Diff view: Add option to pretty print HTML/XML before calculating changes

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

@@ -173,9 +173,10 @@ def pretty_html_tree(container, root):
     for body in root.findall('h:body', namespaces=XPNSMAP):
         pretty_block(body)
 
-    # Handle <script> and <style> tags
+    if container is not None:
-    for child in root.xpath('//*[local-name()="script" or local-name()="style"]'):
+        # Handle <script> and <style> tags
-        pretty_script_or_style(container, child)
+        for child in root.xpath('//*[local-name()="script" or local-name()="style"]'):
+            pretty_script_or_style(container, child)
 
 def fix_html(container, raw):
     root = container.parse_xhtml(raw)

src/calibre/gui2/tweak_book/diff/main.py

@@ -182,6 +182,7 @@ class Diff(Dialog):
 
     def __init__(self, revert_button_msg=None, parent=None, show_open_in_editor=False):
         self.context = 3
+        self.beautify = False
         self.apply_diff_calls = []
         self.show_open_in_editor = show_open_in_editor
         self.revert_button_msg = revert_button_msg
@@ -247,14 +248,17 @@ class Diff(Dialog):
         b.setChecked(True)
         self.pb = b = QToolButton(self)
         b.setIcon(QIcon(I('config.png')))
-        b.setText(_('&Context')), b.setToolButtonStyle(Qt.ToolButtonTextBesideIcon)
+        b.setText(_('&Options')), b.setToolButtonStyle(Qt.ToolButtonTextBesideIcon)
-        b.setToolTip(_('Change the amount of context shown around the changes'))
+        b.setToolTip(_('Change how the differences are displayed'))
         b.setPopupMode(b.InstantPopup)
         m = QMenu(b)
         b.setMenu(m)
+        cm = self.cm = QMenu(_('Lines of context around each change'))
         for i in (3, 5, 10, 50):
-            m.addAction(_('Show %d lines of context around changes') % i, partial(self.change_context, i))
+            cm.addAction(_('Show %d lines of context') % i, partial(self.change_context, i))
-        m.addAction(_('Show all text'), partial(self.change_context, None))
+        cm.addAction(_('Show all text'), partial(self.change_context, None))
+        m.addAction(_('Beautify HTML/XML files before comparing them'), partial(self.change_beautify, True))
+        m.addMenu(cm)
         l.addWidget(b, l.rowCount() - 1, l.columnCount(), 1, 1)
 
         self.bb.setStandardButtons(self.bb.Close)
@@ -286,13 +290,23 @@ class Diff(Dialog):
         if context == self.context:
             return
         self.context = context
+        self.refresh()
+
+    def refresh(self):
         with self:
             self.view.clear()
             for args, kwargs in self.apply_diff_calls:
                 kwargs['context'] = self.context
+                kwargs['beautify'] = self.beautify
                 self.view.add_diff(*args, **kwargs)
             self.view.finalize()
 
+    def change_beautify(self, beautify):
+        if beautify == self.beautify:
+            return
+        self.beautify = beautify
+        self.refresh()
+
     def __enter__(self):
         self.stacks.setCurrentIndex(0)
         self.busy.pi.startAnimation()
@@ -343,31 +357,28 @@ class Diff(Dialog):
             info_dialog(self, _('No changes found'), identical_msg, show=True)
             return True
 
+        kwargs = lambda name: {'context':self.context, 'beautify':self.beautify, 'syntax':syntax_map.get(name, None)}
+
         if isinstance(changed_names, dict):
             for name, other_name in sorted(changed_names.iteritems(), key=lambda x:numeric_sort_key(x[0])):
                 args = (name, other_name, cache.left(name), cache.right(other_name))
-                kwargs = {'syntax':syntax_map.get(name, None), 'context':self.context}
+                add(args, kwargs(name))
-                add(args, kwargs)
         else:
             for name in sorted(changed_names, key=numeric_sort_key):
                 args = (name, name, cache.left(name), cache.right(name))
-                kwargs = {'syntax':syntax_map.get(name, None), 'context':self.context}
+                add(args, kwargs(name))
-                add(args, kwargs)
 
         for name in sorted(added_names, key=numeric_sort_key):
             args = (_('[%s was added]') % name, name, None, cache.right(name))
-            kwargs = {'syntax':syntax_map.get(name, None), 'context':self.context}
+            add(args, kwargs(name))
-            add(args, kwargs)
 
         for name in sorted(removed_names, key=numeric_sort_key):
             args = (name, _('[%s was removed]') % name, cache.left(name), None)
-            kwargs = {'syntax':syntax_map.get(name, None), 'context':self.context}
+            add(args, kwargs(name))
-            add(args, kwargs)
 
         for name, new_name in sorted(renamed_names.iteritems(), key=lambda x:numeric_sort_key(x[0])):
             args = (name, new_name, None, None)
-            kwargs = {'syntax':syntax_map.get(name, None), 'context':self.context}
+            add(args, kwargs(name))
-            add(args, kwargs)
 
     def keyPressEvent(self, ev):
         if not self.view.handle_key(ev):

src/calibre/gui2/tweak_book/diff/view.py

@@ -45,6 +45,20 @@ def get_theme():
         theme = THEMES[default_theme()]
     return theme
 
+def beautify_text(raw, syntax):
+    from lxml import etree
+    from calibre.ebooks.oeb.polish.parsing import parse
+    from calibre.ebooks.oeb.polish.pretty import pretty_xml_tree, pretty_html_tree
+    from calibre.ebooks.chardet import strip_encoding_declarations
+    if syntax == 'xml':
+        root = etree.fromstring(strip_encoding_declarations(raw))
+        pretty_xml_tree(root)
+    else:
+        root = parse(raw, line_numbers=False)
+        pretty_html_tree(None, root)
+    return etree.tostring(root, encoding=unicode)
+
+
 class LineNumberMap(dict):  # {{{
 
     'Map line numbers and keep track of the maximum width of the line numbers'
@@ -518,7 +532,7 @@ class DiffSplit(QSplitter):  # {{{
             v.setTextCursor(c)
         self.update()
 
-    def add_diff(self, left_name, right_name, left_text, right_text, context=None, syntax=None):
+    def add_diff(self, left_name, right_name, left_text, right_text, context=None, syntax=None, beautify=False):
         left_text, right_text = left_text or '', right_text or ''
         is_identical = len(left_text) == len(right_text) and left_text == right_text and left_name == right_name
         is_text = isinstance(left_text, type('')) and isinstance(right_text, type(''))
@@ -542,7 +556,7 @@ class DiffSplit(QSplitter):  # {{{
                 for v in (self.left, self.right):
                     v.appendPlainText('\n')
             elif is_text:
-                self.add_text_diff(left_text, right_text, context, syntax)
+                self.add_text_diff(left_text, right_text, context, syntax, beautify=beautify)
             elif syntax == 'raster_image':
                 self.add_image_diff(left_text, right_text)
             else:
@@ -651,9 +665,11 @@ class DiffSplit(QSplitter):  # {{{
     # }}}
 
     # text diffs {{{
-    def add_text_diff(self, left_text, right_text, context, syntax):
+    def add_text_diff(self, left_text, right_text, context, syntax, beautify=False):
         left_text = unicodedata.normalize('NFC', left_text)
         right_text = unicodedata.normalize('NFC', right_text)
+        if beautify and syntax in {'xml', 'html'}:
+            left_text, right_text = beautify_text(left_text, syntax), beautify_text(right_text, syntax)
         left_lines = self.left_lines = left_text.splitlines()
         right_lines = self.right_lines = right_text.splitlines()