hadith-api/app/routers/narrators.py

"""
Narrator endpoints — search, profiles, network queries.
All queries normalize Arabic input to match post-dedup graph data.
"""
from fastapi import APIRouter, Query, Path, HTTPException
from typing import Optional

from app.services.database import db
from app.utils.arabic import normalize_query, normalize_name
from app.models.schemas import (
    NarratorSummary, NarratorProfile, NarratorInteraction,
    NarratorConnection, NarratorNetwork,
    WhoMetWhoResult, PathNode, PlaceRelation, NameForm, FamilyInfo,
    HadithSummary, PaginatedResponse, PaginationMeta,
)

router = APIRouter(prefix="/narrators", tags=["Narrators"])


def _paginate(total: int, page: int, per_page: int) -> PaginationMeta:
    pages = max(1, (total + per_page - 1) // per_page)
    return PaginationMeta(total=total, page=page, per_page=per_page, pages=pages)


# ── Search narrators by name (paginated, normalized) ───────────────────────

@router.get("/search", response_model=PaginatedResponse)
async def search_narrators(
    q: str = Query(..., min_length=2, description="Narrator name (Arabic). Diacritics stripped automatically."),
    page: int = Query(1, ge=1),
    per_page: int = Query(20, ge=1, le=100),
):
    """
    Search narrators by Arabic name. Input is normalized to match
    the deduplicated graph (diacritics stripped, characters unified).
    """
    q_norm = normalize_query(q)
    skip = (page - 1) * per_page

    total = db.neo4j_count("""
        MATCH (n:Narrator)
        WHERE toLower(n.name_arabic) CONTAINS toLower($q)
        RETURN count(n) AS count
    """, {"q": q_norm})

    rows = db.neo4j_query("""
        MATCH (n:Narrator)
        WHERE toLower(n.name_arabic) CONTAINS toLower($q)
        OPTIONAL MATCH (n)-[:APPEARS_IN]->(h:Hadith)
        RETURN n.name_arabic AS name_arabic,
               n.name_transliterated AS name_transliterated,
               n.entity_type AS entity_type,
               n.generation AS generation,
               n.reliability_grade AS reliability_grade,
               count(DISTINCT h) AS hadith_count
        ORDER BY hadith_count DESC
        SKIP $skip LIMIT $limit
    """, {"q": q_norm, "skip": skip, "limit": per_page})

    data = [NarratorSummary(**r) for r in rows]
    return PaginatedResponse(data=data, meta=_paginate(total, page, per_page))


# ── Full narrator profile ──────────────────────────────────────────────────

@router.get("/profile/{name_arabic}", response_model=NarratorProfile)
async def get_narrator_profile(
    name_arabic: str = Path(..., description="Narrator Arabic name (exact or close match)"),
):
    """
    Complete narrator profile — biography, hadiths, teachers, students, places, tribes.
    This is the mobile app profile page query.
    """
    q_norm = normalize_name(name_arabic)

    # Find the narrator node — exact first, then CONTAINS fallback
    narrator = db.neo4j_query_one("""
        MATCH (n:Narrator)
        WHERE toLower(n.name_arabic) = toLower($q)
        RETURN n
    """, {"q": q_norm})

    if not narrator:
        narrator = db.neo4j_query_one("""
            MATCH (n:Narrator)
            WHERE toLower(n.name_arabic) CONTAINS toLower($q)
            RETURN n
        """, {"q": q_norm})

    if not narrator:
        raise HTTPException(status_code=404, detail=f"Narrator not found: {name_arabic}")

    n = narrator.get("n", {})
    actual_name = n.get("name_arabic", q_norm)

    # Hadith count + collections
    stats = db.neo4j_query_one("""
        MATCH (n:Narrator {name_arabic: $name})-[:APPEARS_IN]->(h:Hadith)
        RETURN count(DISTINCT h) AS hadith_count,
               collect(DISTINCT h.collection) AS collections
    """, {"name": actual_name}) or {}

    # Teachers: narrator NARRATED_FROM teacher + teacher TEACHER_OF narrator
    teachers_nf = db.neo4j_query("""
        MATCH (n:Narrator {name_arabic: $name})-[:NARRATED_FROM]->(t:Narrator)
        OPTIONAL MATCH (t)-[:APPEARS_IN]->(h:Hadith)
        RETURN t.name_arabic AS name_arabic,
               t.name_transliterated AS name_transliterated,
               t.entity_type AS entity_type,
               t.generation AS generation,
               t.reliability_grade AS reliability_grade,
               count(DISTINCT h) AS hadith_count
    """, {"name": actual_name})

    teachers_to = db.neo4j_query("""
        MATCH (t:Narrator)-[:TEACHER_OF]->(n:Narrator {name_arabic: $name})
        OPTIONAL MATCH (t)-[:APPEARS_IN]->(h:Hadith)
        RETURN t.name_arabic AS name_arabic,
               t.name_transliterated AS name_transliterated,
               t.entity_type AS entity_type,
               t.generation AS generation,
               t.reliability_grade AS reliability_grade,
               count(DISTINCT h) AS hadith_count
    """, {"name": actual_name})

    # Deduplicate teachers
    seen_teachers = set()
    teachers = []
    for r in teachers_nf + teachers_to:
        if r["name_arabic"] not in seen_teachers:
            seen_teachers.add(r["name_arabic"])
            teachers.append(NarratorSummary(**r))

    # Students: student NARRATED_FROM narrator + narrator TEACHER_OF student
    students_nf = db.neo4j_query("""
        MATCH (s:Narrator)-[:NARRATED_FROM]->(n:Narrator {name_arabic: $name})
        OPTIONAL MATCH (s)-[:APPEARS_IN]->(h:Hadith)
        RETURN s.name_arabic AS name_arabic,
               s.name_transliterated AS name_transliterated,
               s.entity_type AS entity_type,
               s.generation AS generation,
               s.reliability_grade AS reliability_grade,
               count(DISTINCT h) AS hadith_count
    """, {"name": actual_name})

    students_to = db.neo4j_query("""
        MATCH (n:Narrator {name_arabic: $name})-[:TEACHER_OF]->(s:Narrator)
        OPTIONAL MATCH (s)-[:APPEARS_IN]->(h:Hadith)
        RETURN s.name_arabic AS name_arabic,
               s.name_transliterated AS name_transliterated,
               s.entity_type AS entity_type,
               s.generation AS generation,
               s.reliability_grade AS reliability_grade,
               count(DISTINCT h) AS hadith_count
    """, {"name": actual_name})

    seen_students = set()
    students = []
    for r in students_nf + students_to:
        if r["name_arabic"] not in seen_students:
            seen_students.add(r["name_arabic"])
            students.append(NarratorSummary(**r))

    # Places
    places_rows = db.neo4j_query("""
        MATCH (n:Narrator {name_arabic: $name})-[r]->(p:Place)
        WHERE type(r) IN ['BORN_IN', 'LIVED_IN', 'DIED_IN', 'TRAVELED_TO']
        RETURN p.name_arabic AS place, type(r) AS relation
    """, {"name": actual_name})

    # Tribes
    tribe_rows = db.neo4j_query("""
        MATCH (n:Narrator {name_arabic: $name})-[:BELONGS_TO_TRIBE]->(t:Tribe)
        RETURN t.name_arabic AS name
    """, {"name": actual_name})

    # Name forms (alternative names via RELATED_TO)
    name_form_rows = db.neo4j_query("""
        MATCH (n:Narrator {name_arabic: $name})-[:RELATED_TO]-(alt:Narrator)
        WHERE alt.name_arabic <> $name
        RETURN alt.name_arabic AS name, alt.entity_type AS type
    """, {"name": actual_name})

    # Family info
    family_row = db.neo4j_query_one("""
        MATCH (n:Narrator {name_arabic: $name})
        RETURN n.father AS father, n.mother AS mother,
               n.spouse AS spouse, n.children AS children
    """, {"name": actual_name})
    family = None
    if family_row and any(family_row.get(k) for k in ["father", "mother", "spouse", "children"]):
        family = FamilyInfo(
            father=family_row.get("father"),
            mother=family_row.get("mother"),
            spouse=family_row.get("spouse"),
            children=family_row.get("children") or [],
        )

    # Sample hadiths
    hadith_rows = db.neo4j_query("""
        MATCH (n:Narrator {name_arabic: $name})-[:APPEARS_IN]->(h:Hadith)
        RETURN h.id AS id,
               h.collection AS collection,
               h.hadith_number AS hadith_number,
               h.grade AS grade,
               substring(h.arabic_text, 0, 300) AS arabic_text
        ORDER BY h.collection, h.hadith_number
        LIMIT 50
    """, {"name": actual_name})

    return NarratorProfile(
        name_arabic=n.get("name_arabic", actual_name),
        name_transliterated=n.get("name_transliterated", ""),
        entity_type=n.get("entity_type", ""),
        full_nasab=n.get("full_nasab"),
        kunya=n.get("kunya"),
        nisba=n.get("nisba"),
        laqab=n.get("laqab"),
        generation=n.get("generation"),
        reliability_grade=n.get("reliability_grade"),
        reliability_detail=n.get("reliability_detail"),
        birth_year_hijri=n.get("birth_year_hijri"),
        death_year_hijri=n.get("death_year_hijri"),
        birth_year_ce=n.get("birth_year_ce"),
        death_year_ce=n.get("death_year_ce"),
        biography_summary_arabic=n.get("biography_summary_arabic"),
        biography_summary_english=n.get("biography_summary_english"),
        total_hadiths_narrated_approx=n.get("total_hadiths_narrated_approx"),
        hadith_count=stats.get("hadith_count", 0),
        hadiths=[HadithSummary(**r) for r in hadith_rows],
        teachers=teachers,
        students=students,
        name_forms=[NameForm(**r) for r in name_form_rows],
        family=family,
        places=[PlaceRelation(**r) for r in places_rows],
        tribes=[r["name"] for r in tribe_rows],
        bio_verified=n.get("bio_verified", False),
    )


# ── Narrators by generation (paginated, normalized) ────────────────────────

@router.get("/by-generation/{generation}", response_model=PaginatedResponse)
async def narrators_by_generation(
    generation: str = Path(..., description="Generation: صحابي, تابعي, تابع التابعين, نبي"),
    page: int = Query(1, ge=1),
    per_page: int = Query(20, ge=1, le=100),
):
    """List narrators by generation (e.g. Companions, Successors)."""
    q_norm = normalize_query(generation)
    skip = (page - 1) * per_page

    total = db.neo4j_count("""
        MATCH (n:Narrator)
        WHERE toLower(n.generation) CONTAINS toLower($gen)
        RETURN count(n) AS count
    """, {"gen": q_norm})

    rows = db.neo4j_query("""
        MATCH (n:Narrator)
        WHERE toLower(n.generation) CONTAINS toLower($gen)
        OPTIONAL MATCH (n)-[:APPEARS_IN]->(h:Hadith)
        RETURN n.name_arabic AS name_arabic,
               n.name_transliterated AS name_transliterated,
               n.entity_type AS entity_type,
               n.generation AS generation,
               n.reliability_grade AS reliability_grade,
               count(DISTINCT h) AS hadith_count
        ORDER BY hadith_count DESC
        SKIP $skip LIMIT $limit
    """, {"gen": q_norm, "skip": skip, "limit": per_page})

    data = [NarratorSummary(**r) for r in rows]
    return PaginatedResponse(data=data, meta=_paginate(total, page, per_page))


# ── Narrators by place (paginated, normalized) ─────────────────────────────

@router.get("/by-place/{place_name}", response_model=PaginatedResponse)
async def narrators_by_place(
    place_name: str = Path(..., description="Place name in Arabic (e.g. مكة)"),
    page: int = Query(1, ge=1),
    per_page: int = Query(50, ge=1, le=100),
):
    """
    Narrators associated with a place (born, lived, died, traveled).
    Input is normalized — مكة المكرمة matches مكه المكرمه.
    """
    q_norm = normalize_query(place_name)
    skip = (page - 1) * per_page

    total = db.neo4j_count("""
        MATCH (n:Narrator)-[r]->(p:Place)
        WHERE type(r) IN ['BORN_IN', 'LIVED_IN', 'DIED_IN', 'TRAVELED_TO']
          AND toLower(p.name_arabic) CONTAINS toLower($place)
        RETURN count(DISTINCT n) AS count
    """, {"place": q_norm})

    rows = db.neo4j_query("""
        MATCH (n:Narrator)-[r]->(p:Place)
        WHERE type(r) IN ['BORN_IN', 'LIVED_IN', 'DIED_IN', 'TRAVELED_TO']
          AND toLower(p.name_arabic) CONTAINS toLower($place)
        OPTIONAL MATCH (n)-[:APPEARS_IN]->(h:Hadith)
        RETURN DISTINCT n.name_arabic AS name_arabic,
               n.name_transliterated AS name_transliterated,
               n.entity_type AS entity_type,
               n.generation AS generation,
               n.reliability_grade AS reliability_grade,
               count(DISTINCT h) AS hadith_count
        ORDER BY hadith_count DESC
        SKIP $skip LIMIT $limit
    """, {"place": q_norm, "skip": skip, "limit": per_page})

    data = [NarratorSummary(**r) for r in rows]
    return PaginatedResponse(data=data, meta=_paginate(total, page, per_page))


# ── Narrator interactions ──────────────────────────────────────────────────

@router.get("/interactions/{name_arabic}", response_model=list[NarratorInteraction])
async def narrator_interactions(
    name_arabic: str = Path(..., description="Narrator Arabic name"),
    limit: int = Query(50, ge=1, le=200),
):
    """All direct relationships for a narrator — who they narrated from/to."""
    q_norm = normalize_name(name_arabic)

    rows = db.neo4j_query("""
        MATCH (a:Narrator)-[r]-(b:Narrator)
        WHERE toLower(a.name_arabic) CONTAINS toLower($name)
          AND type(r) IN ['NARRATED_FROM', 'TEACHER_OF']
        WITH a, b, type(r) AS rel_type
        OPTIONAL MATCH (a)-[:APPEARS_IN]->(h:Hadith)<-[:APPEARS_IN]-(b)
        RETURN a.name_arabic AS narrator_a,
               a.name_transliterated AS narrator_a_transliterated,
               b.name_arabic AS narrator_b,
               b.name_transliterated AS narrator_b_transliterated,
               rel_type AS relationship_type,
               count(DISTINCT h) AS shared_hadith_count,
               collect(DISTINCT h.id)[..20] AS hadith_ids
        ORDER BY shared_hadith_count DESC
        LIMIT $limit
    """, {"name": q_norm, "limit": limit})

    return [NarratorInteraction(**r) for r in rows]


# ── Narrator network (graph visualization) ─────────────────────────────────

@router.get("/network/{name_arabic}", response_model=NarratorNetwork)
async def narrator_network(
    name_arabic: str = Path(..., description="Narrator Arabic name"),
    limit: int = Query(50, ge=1, le=200),
):
    """
    Get a narrator's connection network — all incoming/outgoing relationships.
    Useful for network visualization.
    """
    q_norm = normalize_name(name_arabic)

    # Center narrator
    center_row = db.neo4j_query_one("""
        MATCH (n:Narrator)
        WHERE toLower(n.name_arabic) CONTAINS toLower($name)
        OPTIONAL MATCH (n)-[:APPEARS_IN]->(h:Hadith)
        RETURN n.name_arabic AS name_arabic,
               n.name_transliterated AS name_transliterated,
               n.entity_type AS entity_type,
               n.generation AS generation,
               n.reliability_grade AS reliability_grade,
               count(DISTINCT h) AS hadith_count
    """, {"name": q_norm})

    if not center_row:
        raise HTTPException(status_code=404, detail=f"Narrator not found: {name_arabic}")

    # Connections
    conn_rows = db.neo4j_query("""
        MATCH (a:Narrator)-[r]-(b:Narrator)
        WHERE toLower(a.name_arabic) CONTAINS toLower($name)
          AND type(r) IN ['NARRATED_FROM', 'TEACHER_OF']
        RETURN b.name_arabic AS narrator,
               b.name_transliterated AS narrator_transliterated,
               type(r) AS connection_type,
               CASE WHEN startNode(r) = a THEN 'outgoing' ELSE 'incoming' END AS direction
        LIMIT $limit
    """, {"name": q_norm, "limit": limit})

    return NarratorNetwork(
        center=NarratorSummary(**center_row),
        connections=[NarratorConnection(**r) for r in conn_rows],
        total_connections=len(conn_rows),
    )


# ── Who met who (shortest path) ────────────────────────────────────────────

@router.get("/who-met-who", response_model=WhoMetWhoResult)
async def who_met_who(
    narrator_a: str = Query(..., description="First narrator (Arabic)"),
    narrator_b: str = Query(..., description="Second narrator (Arabic)"),
):
    """
    Shortest path between two narrators in the knowledge graph.
    Useful to see how a narrator connects to the Prophet ﷺ.
    """
    a_norm = normalize_name(narrator_a)
    b_norm = normalize_name(narrator_b)

    row = db.neo4j_query_one("""
        MATCH (a:Narrator), (b:Narrator)
        WHERE toLower(a.name_arabic) CONTAINS toLower($a)
          AND toLower(b.name_arabic) CONTAINS toLower($b)
        WITH a, b LIMIT 1
        MATCH path = shortestPath((a)-[*..10]-(b))
        RETURN [n IN nodes(path) |
                {name_arabic: n.name_arabic,
                 name_transliterated: n.name_transliterated,
                 generation: n.generation}] AS path_nodes,
               [r IN relationships(path) | type(r)] AS rel_types,
               length(path) AS path_length
    """, {"a": a_norm, "b": b_norm})

    if not row:
        raise HTTPException(
            status_code=404,
            detail=f"No path found between '{narrator_a}' and '{narrator_b}'",
        )

    return WhoMetWhoResult(
        narrator_a=narrator_a,
        narrator_b=narrator_b,
        path=[PathNode(**n) for n in (row.get("path_nodes") or [])],
        path_length=row.get("path_length"),
        relationship_types=row.get("rel_types", []),
    )