Add malformed markup check when editing ToC as well

src/calibre/ebooks/oeb/polish/choose.coffee

@@ -32,7 +32,22 @@ class AnchorLocator
         # block, if any.
         event.stopPropagation()
         frac = window.pageYOffset/document.body.scrollHeight
-        window.py_bridge.onclick(this, frac)
+        loc = []
+        totals = []
+        parent = this
+        while parent and parent.tagName.toLowerCase() != 'body'
+            totals.push(parent.parentNode.children.length)
+            num = 0
+            sibling = parent.previousElementSibling
+            while sibling
+                num += 1
+                sibling = sibling.previousElementSibling
+            loc.push(num)
+            parent = parent.parentNode
+        loc.reverse()
+        totals.reverse()
+
+        window.py_bridge.onclick(this, JSON.stringify(loc), JSON.stringify(totals), frac)
         return false
 
 calibre_anchor_locator = new AnchorLocator()

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

@@ -359,9 +359,24 @@ def node_from_loc(root, locs, totals=None):
         node = children[locs[0]]
     return node
 
-def add_id(container, name, loc):
+def add_id(container, name, loc, totals=None):
     root = container.parsed(name)
-    node = node_from_loc(root, loc)
+    try:
+        node = node_from_loc(root, loc, totals=totals)
+    except MalformedMarkup:
+        # The webkit HTML parser and the container parser have yielded
+        # different node counts, this can happen if the file is valid XML
+        # but contains constructs like nested <p> tags. So force parse it
+        # with the HTML 5 parser and try again.
+        raw = container.raw_data(name)
+        root = container.parse_xhtml(raw, fname=name, force_html5_parse=True)
+        try:
+            node = node_from_loc(root, loc, totals=totals)
+        except MalformedMarkup:
+            raise MalformedMarkup(_('The file %s has malformed markup. Try running the Fix HTML tool'
+                                    ' before editing.') % name)
+        container.replace(name, root)
+
     node.set('id', node.get('id', uuid_id()))
     container.commit_item(name, keep_parsed=True)
     return node.get('id')

src/calibre/gui2/toc/location.py

@@ -7,6 +7,7 @@ __license__   = 'GPL v3'
 __copyright__ = '2013, Kovid Goyal <kovid at kovidgoyal.net>'
 __docformat__ = 'restructuredtext en'
 
+import json
 from base64 import b64encode
 
 from PyQt4.Qt import (QWidget, QGridLayout, QListWidget, QSize, Qt, QUrl,
@@ -20,7 +21,7 @@ from calibre.utils.logging import default_log
 
 class Page(QWebPage):  # {{{
 
-    elem_clicked = pyqtSignal(object, object, object, object)
+    elem_clicked = pyqtSignal(object, object, object, object, object)
 
     def __init__(self):
         self.log = default_log
@@ -42,21 +43,11 @@ class Page(QWebPage):  # {{{
     def shouldInterruptJavaScript(self):
         return True
 
-    @pyqtSlot(QWebElement, float)
-    def onclick(self, elem, frac):
+    @pyqtSlot(QWebElement, str, str, float)
+    def onclick(self, elem, loc, totals, frac):
         elem_id = unicode(elem.attribute('id')) or None
         tag = unicode(elem.tagName()).lower()
-        parent = elem
-        loc = []
-        while unicode(parent.tagName()).lower() != 'body':
-            num = 0
-            sibling = parent.previousSibling()
-            while not sibling.isNull():
-                num += 1
-                sibling = sibling.previousSibling()
-            loc.insert(0, num)
-            parent = parent.parent()
-        self.elem_clicked.emit(tag, frac, elem_id, tuple(loc))
+        self.elem_clicked.emit(tag, frac, elem_id, json.loads(str(loc)), json.loads(str(totals)))
 
     def load_js(self):
         if self.js is None:
@@ -69,7 +60,7 @@ class Page(QWebPage):  # {{{
 
 class WebView(QWebView):  # {{{
 
-    elem_clicked = pyqtSignal(object, object, object, object)
+    elem_clicked = pyqtSignal(object, object, object, object, object)
 
     def __init__(self, parent):
         QWebView.__init__(self, parent)
@@ -294,8 +285,8 @@ class ItemEdit(QWidget):
             loctext =  _('Approximately %d%% from the top')%frac
         return loctext
 
-    def elem_clicked(self, tag, frac, elem_id, loc):
-        self.current_frag = elem_id or loc
+    def elem_clicked(self, tag, frac, elem_id, loc, totals):
+        self.current_frag = elem_id or (loc, totals)
         base = _('Location: A &lt;%s&gt; tag inside the file')%tag
         loctext = base + ' [%s]'%self.get_loctext(frac)
         self.dest_label.setText(self.base_msg + '<br>' +

src/calibre/gui2/toc/main.py

@@ -743,7 +743,7 @@ class TOCView(QWidget):  # {{{
 
     def update_item(self, item, where, name, frag, title):
         if isinstance(frag, tuple):
-            frag = add_id(self.ebook, name, frag)
+            frag = add_id(self.ebook, name, *frag)
         child = TOC(title, name, frag)
         child.dest_exists = True
         if item is None: