Polish books: Add a tool to add or remove soft hyphens from the book text

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

@@ -0,0 +1,30 @@
+#!/usr/bin/env python2
+# vim:fileencoding=utf-8
+# License: GPL v3 Copyright: 2019, Kovid Goyal <kovid at kovidgoyal.net>
+
+from __future__ import absolute_import, division, print_function, unicode_literals
+
+from calibre.ebooks.oeb.base import OEB_DOCS
+from polyglot.builtins import iteritems
+
+
+def add_soft_hyphens(container, report=None):
+    from calibre.utils.hyphenation.hyphenate import add_soft_hyphens_to_html
+    for name, mt in iteritems(container.mime_map):
+        if mt not in OEB_DOCS:
+            continue
+        add_soft_hyphens_to_html(container.parsed(name), container.mi.language)
+        container.dirty(name)
+    if report is not None:
+        report(_('Soft hyphens added'))
+
+
+def remove_soft_hyphens(container, report=None):
+    from calibre.utils.hyphenation.hyphenate import remove_soft_hyphens_from_html
+    for name, mt in iteritems(container.mime_map):
+        if mt not in OEB_DOCS:
+            continue
+        remove_soft_hyphens_from_html(container.parsed(name))
+        container.dirty(name)
+    if report is not None:
+        report(_('Soft hyphens removed'))

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

@@ -21,6 +21,7 @@ from calibre.ebooks.oeb.polish.replace import smarten_punctuation
 from calibre.ebooks.oeb.polish.jacket import (
     replace_jacket, add_or_replace_jacket, find_existing_jacket, remove_jacket)
 from calibre.ebooks.oeb.polish.css import remove_unused_css
+from calibre.ebooks.oeb.polish.hyphenation import remove_soft_hyphens, add_soft_hyphens
 from calibre.utils.logging import Log
 from polyglot.builtins import iteritems
 
@@ -35,6 +36,8 @@ ALL_OPTS = {
     'remove_unused_css':False,
     'compress_images': False,
     'upgrade_book': False,
+    'add_soft_hyphens': False,
+    'remove_soft_hyphens': False,
 }
 
 CUSTOMIZATION = {
@@ -118,6 +121,15 @@ affecting image quality.</p>
 <p>Upgrade the internal structures of the book, if possible. For instance,
 upgrades EPUB 2 books to EPUB 3 books.</p>
 '''),
+
+'add_soft_hyphens': _('''\
+<p>Add soft hyphens to all words in the book. This allows the book to be rendered
+better when the text is justified, in readers that do not support hyphenation.</p>
+'''),
+
+'remove_soft_hyphens': _('''\
+<p>Remove soft hyphens from all text in the book.</p>
+'''),
 }
 
 
@@ -237,6 +249,15 @@ def polish_one(ebook, opts, report, customization=None):
             changed = True
         report('')
 
+    if opts.remove_soft_hyphens:
+        rt(_('Removing soft hyphens'))
+        remove_soft_hyphens(ebook, report)
+        changed = True
+    elif opts.add_soft_hyphens:
+        rt(_('Adding soft hyphens'))
+        add_soft_hyphens(ebook, report)
+        changed = True
+
     return changed
 
 
@@ -304,6 +325,8 @@ def option_parser():
     o('--smarten-punctuation', '-p', help=CLI_HELP['smarten_punctuation'])
     o('--remove-unused-css', '-u', help=CLI_HELP['remove_unused_css'])
     o('--compress-images', '-i', help=CLI_HELP['compress_images'])
+    o('--add-soft-hyphens', '-H', help=CLI_HELP['add_soft_hyphens'])
+    o('--remove-soft-hyphens', help=CLI_HELP['remove_soft_hyphens'])
     o('--upgrade-book', '-U', help=CLI_HELP['upgrade_book'])
 
     o('--verbose', help=_('Produce more verbose output, useful for debugging.'))

src/calibre/gui2/actions/polish.py

@@ -68,6 +68,8 @@ class Polish(QDialog):  # {{{
             'remove_jacket':_('<h3>Remove book jacket</h3>%s')%HELP['remove_jacket'],
             'remove_unused_css':_('<h3>Remove unused CSS rules</h3>%s')%HELP['remove_unused_css'],
             'compress_images': _('<h3>Losslessly compress images</h3>%s') % HELP['compress_images'],
+            'add_soft_hyphens': _('<h3>Add soft-hyphens</h3>%s') % HELP['add_soft_hyphens'],
+            'remove_soft_hyphens': _('<h3>Remove soft-hyphens</h3>%s') % HELP['remove_soft_hyphens'],
             'upgrade_book': _('<h3>Upgrade book internals</h3>%s') % HELP['upgrade_book'],
         }
 
@@ -88,6 +90,8 @@ class Polish(QDialog):  # {{{
             ('remove_jacket', _('&Remove a previously inserted book jacket')),
             ('remove_unused_css', _('Remove &unused CSS rules from the book')),
             ('compress_images', _('Losslessly &compress images')),
+            ('add_soft_hyphens', _('Add s&oft hyphens')),
+            ('remove_soft_hyphens', _('Remove soft hyphens')),
             ('upgrade_book', _('&Upgrade book internals')),
         ])
         prefs = gprefs.get('polishing_settings', {})