From 1f93448791e78c6465228a9338db518a2ced8cac Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Wed, 20 Mar 2013 17:00:41 +0530
Subject: [PATCH] ToC Editor: Allow creating ToC from links in the book

---
 src/calibre/ebooks/oeb/polish/toc.py | 33 ++++++++++++++++++++++++++++
 src/calibre/gui2/toc/main.py         | 21 ++++++++++++++++--
 2 files changed, 52 insertions(+), 2 deletions(-)

diff --git a/src/calibre/ebooks/oeb/polish/toc.py b/src/calibre/ebooks/oeb/polish/toc.py
index abac6bd7d1..3a72b837c8 100644
--- a/src/calibre/ebooks/oeb/polish/toc.py
+++ b/src/calibre/ebooks/oeb/polish/toc.py
@@ -39,6 +39,10 @@ class TOC(object):
         c.parent = self
         return c
 
+    def remove(self, child):
+        self.children.remove(child)
+        child.parent = None
+
     def __iter__(self):
         for c in self.children:
             yield c
@@ -191,6 +195,8 @@ def elem_to_toc_text(elem):
         text = elem.get('alt', '')
     text = re.sub(r'\s+', ' ', text.strip())
     text = text[:1000].strip()
+    if not text:
+        text = _('(Untitled)')
     return text
 
 def from_xpaths(container, xpaths):
@@ -229,6 +235,33 @@ def from_xpaths(container, xpaths):
 
     return tocroot
 
+def from_links(container):
+    toc = TOC()
+    link_path = XPath('//h:a[@href]')
+    seen_titles, seen_dests = set(), set()
+    for spinepath in container.spine_items:
+        name = container.abspath_to_name(spinepath)
+        root = container.parsed(name)
+        for a in link_path(root):
+            href = a.get('href')
+            if not href or not href.strip():
+                continue
+            dest = container.href_to_name(href, base=name)
+            frag = href.rpartition('#')[-1] or None
+            if (dest, frag) in seen_dests:
+                continue
+            seen_dests.add((dest, frag))
+            text = elem_to_toc_text(a)
+            if text in seen_titles:
+                continue
+            seen_titles.add(text)
+            toc.add(text, dest, frag=frag)
+    verify_toc_destinations(container, toc)
+    for child in toc:
+        if not child.dest_exists:
+            toc.remove(child)
+    return toc
+
 def add_id(container, name, loc):
     root = container.parsed(name)
     body = root.xpath('//*[local-name()="body"]')[0]
diff --git a/src/calibre/gui2/toc/main.py b/src/calibre/gui2/toc/main.py
index d0e2a8e0f6..a9c160a0d4 100644
--- a/src/calibre/gui2/toc/main.py
+++ b/src/calibre/gui2/toc/main.py
@@ -18,7 +18,7 @@ from PyQt4.Qt import (QPushButton, QFrame, QVariant,
 
 from calibre.ebooks.oeb.polish.container import get_container, AZW3Container
 from calibre.ebooks.oeb.polish.toc import (
-    get_toc, add_id, TOC, commit_toc, from_xpaths)
+    get_toc, add_id, TOC, commit_toc, from_xpaths, from_links)
 from calibre.gui2 import Application, error_dialog, gprefs
 from calibre.gui2.progress_indicator import ProgressIndicator
 from calibre.gui2.toc.location import ItemEdit
@@ -33,6 +33,7 @@ class ItemView(QFrame): # {{{
     flatten_item = pyqtSignal()
     go_to_root = pyqtSignal()
     create_from_xpath = pyqtSignal(object)
+    create_from_links = pyqtSignal()
 
     def __init__(self, parent):
         QFrame.__init__(self, parent)
@@ -79,7 +80,15 @@ class ItemView(QFrame): # {{{
             ' heading tags. Uses the <h1-6> tags.')))
         l.addWidget(b)
 
-
+        self.lb = b = QPushButton(_('Generate ToC from &links'))
+        b.clicked.connect(self.create_from_links)
+        b.setToolTip(textwrap.fill(_(
+            'Generate a Table of Contents from all the links in the book.'
+            ' Links that point to destinations that do not exist in the book are'
+            ' ignored. Also multiple links with the same destination or the same'
+            ' text are ignored.'
+        )))
+        l.addWidget(b)
 
         l.addStretch()
         self.w1 = la = QLabel(_('<b>WARNING:</b> calibre only supports the '
@@ -270,6 +279,7 @@ class TOCView(QWidget): # {{{
         self.item_view.delete_item.connect(self.delete_current_item)
         i.add_new_item.connect(self.add_new_item)
         i.create_from_xpath.connect(self.create_from_xpath)
+        i.create_from_links.connect(self.create_from_links)
         i.flatten_item.connect(self.flatten_item)
         i.go_to_root.connect(self.go_to_root)
         l.addWidget(i, 0, 4, col, 1)
@@ -490,6 +500,13 @@ class TOCView(QWidget): # {{{
                 _('No items were found that could be added to the Table of Contents.'), show=True)
         self.insert_toc_fragment(toc)
 
+    def create_from_links(self):
+        toc = from_links(self.ebook)
+        if len(toc) == 0:
+            return error_dialog(self, _('No items found'),
+                _('No links were found that could be added to the Table of Contents.'), show=True)
+        self.insert_toc_fragment(toc)
+
 # }}}
 
 class TOCEditor(QDialog): # {{{