X-Git-Url: https://git.mdrn.pl/redakcja.git/blobdiff_plain/282654ea252af7e2d30740b40bccfc9be61dd3a8..c51d32b7b97016260e4c60dda9fbbd9eb25420ce:/src/catalogue/models.py?ds=sidebyside diff --git a/src/catalogue/models.py b/src/catalogue/models.py index 4544fc68..d4111cb2 100644 --- a/src/catalogue/models.py +++ b/src/catalogue/models.py @@ -1,12 +1,20 @@ from collections import Counter +from datetime import date, timedelta import decimal +import io +import re +from urllib.request import urlopen from django.apps import apps +from django.conf import settings from django.db import models from django.template.loader import render_to_string from django.urls import reverse from django.utils.translation import gettext_lazy as _ from admin_ordering.models import OrderableModel from wikidata.client import Client +from librarian import DCNS +from librarian.cover import make_cover +from librarian.dcparser import BookInfo, Person from .constants import WIKIDATA from .wikidata import WikidataModel from .wikimedia import WikiMedia @@ -16,16 +24,25 @@ class Author(WikidataModel): slug = models.SlugField(max_length=255, null=True, blank=True, unique=True) first_name = models.CharField(_("first name"), max_length=255, blank=True) last_name = models.CharField(_("last name"), max_length=255, blank=True) + genitive = models.CharField( + 'dopełniacz', max_length=255, blank=True, + help_text='utwory … (czyje?)' + ) name_de = models.CharField(_("name (de)"), max_length=255, blank=True) name_lt = models.CharField(_("name (lt)"), max_length=255, blank=True) gender = models.CharField(_("gender"), max_length=255, blank=True) nationality = models.CharField(_("nationality"), max_length=255, blank=True) + year_of_birth = models.SmallIntegerField(_("year of birth"), null=True, blank=True) year_of_birth_inexact = models.BooleanField(_("inexact"), default=False) year_of_birth_range = models.SmallIntegerField(_("year of birth, range end"), null=True, blank=True) date_of_birth = models.DateField(_("date_of_birth"), null=True, blank=True) + century_of_birth = models.SmallIntegerField( + _("century of birth"), null=True, blank=True, + help_text=_('Set if year unknown. Negative for BC.') + ) place_of_birth = models.ForeignKey( 'Place', models.PROTECT, null=True, blank=True, verbose_name=_('place of birth'), @@ -35,6 +52,10 @@ class Author(WikidataModel): year_of_death_inexact = models.BooleanField(_("inexact"), default=False) year_of_death_range = models.SmallIntegerField(_("year of death, range end"), null=True, blank=True) date_of_death = models.DateField(_("date_of_death"), null=True, blank=True) + century_of_death = models.SmallIntegerField( + _("century of death"), null=True, blank=True, + help_text=_('Set if year unknown. Negative for BC.') + ) place_of_death = models.ForeignKey( 'Place', models.PROTECT, null=True, blank=True, verbose_name=_('place of death'), @@ -51,7 +72,7 @@ class Author(WikidataModel): (4, _("Unknown")), ], ) - notes = models.TextField(_("notes"), blank=True) + notes = models.TextField(_("notes"), blank=True, help_text=_('private')) gazeta_link = models.CharField(_("gazeta link"), max_length=255, blank=True) culturepl_link = models.CharField(_("culture.pl link"), max_length=255, blank=True) @@ -60,7 +81,7 @@ class Author(WikidataModel): photo_source = models.CharField(blank=True, max_length=255) photo_attribution = models.CharField(max_length=255, blank=True) - description = models.TextField(_("description"), blank=True) + description = models.TextField(_("description"), blank=True, help_text=_('for publication')) priority = models.PositiveSmallIntegerField( _("priority"), @@ -68,6 +89,8 @@ class Author(WikidataModel): ) collections = models.ManyToManyField("Collection", blank=True, verbose_name=_("collections")) + woblink = models.IntegerField(null=True, blank=True) + class Meta: verbose_name = _('author') verbose_name_plural = _('authors') @@ -83,7 +106,7 @@ class Author(WikidataModel): year_of_death = WIKIDATA.DATE_OF_DEATH place_of_death = WIKIDATA.PLACE_OF_DEATH gender = WIKIDATA.GENDER - notes = "description" + notes = WikiMedia.append("description") plwiki = "plwiki" photo = WikiMedia.download(WIKIDATA.IMAGE) photo_source = WikiMedia.descriptionurl(WIKIDATA.IMAGE) @@ -102,6 +125,18 @@ class Author(WikidataModel): def get_absolute_url(self): return reverse("catalogue_author", args=[self.slug]) + @classmethod + def get_by_literal(cls, literal): + names = literal.split(',', 1) + names = [n.strip() for n in names] + if len(names) == 2: + return cls.objects.filter(last_name=names[0], first_name=names[1]).first() + else: + return cls.objects.filter(last_name_pl=names[0], first_name_pl='').first() or \ + cls.objects.filter(first_name_pl=names[0], last_name_pl='').first() or \ + cls.objects.filter(first_name_pl=literal, last_name_pl='').first() or \ + cls.objects.filter(first_name_pl=literal, last_name_pl=None).first() + @property def name(self): return f"{self.last_name}, {self.first_name}" @@ -122,14 +157,59 @@ class Author(WikidataModel): ) return t + def century_description(self, number): + n = abs(number) + letters = '' + while n > 10: + letters += 'X' + n -= 10 + if n == 9: + letters += 'IX' + n = 0 + elif n >= 5: + letters += 'V' + n -= 5 + if n == 4: + letters += 'IV' + n = 0 + letters += 'I' * n + letters += ' w.' + if number < 0: + letters += ' p.n.e.' + return letters + + def birth_century_description(self): + return self.century_description(self.century_of_birth) + + def death_century_description(self): + return self.century_description(self.century_of_death) + + def year_description(self, number): + n = abs(number) + letters = str(n) + letters += ' r.' + if number < 0: + letters += ' p.n.e.' + return letters + + def year_of_birth_description(self): + return self.year_description(self.year_of_birth) + def year_of_death_description(self): + return self.year_description(self.year_of_death) + + class NotableBook(OrderableModel): author = models.ForeignKey(Author, models.CASCADE) book = models.ForeignKey('Book', models.CASCADE) + def __str__(self): + return self.book.title + class Category(WikidataModel): name = models.CharField(_("name"), max_length=255) slug = models.SlugField(max_length=255, unique=True) + description = models.TextField(_("description"), blank=True, help_text=_('for publication')) class Meta: abstract = True @@ -137,19 +217,46 @@ class Category(WikidataModel): def __str__(self): return self.name + class Epoch(Category): + adjective_feminine_singular = models.CharField( + 'przymiotnik pojedynczy żeński', max_length=255, blank=True, + help_text='twórczość … Adama Mickiewicza' + ) + adjective_nonmasculine_plural = models.CharField( + 'przymiotnik mnogi niemęskoosobowy', max_length=255, blank=True, + help_text='utwory … Adama Mickiewicza' + ) + class Meta: verbose_name = _('epoch') verbose_name_plural = _('epochs') class Genre(Category): + thema = models.CharField( + max_length=32, blank=True, + help_text='Odpowiadający kwalifikator Thema.' + ) + plural = models.CharField( + 'liczba mnoga', max_length=255, blank=True, + ) + is_epoch_specific = models.BooleanField( + default=False, + help_text='Po wskazaniu tego gatunku, dodanie epoki byłoby nadmiarowe, np. „dramat romantyczny”' + ) + class Meta: verbose_name = _('genre') verbose_name_plural = _('genres') class Kind(Category): + collective_noun = models.CharField( + 'określenie zbiorowe', max_length=255, blank=True, + help_text='np. „Liryka” albo „Twórczość dramatyczna”' + ) + class Meta: verbose_name = _('kind') verbose_name_plural = _('kinds') @@ -157,6 +264,8 @@ class Kind(Category): class Book(WikidataModel): slug = models.SlugField(max_length=255, blank=True, null=True, unique=True) + parent = models.ForeignKey('self', models.SET_NULL, null=True, blank=True) + parent_number = models.IntegerField(null=True, blank=True) authors = models.ManyToManyField(Author, blank=True, verbose_name=_("authors")) translators = models.ManyToManyField( Author, @@ -176,13 +285,14 @@ class Book(WikidataModel): ) scans_source = models.CharField(_("scans source"), max_length=255, blank=True) text_source = models.CharField(_("text source"), max_length=255, blank=True) - notes = models.TextField(_("notes"), blank=True) + notes = models.TextField(_("notes"), blank=True, help_text=_('private')) priority = models.PositiveSmallIntegerField( _("priority"), default=0, choices=[(0, _("Low")), (1, _("Medium")), (2, _("High"))] ) original_year = models.IntegerField(_('original publication year'), null=True, blank=True) pd_year = models.IntegerField(_('year of entry into PD'), null=True, blank=True) + plwiki = models.CharField(blank=True, max_length=255) gazeta_link = models.CharField(_("gazeta link"), max_length=255, blank=True) collections = models.ManyToManyField("Collection", blank=True, verbose_name=_("collections")) @@ -193,19 +303,25 @@ class Book(WikidataModel): free_license = models.BooleanField(_('free license'), default=False) polona_missing = models.BooleanField(_('missing on Polona'), default=False) + cover = models.FileField(blank=True, upload_to='catalogue/cover') + + monthly_views_reader = models.IntegerField(default=0) + monthly_views_page = models.IntegerField(default=0) + class Meta: ordering = ("title",) verbose_name = _('book') verbose_name_plural = _('books') class Wikidata: + plwiki = "plwiki" authors = WIKIDATA.AUTHOR translators = WIKIDATA.TRANSLATOR title = WIKIDATA.TITLE language = WIKIDATA.LANGUAGE based_on = WIKIDATA.BASED_ON original_year = WIKIDATA.PUBLICATION_DATE - notes = "description" + notes = WikiMedia.append("description") def __str__(self): txt = self.title @@ -219,34 +335,158 @@ class Book(WikidataModel): txt = f"{txt}, tłum. {tstr}" return txt + def build_cover(self): + width, height = 212, 300 + # TODO: BookInfo shouldn't be required to build a cover. + info = BookInfo(rdf_attrs={}, dc_fields={ + DCNS('creator'): [Person('Mickiewicz', 'Adam')], + DCNS('title'): ['Ziutek'], + DCNS('date'): ['1900-01-01'], + DCNS('publisher'): ['F'], + DCNS('language'): ['pol'], + DCNS('identifier.url'): ['test'], + DCNS('rights'): ['-'], + }) + cover = make_cover(info, width=width, height=height) + out = io.BytesIO() + ext = cover.ext() + cover.save(out) + self.cover.save(f'{self.slug}.{ext}', out, save=False) + type(self).objects.filter(pk=self.pk).update(cover=self.cover) + def get_absolute_url(self): return reverse("catalogue_book", args=[self.slug]) + def is_text_public(self): + return self.free_license or (self.pd_year is not None and self.pd_year <= date.today().year) + + def audio_status(self): + return {} + @property def wluri(self): return f'https://wolnelektury.pl/katalog/lektura/{self.slug}/' def authors_str(self): + if not self.pk: + return '' return ", ".join(str(author) for author in self.authors.all()) authors_str.admin_order_field = 'authors__last_name' authors_str.short_description = _('Author') def translators_str(self): + if not self.pk: + return '' return ", ".join(str(author) for author in self.translators.all()) translators_str.admin_order_field = 'translators__last_name' translators_str.short_description = _('Translator') + def authors_first_names(self): + return ', '.join(a.first_name for a in self.authors.all()) + + def authors_last_names(self): + return ', '.join(a.last_name for a in self.authors.all()) + + def translators_first_names(self): + return ', '.join(a.first_name for a in self.translators.all()) + + def translators_last_names(self): + return ', '.join(a.last_name for a in self.translators.all()) + + def document_book__project(self): + b = self.document_books.first() + if b is None: return '' + if b.project is None: return '' + return b.project.name + + def audience(self): + try: + return self.document_books.first().wldocument().book_info.audience or '' + except: + return '' + def get_estimated_costs(self): return { work_type: work_type.calculate(self) for work_type in WorkType.objects.all() } + def scans_galleries(self): + return [bs.pk for bs in self.booksource_set.all()] + + def is_published(self): + return any(b.is_published() for b in self.document_books.all()) + + def update_monthly_stats(self): + # Find publication date. + # By default, get previous 12 months. + this_month = date.today().replace(day=1) + cutoff = this_month.replace(year=this_month.year - 1) + months = 12 + + # If the book was published later, + # find out the denominator. + pbr = apps.get_model('documents', 'BookPublishRecord').objects.filter( + book__catalogue_book=self).order_by('timestamp').first() + if pbr is not None and pbr.timestamp.date() > cutoff: + months = (this_month - pbr.timestamp.date()).days / 365 * 12 + + if not months: + return + + stats = self.bookmonthlystats_set.filter(date__gte=cutoff).aggregate( + views_page=models.Sum('views_page'), + views_reader=models.Sum('views_reader') + ) + self.monthly_views_page = stats['views_page'] / months + self.monthly_views_reader = stats['views_reader'] / months + self.save(update_fields=['monthly_views_page', 'monthly_views_reader']) + + @property + def content_stats(self): + if hasattr(self, '_content_stats'): + return self._content_stats + try: + stats = self.document_books.first().wldocument(librarian2=True).get_statistics()['total'] + except Exception as e: + stats = {} + self._content_stats = stats + return stats + + @property + def are_sources_ready(self): + if not self.booksource_set.exists(): + return False + for bs in self.booksource_set.all(): + if not bs.source.has_view_files() or not bs.source.has_ocr_files() or bs.source.modified_at > bs.source.processed_at: + return False + return True + + chars = lambda self: self.content_stats.get('chars', '') + chars_with_fn = lambda self: self.content_stats.get('chars_with_fn', '') + words = lambda self: self.content_stats.get('words', '') + words_with_fn = lambda self: self.content_stats.get('words_with_fn', '') + verses = lambda self: self.content_stats.get('verses', '') + verses_with_fn = lambda self: self.content_stats.get('verses_with_fn', '') + chars_out_verse = lambda self: self.content_stats.get('chars_out_verse', '') + chars_out_verse_with_fn = lambda self: self.content_stats.get('chars_out_verse_with_fn', '') + + +class EditorNote(models.Model): + book = models.ForeignKey(Book, models.CASCADE) + user = models.ForeignKey(settings.AUTH_USER_MODEL, models.CASCADE) + created_at = models.DateTimeField(auto_now_add=True) + changed_at = models.DateTimeField(auto_now=True) + note = models.TextField(blank=True) + rate = models.IntegerField(default=3, choices=[ + (n, n) for n in range(1, 6) + ]) + class CollectionCategory(models.Model): name = models.CharField(_("name"), max_length=255) parent = models.ForeignKey('self', models.SET_NULL, related_name='children', null=True, blank=True, verbose_name=_("parent")) - notes = models.TextField(_("notes"), blank=True) + notes = models.TextField(_("notes"), blank=True, help_text=_('private')) class Meta: ordering = ('parent__name', 'name') @@ -264,7 +504,7 @@ class Collection(models.Model): name = models.CharField(_("name"), max_length=255) slug = models.SlugField(max_length=255, unique=True) category = models.ForeignKey(CollectionCategory, models.SET_NULL, null=True, blank=True, verbose_name=_("category")) - notes = models.TextField(_("notes"), blank=True) + notes = models.TextField(_("notes"), blank=True, help_text=_('private')) description = models.TextField(_("description"), blank=True) class Meta: @@ -360,3 +600,93 @@ class Place(WikidataModel): def __str__(self): return self.name + + +class BookMonthlyStats(models.Model): + book = models.ForeignKey('catalogue.Book', models.CASCADE) + date = models.DateField() + views_reader = models.IntegerField(default=0) + views_page = models.IntegerField(default=0) + + @classmethod + def build_for_month(cls, date): + date = date.replace(day=1) + period = 'month' + + date = date.isoformat() + url = f'{settings.PIWIK_URL}?date={date}&filter_limit=-1&format=CSV&idSite={settings.PIWIK_WL_SITE_ID}&language=pl&method=Actions.getPageUrls&module=API&period={period}&segment=&token_auth={settings.PIWIK_TOKEN}&flat=1' + data = urlopen(url).read().decode('utf-16') + lines = data.split('\n')[1:] + for line in lines: + m = re.match('^/katalog/lektura/([^,./]+)\.html,', line) + if m is not None: + which = 'views_reader' + else: + m = re.match('^/katalog/lektura/([^,./]+)/,', line) + if m is not None: + which = 'views_page' + if m is not None: + slug = m.group(1) + _url, _uviews, views, _rest = line.split(',', 3) + views = int(views) + try: + book = Book.objects.get(slug=slug) + except Book.DoesNotExist: + continue + else: + cls.objects.update_or_create( + book=book, date=date, + defaults={which: views} + ) + book.update_monthly_stats() + + +class Thema(models.Model): + code = models.CharField( + max_length=128, unique=True, + help_text='Używamy rozszerzenia .WL- do oznaczania własnych kodów.
' + 'Przykładowo, w przypadku potrzeby stworzenia nowej kategorii „insurekcja kościuszkowska”, ' + 'można by ją utworzyć jako 3MLQ‑PL‑A.WL-A, czyli w ramach już istniejącej wyższej kategorii 3MLQ‑PL‑A „rozbiory Polski”.', + ) + name = models.CharField(max_length=1024) + slug = models.SlugField( + max_length=255, null=True, blank=True, unique=True, + help_text='Element adresu na WL, w postaci: /tag/slug/. Można zmieniać.' + ) + plural = models.CharField( + 'liczba mnoga', max_length=255, blank=True, + ) + description = models.TextField(blank=True) + public_description = models.TextField(blank=True) + usable = models.BooleanField() + usable_as_main = models.BooleanField(default=False) + hidden = models.BooleanField(default=False) + woblink_category = models.IntegerField(null=True, blank=True) + + class Meta: + ordering = ('code',) + verbose_name_plural = 'Thema' + + +class Audience(models.Model): + code = models.CharField( + max_length=128, unique=True, + help_text='Techniczny identifyikator. W miarę możliwości nie należy zmieniać.' + ) + name = models.CharField( + max_length=1024, + help_text='W formie: „dla … (kogo?)”' + ) + slug = models.SlugField( + max_length=255, null=True, blank=True, unique=True, + help_text='Element adresu na WL, w postaci: /dla/slug/. Można zmieniać.' + ) + description = models.TextField(blank=True) + thema = models.CharField( + max_length=32, blank=True, + help_text='Odpowiadający kwalifikator Thema.' + ) + woblink = models.IntegerField(null=True, blank=True) + + class Meta: + ordering = ('code',)