Use the IANA registered mimetypes for fonts in the pipeline

Also use the IANA types for EPUB versions > 3.1
Fixes #1659 (Update fonts MIME following IANA recommendation)

resources/mime.types

@@ -1378,10 +1378,10 @@ application/x-cbc                          cbc
 application/x-koboreader-ebook             kobo
 image/wmf                                  wmf
 application/ereader                        pdb
-# See http://idpf.org/epub/30/spec/epub30-publications.html#sec-core-media-types
+font/otf                                   otf
-application/vnd.ms-opentype                otf
+font/woff                                  woff
-application/font-woff                      woff
+font/woff2                                 woff2
-application/x-font-truetype                ttf
+font/ttf                                   ttf
 text/xml                                   plist
 text/x-markdown                            md markdown
 application/x-ibooks+zip                   ibook ibooks

src/calibre/ebooks/conversion/plugins/oeb_output.py

@@ -45,6 +45,7 @@ class OEBOutput(OutputFormatPlugin):
                             self.log.exception('Something went wrong while trying to'
                                     ' workaround Pocketbook cover bug, ignoring')
                         self.migrate_lang_code(root)
+                        self.adjust_mime_types(root)
                     raw = etree.tostring(root, pretty_print=True,
                             encoding='utf-8', xml_declaration=True)
                     if key == OPF_MIME:
@@ -67,6 +68,15 @@ class OEBOutput(OutputFormatPlugin):
                     f.write(item.bytes_representation)
                 item.unload_data_from_memory(memory=path)
 
+    def adjust_mime_types(self, root):
+        from calibre.ebooks.oeb.polish.utils import adjust_mime_for_epub
+        for x in root.xpath('//*[local-name() = "manifest"]/*[local-name() = "item"]'):
+            mt = x.get('media-type')
+            if mt:
+                nmt = adjust_mime_for_epub(filename=os.path.basename(x.get('href') or ''), mime=mt)
+                if nmt != mt:
+                    x.set('media-type', nmt)
+
     def workaround_nook_cover_bug(self, root):  # {{{
         cov = root.xpath('//*[local-name() = "meta" and @name="cover" and'
                 ' @content != "cover"]')

src/calibre/ebooks/oeb/polish/check/fonts.py

@@ -9,7 +9,7 @@ from css_parser.css import CSSRule
 from calibre import force_unicode
 from calibre.ebooks.oeb.base import OEB_DOCS, OEB_STYLES
 from calibre.ebooks.oeb.polish.check.base import BaseError, WARN
-from calibre.ebooks.oeb.polish.container import OEB_FONTS
+from calibre.ebooks.oeb.polish.utils import OEB_FONTS
 from calibre.ebooks.oeb.polish.pretty import pretty_script_or_style
 from calibre.ebooks.oeb.polish.fonts import change_font_in_declaration
 from calibre.utils.fonts.utils import get_all_font_names, is_font_embeddable, UnsupportedFont

src/calibre/ebooks/oeb/polish/check/links.py

@@ -10,11 +10,10 @@ from threading import Thread
 
 from calibre import browser
 from calibre.ebooks.oeb.base import OEB_DOCS, OEB_STYLES, urlunquote, XHTML_MIME
-from calibre.ebooks.oeb.polish.container import OEB_FONTS
 from calibre.ebooks.oeb.polish.parsing import parse_html5
 from calibre.ebooks.oeb.polish.replace import remove_links_to
 from calibre.ebooks.oeb.polish.cover import get_raster_cover_name
-from calibre.ebooks.oeb.polish.utils import guess_type, actual_case_for_name, corrected_case_for_name
+from calibre.ebooks.oeb.polish.utils import guess_type, actual_case_for_name, corrected_case_for_name, OEB_FONTS
 from calibre.ebooks.oeb.polish.check.base import BaseError, WARN, INFO
 from polyglot.builtins import iteritems, itervalues
 from polyglot.urllib import urlparse

src/calibre/ebooks/oeb/polish/container.py

@@ -41,7 +41,7 @@ from calibre.ebooks.oeb.parse_utils import NotHTML, parse_html
 from calibre.ebooks.oeb.polish.errors import DRMError, InvalidBook
 from calibre.ebooks.oeb.polish.parsing import parse as parse_html_tweak
 from calibre.ebooks.oeb.polish.utils import (
-    CommentFinder, PositionFinder, guess_type, parse_css
+    CommentFinder, PositionFinder, adjust_mime_for_epub, guess_type, parse_css, OEB_FONTS
 )
 from calibre.ptempfile import PersistentTemporaryDirectory, PersistentTemporaryFile
 from calibre.utils.filenames import hardlink_file, nlinks_file, retry_on_fail
@@ -53,10 +53,9 @@ from polyglot.builtins import iteritems
 from polyglot.urllib import urlparse
 
 exists, join, relpath = os.path.exists, os.path.join, os.path.relpath
-
-OEB_FONTS = {guess_type('a.ttf'), guess_type('b.otf'), guess_type('a.woff'), 'application/x-font-ttf', 'application/x-font-otf', 'application/font-sfnt'}
 OPF_NAMESPACES = {'opf':OPF2_NS, 'dc':DC11_NS}
 null = object()
+OEB_FONTS  # for plugin compat
 
 
 class CSSPreProcessor(cssp):
@@ -141,20 +140,7 @@ class ContainerBase:  # {{{
 
     def guess_type(self, name):
         ' Return the expected mimetype for the specified file name based on its extension. '
-        # epubcheck complains if the mimetype for text documents is set to
+        return adjust_mime_for_epub(filename=name, opf_version=self.opf_version_parsed)
-        # text/html in EPUB 2 books. Sigh.
-        ans = guess_type(name)
-        if ans == 'text/html':
-            ans = 'application/xhtml+xml'
-        if ans in {'application/x-font-truetype', 'application/vnd.ms-opentype'}:
-            opfversion = self.opf_version_parsed[:2]
-            if opfversion > (3, 0):
-                return 'application/font-sfnt'
-            if opfversion >= (3, 0):
-                # bloody epubcheck has recently decided it likes this mimetype
-                # for ttf files
-                return 'application/vnd.ms-opentype'
-        return ans
 
     def decode(self, data, normalize_to_nfc=True):
         """

src/calibre/ebooks/oeb/polish/replace.py

@@ -229,8 +229,7 @@ def replace_file(container, name, path, basename, force_mt=None):
 
 
 def mt_to_category(container, mt):
-    from calibre.ebooks.oeb.polish.utils import guess_type
+    from calibre.ebooks.oeb.polish.utils import guess_type, OEB_FONTS
-    from calibre.ebooks.oeb.polish.container import OEB_FONTS
     from calibre.ebooks.oeb.base import OEB_DOCS, OEB_STYLES
     if mt in OEB_DOCS:
         category = 'text'

src/calibre/ebooks/oeb/polish/report.py

@@ -10,7 +10,8 @@ from itertools import chain
 
 from calibre import prepare_string_for_xml, force_unicode
 from calibre.ebooks.oeb.base import XPath, xml2text
-from calibre.ebooks.oeb.polish.container import OEB_DOCS, OEB_STYLES, OEB_FONTS
+from calibre.ebooks.oeb.polish.container import OEB_DOCS, OEB_STYLES
+from calibre.ebooks.oeb.polish.utils import OEB_FONTS
 from calibre.ebooks.oeb.polish.spell import get_all_words, count_all_chars
 from calibre.utils.icu import numeric_sort_key, safe_chr
 from calibre.utils.imghdr import identify

src/calibre/ebooks/oeb/polish/subset.py

@@ -9,8 +9,7 @@ import os, sys
 
 from calibre import prints, as_unicode
 from calibre.ebooks.oeb.base import OEB_STYLES, OEB_DOCS, XPath, css_text
-from calibre.ebooks.oeb.polish.container import OEB_FONTS
+from calibre.ebooks.oeb.polish.utils import guess_type, OEB_FONTS
-from calibre.ebooks.oeb.polish.utils import guess_type
 from calibre.utils.fonts.sfnt.subset import subset
 from calibre.utils.fonts.sfnt.errors import UnsupportedFont
 from calibre.utils.fonts.utils import get_font_names
@@ -34,8 +33,9 @@ def remove_font_face_rules(container, sheet, remove_names, base):
 
 
 def iter_subsettable_fonts(container):
+    woff_font_types = guess_type('a.woff'), guess_type('a.woff2')
     for name, mt in iteritems(container.mime_map):
-        if (mt in OEB_FONTS or name.rpartition('.')[-1].lower() in {'otf', 'ttf'}) and mt != guess_type('a.woff'):
+        if (mt in OEB_FONTS or name.rpartition('.')[-1].lower() in {'otf', 'ttf'}) and mt not in woff_font_types:
             yield name, mt
 
 

src/calibre/ebooks/oeb/polish/upgrade.py

@@ -7,7 +7,7 @@ import sys
 from calibre.ebooks.metadata.opf_2_to_3 import upgrade_metadata
 from calibre.ebooks.oeb.base import EPUB_NS, OEB_DOCS, xpath
 from calibre.ebooks.oeb.parse_utils import ensure_namespace_prefixes
-from calibre.ebooks.oeb.polish.container import OEB_FONTS
+from calibre.ebooks.oeb.polish.utils import OEB_FONTS
 from calibre.ebooks.oeb.polish.opf import get_book_language
 from calibre.ebooks.oeb.polish.toc import (
     commit_nav_toc, find_existing_ncx_toc, get_landmarks, get_toc

src/calibre/ebooks/oeb/polish/utils.py

@@ -14,6 +14,56 @@ def guess_type(x):
     return _guess_type(x)[0] or 'application/octet-stream'
 
 
+# All font mimetypes seen in e-books
+OEB_FONTS = frozenset({
+    'font/otf',
+    'font/woff',
+    'font/woff2',
+    'font/ttf',
+    'application/x-font-ttf',
+    'application/x-font-otf',
+    'application/font-sfnt',
+    'application/vnd.ms-opentype',
+    'application/x-font-truetype',
+})
+
+
+def adjust_mime_for_epub(filename='', mime='', opf_version=(2, 0)):
+    mime = mime or guess_type(filename)
+    if mime == 'text/html':
+        # epubcheck complains if the mimetype for text documents is set to text/html in EPUB 2 books. Sigh.
+        return 'application/xhtml+xml'
+    if mime not in OEB_FONTS:
+        return mime
+    if 'ttf' in mime or 'truetype' in mime:
+        mime = 'font/ttf'
+    elif 'otf' in mime or 'opentype' in mime:
+        mime = 'font/otf'
+    elif mime == 'application/font-sfnt':
+        mime = 'font/otf' if filename.lower().endswith('.otf') else 'font/ttf'
+    elif 'woff2' in mime:
+        mime = 'font/woff2'
+    elif 'woff' in mime:
+        mime = 'font/woff'
+    opf_version = tuple(opf_version[:2])
+    if opf_version == (3, 0):
+        mime = {
+            'font/ttf': 'application/vnd.ms-opentype',  # this is needed by the execrable epubchek
+            'font/otf': 'application/vnd.ms-opentype',
+            'font/woff': 'application/font-woff'}.get(mime, mime)
+    elif opf_version == (3, 1):
+        mime = {
+        'font/ttf': 'application/font-sfnt',
+        'font/otf': 'application/font-sfnt',
+        'font/woff': 'application/font-woff'}.get(mime, mime)
+    elif opf_version < (3, 0):
+        mime = {
+            'font/ttf': 'application/x-font-truetype',
+            'font/otf': 'application/vnd.ms-opentype',
+            'font/woff': 'application/font-woff'}.get(mime, mime)
+    return mime
+
+
 def setup_css_parser_serialization(tab_width=2):
     import css_parser
     prefs = css_parser.ser.prefs

src/calibre/gui2/tweak_book/completion/basic.py

@@ -10,7 +10,8 @@ from collections import namedtuple, OrderedDict
 from qt.core import QObject, pyqtSignal, Qt
 
 from calibre import prepare_string_for_xml
-from calibre.ebooks.oeb.polish.container import OEB_STYLES, OEB_FONTS, name_to_href
+from calibre.ebooks.oeb.polish.container import OEB_STYLES, name_to_href
+from calibre.ebooks.oeb.polish.utils import OEB_FONTS
 from calibre.ebooks.oeb.polish.parsing import parse
 from calibre.ebooks.oeb.polish.report import description_for_anchor
 from calibre.gui2 import is_gui_thread

src/calibre/gui2/tweak_book/file_list.py

@@ -19,7 +19,7 @@ from qt.core import (
 
 from calibre import human_readable, sanitize_file_name
 from calibre.ebooks.oeb.base import OEB_DOCS, OEB_STYLES
-from calibre.ebooks.oeb.polish.container import OEB_FONTS, guess_type
+from calibre.ebooks.oeb.polish.utils import OEB_FONTS, guess_type
 from calibre.ebooks.oeb.polish.cover import (
     get_cover_page_name, get_raster_cover_name, is_raster_image
 )