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/shop/{price}">określonej ceny</a>.',
  24             price=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         if not thema_codes:
 224             if errors is not None:
 225                 errors.append(NoThema())
 226         category_ids = []
 227         for code in thema_codes:
 228             try:
 229                 thema = Thema.objects.get(code=code)
 230             except Thema.DoesNotExist:
 231                 if errors is not None:
 232                     errors.append(UnknownThema(code))
 233             else:
 234                 if thema.woblink_category is None:
 235                     if errors is not None:
 236                         errors.append(ThemaNoWoblink(thema))
 237                 elif thema.woblink_category not in WOBLINK_CATEGORIES:
 238                     if errors is not None:
 239                         errors.append(ThemaUnknownWoblink(thema))
 240                 elif thema.woblink_category not in category_ids:
 241                     category_ids.append(thema.woblink_category)
 242         if not category_ids:
 243             if errors is not None:
 244                 errors.append(NoWoblinkCategory())
 245         return category_ids
 246
 247     def get_series(self, meta, errors=None):
 248         return list(Audience.objects.filter(code__in=meta.audiences).exclude(
 249             woblink=None).values_list('woblink', flat=True))
 250
 251     def get_abstract(self, wldoc, errors=None, description_add=None):
 252         description = self.get_description(wldoc, description_add)
 253         parts = description.split('\n', 1)
 254         if len(parts) == 1 or len(parts[0]) > 240:
 255             # No newline found here.
 256             # Try to find last sentence end..
 257             parts = re.split(r' \.', description[240::-1], 1)
 258             if len(parts) == 2:
 259                 p1 = parts[1][::-1] + '.'
 260                 p2 = description[len(p1) + 1:]
 261             else:
 262                 # No sentence end found.
 263                 # Just find a space.
 264                 p1 = description[:240].rsplit(' ', 1)[0]
 265                 p2 = description[len(p1) + 1:]
 266                 p1 += '…'
 267                 p2 = '…' + p2
 268             parts = [p1, p2]
 269
 270         m = re.search(r'<[^>]+$', parts[0])
 271         if m is not None:
 272             parts[0] = parts[0][:-len(m.group(0))]
 273             parts[1] = m.group(0) + parts[1]
 274
 275         opened = []
 276         for tag in re.findall(r'<[^>]*[^/>]>', parts[0]):
 277             if tag[1] == '/':
 278                 opened.pop()
 279             else:
 280                 opened.append(tag)
 281         for tag in reversed(opened):
 282             parts[0] += '</' + tag[1:-1].split()[0] + '>'
 283             parts[1] = tag + parts[1]
 284         return {
 285             'header': parts[0],
 286             'rest': parts[1],
 287         }
 288
 289     def get_lang2code(self, meta, errors=None):
 290         return lang_code_3to2(meta.language)
 291
 292     def get_price(self, shop, wldoc, errors=None):
 293         try:
 294             stats = wldoc.get_statistics()['total']
 295         except:
 296             if errors:
 297                 errors.append(NoPrice(shop))
 298             return 0
 299         words = stats['words_with_fn']
 300         pages = stats['chars_with_fn'] / 1800
 301         price = shop.get_price(words, pages)
 302         if price is None:
 303             if errors:
 304                 errors.append(NoPrice(shop))
 305             return 0
 306
 307         return price
 308
 309     def can_publish(self, shop, book):
 310         wldoc = book.wldocument(librarian2=True)
 311         d = {
 312             'warnings': [],
 313             'errors': [],
 314             'info': [],
 315         }
 316         errors = []
 317         book_data = self.get_book_data(shop, wldoc, errors)
 318         for error in errors:
 319             if not isinstance(error, Warning):
 320                 errlist = d['errors']
 321             else:
 322                 errlist = d['warnings']
 323             errlist.append(error.as_html())
 324
 325         if book_data.get('genres'):
 326             d['info'].append(format_html(
 327                 'W kategoriach: {cat} ({price} zł)',
 328                 cat=', '.join(self.describe_category(g) for g in book_data['genres']),
 329                 price=book_data['price'],
 330             ))
 331         d['info'].append(mark_safe(
 332             '<strong>' + book_data['abstract']['header'] +
 333             '</strong><br/>' + book_data['abstract']['rest']
 334         ))
 335
 336         return d
 337
 338     def describe_category(self, category):
 339         t = []
 340         while category:
 341             c = WOBLINK_CATEGORIES[category]
 342             t.append(c['name'])
 343             category = c.get('parent')
 344         return ' / '.join(reversed(t))
 345
 346     def create_book(self, isbn):
 347         isbn = ''.join(c for c in isbn if c.isdigit())
 348         assert len(isbn) == 13
 349         response = self.session.post(
 350             self.ADD_URL,
 351             data={
 352                 'AddPublication[pubType]': 'ebook',
 353                 'AddPublication[pubHasIsbn]': '1',
 354                 'AddPublication[pubIsbn]': isbn,
 355                  ##AddPubation[save]
 356             }
 357         )
 358         m = re.search(r'/(\d+)$', response.url)
 359         if m is not None:
 360             return m.group(1)
 361
 362     def send_book(self, shop, book, changes=None):
 363         wldoc = book.wldocument(librarian2=True, changes=changes, publishable=False) # TODO pub
 364         meta = wldoc.meta
 365
 366         book_data = self.get_book_data(shop, wldoc)
 367
 368         if not book.woblink_id:
 369             #book.woblink_id = 2959868
 370             woblink_id = self.create_book(book_data['isbn'])
 371             assert woblink_id
 372             book.woblink_id = woblink_id
 373             book.save(update_fields=['woblink_id'])
 374
 375         self.edit_step1(book.woblink_id, book_data)
 376         self.edit_step2(book.woblink_id, book_data)
 377         self.edit_step3(book.woblink_id, book_data)
 378         cover_id = self.send_cover(book.woblink_id, wldoc)
 379
 380         texts = shop.get_texts()
 381         epub_id, epub_demo = self.send_epub(
 382             book.woblink_id, wldoc, book.gallery_path(),
 383             fundraising=texts
 384         )
 385         mobi_id, mobi_demo = self.send_mobi(
 386             book.woblink_id, wldoc, book.gallery_path(),
 387             fundraising=texts
 388         )
 389         self.edit_step4(
 390             book.woblink_id, book_data,
 391             cover_id, epub_id, epub_demo, mobi_id, mobi_demo,
 392         )
 393         self.edit_step5(book.woblink_id, book_data)
 394
 395     def get_book_data(self, shop, wldoc, errors=None):
 396         return {
 397             "title": wldoc.meta.title,
 398             "isbn": self.get_isbn(wldoc.meta, errors=errors),
 399             "authors": self.get_authors_data(wldoc.meta, errors=errors),
 400             "abstract": self.get_abstract(
 401                 wldoc, errors=errors, description_add=shop.description_add
 402             ),
 403             "lang2code": self.get_lang2code(wldoc.meta, errors=errors),
 404             "genres": self.get_genres(wldoc.meta, errors=errors),
 405             "price": self.get_price(shop, wldoc, errors=errors),
 406             "series": self.get_series(wldoc.meta, errors=errors),
 407         }
 408
 409     def with_form_name(self, data, name):
 410         return {
 411             f"{name}[{k}]": v
 412             for (k, v) in data.items()
 413         }
 414
 415     def edit_step1(self, woblink_id, book_data):
 416         data = book_data
 417
 418         authors_data = [
 419             {
 420                 "AhpPubId": woblink_id,
 421                 "AhpAutId": author_id,
 422                 "AhpType": author_type,
 423             }
 424             for (author_type, author_id) in data['authors']
 425         ]
 426
 427         series_data = [
 428             {
 429                 'PublicationId': woblink_id,
 430                 'SeriesId': series_id,
 431             }
 432             for series_id in data['series']
 433         ]
 434
 435         d = {
 436             'pubTitle': book_data['title'],
 437             'npwAuthorHasPublications': json.dumps(authors_data),
 438             'pubShortNote': data['abstract']['header'],
 439             'pubNote': data['abstract']['rest'],
 440             'pubCulture': data['lang2code'],
 441             'npwPublicationHasAwards': '[]',
 442             'npwPublicationHasSeriess': json.dumps(series_data),
 443         }
 444         d = self.with_form_name(d, 'EditPublicationStep1')
 445         d['roles'] = [author_type for (author_type, author_id) in data['authors']]
 446         r = self.session.post(self.STEP1_URL % woblink_id, data=d)
 447         return r
 448
 449
 450     def edit_step2(self, woblink_id, book_data):
 451         gd = {}
 452         legacy = None
 453         for i, g in enumerate(book_data['genres']):
 454             gdata = WOBLINK_CATEGORIES[g]
 455             if legacy is None:
 456                 legacy = gdata.get('legacy')
 457             if p := gdata.get('parent'):
 458                 gd.setdefault(p, {'isMain': False})
 459                 gd[p].setdefault('children', [])
 460                 gd[p]['children'].append(str(g))
 461                 gd[p].setdefault('mainChild', str(g))
 462                 if legacy is None:
 463                     legacy = WOBLINK_CATEGORIES[p].get('legacy')
 464             else:
 465                 gd.setdefault(g, {})
 466                 gd[g]['isMain'] = True
 467         gd = [
 468             {
 469                 "pubId": woblink_id,
 470                 "category": str(k),
 471                 **v
 472             }
 473             for k, v in gd.items()
 474         ]
 475
 476         data = {
 477             'npwPublicationHasNewGenres': json.dumps(gd),
 478             'genre': legacy or '',
 479         }
 480         data = self.with_form_name(data, 'AddPublicationStep2')
 481         return self.session.post(self.STEP2_URL % woblink_id, data=data)
 482
 483     def edit_step3(self, woblink_id, book_data):
 484         d = {
 485             'pubBasePrice': book_data['price'],
 486             'pubPremiereDate': date.today().isoformat(),
 487             'pubIsLicenseIndefinite': '1',
 488             'pubFileFormat': 'epub+mobi',
 489             'pubIsAcs': '0',
 490             'pubPublisherIndex': '',
 491         }
 492         d = self.with_form_name(d, 'EditPublicationStep3')
 493         return self.session.post(self.STEP3_URL % woblink_id, data=d)
 494
 495     def edit_step4(self, woblink_id, book_data, cover_id, epub_id, epub_demo, mobi_id, mobi_demo):
 496         d = {
 497             'pubCoverResId': cover_id,
 498             'pubEpubResId': epub_id,
 499             'pubEpubDemoResId': epub_demo,
 500             'pubMobiResId': mobi_id,
 501             'pubMobiDemoResId': mobi_demo,
 502             'pubFileFormat': 'epub+mobi',
 503             'pubId': woblink_id,
 504         }
 505         d = self.with_form_name(d, 'EditPublicationStep4')
 506         return self.session.post(self.STEP4_URL % woblink_id, data=d)
 507
 508     def edit_step5(self, woblink_id, book_data):
 509         d = {}
 510         d = self.with_form_name(d, 'EditPublicationStep5')
 511         return self.session.post(self.STEP5_URL % woblink_id, data=d)
 512
 513     def wait_for_job(self, job_id):
 514         while True:
 515             response = self.session.post(
 516                 self.JOB_STATUS_URL,
 517                 data={'ids[]': job_id}
 518             )
 519             data = response.json()[job_id]
 520             if data['ready']:
 521                 assert data['successful']
 522                 return data.get('returnValue')
 523             sleep(2)
 524
 525     def upload_file(self, woblink_id, filename, content, field_name, mime_type):
 526         data = {
 527             'pubId': woblink_id,
 528         }
 529         files = {
 530             field_name: (filename, content, mime_type)
 531         }
 532
 533         form_name = f'Upload{field_name}'
 534         id_field = f'pub{field_name}ResId'
 535         field_name = field_name.lower()
 536
 537         response = self.session.post(
 538             self.UPLOAD_URL % field_name,
 539             data=self.with_form_name(data, form_name),
 540             files=self.with_form_name(files, form_name),
 541         )
 542         resp_data = response.json()
 543         assert resp_data['success'] is True
 544         file_id = resp_data[id_field]
 545         if 'jobId' in resp_data:
 546             self.wait_for_job(resp_data['jobId'])
 547         return file_id
 548
 549     def generate_demo(self, woblink_id, file_format, check=True):
 550         percent = 10
 551         while True:
 552             job_id = self.session.get(
 553                 self.GENERATE_DEMO_URL % (file_format, woblink_id, percent),
 554             ).json()['jobId']
 555             try:
 556                 file_id = self.wait_for_job(job_id)
 557             except AssertionError:
 558                 if percent < 50:
 559                     percent += 10
 560                 else:
 561                     raise
 562             else:
 563                 break
 564
 565         if check:
 566             self.wait_for_job(
 567                 self.session.get(
 568                     self.CHECK_DEMO_URL % (file_format, woblink_id)
 569                 ).json()['jobId']
 570             )
 571         return file_id
 572
 573     def send_epub(self, woblink_id, doc, gallery_path, fundraising=None):
 574         from librarian.builders import EpubBuilder
 575         content = EpubBuilder(
 576             base_url='file://' + gallery_path + '/',
 577             fundraising=fundraising or [],
 578         ).build(doc).get_file()
 579         file_id = self.upload_file(
 580             woblink_id,
 581             doc.meta.url.slug + '.epub',
 582             content,
 583             'UploadEpub',
 584             'epub',
 585             'application/epub+zip'
 586         )
 587         demo_id = self.generate_demo(woblink_id, 'epub')
 588         return file_id, demo_id
 589
 590     def send_mobi(self, woblink_id, doc, gallery_path, fundraising=None):
 591         from librarian.builders import MobiBuilder
 592         content = MobiBuilder(
 593             base_url='file://' + gallery_path + '/',
 594             fundraising=fundraising or [],
 595         ).build(doc).get_file()
 596         file_id = self.upload_file(
 597             woblink_id,
 598             doc.meta.url.slug + '.mobi',
 599             content,
 600             'UploadMobi',
 601             'mobi',
 602             'application/x-mobipocket-ebook'
 603         )
 604         demo_id = self.generate_demo(woblink_id, 'mobi', check=False)
 605         return file_id, demo_id
 606
 607     def send_cover(self, woblink_id, doc):
 608         from librarian.cover import make_cover
 609         # TODO Labe
 610         # A5 @ 300ppi.
 611         cover = make_cover(doc.meta, cover_class='m-label', width=1748, height=2480)
 612         content = io.BytesIO()
 613         cover.final_image().save(content, cover.format)
 614         content.seek(0)
 615         file_id = self.upload_file(
 616             woblink_id,
 617             doc.meta.url.slug + '.jpeg',
 618             content,
 619             'UploadCover',
 620             'cover',
 621             cover.mime_type()
 622         )
 623         return file_id