Edit book: Remove unused CSS: Add an option to also remove unused class attributes from the HTML

Edit Book: Remove unused CSS: Handle @import rules when checking for unused CSS

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

@@ -9,8 +9,10 @@ __copyright__ = '2014, Kovid Goyal <kovid at kovidgoyal.net>'
 import re
 
 from lxml import etree
-from cssselect import HTMLTranslator
+from cssutils.css import CSSRule
+from cssselect import HTMLTranslator, parse
 from cssselect.xpath import XPathExpr, is_safe_name
+from cssselect.parser import SelectorSyntaxError
 
 from calibre import force_unicode
 from calibre.ebooks.oeb.base import OEB_STYLES, OEB_DOCS, XPNSMAP, XHTML_NS
@@ -118,8 +120,21 @@ def preserve_htmlns_prefix(sheet, prefix):
     else:
         sheet.namespaces[prefix] = XHTML_NS
 
-def remove_unused_css(container, report):
+def get_imported_sheets(name, container, sheets, recursion_level=10, sheet=None):
-    from cssutils.css import CSSRule
+    ans = set()
+    sheet = sheet or sheets[name]
+    for rule in sheet.cssRules.rulesOfType(CSSRule.IMPORT_RULE):
+        if rule.href:
+            iname = container.href_to_name(rule.href, name)
+            if iname in sheets:
+                ans.add(iname)
+    if recursion_level > 0:
+        for imported_sheet in tuple(ans):
+            ans |= get_imported_sheets(imported_sheet, container, sheets, recursion_level=recursion_level-1)
+    ans.discard(name)
+    return ans
+
+def remove_unused_css(container, report, remove_unused_classes=False):
     def safe_parse(name):
         try:
             return container.parsed(name)
@@ -127,13 +142,16 @@ def remove_unused_css(container, report):
             pass
     sheets = {name:safe_parse(name) for name, mt in container.mime_map.iteritems() if mt in OEB_STYLES}
     sheets = {k:v for k, v in sheets.iteritems() if v is not None}
+    import_map = {name:get_imported_sheets(name, container, sheets) for name in sheets}
+    if remove_unused_classes:
+        class_map = {name:{icu_lower(x) for x in classes_in_rule_list(sheet.cssRules)} for name, sheet in sheets.iteritems()}
     sheet_namespace = {}
     for sheet in sheets.itervalues():
         sheet_namespace[sheet] = process_namespaces(sheet)
         sheet.namespaces['h'] = XHTML_NS
     style_rules = {name:tuple(sheet.cssRules.rulesOfType(CSSRule.STYLE_RULE)) for name, sheet in sheets.iteritems()}
 
-    num_of_removed_rules = 0
+    num_of_removed_rules = num_of_removed_classes = 0
     pseudo_pat = re.compile(r':(first-letter|first-line|link|hover|visited|active|focus|before|after)', re.I)
     cache = {}
 
@@ -141,9 +159,17 @@ def remove_unused_css(container, report):
         if mt not in OEB_DOCS:
             continue
         root = container.parsed(name)
+        used_classes = set()
         for style in root.xpath('//*[local-name()="style"]'):
             if style.get('type', 'text/css') == 'text/css' and style.text:
                 sheet = container.parse_css(style.text)
+                if remove_unused_classes:
+                    used_classes |= {icu_lower(x) for x in classes_in_rule_list(sheet.cssRules)}
+                imports = get_imported_sheets(name, container, sheets, sheet=sheet)
+                for imported_sheet in imports:
+                    style_rules[imported_sheet] = tuple(filter_used_rules(root, style_rules[imported_sheet], container.log, pseudo_pat, cache))
+                    if remove_unused_classes:
+                        used_classes |= class_map[imported_sheet]
                 ns = process_namespaces(sheet)
                 sheet.namespaces['h'] = XHTML_NS
                 rules = tuple(sheet.cssRules.rulesOfType(CSSRule.STYLE_RULE))
@@ -160,6 +186,27 @@ def remove_unused_css(container, report):
             sname = container.href_to_name(link.get('href'), name)
             if sname in sheets:
                 style_rules[sname] = tuple(filter_used_rules(root, style_rules[sname], container.log, pseudo_pat, cache))
+            if remove_unused_classes:
+                used_classes |= class_map[sname]
+
+            for iname in import_map[sname]:
+                style_rules[iname] = tuple(filter_used_rules(root, style_rules[iname], container.log, pseudo_pat, cache))
+                if remove_unused_classes:
+                    used_classes |= class_map[iname]
+
+        if remove_unused_classes:
+            for elem in root.xpath('//*[@class]'):
+                original_classes, classes = elem.get('class', '').split(), []
+                for x in original_classes:
+                    if icu_lower(x) in used_classes:
+                        classes.append(x)
+                if len(classes) != len(original_classes):
+                    if classes:
+                        elem.set('class', ' '.join(classes))
+                    else:
+                        del elem.attrib['class']
+                    num_of_removed_classes += len(original_classes) - len(classes)
+                    container.dirty(name)
 
     for name, sheet in sheets.iteritems():
         preserve_htmlns_prefix(sheet, sheet_namespace[sheet])
@@ -170,10 +217,17 @@ def remove_unused_css(container, report):
             container.dirty(name)
 
     if num_of_removed_rules > 0:
-        report(_('Removed %d unused CSS style rules') % num_of_removed_rules)
+        report(ngettext('Removed %d unused CSS style rule', 'Removed %d unused CSS style rules',
+                        num_of_removed_rules) % num_of_removed_rules)
     else:
         report(_('No unused CSS style rules found'))
-    return num_of_removed_rules > 0
+    if remove_unused_classes:
+        if num_of_removed_classes > 0:
+            report(ngettext('Removed %d unused class from the HTML', 'Removed %d unused classes from the HTML',
+                   num_of_removed_classes) % num_of_removed_classes)
+        else:
+            report(_('No unused class attributes found'))
+    return num_of_removed_rules + num_of_removed_classes > 0
 
 def filter_declaration(style, properties):
     changed = False
@@ -251,3 +305,30 @@ def filter_css(container, properties, names=()):
 
     return doc_changed
 
+def _classes_in_selector(selector, classes):
+    for attr in ('selector', 'subselector', 'parsed_tree'):
+        s = getattr(selector, attr, None)
+        if s is not None:
+            _classes_in_selector(s, classes)
+    cn = getattr(selector, 'class_name', None)
+    if cn is not None:
+        classes.add(cn)
+
+def classes_in_selector(text):
+    classes = set()
+    try:
+        for selector in parse(text):
+            _classes_in_selector(selector, classes)
+    except SelectorSyntaxError:
+        pass
+    return classes
+
+def classes_in_rule_list(css_rules):
+    classes = set()
+    for rule in css_rules:
+        if rule.type == rule.STYLE_RULE:
+            classes |= classes_in_selector(rule.selectorText)
+        elif hasattr(rule, 'cssRules'):
+            classes |= classes_in_rule_list(rule.cssRules)
+    return classes
+

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

@@ -33,6 +33,10 @@ ALL_OPTS = {
     'remove_unused_css':False,
 }
 
+CUSTOMIZATION = {
+    'remove_unused_classes': False,
+}
+
 SUPPORTED = {'EPUB', 'AZW3'}
 
 # Help {{{
@@ -129,10 +133,11 @@ def update_metadata(ebook, new_opf):
         stream.truncate()
         stream.write(opf.render())
 
-def polish_one(ebook, opts, report):
+def polish_one(ebook, opts, report, customization=None):
     rt = lambda x: report('\n### ' + x)
     jacket = None
     changed = False
+    customization = customization or CUSTOMIZATION.copy()
 
     if opts.subset or opts.embed:
         stats = StatsCollector(ebook, do_embed=opts.embed)
@@ -194,7 +199,7 @@ def polish_one(ebook, opts, report):
 
     if opts.remove_unused_css:
         rt(_('Removing unused CSS rules'))
-        if remove_unused_css(ebook, report):
+        if remove_unused_css(ebook, report, remove_unused_classes=customization['remove_unused_classes']):
             changed = True
         report('')
 
@@ -233,13 +238,13 @@ def gui_polish(data):
         log(msg)
     return '\n\n'.join(report)
 
-def tweak_polish(container, actions):
+def tweak_polish(container, actions, customization=None):
     opts = ALL_OPTS.copy()
     opts.update(actions)
     O = namedtuple('Options', ' '.join(ALL_OPTS.iterkeys()))
     opts = O(**opts)
     report = []
-    changed = polish_one(container, opts, report.append)
+    changed = polish_one(container, opts, report.append, customization=customization)
     return report, changed
 
 def option_parser():

src/calibre/gui2/tweak_book/__init__.py

@@ -48,6 +48,7 @@ d['insert_tag_mru'] = ['p', 'div', 'li', 'h1', 'h2', 'h3', 'h4', 'em', 'strong',
 d['spell_check_case_sensitive_sort'] = False
 d['inline_spell_check'] = True
 d['custom_themes'] = {}
+d['remove_unused_classes'] = False
 
 del d
 

src/calibre/gui2/tweak_book/boss.py

@@ -11,8 +11,7 @@ from functools import partial, wraps
 
 from PyQt4.Qt import (
     QObject, QApplication, QDialog, QGridLayout, QLabel, QSize, Qt,
-    QDialogButtonBox, QIcon, QTimer, QPixmap, QTextBrowser, QVBoxLayout,
+    QDialogButtonBox, QIcon, QTimer, QPixmap, QInputDialog)
-    QInputDialog)
 
 from calibre import prints, isbytestring
 from calibre.ptempfile import PersistentTemporaryDirectory, TemporaryDirectory
@@ -425,36 +424,22 @@ class Boss(QObject):
         self.edit_file(name, 'html')
 
     def polish(self, action, name, parent=None):
+        from calibre.gui2.tweak_book.polish import get_customization, show_report
+        customization = get_customization(action, name, parent or self.gui)
+        if customization is None:
+            return
         with BusyCursor():
             self.add_savepoint(_('Before: %s') % name)
             try:
-                report, changed = tweak_polish(current_container(), {action:True})
+                report, changed = tweak_polish(current_container(), {action:True}, customization=customization)
             except:
                 self.rewind_savepoint()
                 raise
             if changed:
                 self.apply_container_update_to_gui()
-            from calibre.ebooks.markdown import markdown
-            report = markdown('# %s\n\n'%self.current_metadata.title + '\n\n'.join(report), output_format='html4')
         if not changed:
             self.rewind_savepoint()
-        d = QDialog(parent or self.gui)
+        show_report(changed, self.current_metadata.title, report, parent or self.gui, self.show_current_diff)
-        d.l = QVBoxLayout()
-        d.setLayout(d.l)
-        d.e = QTextBrowser(d)
-        d.l.addWidget(d.e)
-        d.e.setHtml(report)
-        d.bb = QDialogButtonBox(QDialogButtonBox.Close)
-        if changed:
-            b = d.b = d.bb.addButton(_('See what &changed'), d.bb.AcceptRole)
-            b.setIcon(QIcon(I('diff.png'))), b.setAutoDefault(False)
-            b.clicked.connect(partial(self.show_current_diff, allow_revert=True))
-        d.bb.button(d.bb.Close).setDefault(True)
-        d.l.addWidget(d.bb)
-        d.bb.rejected.connect(d.reject)
-        d.bb.accepted.connect(d.accept)
-        d.resize(600, 400)
-        d.exec_()
 
     def manage_fonts(self):
         self.commit_all_editors_to_container()

src/calibre/gui2/tweak_book/polish.py

@@ -0,0 +1,72 @@
+#!/usr/bin/env python
+# vim:fileencoding=utf-8
+from __future__ import (unicode_literals, division, absolute_import,
+                        print_function)
+
+__license__ = 'GPL v3'
+__copyright__ = '2014, Kovid Goyal <kovid at kovidgoyal.net>'
+
+from functools import partial
+
+from PyQt4.Qt import (
+    QTextBrowser, QVBoxLayout, QDialog, QDialogButtonBox, QIcon, QLabel, QCheckBox)
+
+from calibre.ebooks.oeb.polish.main import CUSTOMIZATION
+from calibre.gui2.tweak_book import tprefs
+
+class Abort(Exception):
+    pass
+
+def customize_remove_unused_css(name, parent, ans):
+    d = QDialog(parent)
+    d.l = l = QVBoxLayout()
+    d.setLayout(d.l)
+    d.setWindowTitle(_('Remove unused CSS'))
+    d.la = la = QLabel(_(
+        'This will remove all CSS rules that do not match any actual content. You'
+        ' can also have it automatically remove any class attributes from the HTML'
+        ' that do not match any CSS rules, by using the check box below:'))
+    la.setWordWrap(True), l.addWidget(la)
+    d.c = c = QCheckBox(_('Remove unused &class attributes'))
+    c.setChecked(tprefs['remove_unused_classes'])
+    l.addWidget(c)
+    d.bb = QDialogButtonBox(QDialogButtonBox.Ok | QDialogButtonBox.Cancel)
+    d.l.addWidget(d.bb)
+    d.bb.rejected.connect(d.reject)
+    d.bb.accepted.connect(d.accept)
+    if d.exec_() != d.Accepted:
+        raise Abort()
+    ans['remove_unused_classes'] = tprefs['remove_unused_classes'] = c.isChecked()
+
+def get_customization(action, name, parent):
+    ans = CUSTOMIZATION.copy()
+    try:
+        if action == 'remove_unused_css':
+            customize_remove_unused_css(name, parent, ans)
+    except Abort:
+        return None
+    return ans
+
+def format_report(title, report):
+    from calibre.ebooks.markdown import markdown
+    return markdown('# %s\n\n'%title + '\n\n'.join(report), output_format='html4')
+
+def show_report(changed, title, report, parent, show_current_diff):
+    report = format_report(title, report)
+    d = QDialog(parent)
+    d.l = QVBoxLayout()
+    d.setLayout(d.l)
+    d.e = QTextBrowser(d)
+    d.l.addWidget(d.e)
+    d.e.setHtml(report)
+    d.bb = QDialogButtonBox(QDialogButtonBox.Close)
+    if changed:
+        b = d.b = d.bb.addButton(_('See what &changed'), d.bb.AcceptRole)
+        b.setIcon(QIcon(I('diff.png'))), b.setAutoDefault(False)
+        b.clicked.connect(partial(show_current_diff, allow_revert=True))
+    d.bb.button(d.bb.Close).setDefault(True)
+    d.l.addWidget(d.bb)
+    d.bb.rejected.connect(d.reject)
+    d.bb.accepted.connect(d.accept)
+    d.resize(600, 400)
+    d.exec_()