Improve fb2 parsing

Miscellaneous fixes to navigation

TODO

@@ -3,6 +3,7 @@ TODO
         ✓ Internationalization
         ✓ Application icon
         ✓ .desktop file
+        ✓ Shift to logging instead of print statements
         Flatpak and AppImage support
     Options:
         ✓ Automatic library management
@@ -93,9 +94,13 @@ TODO
         Last line in QTextBrowser should never be cut off
         Does image alignment need to be centered?
         Bookmark name for a page that's not on the TOC or has nothing before
+        Screen position still keeps jumping when inside a paragraph
+        Better recursion needed for fb2 toc
 
     Secondary:
-        The cover page needs to be marked separately
+        Special formatting for each chapter's title
+        Create covers for books without them
+        Signal end of chapter with some text
         Graphical themes
         Change focus rectangle dimensions
         Universal Ctrl + Tab
@@ -103,7 +108,6 @@ TODO
         Goodreads API: Ratings, Read, Recommendations
         Get ISBN using python-isbnlib
         Use embedded fonts + CSS
-        Shift to logging instead of print statements
         txt, doc, chm support
         Include icons for filetype emblems
         Comic view modes

lector/__main__.py

@@ -724,7 +724,7 @@ class MainUI(QtWidgets.QMainWindow, mainwindow.Ui_MainWindow):
         # The set_content method is universal
         # It's going to do position tracking
         current_tab = self.tabWidget.currentWidget()
-        current_tab.set_content(required_position)
+        current_tab.set_content(required_position, False, True)
 
     def library_doubleclick(self, index):
         sender = self.sender().objectName()

lector/contentwidgets.py

@@ -737,7 +737,7 @@ class PliantWidgetsCommonFunctions:
                 return direction
 
         current_tab.set_content(
-            current_position + direction + get_modifier(), True)
+            current_position + direction + get_modifier(), True, True)
 
         # Set page position depending on if the chapter number is increasing or decreasing
         if direction == 1 or was_button_pressed:
@@ -853,7 +853,7 @@ class PliantWidgetsCommonFunctions:
         def set_toc_position(tocTree):
             currentIndex = tocTree.currentIndex()
             required_position = currentIndex.data(QtCore.Qt.UserRole)
-            self.pw.parent.set_content(required_position, True)
+            self.pw.parent.set_content(required_position, True, True)
 
         # Create the Combobox / Treeview combination
         tocComboBox = QtWidgets.QComboBox()

lector/dockwidgets.py

@@ -220,7 +220,7 @@ class Bookmarks:
         if is_parent:
             chapter_number = self.parent.bookmarkProxyModel.data(
                 index, QtCore.Qt.UserRole)
-            self.parentTab.set_content(chapter_number, True)
+            self.parentTab.set_content(chapter_number, True, True)
             return
 
         chapter = self.parent.bookmarkProxyModel.data(
@@ -228,7 +228,7 @@ class Bookmarks:
         cursor_position = self.parent.bookmarkProxyModel.data(
             index, QtCore.Qt.UserRole + 1)
 
-        self.parentTab.set_content(chapter, True)
+        self.parentTab.set_content(chapter, True, True)
         if not self.parentTab.are_we_doing_images_only:
             self.parentTab.set_cursor_position(cursor_position)
 
@@ -500,7 +500,7 @@ class Search:
         cursor_position = self.parent.searchResultsModel.data(index, QtCore.Qt.UserRole + 2)
         search_term = self.parent.searchResultsModel.data(index, QtCore.Qt.UserRole + 4)
 
-        self.parentTab.set_content(chapter_number, True)
+        self.parentTab.set_content(chapter_number, True, True)
         if not self.parentTab.are_we_doing_images_only:
             self.parentTab.set_cursor_position(
                 cursor_position, len(search_term))

lector/parsers/fb2.py

@@ -64,8 +64,8 @@ class ParseFB2:
         toc = []
         content = []
         for count, i in enumerate(self.book['book_list']):
-            toc.append((1, i[0], count + 1))
+            toc.append((i[0], i[1], count + 1))
-            content.append(i[1])
+            content.append(i[2])
 
         # Return toc, content, images_only
         return toc, content, False

lector/readers/read_fb2.py

@@ -92,18 +92,50 @@ class FB2:
             self.book['cover'] = None
 
     def parse_chapters(self, temp_dir):
-        for i in self.xml.find_all('section'):
+        # TODO
-            this_title = '<No title>'
+        # Check what's up with recursion levels
-            for j in i:
+        # Why is the TypeError happening in get_title
-                if j.name == 'title':
-                    this_title = j.getText(separator=' ')
-                    this_title = this_title.replace('\n', '').strip()
-                    # This comes later because the tag is changed in place
-                    title_xml = j.unwrap()
-                    break
 
+        def get_title(element):
+            this_title = '<No title>'
+            title_xml = '<No title xml>'
+            try:
+                for i in element:
+                    if i.name == 'title':
+                        this_title = i.getText(separator=' ')
+                        this_title = this_title.replace('\n', '').strip()
+                        title_xml = str(i.unwrap())
+                        break
+            except TypeError:
+                return None, None
+            return this_title, title_xml
+
+        def recursor(level, element):
+            children = element.findChildren('section', recursive=False)
+            if not children and level != 1:
+                this_title, title_xml = get_title(element)
                 self.book['book_list'].append(
-                [this_title, str(title_xml) + str(i)])
+                    [level, this_title, title_xml + str(element)])
+            else:
+                for i in children:
+                    recursor(level + 1, i)
+
+        first_element = self.xml.find('section')  # Recursive find
+        siblings = list(first_element.findNextSiblings('section', recursive=False))
+        siblings.insert(0, first_element)
+
+        for this_element in siblings:
+            this_title, title_xml = get_title(this_element)
+            # Do not add chapter content in case it has sections
+            # inside it. This prevents having large Book sections that
+            # have duplicated content
+            section_children = this_element.findChildren('section')
+            chapter_text = str(this_element)
+            if section_children:
+                chapter_text = this_title
+
+            self.book['book_list'].append([1, this_title, chapter_text])
+            recursor(1, this_element)
 
         # Extract all images to the temp_dir
         for i in self.xml.find_all('binary'):
@@ -113,7 +145,7 @@ class FB2:
             replacement_string = f'<p></p><img src=\"{image_path}\"'
 
             for j in self.book['book_list']:
-                j[1] = j[1].replace(
+                j[2] = j[2].replace(
                     image_string, replacement_string)
             try:
                 image_data = base64.decodebytes(i.text.encode())
@@ -128,4 +160,4 @@ class FB2:
             with open(cover_path, 'wb') as outimage:
                 outimage.write(self.book['cover'])
             self.book['book_list'].insert(
-                0, ('Cover', f'<center><img src="{cover_path}" alt="Cover"></center>'))
+                0, (1, 'Cover', f'<center><img src="{cover_path}" alt="Cover"></center>'))

lector/widgets.py

@@ -109,7 +109,7 @@ class Tab(QtWidgets.QWidget):
             self.hiddenButton.clicked.connect(self.set_cursor_position)
 
         # All content must be set through this function
-        self.set_content(current_chapter, True)
+        self.set_content(current_chapter, True, False)
         if not self.are_we_doing_images_only:
             # Setting this later breaks cursor positioning for search results
             self.hiddenButton.animateClick(50)
@@ -249,7 +249,7 @@ class Tab(QtWidgets.QWidget):
 
         # Finally, to make sure the cover image isn't
         # scrolled halfway through on first open,
-        if self.main_window.bookToolBar.tocBox.currentIndex() == 0:
+        if self.metadata['position']['current_chapter'] == 1:
             self.contentView.verticalScrollBar().setValue(0)
 
     def generate_position(self, is_read=False):
@@ -266,10 +266,8 @@ class Tab(QtWidgets.QWidget):
 
         if not self.are_we_doing_images_only:
             for i in self.metadata['content']:
-                chapter_html = i[1]
-
                 textDocument = QtGui.QTextDocument(None)
-                textDocument.setHtml(chapter_html)
+                textDocument.setHtml(i)
                 block_count = textDocument.blockCount()
 
                 blocks_per_chapter.append(block_count)
@@ -418,7 +416,7 @@ class Tab(QtWidgets.QWidget):
         self.mouse_hide_timer.start(2000)
         self.contentView.setFocus()
 
-    def set_content(self, required_position, tocBox_readjust=False):
+    def set_content(self, required_position, tocBox_readjust=False, record_position=False):
         # All content changes must come through here
         # This function will decide how to relate
         # entries in the toc to the actual content
@@ -433,7 +431,8 @@ class Tab(QtWidgets.QWidget):
 
         # Update the metadata dictionary to save position
         self.metadata['position']['current_chapter'] = required_position
-        self.metadata['position']['is_read'] = False
+        if record_position:
+            self.contentView.record_position()
 
         if self.are_we_doing_images_only:
             self.contentView.loadImage(required_content)