Edit book: Add a tool to transform HTML tags based on rules (Tools->Transform HTML)

src/calibre/gui2/html_transform_rules.py

@@ -293,7 +293,7 @@ class RuleItem(RuleItemBase):  # {{{
                 text += '<br>' + ACTION_MAP[action['type']].short_text
                 if action.get('data'):
                     ad = elided_text(action['data'], font=parent.font(), width=200, pos='right')
-                    text += f'<code>{prepare_string_for_xml(ad)}</code>'
+                    text += f' <code>{prepare_string_for_xml(ad)}</code>'
         except Exception:
             import traceback
             traceback.print_exc()
@@ -336,6 +336,24 @@ class RulesDialog(RulesDialogBase):  # {{{
     PREFS_OBJECT_NAME = 'html-transform-rules'
     RulesClass = Rules
     TesterClass = Tester
+
+    def extra_bottom_widget(self):
+        self.scope_cb = cb = QComboBox()
+        cb.addItem(_('Current HTML file'), 'current')
+        cb.addItem(_('All HTML files'), 'all')
+        cb.addItem(_('Open HTML files'), 'open')
+        cb.addItem(_('Selected HTML files'), 'selected')
+        return cb
+
+    @property
+    def transform_scope(self):
+        return self.scope_cb.currentData()
+
+    @transform_scope.setter
+    def transform_scope(self, val):
+        idx = self.scope_cb.findData(val)
+        self.scope_cb.setCurrentIndex(max(0, idx))
+
 # }}}
 
 

src/calibre/gui2/tag_mapper.py

@@ -495,7 +495,13 @@ class RulesDialog(Dialog, SaveLoadMixin):
         self.l = l = QVBoxLayout(self)
         self.edit_widget = w = self.RulesClass(self)
         l.addWidget(w)
-        l.addWidget(self.bb)
+        ebw = self.extra_bottom_widget()
+        if ebw is None:
+            l.addWidget(self.bb)
+        else:
+            self.h = h = QHBoxLayout()
+            l.addLayout(h)
+            h.addWidget(ebw), h.addStretch(10), h.addWidget(self.bb)
         self.save_button = b = self.bb.addButton(_('&Save'), QDialogButtonBox.ButtonRole.ActionRole)
         b.setToolTip(_('Save this ruleset for later re-use'))
         b.clicked.connect(self.save_ruleset)
@@ -507,6 +513,9 @@ class RulesDialog(Dialog, SaveLoadMixin):
         self.test_button = b = self.bb.addButton(_('&Test rules'), QDialogButtonBox.ButtonRole.ActionRole)
         b.clicked.connect(self.test_rules)
 
+    def extra_bottom_widget(self):
+        pass
+
     @property
     def rules(self):
         return self.edit_widget.rules
@@ -518,6 +527,11 @@ class RulesDialog(Dialog, SaveLoadMixin):
     def test_rules(self):
         self.TesterClass(self.rules, self).exec_()
 
+    def sizeHint(self):
+        ans = super().sizeHint()
+        ans.setWidth(ans.width() + 100)
+        return ans
+
 
 if __name__ == '__main__':
     app = Application([])

src/calibre/gui2/tweak_book/__init__.py

@@ -85,6 +85,7 @@ d['file_list_shows_full_pathname'] = False
 d['auto_link_stylesheets'] = True
 d['check_external_link_anchors'] = True
 d['remove_ncx'] = True
+d['html_transform_scope'] = 'current'
 del d
 
 ucase_map = {l:string.ascii_uppercase[i] for i, l in enumerate(string.ascii_lowercase)}

src/calibre/gui2/tweak_book/boss.py

@@ -74,6 +74,7 @@ from polyglot.urllib import urlparse
 
 _diff_dialogs = []
 last_used_transform_rules = []
+last_used_html_transform_rules = []
 
 
 def get_container(*args, **kwargs):
@@ -626,6 +627,51 @@ class Boss(QObject):
             self.rewind_savepoint()
         show_report(changed, self.current_metadata.title, report, parent or self.gui, self.show_current_diff)
 
+    def transform_html(self):
+        global last_used_html_transform_rules
+        if not self.ensure_book(_('You must first open a book in order to transform styles.')):
+            return
+        from calibre.gui2.html_transform_rules import RulesDialog
+        from calibre.ebooks.html_transform_rules import transform_container
+        d = RulesDialog(self.gui)
+        d.rules = last_used_html_transform_rules
+        d.transform_scope = scope = tprefs['html_transform_scope']
+        ret = d.exec_()
+        last_used_html_transform_rules = d.rules
+        tprefs.set('html_transform_scope', d.transform_scope)
+        if ret != QDialog.DialogCode.Accepted:
+            return
+
+        cc = current_container()
+        names = ()
+        if scope == 'current':
+            if not self.currently_editing or cc.mime_map.get(self.currently_editing) not in OEB_DOCS:
+                return error_dialog(self.gui, _('No HTML file'), _('Not currently editing an HTML file'), show=True)
+            names = (self.currently_editing,)
+        elif scope == 'open':
+            names = tuple(name for name in editors if cc.mime_map.get(name) in OEB_DOCS)
+            if not names:
+                return error_dialog(self.gui, _('No HTML files'), _('Not currently editing any HTML files'), show=True)
+        elif scope == 'selected':
+            names = tuple(name for name in self.gui.file_list.file_list.selected_names if cc.mime_map.get(name) in OEB_DOCS)
+            if not names:
+                return error_dialog(self.gui, _('No HTML files'), _('No HTML files are currently selected in the File browser'), show=True)
+        with BusyCursor():
+            self.add_savepoint(_('Before HTML transformation'))
+            try:
+                changed = transform_container(cc, last_used_html_transform_rules, names)
+            except:
+                self.rewind_savepoint()
+                raise
+            if changed:
+                self.apply_container_update_to_gui()
+        if not changed:
+            self.rewind_savepoint()
+            info_dialog(self.gui, _('No changes'), _(
+                'No HTML was changed.'), show=True)
+            return
+        self.show_current_diff()
+
     def transform_styles(self):
         global last_used_transform_rules
         if not self.ensure_book(_('You must first open a book in order to transform styles.')):

src/calibre/gui2/tweak_book/ui.py

@@ -500,6 +500,8 @@ class Main(MainWindow):
             'Compress images losslessly'))
         self.action_transform_styles = treg('wizard.png', _('Transform &styles'), self.boss.transform_styles, 'transform-styles', (), _(
             'Transform styles used in the book'))
+        self.action_transform_html = treg('wizard.png', _('Transform &HTML'), self.boss.transform_html, 'transform-html', (), _(
+            'Transform HTML used in the book'))
         self.action_get_ext_resources = treg('download-metadata.png', _('Download external &resources'),
                                              self.boss.get_external_resources, 'get-external-resources', (), _(
             'Download external resources in the book (images/stylesheets/etc/ that are not included in the book)'))
@@ -666,6 +668,7 @@ class Main(MainWindow):
         e.addAction(self.action_smarten_punctuation)
         e.addAction(self.action_remove_unused_css)
         e.addAction(self.action_transform_styles)
+        e.addAction(self.action_transform_html)
         e.addAction(self.action_fix_html_all)
         e.addAction(self.action_pretty_all)
         e.addAction(self.action_rationalize_folders)