src/depot/publishers/woblink.py

   1 from datetime import date
   2 import io
   3 import json
   4 import re
   5 from time import sleep
   6 from django.conf import settings
   7 from django.utils.html import escape, format_html
   8 from django.utils.safestring import mark_safe
   9 from librarian.builders.html import SnippetHtmlBuilder
  10 from librarian.functions import lang_code_3to2
  11 from catalogue.models import Audience, Author, Thema
  12 from .. import models
  13 from .base import BasePublisher
  14 from .woblink_constants import WOBLINK_CATEGORIES
  15
  16
  17 class WoblinkError(ValueError):
  18     pass
  19
  20 class NoPrice(WoblinkError):
  21     def as_html(self):
  22         return format_html(
  23             'Brak <a href="/admin/depot/site/{site}">określonej ceny</a>.',
  24             site=self.args[0].id
  25         )
  26
  27 class NoIsbn(WoblinkError):
  28     def as_html(self):
  29         return 'Brak ISBN.'
  30
  31 class AuthorLiteralForeign(WoblinkError):
  32     def as_html(self):
  33         return format_html(
  34             'Nie obsługiwane: autor „{author}” w języku {lang}.',
  35             author=str(self.args[0]),
  36             lang=self.args[0].lang,
  37         )
  38
  39 class AuthorNotInCatalogue(WoblinkError):
  40     def as_html(self):
  41         return format_html(
  42             'Brak autora „{author}” w katalogu.',
  43             author=str(self.args[0])
  44         )
  45
  46 class AuthorNoWoblink(WoblinkError):
  47     def as_html(self):
  48         return format_html(
  49             'Autor <a href="/admin/catalogue/author/{author_id}/">{author}</a> bez identyfikatora Woblink.',
  50             author_id=self.args[0].id,
  51             author=self.args[0].name
  52         )
  53
  54 class NoThema(WoblinkError):
  55     def as_html(self):
  56         return format_html('Brak Thema.')
  57
  58 class UnknownThema(WoblinkError):
  59     def as_html(self):
  60         return format_html(
  61             'Nieznana Thema {code}.',
  62             code=self.args[0]
  63         )
  64
  65
  66 class ThemaUnknownWoblink(WoblinkError):
  67     def as_html(self):
  68         return format_html(
  69             'Thema <a href="/admin/catalogue/thema/{id}/">{code}</a> przypisana do nieznanej kategorii Woblink.',
  70             id=self.args[0].id,
  71             code=self.args[0].code,
  72         )
  73
  74 class NoWoblinkCategory(WoblinkError):
  75     def as_html(self):
  76         return 'Brak kategorii Woblink.'
  77
  78 class WoblinkWarning(Warning):
  79     pass
  80
  81 class NoMainThemaWarning(WoblinkWarning):
  82     def as_html(self):
  83         return format_html(
  84             'Brak głównej kategorii Thema.'
  85         )
  86
  87 class ThemaNoWoblink(WoblinkWarning):
  88     def as_html(self):
  89         return format_html(
  90             'Thema <a href="/admin/catalogue/thema/{id}/">{code}</a> nie przypisana do kategorii Woblink.',
  91             id=self.args[0].id,
  92             code=self.args[0].code,
  93         )
  94
  95 class AuthorLiteralForeignWarning(WoblinkWarning):
  96     def as_html(self):
  97         return format_html(
  98             'Nie obsługiwane: autor „{author}” w języku {lang}.',
  99             author=str(self.args[0]),
 100             lang=self.args[0].lang,
 101         )
 102
 103 class AuthorNotInCatalogueWarning(WoblinkWarning):
 104     def as_html(self):
 105         return format_html(
 106             'Brak autora „{author}” w katalogu.',
 107             author=str(self.args[0])
 108         )
 109
 110 class AuthorNoWoblinkWarning(WoblinkWarning):
 111     def as_html(self):
 112         return format_html(
 113             'Autor <a href="/admin/catalogue/author/{author_id}/">{author}</a> bez identyfikatora Woblink.',
 114             author_id=self.args[0].id,
 115             author=self.args[0].name
 116         )
 117
 118
 119
 120
 121 class Woblink(BasePublisher):
 122     BASE_URL = 'https://publisher.woblink.com/'
 123     ADD_URL = BASE_URL + 'catalog/add'
 124     STEP1_URL = BASE_URL + 'catalog/edit/%s'
 125     STEP2_URL = BASE_URL + 'catalog/edit/%s/2'
 126     STEP3_URL = BASE_URL + 'catalog/edit/%s/3'
 127     STEP4_URL = BASE_URL + 'catalog/edit/%s/4'
 128     STEP5_URL = BASE_URL + 'catalog/edit/%s/5'
 129     UPLOAD_URL = BASE_URL + 'file/upload-%s'
 130     JOB_STATUS_URL = BASE_URL + 'task/status'
 131     GENERATE_DEMO_URL = BASE_URL + 'task/run/generate-%s-demo/%s/%d'
 132     CHECK_DEMO_URL = BASE_URL + 'task/run/check-%s-demo/%s'
 133
 134     SEARCH_CATALOGUE_URL = BASE_URL + '{category}/autocomplete/{term}'
 135
 136     ROLE_AUTHOR = 1
 137     ROLE_TRANSLATOR = 4
 138
 139     def login(self):
 140         response = self.session.get('https://publisher.woblink.com/login')
 141         token = re.search(
 142             r'name="_csrf_token" value="([^"]+)"',
 143             response.text
 144         ).group(1)
 145         data = {
 146             '_csrf_token': token,
 147             '_username': self.username,
 148             '_password': self.password,
 149         }
 150         response = self.session.post(
 151             'https://publisher.woblink.com/login_check',
 152             data=data,
 153         )
 154
 155     def search_catalogue(self, category, term):
 156         return self.session.get(
 157             self.SEARCH_CATALOGUE_URL.format(category=category, term=term)
 158         ).json()
 159
 160     def search_author_catalogue(self, term):
 161         return [
 162             {
 163                 'id': item['autId'],
 164                 'text': item['autFullname']
 165             }
 166             for item in self.search_catalogue('author', term)
 167         ]
 168     def search_series_catalogue(self, term):
 169         return [
 170             {
 171                 'id': item['id'],
 172                 'text': item['name']
 173             }
 174             for item in self.search_catalogue('series', term)
 175         ]
 176
 177     def get_isbn(self, meta, errors=None):
 178         if not meta.isbn_epub:
 179             if errors is not None:
 180                 errors.append(NoIsbn())
 181         return meta.isbn_epub
 182
 183     def get_authors_data(self, meta, errors=None):
 184         authors = []
 185         for role, items, obligatory in [
 186                 (self.ROLE_AUTHOR, meta.authors, True),
 187                 (self.ROLE_TRANSLATOR, meta.translators, False)
 188         ]:
 189             for person_literal in items:
 190                 if person_literal.lang != 'pl':
 191                     if errors is not None:
 192                         if obligatory:
 193                              errors.append(AuthorLiteralForeign(person_literal))
 194                         else:
 195                             errors.append(AuthorLiteralForeignWarning(person_literal))
 196                     continue
 197                 aobj = Author.get_by_literal(str(person_literal))
 198                 if aobj is None:
 199                     if errors is not None:
 200                         if obligatory:
 201                              errors.append(AuthorNotInCatalogue(person_literal))
 202                         else:
 203                             errors.append(AuthorNotInCatalogueWarning(person_literal))
 204                     continue
 205                 if not aobj.woblink:
 206                     if errors is not None:
 207                         if obligatory:
 208                              errors.append(AuthorNoWoblink(aobj))
 209                         else:
 210                             errors.append(AuthorNoWoblinkWarning(aobj))
 211                     continue
 212                 authors.append((role, aobj.woblink))
 213         return authors
 214
 215     def get_genres(self, meta, errors=None):
 216         thema_codes = []
 217         if meta.thema_main:
 218             thema_codes.append(meta.thema_main)
 219         else:
 220             if errors is not None:
 221                 errors.append(NoMainThemaWarning())
 222         thema_codes.extend(meta.thema)
 223
 224         thema_codes.extend(
 225             Audience.objects.filter(code__in=meta.audiences).exclude(
 226                 thema='').values_list('thema', flat=True)
 227         )
 228
 229         if not thema_codes:
 230             if errors is not None:
 231                 errors.append(NoThema())
 232         category_ids = []
 233         for code in thema_codes:
 234             try:
 235                 thema = Thema.objects.get(code=code)
 236             except Thema.DoesNotExist:
 237                 if errors is not None:
 238                     errors.append(UnknownThema(code))
 239             else:
 240                 if thema.woblink_category is None:
 241                     if errors is not None:
 242                         errors.append(ThemaNoWoblink(thema))
 243                 elif thema.woblink_category not in WOBLINK_CATEGORIES:
 244                     if errors is not None:
 245                         errors.append(ThemaUnknownWoblink(thema))
 246                 elif thema.woblink_category not in category_ids:
 247                     category_ids.append(thema.woblink_category)
 248         if not category_ids:
 249             if errors is not None:
 250                 errors.append(NoWoblinkCategory())
 251         return category_ids
 252
 253     def get_series(self, meta, errors=None):
 254         return list(Audience.objects.filter(code__in=meta.audiences).exclude(
 255             woblink=None).values_list('woblink', flat=True))
 256
 257     def get_abstract(self, wldoc, errors=None, description_add=None):
 258         description = self.get_description(wldoc, description_add)
 259         parts = description.split('\n', 1)
 260         if len(parts) == 1 or len(parts[0]) > 240:
 261             # No newline found here.
 262             # Try to find last sentence end..
 263             parts = re.split(r' \.', description[240::-1], 1)
 264             if len(parts) == 2:
 265                 p1 = parts[1][::-1] + '.'
 266                 p2 = description[len(p1) + 1:]
 267             else:
 268                 # No sentence end found.
 269                 # Just find a space.
 270                 p1 = description[:240].rsplit(' ', 1)[0]
 271                 p2 = description[len(p1) + 1:]
 272                 p1 += '…'
 273                 p2 = '…' + p2
 274             parts = [p1, p2]
 275
 276         m = re.search(r'<[^>]+$', parts[0])
 277         if m is not None:
 278             parts[0] = parts[0][:-len(m.group(0))]
 279             parts[1] = m.group(0) + parts[1]
 280
 281         opened = []
 282         for tag in re.findall(r'<[^>]*[^/>]>', parts[0]):
 283             if tag[1] == '/':
 284                 opened.pop()
 285             else:
 286                 opened.append(tag)
 287         for tag in reversed(opened):
 288             parts[0] += '</' + tag[1:-1].split()[0] + '>'
 289             parts[1] = tag + parts[1]
 290         return {
 291             'header': parts[0],
 292             'rest': parts[1],
 293         }
 294
 295     def get_lang2code(self, meta, errors=None):
 296         return lang_code_3to2(meta.language)
 297
 298     def get_price(self, site, wldoc, errors=None):
 299         try:
 300             stats = wldoc.get_statistics()['total']
 301         except:
 302             if errors:
 303                 errors.append(NoPrice(site))
 304             return 0
 305         words = stats['words_with_fn']
 306         pages = stats['chars_with_fn'] / 1800
 307         price = site.get_price(words, pages)
 308         if price is None:
 309             if errors:
 310                 errors.append(NoPrice(site))
 311             return 0
 312
 313         return price
 314
 315     def can_publish(self, site, book):
 316         wldoc = book.wldocument(librarian2=True)
 317         d = {
 318             'warnings': [],
 319             'errors': [],
 320             'info': [],
 321         }
 322         errors = []
 323         book_data = self.get_book_data(site, wldoc, errors)
 324         for error in errors:
 325             if not isinstance(error, Warning):
 326                 errlist = d['errors']
 327             else:
 328                 errlist = d['warnings']
 329             errlist.append(error.as_html())
 330
 331         if book_data.get('genres'):
 332             d['info'].append(format_html(
 333                 'W kategoriach: {cat} ({price} zł)',
 334                 cat=', '.join(self.describe_category(g) for g in book_data['genres']),
 335                 price=book_data['price'],
 336             ))
 337         d['info'].append(mark_safe(
 338             '<strong>' + book_data['abstract']['header'] +
 339             '</strong><br/>' + book_data['abstract']['rest']
 340         ))
 341
 342         return d
 343
 344     def describe_category(self, category):
 345         t = []
 346         while category:
 347             c = WOBLINK_CATEGORIES[category]
 348             t.append(c['name'])
 349             category = c.get('parent')
 350         return ' / '.join(reversed(t))
 351
 352     def create_book(self, isbn):
 353         isbn = ''.join(c for c in isbn if c.isdigit())
 354         assert len(isbn) == 13
 355         response = self.session.post(
 356             self.ADD_URL,
 357             data={
 358                 'AddPublication[pubType]': 'ebook',
 359                 'AddPublication[pubHasIsbn]': '1',
 360                 'AddPublication[pubIsbn]': isbn,
 361                  ##AddPubation[save]
 362             }
 363         )
 364         m = re.search(r'/(\d+)$', response.url)
 365         if m is not None:
 366             return m.group(1)
 367
 368     def send_book(self, site_book_publish, changes=None):
 369         site_book = site_book_publish.site_book
 370         book = site_book.book
 371         site = site_book.site
 372         wldoc = book.wldocument(librarian2=True, changes=changes, publishable=False) # TODO pub
 373         meta = wldoc.meta
 374
 375         book_data = self.get_book_data(site, wldoc)
 376
 377         if not site_book.external_id:
 378             woblink_id = self.create_book(book_data['isbn'])
 379             assert woblink_id
 380             site_book.external_id = woblink_id
 381             site_book.save(update_fields=['external_id'])
 382         woblink_id = site_book.external_id
 383
 384         self.edit_step1(woblink_id, book_data)
 385         self.edit_step2(woblink_id, book_data)
 386         self.edit_step3(woblink_id, book_data)
 387         cover_id = self.send_cover(woblink_id, wldoc)
 388
 389         texts = site.get_texts()
 390         epub_id, epub_demo = self.send_epub(
 391             woblink_id, wldoc, book.gallery_path(),
 392             fundraising=texts
 393         )
 394         mobi_id, mobi_demo = self.send_mobi(
 395             woblink_id, wldoc, book.gallery_path(),
 396             fundraising=texts
 397         )
 398         self.edit_step4(
 399             woblink_id, book_data,
 400             cover_id, epub_id, epub_demo, mobi_id, mobi_demo,
 401         )
 402         self.edit_step5(woblink_id, book_data)
 403
 404     def get_book_data(self, site, wldoc, errors=None):
 405         return {
 406             "title": wldoc.meta.title,
 407             "isbn": self.get_isbn(wldoc.meta, errors=errors),
 408             "authors": self.get_authors_data(wldoc.meta, errors=errors),
 409             "abstract": self.get_abstract(
 410                 wldoc, errors=errors, description_add=site.description_add
 411             ),
 412             "lang2code": self.get_lang2code(wldoc.meta, errors=errors),
 413             "genres": self.get_genres(wldoc.meta, errors=errors),
 414             "price": self.get_price(site, wldoc, errors=errors),
 415             "series": self.get_series(wldoc.meta, errors=errors),
 416         }
 417
 418     def with_form_name(self, data, name):
 419         return {
 420             f"{name}[{k}]": v
 421             for (k, v) in data.items()
 422         }
 423
 424     def edit_step1(self, woblink_id, book_data):
 425         data = book_data
 426
 427         authors_data = [
 428             {
 429                 "AhpPubId": woblink_id,
 430                 "AhpAutId": author_id,
 431                 "AhpType": author_type,
 432             }
 433             for (author_type, author_id) in data['authors']
 434         ]
 435
 436         series_data = [
 437             {
 438                 'PublicationId': woblink_id,
 439                 'SeriesId': series_id,
 440             }
 441             for series_id in data['series']
 442         ]
 443
 444         d = {
 445             'pubTitle': book_data['title'],
 446             'npwAuthorHasPublications': json.dumps(authors_data),
 447             'pubShortNote': data['abstract']['header'],
 448             'pubNote': data['abstract']['rest'],
 449             'pubCulture': data['lang2code'],
 450             'npwPublicationHasAwards': '[]',
 451             'npwPublicationHasSeriess': json.dumps(series_data),
 452         }
 453         d = self.with_form_name(d, 'EditPublicationStep1')
 454         d['roles'] = [author_type for (author_type, author_id) in data['authors']]
 455         r = self.session.post(self.STEP1_URL % woblink_id, data=d)
 456         return r
 457
 458
 459     def edit_step2(self, woblink_id, book_data):
 460         gd = {}
 461         legacy = None
 462         for i, g in enumerate(book_data['genres']):
 463             gdata = WOBLINK_CATEGORIES[g]
 464             if legacy is None:
 465                 legacy = gdata.get('legacy')
 466             if p := gdata.get('parent'):
 467                 gd.setdefault(p, {'isMain': False})
 468                 gd[p].setdefault('children', [])
 469                 gd[p]['children'].append(str(g))
 470                 gd[p].setdefault('mainChild', str(g))
 471                 if legacy is None:
 472                     legacy = WOBLINK_CATEGORIES[p].get('legacy')
 473             else:
 474                 gd.setdefault(g, {})
 475                 gd[g]['isMain'] = True
 476         gd = [
 477             {
 478                 "pubId": woblink_id,
 479                 "category": str(k),
 480                 **v
 481             }
 482             for k, v in gd.items()
 483         ]
 484
 485         data = {
 486             'npwPublicationHasNewGenres': json.dumps(gd),
 487             'genre': legacy or '',
 488         }
 489         data = self.with_form_name(data, 'AddPublicationStep2')
 490         return self.session.post(self.STEP2_URL % woblink_id, data=data)
 491
 492     def edit_step3(self, woblink_id, book_data):
 493         d = {
 494             'pubBasePrice': book_data['price'],
 495             'pubPremiereDate': date.today().isoformat(),
 496             'pubIsLicenseIndefinite': '1',
 497             'pubFileFormat': 'epub+mobi',
 498             'pubIsAcs': '0',
 499             'pubPublisherIndex': '',
 500             'save_and_continue': '',
 501         }
 502         d = self.with_form_name(d, 'EditPublicationStep3')
 503         return self.session.post(self.STEP3_URL % woblink_id, data=d)
 504
 505     def edit_step4(self, woblink_id, book_data, cover_id, epub_id, epub_demo, mobi_id, mobi_demo):
 506         d = {
 507             'pubCoverResId': cover_id,
 508             'pubEpubResId': epub_id,
 509             'pubEpubDemoResId': epub_demo,
 510             'pubMobiResId': mobi_id,
 511             'pubMobiDemoResId': mobi_demo,
 512             'pubFileFormat': 'epub+mobi',
 513             'pubId': woblink_id,
 514             'save_and_continue': '',
 515         }
 516         d = self.with_form_name(d, 'EditPublicationStep4')
 517         return self.session.post(self.STEP4_URL % woblink_id, data=d)
 518
 519     def edit_step5(self, woblink_id, book_data):
 520         d = {'save': ''}
 521         d = self.with_form_name(d, 'EditPublicationStep5')
 522         return self.session.post(self.STEP5_URL % woblink_id, data=d)
 523
 524     def wait_for_job(self, job_id):
 525         while True:
 526             response = self.session.post(
 527                 self.JOB_STATUS_URL,
 528                 data={'ids[]': job_id}
 529             )
 530             data = response.json()[job_id]
 531             if data['ready']:
 532                 assert data['successful']
 533                 return data.get('returnValue')
 534             sleep(2)
 535
 536     def upload_file(self, woblink_id, filename, content, field_name, mime_type):
 537         form_name = f'Upload{field_name}'
 538         id_field = f'pub{field_name}ResId'
 539         field_name = field_name.lower()
 540
 541         data = {
 542             'pubId': woblink_id,
 543         }
 544         files = {
 545             field_name: (filename, content, mime_type)
 546         }
 547
 548         response = self.session.post(
 549             self.UPLOAD_URL % field_name,
 550             data=self.with_form_name(data, form_name),
 551             files=self.with_form_name(files, form_name),
 552         )
 553         resp_data = response.json()
 554         assert resp_data['success'] is True
 555         file_id = resp_data[id_field]
 556         if 'jobId' in resp_data:
 557             self.wait_for_job(resp_data['jobId'])
 558         return file_id
 559
 560     def generate_demo(self, woblink_id, file_format, check=True):
 561         percent = 10
 562         while True:
 563             job_id = self.session.get(
 564                 self.GENERATE_DEMO_URL % (file_format, woblink_id, percent),
 565             ).json()['jobId']
 566             try:
 567                 file_id = self.wait_for_job(job_id)
 568             except AssertionError:
 569                 if percent < 50:
 570                     percent += 10
 571                 else:
 572                     raise
 573             else:
 574                 break
 575
 576         if check:
 577             self.wait_for_job(
 578                 self.session.get(
 579                     self.CHECK_DEMO_URL % (file_format, woblink_id)
 580                 ).json()['jobId']
 581             )
 582         return file_id
 583
 584     def send_epub(self, woblink_id, doc, gallery_path, fundraising=None):
 585         from librarian.builders import EpubBuilder
 586         content = EpubBuilder(
 587             base_url='file://' + gallery_path + '/',
 588             fundraising=fundraising or [],
 589         ).build(doc).get_file()
 590         file_id = self.upload_file(
 591             woblink_id,
 592             doc.meta.url.slug + '.epub',
 593             content,
 594             'Epub',
 595             'application/epub+zip'
 596         )
 597         demo_id = self.generate_demo(woblink_id, 'epub')
 598         return file_id, demo_id
 599
 600     def send_mobi(self, woblink_id, doc, gallery_path, fundraising=None):
 601         from librarian.builders import MobiBuilder
 602         content = MobiBuilder(
 603             base_url='file://' + gallery_path + '/',
 604             fundraising=fundraising or [],
 605         ).build(doc).get_file()
 606         file_id = self.upload_file(
 607             woblink_id,
 608             doc.meta.url.slug + '.mobi',
 609             content,
 610             'Mobi',
 611             'application/x-mobipocket-ebook'
 612         )
 613         demo_id = self.generate_demo(woblink_id, 'mobi', check=False)
 614         return file_id, demo_id
 615
 616     def send_cover(self, woblink_id, doc):
 617         from librarian.cover import make_cover
 618         # TODO Labe
 619         # A5 @ 300ppi.
 620         cover = make_cover(doc.meta, cover_class='m-label', width=1748, height=2480)
 621         content = io.BytesIO()
 622         cover.final_image().save(content, cover.format)
 623         content.seek(0)
 624         file_id = self.upload_file(
 625             woblink_id,
 626             doc.meta.url.slug + '.jpeg',
 627             content,
 628             'Cover',
 629             cover.mime_type()
 630         )
 631         return file_id