile(user=user)  # empty defaults

    query_tokens = tokenize_text((me_profile.bio or "") + " " + (me_profile.interests or "") + " " + (me_profile.industry or ""))

    ranked: List[Tuple[float, Opportunity]] = []
    for o in Opportunity.objects.all().iterator():
        doc_tokens = tokenize_text(o.title + " " + (o.description or ""))
        sim = cosine(query_tokens, doc_tokens)
        freshness = time_decay(o.created_at, half_life_days=10.0)
        score = 0.6 * sim + 0.4 * freshness
        if score > 0:
            ranked.append((score, o))

    ranked.sort(key=lambda t: t[0], reverse=True)
    top = ranked[:k]
    return [RankedOpportunity(opportunity=o, score=s) for s, o in top]