feat(wordlist): add wordlist feature to kobuddy

This adds a new flag 'wordlist' which allows to retrieve the wordlist of
the connected kobo device.

src/kobuddy/__init__.py

@@ -883,9 +883,28 @@ def _load_highlights(bfile: Path, books: Books):
         yield Highlight(bm, book=book)
 
 
+def _load_wordlist(bfile: Path):
+    logger = get_logger()
+    logger.info(f"Using %s for highlights", bfile)
+    db = dataset_connect_ro(bfile)
+    for bm in db.query('SELECT * FROM WordList'):
+        yield bm['Text']
+
+
 def get_highlights(**kwargs) -> List[Highlight]:
     return list(sorted(_iter_highlights(**kwargs), key=lambda h: h.created))
 
+
+def get_wordlist() -> Iterator[str]:
+    yielded: Set[str] = set()
+
+    for bfile in DATABASES:
+        for h in _load_wordlist(bfile):
+            if h not in yielded:
+                yield h
+                yielded.add(h)
+
+
 # TODO Activity -- sort of interesting (e.g RecentBook). wonder what is Action (it's always 2)
 
 # TODO not sure if need to be exposed
@@ -1027,3 +1046,8 @@ def print_annotations():
 """.strip('\n')
         print(h)
         print("------")
+
+
+def print_wordlist() -> None:
+    for i in get_wordlist():
+        print(i)

src/kobuddy/__main__.py

@@ -3,7 +3,7 @@
 import argparse
 from pathlib import Path
 
-from kobuddy import set_databases, print_progress, print_books, print_annotations, get_logger
+from kobuddy import set_databases, print_progress, print_books, print_annotations, print_wordlist, get_logger
 
 def setup_logger(logger, level=None, format=None, datefmt=None):
     import logging
@@ -42,6 +42,7 @@ def main():
     sp.add_parser('books'      , help='print all books')
     sp.add_parser('progress'   , help='print all book reading progress')
     sp.add_parser('annotations', help='print all annotations (bookmarks/highlights/comments)')
+    sp.add_parser('wordlist'   , help='print all words from the wordlist')
     bp = sp.add_parser('backup', help='backup the database from your Kobo device', description='''
 You can run it via cron, for example every minute. When you connect your device via USB, the database will be backed up.
 
@@ -65,6 +66,8 @@ def main():
             print_books()
         elif args.mode == 'annotations':
             print_annotations()
+        elif args.mode == 'wordlist':
+            print_wordlist()
         else:
             raise RuntimeError(f'Unexpected mode {args.mode}')