Allow using HTML transform rules during conversion as well

src/calibre/ebooks/conversion/cli.py

@@ -101,7 +101,7 @@ def option_recommendation_to_cli_option(add_option, rec):
         attrs.pop('type', '')
     if opt.name == 'read_metadata_from_opf':
         switches.append('--from-opf')
-    if opt.name == 'transform_css_rules':
+    elif opt.name == 'transform_css_rules':
         attrs['help'] = _(
             'Path to a file containing rules to transform the CSS styles'
             ' in this book. The easiest way to create such a file is to'
@@ -110,6 +110,15 @@ def option_recommendation_to_cli_option(add_option, rec):
             ' dialog. Once you create the rules, you can use the "Export" button'
             ' to save them to a file.'
         )
+    elif opt.name == 'transform_html_rules':
+        attrs['help'] = _(
+            'Path to a file containing rules to transform the HTML'
+            ' in this book. The easiest way to create such a file is to'
+            ' use the wizard for creating rules in the calibre GUI. Access'
+            ' it in the "Look & feel->Transform HTML" section of the conversion'
+            ' dialog. Once you create the rules, you can use the "Export" button'
+            ' to save them to a file.'
+        )
     if opt.name in DEFAULT_TRUE_OPTIONS and rec.recommended_value is True:
         switches = ['--disable-'+opt.long_switch]
     add_option(Option(*switches, **attrs))
@@ -192,7 +201,7 @@ def add_pipeline_options(parser, plumber):
                       'font_size_mapping', 'embed_font_family',
                       'subset_embedded_fonts', 'embed_all_fonts',
                       'line_height', 'minimum_line_height',
-                      'linearize_tables',
+                      'linearize_tables', 'transform_html_rules',
                       'extra_css', 'filter_css', 'transform_css_rules', 'expand_css',
                       'smarten_punctuation', 'unsmarten_punctuation',
                       'margin_top', 'margin_left', 'margin_right',
@@ -388,6 +397,17 @@ def main(args=sys.argv):
                     log.error(title)
                     log.error(msg)
                     return 1
+    if opts.transform_html_rules:
+        from calibre.ebooks.html_transform_rules import import_rules, validate_rule
+        with open(opts.transform_html_rules, 'rb') as tcr:
+            opts.transform_html_rules = rules = list(import_rules(tcr.read()))
+            for rule in rules:
+                title, msg = validate_rule(rule)
+                if title and msg:
+                    log.error('Failed to parse HTML transform rules')
+                    log.error(title)
+                    log.error(msg)
+                    return 1
 
     recommendations = [(n.dest, getattr(opts, n.dest),
                         OptionRecommendation.HIGH)

src/calibre/ebooks/conversion/config.py

@@ -246,7 +246,7 @@ def get_sorted_output_formats(preferred_fmt=None):
             'remove_paragraph_spacing', 'remove_paragraph_spacing_indent_size',
             'insert_blank_line_size', 'input_encoding', 'filter_css',
             'expand_css', 'asciiize', 'keep_ligatures', 'linearize_tables',
-            'transform_css_rules'),
+            'transform_css_rules', 'transform_html_rules'),
 
         'metadata': ('prefer_metadata_cover',),
 

src/calibre/ebooks/conversion/plumber.py

@@ -369,6 +369,12 @@ def __init__(self, input, output, log, report_progress=DummyReporter(),
                    ' rules are applied after all other CSS processing is done.')
         ),
 
+OptionRecommendation(name='transform_html_rules',
+            recommended_value=None, level=OptionRecommendation.LOW,
+            help=_('Rules for transforming the HTML in this book. These'
+                   ' rules are applied after the HTML is parsed, but before any other transformations.')
+        ),
+
 OptionRecommendation(name='filter_css',
             recommended_value=None, level=OptionRecommendation.LOW,
             help=_('A comma separated list of CSS properties that '
@@ -881,7 +887,7 @@ def eq(name, a, b):
             if name in {'sr1_search', 'sr1_replace', 'sr2_search', 'sr2_replace', 'sr3_search', 'sr3_replace', 'filter_css', 'comments'}:
                 if not a and not b:
                     return True
-            if name in {'transform_css_rules', 'search_replace'}:
+            if name in {'transform_css_rules', 'transform_html_rules', 'search_replace'}:
                 if b == '[]':
                     b = None
             return a == b
@@ -1133,6 +1139,13 @@ def run(self):
 
         self.oeb.plumber_output_format = self.output_fmt or ''
 
+        if self.opts.transform_html_rules:
+            transform_html_rules = self.opts.transform_html_rules
+            if isinstance(transform_html_rules, string_or_bytes):
+                transform_html_rules = json.loads(transform_html_rules)
+            from calibre.ebooks.html_transform_rules import transform_conversion_book
+            transform_conversion_book(self.oeb, self.opts, transform_html_rules)
+
         from calibre.ebooks.oeb.transforms.data_url import DataURL
         DataURL()(self.oeb, self.opts)
         from calibre.ebooks.oeb.transforms.guide import Clean

src/calibre/ebooks/html_transform_rules.py

@@ -438,6 +438,15 @@ def transform_container(container, serialized_rules, names=()):
     return doc_changed
 
 
+def transform_conversion_book(oeb, opts, serialized_rules):
+    rules = tuple(Rule(r) for r in serialized_rules)
+    for item in oeb.spine:
+        root = item.data
+        if not hasattr(root, 'xpath'):
+            continue
+        transform_doc(root, rules)
+
+
 def rule_to_text(rule):
     text = _('If the tag {match_type} {query}').format(
         match_type=MATCH_TYPE_MAP[rule['match_type']].text, query=rule.get('query') or '')

src/calibre/gui2/convert/look_and_feel.py

@@ -70,7 +70,7 @@ def get_value_handler(self, g):
             val = str(g.text()).strip()
             val = [x.strip() for x in val.split(',' if ',' in val else ' ') if x.strip()]
             return ', '.join(val) or None
-        if g is self.opt_transform_css_rules:
+        if g is self.opt_transform_css_rules or g is self.opt_transform_html_rules:
             return json.dumps(g.rules)
         return Widget.get_value_handler(self, g)
 
@@ -95,7 +95,7 @@ def set_value_handler(self, g, val):
                     w.setChecked(False)
             self.filter_css_others.setText(', '.join(items))
             return True
-        if g is self.opt_transform_css_rules:
+        if g is self.opt_transform_css_rules or g is self.opt_transform_html_rules:
             g.rules = json.loads(val) if val else []
             return True
 
@@ -106,7 +106,7 @@ def connect_gui_obj_handler(self, gui_obj, slot):
                 w.stateChanged.connect(slot)
             self.filter_css_others.textChanged.connect(slot)
             return
-        if gui_obj is self.opt_transform_css_rules:
+        if gui_obj is self.opt_transform_css_rules or gui_obj is self.opt_transform_html_rules:
             gui_obj.changed.connect(slot)
             return
         raise NotImplementedError()

src/calibre/gui2/convert/look_and_feel.ui

@@ -10,7 +10,7 @@
     <height>619</height>
    </rect>
   </property>
-  <layout class="QVBoxLayout" name="verticalLayout_2">
+  <layout class="QHBoxLayout" name="horizontalLayout_2">
    <property name="spacing">
     <number>0</number>
    </property>
@@ -503,6 +503,16 @@
        </item>
       </layout>
      </widget>
+     <widget class="QWidget" name="tab_5">
+      <attribute name="title">
+       <string>Transform &amp;HTML</string>
+      </attribute>
+      <layout class="QVBoxLayout" name="verticalLayout_2">
+       <item>
+        <widget class="HtmlRulesWidget" name="opt_transform_html_rules" native="true"/>
+       </item>
+      </layout>
+     </widget>
     </widget>
    </item>
   </layout>
@@ -524,6 +534,12 @@
    <header>calibre/gui2/css_transform_rules.h</header>
    <container>1</container>
   </customwidget>
+  <customwidget>
+   <class>HtmlRulesWidget</class>
+   <extends>QWidget</extends>
+   <header>calibre/gui2/html_transform_rules.h</header>
+   <container>1</container>
+  </customwidget>
  </customwidgets>
  <resources>
   <include location="../../../../resources/images.qrc"/>

src/calibre/gui2/css_transform_rules.py

@@ -317,15 +317,17 @@ def export_rules(self):
                 'There are no rules to export'), show=True)
         path = choose_save_file(self, self.DIR_SAVE_NAME, _('Choose file for exported rules'), initial_filename=self.INITIAL_FILE_NAME)
         if path:
-            raw = self.export_func(rules)
+            f = self.__class__.export_func
+            raw = f(rules)
             with open(path, 'wb') as f:
                 f.write(raw)
 
     def import_rules(self):
         paths = choose_files(self, self.DIR_SAVE_NAME, _('Choose file to import rules from'), select_only_single_file=True)
         if paths:
+            func = self.__class__.import_func
             with open(paths[0], 'rb') as f:
-                rules = self.import_func(f.read())
+                rules = func(f.read())
             self.rules_widget.rules = list(rules) + list(self.rules_widget.rules)
             self.changed.emit()
 

src/calibre/gui2/html_transform_rules.py

@@ -357,7 +357,7 @@ def transform_scope(self, val):
 # }}}
 
 
-class RulesWidget(RulesWidgetBase):  # {{{
+class HtmlRulesWidget(RulesWidgetBase):  # {{{
     PREFS_NAME = 'html-transform-rules'
     INITIAL_FILE_NAME = 'html-rules.txt'
     DIR_SAVE_NAME = 'export-html-transform-rules'