diora-web/books/views.py

import base64
import json
import re
import time
import xml.etree.ElementTree as ET
from datetime import datetime, timezone as dt_timezone

import requests
from django.conf import settings
from django.db import transaction
from django.http import JsonResponse
from django.views.decorators.csrf import csrf_exempt
from django.views.decorators.http import require_http_methods

from .models import EBook, EBookProgress, EBookHighlights, EBookBookmarks


def _require_auth(request):
    if not request.user.is_authenticated:
        return JsonResponse({'error': 'authentication required'}, status=401)
    return None


# ---------------------------------------------------------------------------
# Library-shelf metadata lookup (DNB, falling back to Open Library)
#
# Only ever called on explicit user request (the "Metadaten" book action), never
# automatically — the server briefly sees the plaintext ISBN for this one proxied
# request, which is a deliberate, narrow exception to the "server never sees book
# content" rule (see CLAUDE.md), made because the encryption's real purpose here is
# to keep the operator from being able to see what's on the platform (piracy
# liability), not strict user privacy — an ISBN lookup against public library
# catalogs doesn't undermine that. Nothing from this lookup is persisted server-side;
# the resulting label is stored only in the client's encrypted meta blob.
# ---------------------------------------------------------------------------

# DNB "Sachgruppen der Deutschen Nationalbibliografie" / DDC divisions (hundreds -> tens),
# used to turn a raw DDC notation like "833.92" into a short shelf label.
_DDC_DIVISIONS = {
    '000': 'Allgemeines, Informatik', '010': 'Bibliografien', '020': 'Bibliotheks- und Informationswissenschaft',
    '030': 'Enzyklopädien', '050': 'Zeitschriften, fortlaufende Sammelwerke', '060': 'Organisationen, Museumswissenschaft',
    '070': 'Nachrichtenmedien, Journalismus, Verlagswesen', '090': 'Handschriften, seltene Bücher',
    '100': 'Philosophie', '130': 'Parapsychologie, Okkultismus', '150': 'Psychologie',
    '200': 'Religion', '230': 'Christentum, Christliche Theologie', '290': 'Andere Religionen',
    '300': 'Sozialwissenschaften, Soziologie', '310': 'Statistiken', '320': 'Politikwissenschaft',
    '330': 'Wirtschaft', '340': 'Recht', '350': 'Öffentliche Verwaltung, Militärwissenschaft',
    '360': 'Soziale Probleme, Sozialdienste, Versicherungen', '370': 'Erziehung, Schul- und Bildungswesen',
    '380': 'Handel, Kommunikation, Verkehr', '390': 'Gebräuche, Etikette, Folklore',
    '400': 'Sprache, Linguistik', '420': 'Englisch', '430': 'Deutsch, Germanische Sprachen',
    '440': 'Französisch, Romanische Sprachen', '450': 'Italienisch, Rumänisch, Rätoromanisch',
    '460': 'Spanisch, Portugiesisch', '470': 'Latein, Italische Sprachen', '480': 'Griechisch', '490': 'Andere Sprachen',
    '500': 'Naturwissenschaften', '510': 'Mathematik', '520': 'Astronomie', '530': 'Physik', '540': 'Chemie',
    '550': 'Geowissenschaften', '560': 'Paläontologie', '570': 'Biowissenschaften, Biologie',
    '580': 'Pflanzen (Botanik)', '590': 'Tiere (Zoologie)',
    '600': 'Technik', '610': 'Medizin, Gesundheit', '620': 'Ingenieurwissenschaften', '630': 'Landwirtschaft',
    '640': 'Hauswirtschaft', '650': 'Management', '660': 'Chemische Technik', '670': 'Industrielle Fertigung',
    '680': 'Fertigung für spezielle Zwecke', '690': 'Hausbau, Bauhandwerk',
    '700': 'Künste', '710': 'Landschaftsgestaltung, Raumplanung', '720': 'Architektur',
    '730': 'Plastik, Keramik, Metallkunst', '740': 'Zeichnung, angewandte Kunst', '750': 'Malerei',
    '760': 'Grafik, Druckgrafik, Fotografie', '780': 'Musik', '790': 'Freizeit, Darstellende Kunst, Sport',
    '800': 'Literatur', '810': 'Amerikanische Literatur', '820': 'Englische Literatur',
    '830': 'Deutsche Literatur', '840': 'Französische Literatur', '850': 'Italienische Literatur',
    '860': 'Spanische, Portugiesische Literatur', '870': 'Lateinische Literatur', '880': 'Griechische Literatur',
    '890': 'Literaturen in anderen Sprachen',
    '900': 'Geschichte', '910': 'Geografie, Reisen', '920': 'Biografie, Genealogie',
    '930': 'Geschichte des Altertums', '940': 'Geschichte Europas', '950': 'Geschichte Asiens',
    '960': 'Geschichte Afrikas', '970': 'Geschichte Nordamerikas', '980': 'Geschichte Südamerikas',
    '990': 'Geschichte der übrigen Welt',
}


def _ddc_label(ddc_raw):
    """Coarsen a DDC notation ('833.92') down to its nearest known division/class label."""
    digits = re.sub(r'\D', '', ddc_raw or '')
    if len(digits) < 3:
        return None
    tens = digits[:2] + '0'
    if tens in _DDC_DIVISIONS:
        return _DDC_DIVISIONS[tens]
    return _DDC_DIVISIONS.get(digits[0] + '00')


_MARC_NS = '{http://www.loc.gov/MARC21/slim}'


def _lookup_dnb_shelf(isbn):
    url = 'https://services.dnb.de/sru/dnb'
    params = {
        'version': '1.1',
        'operation': 'searchRetrieve',
        'query': f'dnb.num={isbn}',
        'recordSchema': 'MARC21-xml',
        'maximumRecords': '1',
    }
    resp = requests.get(url, params=params, timeout=getattr(settings, 'BOOK_METADATA_TIMEOUT', 6))
    resp.raise_for_status()
    root = ET.fromstring(resp.content)
    for datafield in root.iter(f'{_MARC_NS}datafield'):
        if datafield.get('tag') != '082':
            continue
        for subfield in datafield.findall(f'{_MARC_NS}subfield'):
            if subfield.get('code') == 'a' and subfield.text:
                label = _ddc_label(subfield.text)
                if label:
                    return label
    return None


def _lookup_openlibrary_shelf(isbn):
    url = 'https://openlibrary.org/api/books'
    params = {'bibkeys': f'ISBN:{isbn}', 'jscmd': 'data', 'format': 'json'}
    resp = requests.get(url, params=params, timeout=getattr(settings, 'BOOK_METADATA_TIMEOUT', 6))
    resp.raise_for_status()
    data = resp.json().get(f'ISBN:{isbn}', {})
    ddc = (data.get('classifications') or {}).get('dewey_decimal_class') or []
    if ddc:
        label = _ddc_label(ddc[0])
        if label:
            return label
    subjects = data.get('subjects') or []
    if subjects:
        return subjects[0].get('name')
    return None


def _cql_phrase(value):
    # CQL string literals are quoted; strip embedded quotes rather than escaping them
    # (this only feeds a lookup heuristic, not a stored/displayed value).
    return value.replace('"', ' ').strip()


def _lookup_dnb_shelf_by_title(title, author):
    url = 'https://services.dnb.de/sru/dnb'
    query = f'dnb.tit="{_cql_phrase(title)}"'
    if author:
        query += f' and dnb.per="{_cql_phrase(author)}"'
    params = {
        'version': '1.1',
        'operation': 'searchRetrieve',
        'query': query,
        'recordSchema': 'MARC21-xml',
        'maximumRecords': '1',
    }
    resp = requests.get(url, params=params, timeout=getattr(settings, 'BOOK_METADATA_TIMEOUT', 6))
    resp.raise_for_status()
    root = ET.fromstring(resp.content)
    for datafield in root.iter(f'{_MARC_NS}datafield'):
        if datafield.get('tag') != '082':
            continue
        for subfield in datafield.findall(f'{_MARC_NS}subfield'):
            if subfield.get('code') == 'a' and subfield.text:
                label = _ddc_label(subfield.text)
                if label:
                    return label
    return None


def _lookup_openlibrary_shelf_by_title(title, author):
    url = 'https://openlibrary.org/search.json'
    params = {'title': title, 'limit': 1, 'fields': 'ddc,subject'}
    if author:
        params['author'] = author
    resp = requests.get(url, params=params, timeout=getattr(settings, 'BOOK_METADATA_TIMEOUT', 6))
    resp.raise_for_status()
    docs = resp.json().get('docs') or []
    if not docs:
        return None
    doc = docs[0]
    ddc = doc.get('ddc') or []
    if ddc:
        label = _ddc_label(ddc[0])
        if label:
            return label
    subjects = doc.get('subject') or []
    if subjects:
        return subjects[0]
    return None


@require_http_methods(['GET'])
def lookup_book_metadata(request):
    err = _require_auth(request)
    if err:
        return err

    isbn_raw = request.GET.get('isbn', '').strip()
    title = request.GET.get('title', '').strip()
    author = request.GET.get('author', '').strip()
    isbn = re.sub(r'[^0-9Xx]', '', isbn_raw)

    if isbn_raw and len(isbn) not in (10, 13):
        return JsonResponse({'error': 'invalid ISBN'}, status=400)
    if not isbn and not title:
        return JsonResponse({'error': 'isbn or title required'}, status=400)

    label = None
    source = None

    # ISBN is the precise path — try it first when we have one.
    if isbn:
        try:
            label = _lookup_dnb_shelf(isbn)
            if label:
                source = 'dnb'
        except Exception:
            pass
        if not label:
            try:
                label = _lookup_openlibrary_shelf(isbn)
                if label:
                    source = 'openlibrary'
            except Exception:
                pass

    # No ISBN (or it drew a blank) — fall back to a title/author text search. Less
    # precise (wrong edition/translation is possible), so the source is tagged
    # distinctly for the client to hint at that if it wants to.
    if not label and title:
        try:
            label = _lookup_dnb_shelf_by_title(title, author)
            if label:
                source = 'dnb-title'
        except Exception:
            pass
        if not label:
            try:
                label = _lookup_openlibrary_shelf_by_title(title, author)
                if label:
                    source = 'openlibrary-title'
            except Exception:
                pass

    return JsonResponse({'label': label, 'source': source})


def _anchor_parts(anchor):
    """Split a position anchor 'blockIndex:innerFraction' into (block, inner).

    Returns (-1, 0.0) for empty/invalid anchors (e.g. PDF progress)."""
    if not isinstance(anchor, str) or ':' not in anchor:
        return (-1, 0.0)
    block, _, inner = anchor.partition(':')
    try:
        return (int(block), float(inner))
    except ValueError:
        return (-1, 0.0)


def _progress_is_further(new_anchor, new_frac, old_anchor, old_frac):
    """True if the new reading position is at least as far into the book as the
    old one. Compares by anchor block index (precise, decoupled from layout);
    falls back to scroll_fraction only when an anchor is missing (PDFs)."""
    return _cmp_progress(new_anchor, new_frac, old_anchor, old_frac) >= 0


def _cmp_progress(a_anchor, a_frac, b_anchor, b_frac):
    """> 0 if A is further into the book than B, < 0 if behind, 0 if level."""
    ab, ai = _anchor_parts(a_anchor)
    bb, bi = _anchor_parts(b_anchor)
    if ab >= 0 and bb >= 0:
        if ab != bb:
            return ab - bb
        return (ai > bi) - (ai < bi)
    return (a_frac > b_frac) - (a_frac < b_frac)


def _now_ms():
    return int(time.time() * 1000)


def progress_state(p):
    """A book's reading state as every client sees it (docs/progress-sync.md);
    p may be None (never opened)."""
    if p is None:
        return {'scroll_fraction': 0.0, 'position_anchor': '', 'read_at': None, 'device': '',
                'furthest_fraction': 0.0, 'furthest_anchor': '', 'last_read': None}
    read_at = p.read_at if p.read_at is not None else int(p.updated_at.timestamp() * 1000)
    return {
        'scroll_fraction': p.scroll_fraction,
        'position_anchor': p.position_anchor,
        'read_at': read_at,
        'device': p.device,
        'furthest_fraction': p.furthest_fraction,
        'furthest_anchor': p.furthest_anchor,
        'last_read': datetime.fromtimestamp(read_at / 1000, tz=dt_timezone.utc).isoformat(),
    }


@require_http_methods(['GET'])
def book_list(request):
    err = _require_auth(request)
    if err:
        return err
    books = list(
        request.user.ebooks.values('id', 'meta_ct', 'meta_iv', 'uploaded_at', 'is_read')
    )
    for b in books:
        b['uploaded_at'] = b['uploaded_at'].isoformat()
    progress_map = {p.book_id: p for p in EBookProgress.objects.filter(user=request.user)}
    highlighted_ids = set(
        EBookHighlights.objects.filter(user=request.user).values_list('book_id', flat=True)
    )
    for b in books:
        b.update(progress_state(progress_map.get(b['id'])))
        b['has_highlights'] = b['id'] in highlighted_ids
    response = JsonResponse(books, safe=False)
    # The list is a bare array (old clients expect that): the server clock goes in a header.
    response['X-Server-Now'] = str(_now_ms())
    return response


@csrf_exempt
@require_http_methods(['POST'])
def set_book_read(request, pk):
    err = _require_auth(request)
    if err:
        return err

    try:
        book = EBook.objects.get(pk=pk, user=request.user)
    except EBook.DoesNotExist:
        return JsonResponse({'error': 'not found'}, status=404)

    try:
        body = json.loads(request.body)
    except (json.JSONDecodeError, ValueError):
        return JsonResponse({'error': 'invalid JSON'}, status=400)

    book.is_read = bool(body.get('is_read', True))
    book.save(update_fields=['is_read'])
    return JsonResponse({'ok': True, 'is_read': book.is_read})


@csrf_exempt
@require_http_methods(['POST'])
def update_book_meta(request, pk):
    """Update only the encrypted metadata blob (e.g. to assign a folder) without touching book bytes."""
    err = _require_auth(request)
    if err:
        return err

    try:
        book = EBook.objects.get(pk=pk, user=request.user)
    except EBook.DoesNotExist:
        return JsonResponse({'error': 'not found'}, status=404)

    try:
        body = json.loads(request.body)
    except (json.JSONDecodeError, ValueError):
        return JsonResponse({'error': 'invalid JSON'}, status=400)

    meta_ct = body.get('meta_ct', '')
    meta_iv = body.get('meta_iv', '')
    if not meta_ct or not meta_iv:
        return JsonResponse({'error': 'meta_ct, meta_iv required'}, status=400)

    book.meta_ct = meta_ct
    book.meta_iv = meta_iv
    book.save(update_fields=['meta_ct', 'meta_iv'])
    return JsonResponse({'ok': True})


@csrf_exempt
@require_http_methods(['POST'])
def upload_book(request):
    err = _require_auth(request)
    if err:
        return err

    try:
        body = json.loads(request.body)
    except (json.JSONDecodeError, ValueError):
        return JsonResponse({'error': 'invalid JSON'}, status=400)

    meta_ct = body.get('meta_ct', '')
    meta_iv = body.get('meta_iv', '')
    data_ct = body.get('data_ct', '')
    data_iv = body.get('data_iv', '')

    if not all([meta_ct, meta_iv, data_ct, data_iv]):
        return JsonResponse({'error': 'meta_ct, meta_iv, data_ct, data_iv required'}, status=400)

    # Enforce size limit: ciphertext is plaintext + 16-byte GCM tag
    max_bytes = getattr(settings, 'EBOOK_MAX_BYTES', 10 * 1024 * 1024) + 32
    try:
        raw_size = len(base64.b64decode(data_ct))
    except Exception:
        return JsonResponse({'error': 'invalid base64 in data_ct'}, status=400)

    if raw_size > max_bytes:
        return JsonResponse({'error': 'file too large (max 50 MB)'}, status=400)

    book = EBook.objects.create(
        user=request.user,
        meta_ct=meta_ct,
        meta_iv=meta_iv,
        data_ct=data_ct,
        data_iv=data_iv,
    )
    EBookProgress.objects.create(user=request.user, book=book)
    return JsonResponse({'ok': True, 'id': book.id})


@require_http_methods(['GET'])
def get_book_data(request, pk):
    err = _require_auth(request)
    if err:
        return err

    try:
        book = EBook.objects.get(pk=pk, user=request.user)
    except EBook.DoesNotExist:
        return JsonResponse({'error': 'not found'}, status=404)

    return JsonResponse({'data_ct': book.data_ct, 'data_iv': book.data_iv})


@csrf_exempt
@require_http_methods(['POST'])
def replace_book_data(request, pk):
    err = _require_auth(request)
    if err:
        return err

    try:
        book = EBook.objects.get(pk=pk, user=request.user)
    except EBook.DoesNotExist:
        return JsonResponse({'error': 'not found'}, status=404)

    try:
        body = json.loads(request.body)
    except (json.JSONDecodeError, ValueError):
        return JsonResponse({'error': 'invalid JSON'}, status=400)

    data_ct  = body.get('data_ct', '')
    data_iv  = body.get('data_iv', '')
    meta_ct  = body.get('meta_ct', '')
    meta_iv  = body.get('meta_iv', '')

    if not all([data_ct, data_iv]):
        return JsonResponse({'error': 'data_ct and data_iv required'}, status=400)

    update_fields = ['data_ct', 'data_iv']
    book.data_ct = data_ct
    book.data_iv = data_iv
    if meta_ct and meta_iv:
        book.meta_ct = meta_ct
        book.meta_iv = meta_iv
        update_fields += ['meta_ct', 'meta_iv']
    book.save(update_fields=update_fields)
    return JsonResponse({'ok': True})


@csrf_exempt
@require_http_methods(['POST'])
def rekey_book(request, pk):
    err = _require_auth(request)
    if err:
        return err

    try:
        book = EBook.objects.get(pk=pk, user=request.user)
    except EBook.DoesNotExist:
        return JsonResponse({'error': 'not found'}, status=404)

    try:
        body = json.loads(request.body)
    except (json.JSONDecodeError, ValueError):
        return JsonResponse({'error': 'invalid JSON'}, status=400)

    meta_ct = body.get('meta_ct', '')
    meta_iv = body.get('meta_iv', '')
    data_ct = body.get('data_ct', '')
    data_iv = body.get('data_iv', '')

    if not all([meta_ct, meta_iv, data_ct, data_iv]):
        return JsonResponse({'error': 'meta_ct, meta_iv, data_ct, data_iv required'}, status=400)

    book.meta_ct = meta_ct
    book.meta_iv = meta_iv
    book.data_ct = data_ct
    book.data_iv = data_iv
    book.save(update_fields=['meta_ct', 'meta_iv', 'data_ct', 'data_iv'])
    return JsonResponse({'ok': True})


@csrf_exempt
@require_http_methods(['POST'])
def delete_book(request, pk):
    err = _require_auth(request)
    if err:
        return err

    try:
        book = EBook.objects.get(pk=pk, user=request.user)
    except EBook.DoesNotExist:
        return JsonResponse({'error': 'not found'}, status=404)

    book.delete()
    return JsonResponse({'ok': True})


@csrf_exempt
@require_http_methods(['POST'])
def save_progress(request, pk):
    err = _require_auth(request)
    if err:
        return err

    try:
        book = EBook.objects.get(pk=pk, user=request.user)
    except EBook.DoesNotExist:
        return JsonResponse({'error': 'not found'}, status=404)

    try:
        body = json.loads(request.body)
    except (json.JSONDecodeError, ValueError):
        return JsonResponse({'error': 'invalid JSON'}, status=400)

    scroll_fraction = float(body.get('scroll_fraction', 0.0))
    scroll_fraction = max(0.0, min(1.0, scroll_fraction))

    raw_anchor = body.get('position_anchor', '')
    position_anchor = ''
    if isinstance(raw_anchor, str) and re.fullmatch(r'\d{1,7}:\d(\.\d{1,6})?', raw_anchor):
        position_anchor = raw_anchor

    force = body.get('force') is True
    device = body.get('device') if isinstance(body.get('device'), str) else ''
    now = _now_ms()
    read_at = _client_ms(body.get('read_at'))
    sent_at = _client_ms(body.get('sent_at'))

    with transaction.atomic():
        progress, created = EBookProgress.objects.select_for_update().get_or_create(user=request.user, book=book)
        if created:
            progress.read_at = None
        if force:
            when, store = now, True
        elif read_at is not None:
            # The device's clock may be off: move its time onto ours by the
            # difference at sending, and nothing counts as later than now.
            when = read_at + (now - sent_at) if sent_at is not None else read_at
            when = min(when, now)
            store = progress.read_at is None or when >= progress.read_at
        else:
            # An app from before 2026-10-12: only forward, as it used to be — but
            # strictly, so re-saving the same place changes nothing.
            when = now
            store = created or _cmp_progress(position_anchor, scroll_fraction,
                                             progress.position_anchor, progress.scroll_fraction) > 0
        if store:
            progress.scroll_fraction = scroll_fraction
            progress.position_anchor = position_anchor
            progress.read_at = when
            progress.device = device[:40]
        if _cmp_progress(position_anchor, scroll_fraction,
                         progress.furthest_anchor, progress.furthest_fraction) > 0 or created:
            progress.furthest_fraction = scroll_fraction
            progress.furthest_anchor = position_anchor
        progress.save()

    return JsonResponse({'ok': True, 'stored': store, 'server_now': now, **progress_state(progress)})


def _client_ms(value):
    """A time from a client (ms since epoch) or None if it isn't a plausible one."""
    if isinstance(value, bool) or not isinstance(value, (int, float)):
        return None
    value = int(value)
    return value if 10 ** 12 <= value < 10 ** 14 else None


@csrf_exempt
@require_http_methods(['GET', 'POST'])
def book_highlights(request, pk):
    err = _require_auth(request)
    if err:
        return err

    try:
        book = EBook.objects.get(pk=pk, user=request.user)
    except EBook.DoesNotExist:
        return JsonResponse({'error': 'not found'}, status=404)

    if request.method == 'GET':
        try:
            row = EBookHighlights.objects.get(user=request.user, book=book)
            return JsonResponse({'ct': row.ct, 'iv': row.iv})
        except EBookHighlights.DoesNotExist:
            return JsonResponse({'ct': None, 'iv': None})

    # POST — upsert
    try:
        body = json.loads(request.body)
    except (json.JSONDecodeError, ValueError):
        return JsonResponse({'error': 'invalid JSON'}, status=400)

    ct = body.get('ct', '')
    iv = body.get('iv', '')
    if not ct or not iv:
        return JsonResponse({'error': 'ct and iv required'}, status=400)

    # Size guard: highlights ≤ 700 KB base64
    try:
        raw_size = len(base64.b64decode(ct))
    except Exception:
        return JsonResponse({'error': 'invalid base64 in ct'}, status=400)
    if raw_size > getattr(settings, 'HIGHLIGHTS_MAX_BYTES', 700 * 1024):
        return JsonResponse({'error': 'highlights data too large (max 700 KB)'}, status=400)

    row, _ = EBookHighlights.objects.get_or_create(user=request.user, book=book)
    row.ct = ct
    row.iv = iv
    row.save(update_fields=['ct', 'iv', 'updated_at'])
    return JsonResponse({'ok': True})


@csrf_exempt
@require_http_methods(['GET', 'POST'])
def book_bookmarks(request, pk):
    err = _require_auth(request)
    if err:
        return err

    try:
        book = EBook.objects.get(pk=pk, user=request.user)
    except EBook.DoesNotExist:
        return JsonResponse({'error': 'not found'}, status=404)

    if request.method == 'GET':
        try:
            row = EBookBookmarks.objects.get(user=request.user, book=book)
            return JsonResponse({'ct': row.ct, 'iv': row.iv})
        except EBookBookmarks.DoesNotExist:
            return JsonResponse({'ct': None, 'iv': None})

    # POST — upsert
    try:
        body = json.loads(request.body)
    except (json.JSONDecodeError, ValueError):
        return JsonResponse({'error': 'invalid JSON'}, status=400)

    ct = body.get('ct', '')
    iv = body.get('iv', '')
    if not ct or not iv:
        return JsonResponse({'error': 'ct and iv required'}, status=400)

    # Size guard: bookmarks ≤ 100 KB base64
    try:
        raw_size = len(base64.b64decode(ct))
    except Exception:
        return JsonResponse({'error': 'invalid base64 in ct'}, status=400)
    if raw_size > getattr(settings, 'BOOKMARKS_MAX_BYTES', 100 * 1024):
        return JsonResponse({'error': 'bookmarks data too large (max 100 KB)'}, status=400)

    row, _ = EBookBookmarks.objects.get_or_create(user=request.user, book=book)
    row.ct = ct
    row.iv = iv
    row.save(update_fields=['ct', 'iv', 'updated_at'])
    return JsonResponse({'ok': True})