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         d = {
 317             'warnings': [],
 318             'errors': [],
 319             'info': [],
 320         }
 321         try:
 322             wldoc = book.wldocument(librarian2=True)
 323         except:
 324             d['errors'].append('Nieprawidłowy dokument.')
 325             return d
 326         errors = []
 327         book_data = self.get_book_data(site, wldoc, errors)
 328         for error in errors:
 329             if not isinstance(error, Warning):
 330                 errlist = d['errors']
 331             else:
 332                 errlist = d['warnings']
 333             errlist.append(error.as_html())
 334
 335         if book_data.get('genres'):
 336             d['info'].append(format_html(
 337                 'W kategoriach: {cat} ({price} zł)',
 338                 cat=', '.join(self.describe_category(g) for g in book_data['genres']),
 339                 price=book_data['price'],
 340             ))
 341         d['info'].append(mark_safe(
 342             '<strong>' + book_data['abstract']['header'] +
 343             '</strong><br/>' + book_data['abstract']['rest']
 344         ))
 345
 346         return d
 347
 348     def describe_category(self, category):
 349         t = []
 350         while category:
 351             c = WOBLINK_CATEGORIES[category]
 352             t.append(c['name'])
 353             category = c.get('parent')
 354         return ' / '.join(reversed(t))
 355
 356     def create_book(self, isbn):
 357         isbn = ''.join(c for c in isbn if c.isdigit())
 358         assert len(isbn) == 13
 359         response = self.session.post(
 360             self.ADD_URL,
 361             data={
 362                 'AddPublication[pubType]': 'ebook',
 363                 'AddPublication[pubHasIsbn]': '1',
 364                 'AddPublication[pubIsbn]': isbn,
 365                  ##AddPubation[save]
 366             }
 367         )
 368         m = re.search(r'/(\d+)$', response.url)
 369         if m is not None:
 370             return m.group(1)
 371
 372     def send_book(self, site_book_publish, changes=None):
 373         site_book = site_book_publish.site_book
 374         book = site_book.book
 375         site = site_book.site
 376         wldoc = book.wldocument(librarian2=True, changes=changes, publishable=False) # TODO pub
 377         meta = wldoc.meta
 378
 379         book_data = self.get_book_data(site, wldoc)
 380
 381         if not site_book.external_id:
 382             woblink_id = self.create_book(book_data['isbn'])
 383             assert woblink_id
 384             site_book.external_id = woblink_id
 385             site_book.save(update_fields=['external_id'])
 386         woblink_id = site_book.external_id
 387
 388         self.edit_step1(woblink_id, book_data)
 389         self.edit_step2(woblink_id, book_data)
 390         self.edit_step3(woblink_id, book_data)
 391         cover_id = self.send_cover(woblink_id, wldoc)
 392
 393         texts = site.get_texts()
 394         epub_id, epub_demo = self.send_epub(
 395             woblink_id, wldoc, book.gallery_path(),
 396             fundraising=texts
 397         )
 398         mobi_id, mobi_demo = self.send_mobi(
 399             woblink_id, wldoc, book.gallery_path(),
 400             fundraising=texts
 401         )
 402         self.edit_step4(
 403             woblink_id, book_data,
 404             cover_id, epub_id, epub_demo, mobi_id, mobi_demo,
 405         )
 406         self.edit_step5(woblink_id, book_data)
 407
 408     def get_book_data(self, site, wldoc, errors=None):
 409         return {
 410             "title": wldoc.meta.title,
 411             "isbn": self.get_isbn(wldoc.meta, errors=errors),
 412             "authors": self.get_authors_data(wldoc.meta, errors=errors),
 413             "abstract": self.get_abstract(
 414                 wldoc, errors=errors, description_add=site.description_add
 415             ),
 416             "lang2code": self.get_lang2code(wldoc.meta, errors=errors),
 417             "genres": self.get_genres(wldoc.meta, errors=errors),
 418             "price": self.get_price(site, wldoc, errors=errors),
 419             "series": self.get_series(wldoc.meta, errors=errors),
 420         }
 421
 422     def with_form_name(self, data, name):
 423         return {
 424             f"{name}[{k}]": v
 425             for (k, v) in data.items()
 426         }
 427
 428     def edit_step1(self, woblink_id, book_data):
 429         data = book_data
 430
 431         authors_data = [
 432             {
 433                 "AhpPubId": woblink_id,
 434                 "AhpAutId": author_id,
 435                 "AhpType": author_type,
 436             }
 437             for (author_type, author_id) in data['authors']
 438         ]
 439
 440         series_data = [
 441             {
 442                 'PublicationId': woblink_id,
 443                 'SeriesId': series_id,
 444             }
 445             for series_id in data['series']
 446         ]
 447
 448         d = {
 449             'pubTitle': book_data['title'],
 450             'npwAuthorHasPublications': json.dumps(authors_data),
 451             'pubShortNote': data['abstract']['header'],
 452             'pubNote': data['abstract']['rest'],
 453             'pubCulture': data['lang2code'],
 454             'npwPublicationHasAwards': '[]',
 455             'npwPublicationHasSeriess': json.dumps(series_data),
 456         }
 457         d = self.with_form_name(d, 'EditPublicationStep1')
 458         d['roles'] = [author_type for (author_type, author_id) in data['authors']]
 459         r = self.session.post(self.STEP1_URL % woblink_id, data=d)
 460         return r
 461
 462
 463     def edit_step2(self, woblink_id, book_data):
 464         gd = {}
 465         legacy = None
 466         for i, g in enumerate(book_data['genres']):
 467             gdata = WOBLINK_CATEGORIES[g]
 468             if legacy is None:
 469                 legacy = gdata.get('legacy')
 470             if p := gdata.get('parent'):
 471                 gd.setdefault(p, {'isMain': False})
 472                 gd[p].setdefault('children', [])
 473                 gd[p]['children'].append(str(g))
 474                 gd[p].setdefault('mainChild', str(g))
 475                 if legacy is None:
 476                     legacy = WOBLINK_CATEGORIES[p].get('legacy')
 477             else:
 478                 gd.setdefault(g, {})
 479                 gd[g]['isMain'] = True
 480         gd = [
 481             {
 482                 "pubId": woblink_id,
 483                 "category": str(k),
 484                 **v
 485             }
 486             for k, v in gd.items()
 487         ]
 488
 489         data = {
 490             'npwPublicationHasNewGenres': json.dumps(gd),
 491             'genre': legacy or '',
 492         }
 493         data = self.with_form_name(data, 'AddPublicationStep2')
 494         return self.session.post(self.STEP2_URL % woblink_id, data=data)
 495
 496     def edit_step3(self, woblink_id, book_data):
 497         d = {
 498             'pubBasePrice': book_data['price'],
 499             'pubPremiereDate': date.today().isoformat(),
 500             'pubIsLicenseIndefinite': '1',
 501             'pubFileFormat': 'epub+mobi',
 502             'pubIsAcs': '0',
 503             'pubPublisherIndex': '',
 504             'save_and_continue': '',
 505         }
 506         d = self.with_form_name(d, 'EditPublicationStep3')
 507         return self.session.post(self.STEP3_URL % woblink_id, data=d)
 508
 509     def edit_step4(self, woblink_id, book_data, cover_id, epub_id, epub_demo, mobi_id, mobi_demo):
 510         d = {
 511             'pubCoverResId': cover_id,
 512             'pubEpubResId': epub_id,
 513             'pubEpubDemoResId': epub_demo,
 514             'pubMobiResId': mobi_id,
 515             'pubMobiDemoResId': mobi_demo,
 516             'pubFileFormat': 'epub+mobi',
 517             'pubId': woblink_id,
 518             'save_and_continue': '',
 519         }
 520         d = self.with_form_name(d, 'EditPublicationStep4')
 521         return self.session.post(self.STEP4_URL % woblink_id, data=d)
 522
 523     def edit_step5(self, woblink_id, book_data):
 524         d = {'save': ''}
 525         d = self.with_form_name(d, 'EditPublicationStep5')
 526         return self.session.post(self.STEP5_URL % woblink_id, data=d)
 527
 528     def wait_for_job(self, job_id):
 529         while True:
 530             response = self.session.post(
 531                 self.JOB_STATUS_URL,
 532                 data={'ids[]': job_id}
 533             )
 534             data = response.json()[job_id]
 535             if data['ready']:
 536                 assert data['successful']
 537                 return data.get('returnValue')
 538             sleep(2)
 539
 540     def upload_file(self, woblink_id, filename, content, field_name, mime_type):
 541         form_name = f'Upload{field_name}'
 542         id_field = f'pub{field_name}ResId'
 543         field_name = field_name.lower()
 544
 545         data = {
 546             'pubId': woblink_id,
 547         }
 548         files = {
 549             field_name: (filename, content, mime_type)
 550         }
 551
 552         response = self.session.post(
 553             self.UPLOAD_URL % field_name,
 554             data=self.with_form_name(data, form_name),
 555             files=self.with_form_name(files, form_name),
 556         )
 557         resp_data = response.json()
 558         assert resp_data['success'] is True
 559         file_id = resp_data[id_field]
 560         if 'jobId' in resp_data:
 561             self.wait_for_job(resp_data['jobId'])
 562         return file_id
 563
 564     def generate_demo(self, woblink_id, file_format, check=True):
 565         percent = 10
 566         while True:
 567             job_id = self.session.get(
 568                 self.GENERATE_DEMO_URL % (file_format, woblink_id, percent),
 569             ).json()['jobId']
 570             try:
 571                 file_id = self.wait_for_job(job_id)
 572                 if check:
 573                     self.wait_for_job(
 574                         self.session.get(
 575                             self.CHECK_DEMO_URL % (file_format, woblink_id)
 576                         ).json()['jobId']
 577                     )
 578             except AssertionError:
 579                 if percent < 50:
 580                     percent += 10
 581                 else:
 582                     raise
 583             else:
 584                 break
 585
 586         return file_id
 587
 588     def send_epub(self, woblink_id, doc, gallery_path, fundraising=None):
 589         from librarian.builders import EpubBuilder
 590         content = EpubBuilder(
 591             base_url='file://' + gallery_path + '/',
 592             fundraising=fundraising or [],
 593         ).build(doc).get_file()
 594         file_id = self.upload_file(
 595             woblink_id,
 596             doc.meta.url.slug + '.epub',
 597             content,
 598             'Epub',
 599             'application/epub+zip'
 600         )
 601         demo_id = self.generate_demo(woblink_id, 'epub')
 602         return file_id, demo_id
 603
 604     def send_mobi(self, woblink_id, doc, gallery_path, fundraising=None):
 605         from librarian.builders import MobiBuilder
 606         content = MobiBuilder(
 607             base_url='file://' + gallery_path + '/',
 608             fundraising=fundraising or [],
 609         ).build(doc).get_file()
 610         file_id = self.upload_file(
 611             woblink_id,
 612             doc.meta.url.slug + '.mobi',
 613             content,
 614             'Mobi',
 615             'application/x-mobipocket-ebook'
 616         )
 617         demo_id = self.generate_demo(woblink_id, 'mobi', check=False)
 618         return file_id, demo_id
 619
 620     def send_cover(self, woblink_id, doc):
 621         from librarian.cover import make_cover
 622         # TODO Labe
 623         # A5 @ 300ppi.
 624         cover = make_cover(doc.meta, cover_class='m-label', width=1748, height=2480)
 625         content = io.BytesIO()
 626         cover.final_image().save(content, cover.format)
 627         content.seek(0)
 628         file_id = self.upload_file(
 629             woblink_id,
 630             doc.meta.url.slug + '.jpeg',
 631             content,
 632             'Cover',
 633             cover.mime_type()
 634         )
 635         return file_id