ToC Editor: Allow generating the ToC directly from individual files inside the ebook. Useful for EPUBs that have individual chapters in single files. Fixes #1163520 (Request for new method to generate entries in ToC editor)

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

@@ -262,6 +262,35 @@ def from_links(container):
             toc.remove(child)
     return toc
 
+def find_text(node):
+    LIMIT = 200
+    pat = re.compile(r'\s+')
+    for child in node:
+        if isinstance(child, etree._Element):
+            text = xml2text(child).strip()
+            text = pat.sub(' ', text)
+            if len(text) < 1:
+                continue
+            if len(text) > LIMIT:
+                # Look for less text in a child of this node, recursively
+                ntext = find_text(child)
+                return ntext or (text[:LIMIT] + '...')
+            else:
+                return text
+
+def from_files(container):
+    toc = TOC()
+    for spinepath in container.spine_items:
+        name = container.abspath_to_name(spinepath)
+        root = container.parsed(name)
+        body = XPath('//h:body')(root)
+        if not body:
+            continue
+        text = find_text(body[0])
+        if text:
+            toc.add(text, name)
+    return toc
+
 def add_id(container, name, loc):
     root = container.parsed(name)
     body = root.xpath('//*[local-name()="body"]')[0]

src/calibre/gui2/toc/main.py

@@ -18,7 +18,7 @@ from PyQt4.Qt import (QPushButton, QFrame, QVariant, QMenu, QInputDialog,
 
 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, from_links)
+    get_toc, add_id, TOC, commit_toc, from_xpaths, from_links, from_files)
 from calibre.gui2 import Application, error_dialog, gprefs
 from calibre.gui2.progress_indicator import ProgressIndicator
 from calibre.gui2.toc.location import ItemEdit
@@ -126,6 +126,7 @@ class ItemView(QFrame): # {{{
     go_to_root = pyqtSignal()
     create_from_xpath = pyqtSignal(object)
     create_from_links = pyqtSignal()
+    create_from_files = pyqtSignal()
     flatten_toc = pyqtSignal()
 
     def __init__(self, parent):
@@ -183,6 +184,15 @@ class ItemView(QFrame): # {{{
         )))
         l.addWidget(b)
 
+        self.cfb = b = QPushButton(_('Generate ToC from &files'))
+        b.clicked.connect(self.create_from_files)
+        b.setToolTip(textwrap.fill(_(
+            'Generate a Table of Contents from individual files in the book.'
+            ' Each entry in the ToC will point to the start of the file, the'
+            ' text of the entry will be the "first line" of text from the file.'
+        )))
+        l.addWidget(b)
+
         self.xpb = b = QPushButton(_('Generate ToC from &XPath'))
         b.clicked.connect(self.create_from_user_xpath)
         b.setToolTip(textwrap.fill(_(
@@ -577,6 +587,7 @@ class TOCView(QWidget): # {{{
         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.create_from_files.connect(self.create_from_files)
         i.flatten_item.connect(self.flatten_item)
         i.flatten_toc.connect(self.flatten_toc)
         i.go_to_root.connect(self.go_to_root)
@@ -778,6 +789,14 @@ class TOCView(QWidget): # {{{
                 _('No links were found that could be added to the Table of Contents.'), show=True)
         self.insert_toc_fragment(toc)
 
+    def create_from_files(self):
+        toc = from_files(self.ebook)
+        if len(toc) == 0:
+            return error_dialog(self, _('No items found'),
+                _('No files were found that could be added to the Table of Contents.'), show=True)
+        self.insert_toc_fragment(toc)
+
+
 # }}}
 
 class TOCEditor(QDialog): # {{{