get rid of azw3 offset, thanks to jhowell

README.md

@@ -14,8 +14,6 @@ Add the plugin to "The context menu for the books in the calibre library" in "Pr
 
 If your Kindle device is connected, it will send the file to your device.
 
-For AZW3 books you may have to adjust the offset manually. For example, if you find the word is 16 characters(include white space) behind where it should be, set -16 in the plugin settings.
-
 ## License
 
 This work is licensed under GPL version 3 or later.

__init__.py

@@ -7,7 +7,7 @@ class WordDumbDumb(InterfaceActionBase):
     description = 'Create Kindle Word Wise file.'
     supported_platforms = ['linux', 'osx', 'windows']
     author = 'xxyzz'
-    version = (1, 2, 0)
+    version = (1, 3, 0)
     minimum_calibre_version = (5, 0, 0) # Python3
     actual_plugin = 'calibre_plugins.worddumb.ui:WordDumb'
 
@@ -19,4 +19,4 @@ def config_widget(self):
         return ConfigWidget()
 
     def save_settings(self, config_widget):
-        config_widget.save_settings()
+        pass

config.py

@@ -1,29 +1,17 @@
 #!/usr/bin/env python3
 
 from calibre.utils.config import JSONConfig
-from PyQt5.Qt import QWidget, QHBoxLayout, QLabel, QLineEdit, QPushButton, QVBoxLayout
+from PyQt5.Qt import QWidget, QPushButton, QVBoxLayout
 import webbrowser
 
 prefs = JSONConfig('plugins/worddumb')
 
-prefs.defaults['offset'] = '0'
-
 class ConfigWidget(QWidget):
     def __init__(self):
         QWidget.__init__(self)
 
         self.vl = QVBoxLayout()
         self.setLayout(self.vl)
-        self.hl = QHBoxLayout()
-        self.vl.addLayout(self.hl)
-
-        self.label = QLabel('Offset:')
-        self.hl.addWidget(self.label)
-
-        self.offset = QLineEdit(self)
-        self.offset.setText(prefs['offset'])
-        self.hl.addWidget(self.offset)
-        self.label.setBuddy(self.offset)
 
         self.donate_button = QPushButton('Donate', self)
         self.donate_button.clicked.connect(self.donate)
@@ -33,9 +21,6 @@ def __init__(self):
         self.github_button.clicked.connect(self.github)
         self.vl.addWidget(self.github_button)
 
-    def save_settings(self):
-        prefs['offset'] = self.offset.text()
-
     def donate(self):
         webbrowser.open('https://liberapay.com/xxyzz/donate')
 

parse_job.py

@@ -1,11 +1,10 @@
 #!/usr/bin/env python3
 
 from calibre.ebooks.mobi.reader.mobi6 import MobiReader
+from calibre.ebooks.mobi.reader.mobi8 import Mobi8Reader
 from calibre.utils.logging import default_log
 from calibre_plugins.worddumb.database import connect_ww_database, \
     create_lang_layer, match_word
-from calibre_plugins.worddumb.config import prefs
-from pathlib import Path
 import re
 
 def do_job(gui, books, abort, log, notifications):
@@ -33,6 +32,12 @@ def parse_book(pathtoebook, book_fmt):
         offset = mobiReader.kf8_boundary + 2
     mobiReader.extract_text(offset=offset)
     html = mobiReader.mobi_html
+    if book_fmt.lower() == 'azw3':
+        m8r = Mobi8Reader(mobiReader, default_log)
+        m8r.kf8_sections = mobiReader.sections[offset-1:]
+        m8r.read_indices()
+        m8r.build_parts()
+        html = b''.join(m8r.parts)
 
     # match text between HTML tags
     for match_text in re.finditer(b">[^<>]+<", html):
@@ -41,7 +46,4 @@ def parse_book(pathtoebook, book_fmt):
         for match_word in re.finditer(b"[a-zA-Z]+", text):
             lemma = text[match_word.start():match_word.end()]
             start = match_text.start() + match_word.start()
-            if book_fmt.lower() == 'azw3':
-                start -= 14 # this value works for some books of mine
-                start += int(prefs['offset'])
             yield (start, lemma.decode('utf-8'))