From 1767c90680afa567d2ae419b10a39e294811fe55 Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Wed, 7 May 2014 14:34:11 +0530
Subject: [PATCH] Edit Book: A new tool to filter style information
 (Tools->Filter style information). Useful to easily remove some CSS
 properties from the entire book.

---
 manual/edit.rst                        | 11 ++++
 src/calibre/ebooks/oeb/polish/css.py   | 78 ++++++++++++++++++++++++++
 src/calibre/gui2/tweak_book/boss.py    | 18 +++++-
 src/calibre/gui2/tweak_book/ui.py      |  3 +
 src/calibre/gui2/tweak_book/widgets.py | 64 ++++++++++++++++++++-
 5 files changed, 170 insertions(+), 4 deletions(-)

diff --git a/manual/edit.rst b/manual/edit.rst
index d4a2269156..a43992b375 100644
--- a/manual/edit.rst
+++ b/manual/edit.rst
@@ -400,6 +400,17 @@ If you use this tool multiple times, each invocation will cause the previously
 created inline Table of Contents to be replaced. The tool can be accessed via
 :guilabel:`Tools->Table of Contents->Insert inline Table of Contents`.
 
+Filter style information
+^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+This tool can be used to easily remove specified CSS style properties from the
+entire book. You can tell it what properties you want removed, for example,
+``color, background-color, line-height`` and it will remove them from
+everywhere they occur --- stylesheets, ``<style>`` tags and inline ``style``
+attributes. After removing the style information, a summary of all the changes
+made is displayed so you can see exactly what was changed. The tool can be
+accessed via :guilabel:`Tools->Filter style information`.
+
 .. _checkpoints:
 
 Checkpoints
diff --git a/src/calibre/ebooks/oeb/polish/css.py b/src/calibre/ebooks/oeb/polish/css.py
index 6f82cacd4c..8fd885861f 100644
--- a/src/calibre/ebooks/oeb/polish/css.py
+++ b/src/calibre/ebooks/oeb/polish/css.py
@@ -14,6 +14,7 @@ from cssselect.xpath import XPathExpr, is_safe_name
 
 from calibre import force_unicode
 from calibre.ebooks.oeb.base import OEB_STYLES, OEB_DOCS, XPNSMAP, XHTML_NS
+from calibre.ebooks.oeb.normalize_css import normalize_filter_css, normalizers
 from calibre.ebooks.oeb.stylizer import MIN_SPACE_RE, is_non_whitespace, xpath_lower_case, fix_namespace
 from calibre.ebooks.oeb.polish.pretty import pretty_script_or_style
 
@@ -167,3 +168,80 @@ def remove_unused_css(container, report):
     else:
         report(_('No unused CSS style rules found'))
     return num_of_removed_rules > 0
+
+def filter_declaration(style, properties):
+    changed = False
+    for prop in properties:
+        if style.removeProperty(prop) != '':
+            changed = True
+    all_props = set(style.keys())
+    for prop in style.getProperties():
+        n = normalizers.get(prop.name, None)
+        if n is not None:
+            normalized = n(prop.name, prop.propertyValue)
+            removed = properties.intersection(set(normalized))
+            if removed:
+                changed = True
+                style.removeProperty(prop.name)
+                for prop in set(normalized) - removed - all_props:
+                    style.setProperty(prop, normalized[prop])
+    return changed
+
+def filter_sheet(sheet, properties):
+    from cssutils.css import CSSRule
+    changed = False
+    remove = []
+    for rule in sheet.cssRules.rulesOfType(CSSRule.STYLE_RULE):
+        if filter_declaration(rule.style, properties):
+            changed = True
+            if rule.style.length == 0:
+                remove.append(rule)
+    for rule in remove:
+        sheet.cssRules.remove(rule)
+    return changed
+
+
+def filter_css(container, properties, names=()):
+    if not names:
+        types = OEB_STYLES | OEB_DOCS
+        names = []
+        for name, mt in container.mime_map.iteritems():
+            if mt in types:
+                names.append(name)
+    properties = normalize_filter_css(properties)
+    doc_changed = False
+
+    for name in names:
+        mt = container.mime_map[name]
+        if mt in OEB_STYLES:
+            sheet = container.parsed(name)
+            filtered = filter_sheet(sheet, properties)
+            if filtered:
+                container.dirty(name)
+                doc_changed = True
+        elif mt in OEB_DOCS:
+            root = container.parsed(name)
+            changed = False
+            for style in root.xpath('//*[local-name()="style"]'):
+                if style.text and style.get('type', 'text/css') in {None, '', 'text/css'}:
+                    sheet = container.parse_css(style.text)
+                    if filter_sheet(sheet, properties):
+                        changed = True
+                        style.text = force_unicode(sheet.cssText, 'utf-8')
+                        pretty_script_or_style(container, style)
+            for elem in root.xpath('//*[@style]'):
+                text = elem.get('style', None)
+                if text:
+                    style = container.parse_css(text, is_declaration=True)
+                    if filter_declaration(style, properties):
+                        changed = True
+                        if style.length == 0:
+                            del elem.attrib['style']
+                        else:
+                            elem.set('style', force_unicode(style.getCssText(separator=' '), 'utf-8'))
+            if changed:
+                container.dirty(name)
+                doc_changed = True
+
+    return doc_changed
+
diff --git a/src/calibre/gui2/tweak_book/boss.py b/src/calibre/gui2/tweak_book/boss.py
index 322ab81f6b..c1f36621c8 100644
--- a/src/calibre/gui2/tweak_book/boss.py
+++ b/src/calibre/gui2/tweak_book/boss.py
@@ -20,6 +20,7 @@ from calibre.ebooks.oeb.base import urlnormalize
 from calibre.ebooks.oeb.polish.main import SUPPORTED, tweak_polish
 from calibre.ebooks.oeb.polish.container import get_container as _gc, clone_container, guess_type, OEB_FONTS
 from calibre.ebooks.oeb.polish.cover import mark_as_cover, mark_as_titlepage
+from calibre.ebooks.oeb.polish.css import filter_css
 from calibre.ebooks.oeb.polish.pretty import fix_all_html, pretty_all
 from calibre.ebooks.oeb.polish.replace import rename_files, replace_file, get_recommended_folders, rationalize_folders
 from calibre.ebooks.oeb.polish.split import split, merge, AbortError, multisplit
@@ -41,7 +42,7 @@ from calibre.gui2.tweak_book.search import validate_search_request, run_search
 from calibre.gui2.tweak_book.spell import find_next as find_next_word
 from calibre.gui2.tweak_book.widgets import (
     RationalizeFolders, MultiSplit, ImportForeign, QuickOpen, InsertLink,
-    InsertSemantics, BusyCursor, InsertTag)
+    InsertSemantics, BusyCursor, InsertTag, FilterCSS)
 
 _diff_dialogs = []
 
@@ -674,6 +675,21 @@ class Boss(QObject):
             d.apply_changes(current_container())
             self.apply_container_update_to_gui()
 
+    def filter_css(self):
+        self.commit_all_editors_to_container()
+        d = FilterCSS(parent=self.gui)
+        if d.exec_() == d.Accepted and d.filtered_properties:
+            self.add_savepoint(_('Before: Filter style information'))
+            with BusyCursor():
+                changed = filter_css(current_container(), d.filtered_properties)
+            if changed:
+                self.apply_container_update_to_gui()
+                self.show_current_diff()
+            else:
+                self.rewind_savepoint()
+                return info_dialog(self.gui, _('No matches'), _(
+                    'No matching style rules were found'), show=True)
+
     def show_find(self):
         self.gui.central.show_find()
         ed = self.gui.central.current_editor
diff --git a/src/calibre/gui2/tweak_book/ui.py b/src/calibre/gui2/tweak_book/ui.py
index cc7832c906..b4ac60f893 100644
--- a/src/calibre/gui2/tweak_book/ui.py
+++ b/src/calibre/gui2/tweak_book/ui.py
@@ -349,6 +349,8 @@ class Main(MainWindow):
                                       _('Arrange into folders'))
         self.action_set_semantics = reg('tags.png', _('Set &Semantics'), self.boss.set_semantics, 'set-semantics', (),
                                         _('Set Semantics'))
+        self.action_filter_css = reg('filter.png', _('&Filter style information'), self.boss.filter_css, 'filter-css', (),
+                                     _('Filter style information'))
 
         # Polish actions
         group = _('Polish Book')
@@ -482,6 +484,7 @@ class Main(MainWindow):
         e.addAction(self.action_pretty_all)
         e.addAction(self.action_rationalize_folders)
         e.addAction(self.action_set_semantics)
+        e.addAction(self.action_filter_css)
         e.addAction(self.action_spell_check_book)
         e.addAction(self.action_check_book)
 
diff --git a/src/calibre/gui2/tweak_book/widgets.py b/src/calibre/gui2/tweak_book/widgets.py
index d2b9b1da7f..d82ea589e1 100644
--- a/src/calibre/gui2/tweak_book/widgets.py
+++ b/src/calibre/gui2/tweak_book/widgets.py
@@ -15,7 +15,7 @@ from PyQt4.Qt import (
     QFormLayout, QHBoxLayout, QToolButton, QIcon, QApplication, Qt, QWidget,
     QPoint, QSizePolicy, QPainter, QStaticText, pyqtSignal, QTextOption,
     QAbstractListModel, QModelIndex, QVariant, QStyledItemDelegate, QStyle,
-    QListView, QTextDocument, QSize, QComboBox, QFrame, QCursor)
+    QListView, QTextDocument, QSize, QComboBox, QFrame, QCursor, QCheckBox)
 
 from calibre import prepare_string_for_xml
 from calibre.ebooks.oeb.polish.utils import lead_text
@@ -689,7 +689,7 @@ class InsertLink(Dialog):
                 frag = item.get('id', None) or item.get('name')
                 text = lead_text(item, num_words=4)
                 ac.append((text, frag))
-            ac.sort(key=lambda (text, frag): primary_sort_key(text))
+            ac.sort(key=lambda text_frag: primary_sort_key(text_frag[0]))
         self.anchor_names.model().set_names(self.anchor_cache[name])
         self.update_target()
 
@@ -924,6 +924,64 @@ class InsertSemantics(Dialog):
 
 # }}}
 
+class FilterCSS(Dialog):  # {{{
+
+    def __init__(self, parent=None):
+        Dialog.__init__(self, _('Filter Style Information'), 'filter-css', parent=parent)
+
+    def setup_ui(self):
+        from calibre.gui2.convert.look_and_feel_ui import Ui_Form
+        f, w = Ui_Form(), QWidget()
+        f.setupUi(w)
+        self.l = l = QFormLayout(self)
+        self.setLayout(l)
+
+        l.addRow(QLabel(_('Select what style information you want completely removed:')))
+        self.h = h = QHBoxLayout()
+
+        for name, text in {
+                'fonts':_('&Fonts'), 'margins':_('&Margins'), 'padding':_('&Padding'), 'floats':_('Flo&ats'), 'colors':_('&Colors')}.iteritems():
+            c = QCheckBox(text)
+            setattr(self, 'opt_' + name, c)
+            h.addWidget(c)
+            c.setToolTip(getattr(f, 'filter_css_' + name).toolTip())
+        l.addRow(h)
+
+        self.others = o = QLineEdit(self)
+        l.addRow(_('&Other CSS properties:'), o)
+        o.setToolTip(f.filter_css_others.toolTip())
+
+        l.addRow(self.bb)
+
+    @property
+    def filtered_properties(self):
+        ans = set()
+        a = ans.add
+        if self.opt_fonts.isChecked():
+            a('font-family')
+        if self.opt_margins.isChecked():
+            a('margin')
+        if self.opt_padding.isChecked():
+            a('padding')
+        if self.opt_floats.isChecked():
+            a('float'), a('clear')
+        if self.opt_colors.isChecked():
+            a('color'), a('background-color')
+        for x in unicode(self.others.text()).split(','):
+            x = x.strip()
+            if x:
+                a(x)
+        return ans
+
+    @classmethod
+    def test(cls):
+        d = cls()
+        if d.exec_() == d.Accepted:
+            print (d.filtered_properties)
+
+# }}}
+
+
 if __name__ == '__main__':
     app = QApplication([])
-    InsertLink.test()
+    FilterCSS.test()