Fix #7207 ("Library Check" to also clean up and delete files/folders.)

src/calibre/gui2/dialogs/check_library.py

@@ -3,11 +3,16 @@ __copyright__ = '2008, Kovid Goyal kovid@kovidgoyal.net'
 __docformat__ = 'restructuredtext en'
 __license__   = 'GPL v3'
 
+import os
+
 from PyQt4.Qt import QDialog, QVBoxLayout, QHBoxLayout, QTreeWidget, QLabel, \
             QPushButton, QDialogButtonBox, QApplication, QTreeWidgetItem, \
-            QLineEdit
+            QLineEdit, Qt
 
+from calibre.gui2.dialogs.confirm_delete import confirm
 from calibre.library.check_library import CheckLibrary, CHECKS
+from calibre.library.database2 import delete_file
+from calibre import prints
 
 class Item(QTreeWidgetItem):
     pass
@@ -24,23 +29,28 @@ class CheckLibraryDialog(QDialog):
         self.setLayout(self._layout)
 
         self.log = QTreeWidget(self)
+        self.log.itemChanged.connect(self.item_changed)
         self._layout.addWidget(self.log)
 
-        self.check = QPushButton(_('Run the check'))
+        self.check = QPushButton(_('&Run the check'))
         self.check.setDefault(False)
         self.check.clicked.connect(self.run_the_check)
-        self.copy = QPushButton(_('Copy to clipboard'))
+        self.copy = QPushButton(_('Copy &to clipboard'))
         self.copy.setDefault(False)
         self.copy.clicked.connect(self.copy_to_clipboard)
         self.ok = QPushButton('&Done')
         self.ok.setDefault(True)
         self.ok.clicked.connect(self.accept)
+        self.delete = QPushButton('Delete &marked')
+        self.delete.setDefault(False)
+        self.delete.clicked.connect(self.delete_marked)
         self.cancel = QPushButton('&Cancel')
         self.cancel.setDefault(False)
         self.cancel.clicked.connect(self.reject)
         self.bbox = QDialogButtonBox(self)
         self.bbox.addButton(self.copy, QDialogButtonBox.ActionRole)
         self.bbox.addButton(self.check, QDialogButtonBox.ActionRole)
+        self.bbox.addButton(self.delete, QDialogButtonBox.ActionRole)
         self.bbox.addButton(self.cancel, QDialogButtonBox.RejectRole)
         self.bbox.addButton(self.ok, QDialogButtonBox.AcceptRole)
 
@@ -83,35 +93,66 @@ class CheckLibraryDialog(QDialog):
         plaintext = []
 
         def builder(tree, checker, check):
-            attr = check[0]
+            attr, h, checkable = check
             list = getattr(checker, attr, None)
             if list is None:
                 return
 
-            h = check[1]
             tl = Item([h])
             for problem in list:
                 it = Item()
+                if checkable:
+                    it.setFlags(Qt.ItemIsEnabled | Qt.ItemIsUserCheckable)
+                    it.setCheckState(1, False)
+                else:
+                    it.setFlags(Qt.ItemIsEnabled)
                 it.setText(0, problem[0])
                 it.setText(1, problem[1])
-                p = ', '.join(problem[2])
-                it.setText(2, p)
                 tl.addChild(it)
-                plaintext.append(','.join([h, problem[0], problem[1], p]))
+                self.all_items.append(it)
+                plaintext.append(','.join([h, problem[0], problem[1]]))
             tree.addTopLevelItem(tl)
 
         t = self.log
         t.clear()
-        t.setColumnCount(3);
+        t.setColumnCount(2);
-        t.setHeaderLabels([_('Name'), _('Path from library'), _('Additional Information')])
+        t.setHeaderLabels([_('Name'), _('Path from library')])
+        self.all_items = []
         for check in CHECKS:
             builder(t, checker, check)
 
         t.setColumnWidth(0, 200)
         t.setColumnWidth(1, 400)
-
+        self.delete.setEnabled(False)
         self.text_results = '\n'.join(plaintext)
 
+    def item_changed(self, item, column):
+        for it in self.all_items:
+            if it.checkState(1):
+                self.delete.setEnabled(True)
+                return
+
+    def delete_marked(self):
+        if not confirm('<p>'+_('The marked files and folders will be '
+               '<b>permanently deleted</b>. Are you sure?')
+               +'</p>', 'check_library_editor_delete', self):
+            return
+
+        # Sort the paths in reverse length order so that we can be sure that
+        # if an item is in another item, the sub-item will be deleted first.
+        items = sorted(self.all_items,
+                       key=lambda x: len(x.text(1)),
+                       reverse=True)
+        for it in items:
+            if it.checkState(1):
+                try:
+                    delete_file(os.path.join(self.db.library_path, unicode(it.text(1))))
+                except:
+                    prints('failed to delete',
+                            os.path.join(self.db.library_path,
+                                unicode(it.text(1))))
+        self.run_the_check()
+
     def copy_to_clipboard(self):
         QApplication.clipboard().setText(self.text_results)
 

src/calibre/library/check_library.py

@@ -14,14 +14,14 @@ from calibre.ebooks import BOOK_EXTENSIONS
 EBOOK_EXTENSIONS = frozenset(BOOK_EXTENSIONS)
 NORMALS = frozenset(['metadata.opf', 'cover.jpg'])
 
-CHECKS = [('invalid_titles',    _('Invalid titles')),
+CHECKS = [('invalid_titles',    _('Invalid titles'), True),
-          ('extra_titles',      _('Extra titles')),
+          ('extra_titles',      _('Extra titles'), True),
-          ('invalid_authors',   _('Invalid authors')),
+          ('invalid_authors',   _('Invalid authors'), True),
-          ('extra_authors',     _('Extra authors')),
+          ('extra_authors',     _('Extra authors'), True),
-          ('missing_formats',   _('Missing book formats')),
+          ('missing_formats',   _('Missing book formats'), False),
-          ('extra_formats',     _('Extra book formats')),
+          ('extra_formats',     _('Extra book formats'), True),
-          ('extra_files',       _('Unknown files in books')),
+          ('extra_files',       _('Unknown files in books'), True),
-          ('failed_folders',    _('Folders raising exception'))
+          ('failed_folders',    _('Folders raising exception'), False)
       ]
 
 
@@ -41,7 +41,6 @@ class CheckLibrary(object):
         self.all_lc_dbpaths = frozenset([f.lower() for f in self.all_dbpaths])
 
         self.db_id_regexp = re.compile(r'^.* \((\d+)\)$')
-        self.bad_ext_pat = re.compile(r'[^a-z0-9]+')
 
         self.dirs = []
         self.book_dirs = []
@@ -78,7 +77,7 @@ class CheckLibrary(object):
             auth_path = os.path.join(lib, auth_dir)
             # First check: author must be a directory
             if not os.path.isdir(auth_path):
-                self.invalid_authors.append((auth_dir, auth_dir, []))
+                self.invalid_authors.append((auth_dir, auth_dir))
                 continue
 
             self.potential_authors[auth_dir] = {}
@@ -93,7 +92,7 @@ class CheckLibrary(object):
                 m = self.db_id_regexp.search(title_dir)
                 # Second check: title must have an ID and must be a directory
                 if m is None or not os.path.isdir(title_path):
-                    self.invalid_titles.append((auth_dir, db_path, [title_dir]))
+                    self.invalid_titles.append((auth_dir, db_path))
                     continue
 
                 id = m.group(1)
@@ -101,12 +100,12 @@ class CheckLibrary(object):
                 if self.is_case_sensitive:
                     if int(id) not in self.all_ids or \
                             db_path not in self.all_dbpaths:
-                        self.extra_titles.append((title_dir, db_path, []))
+                        self.extra_titles.append((title_dir, db_path))
                         continue
                 else:
                     if int(id) not in self.all_ids or \
                             db_path.lower() not in self.all_lc_dbpaths:
-                        self.extra_titles.append((title_dir, db_path, []))
+                        self.extra_titles.append((title_dir, db_path))
                         continue
 
                 # Record the book to check its formats
@@ -115,7 +114,7 @@ class CheckLibrary(object):
 
             # Fourth check: author directories that contain no titles
             if not found_titles:
-                self.extra_authors.append((auth_dir, auth_dir, []))
+                self.extra_authors.append((auth_dir, auth_dir))
 
         for x in self.book_dirs:
             try:
@@ -132,9 +131,7 @@ class CheckLibrary(object):
         ext = ext[1:].lower()
         if ext in EBOOK_EXTENSIONS:
             return True
-        if self.bad_ext_pat.search(ext) is not None:
         return False
-        return True
 
     def process_book(self, lib, book_info):
         (db_path, title_dir, book_id) = book_info
@@ -148,18 +145,18 @@ class CheckLibrary(object):
         if self.is_case_sensitive:
             unknowns = frozenset(filenames-formats-NORMALS)
             # Check: any books that aren't formats or normally there?
-            if unknowns:
+            for u in unknowns:
-                self.extra_files.append((title_dir, db_path, unknowns))
+                self.extra_files.append((title_dir, os.path.join(db_path, u)))
 
             # Check: any book formats that should be there?
             missing = book_formats - formats
-            if missing:
+            for m in  missing:
-                self.missing_formats.append((title_dir, db_path, missing))
+                self.missing_formats.append((title_dir, os.path.join(db_path, m)))
 
             # Check: any book formats that shouldn't be there?
             extra = formats - book_formats - NORMALS
-            if extra:
+            for e in extra:
-                self.extra_formats.append((title_dir, db_path, extra))
+                self.extra_formats.append((title_dir, os.path.join(db_path, e)))
         else:
             def lc_map(fnames, fset):
                 m = {}
@@ -171,19 +168,16 @@ class CheckLibrary(object):
             formats_lc = frozenset([f.lower() for f in formats])
             unknowns = frozenset(filenames_lc-formats_lc-NORMALS)
             # Check: any books that aren't formats or normally there?
-            if unknowns:
+            for f in lc_map(filenames, unknowns):
-                self.extra_files.append((title_dir, db_path,
+                self.extra_files.append((title_dir, os.path.join(db_path, f)))
-                                         lc_map(filenames, unknowns)))
 
             book_formats_lc = frozenset([f.lower() for f in book_formats])
             # Check: any book formats that should be there?
             missing = book_formats_lc - formats_lc
-            if missing:
+            for m in lc_map(book_formats, missing):
-                self.missing_formats.append((title_dir, db_path,
+                self.missing_formats.append((title_dir, os.path.join(db_path, m)))
-                                             lc_map(book_formats, missing)))
 
             # Check: any book formats that shouldn't be there?
             extra = formats_lc - book_formats_lc - NORMALS
-            if extra:
+            for e in lc_map(formats, extra):
-                self.extra_formats.append((title_dir, db_path,
+                self.extra_formats.append((title_dir, os.path.join(db_path, e)))
-                                           lc_map(formats, extra)))

src/calibre/library/cli.py

@@ -943,11 +943,11 @@ def command_check_library(args, dbpath):
             return
         if opts.csv:
             for i in list:
-                print check[1] + ',' + i[0] + ',' + i[1] + ',' + '|'.join(i[2])
+                print check[1] + ',' + i[0] + ',' + i[1]
         else:
             print check[1]
             for i in list:
-                print '    %-30.30s - %-30.30s - %s'%(i[0], i[1], ', '.join(i[2]))
+                print '    %-40.40s - %-40.40s'%(i[0], i[1])
 
     db = LibraryDatabase2(dbpath)
     checker = CheckLibrary(dbpath, db)