Instantiate Scanner on-the-fly

Previously, it would be instantiated on startup.

This will make our job easier for an upcoming SE/ME/PE merge.

core/app.py

@@ -23,8 +23,9 @@ from hscommon.trans import tr
 from hscommon.plat import ISWINDOWS
 from hscommon import desktop
 
-from . import directories, results, scanner, export, fs
+from . import directories, results, export, fs
 from .ignore import IgnoreList
+from .scanner import ScanType, Scanner
 from .gui.deletion_options import DeletionOptions
 from .gui.details_panel import DetailsPanel
 from .gui.directory_tree import DirectoryTree
@@ -113,7 +114,7 @@ class DupeGuru(Broadcaster):
     """Holds everything together.
 
     Instantiated once per running application, it holds a reference to every high-level object
-    whose reference needs to be held: :class:`~core.results.Results`, :class:`Scanner`,
+    whose reference needs to be held: :class:`~core.results.Results`,
     :class:`~core.directories.Directories`, :mod:`core.gui` instances, etc..
 
     It also hosts high level methods and acts as a coordinator for all those elements. This is why
@@ -154,7 +155,7 @@ class DupeGuru(Broadcaster):
     # select_dest_file(prompt: str, ext: str) --> str
 
     PROMPT_NAME = "dupeGuru"
-    SCANNER_CLASS = scanner.Scanner
+    SCANNER_CLASS = Scanner
 
     def __init__(self, view):
         if view.get_default(DEBUG_MODE_PREFERENCE):
@@ -165,10 +166,13 @@ class DupeGuru(Broadcaster):
         self.appdata = desktop.special_folder_path(desktop.SpecialFolder.AppData, appname=self.NAME)
         if not op.exists(self.appdata):
             os.makedirs(self.appdata)
+        self.discarded_file_count = 0
         self.directories = directories.Directories()
         self.results = results.Results(self)
         self.ignore_list = IgnoreList()
-        self.scanner = self.SCANNER_CLASS()
+        # In addition to "app-level" options, this dictionary also holds options that will be
+        # sent to the scanner. They don't have default values because those defaults values are
+        # defined in the scanner class.
         self.options = {
             'escape_filter_regexp': True,
             'clean_empty_dirs': False,
@@ -360,7 +364,7 @@ class DupeGuru(Broadcaster):
             self.view.show_message(tr("'{}' does not exist.").format(d))
 
     def add_selected_to_ignore_list(self):
-        """Adds :attr:`selected_dupes` to :attr:`scanner`'s ignore list.
+        """Adds :attr:`selected_dupes` to :attr:`ignore_list`.
         """
         dupes = self.without_ref(self.selected_dupes)
         if not dupes:
@@ -731,22 +735,29 @@ class DupeGuru(Broadcaster):
 
         Scans folders selected in :attr:`directories` and put the results in :attr:`results`
         """
+        scanner = self.SCANNER_CLASS()
+        if not self.directories.has_any_file():
+            self.view.show_message(tr("The selected directories contain no scannable file."))
+            return
+        # Send relevant options down to the scanner instance
+        for k, v in self.options.items():
+            if hasattr(scanner, k):
+                setattr(scanner, k, v)
+        self.results.groups = []
+        self._results_changed()
+
         def do(j):
             j.set_progress(0, tr("Collecting files to scan"))
-            if self.scanner.scan_type == scanner.ScanType.Folders:
+            if scanner.scan_type == ScanType.Folders:
                 files = list(self.directories.get_folders(j))
             else:
                 files = list(self.directories.get_files(j))
             if self.options['ignore_hardlink_matches']:
                 files = self._remove_hardlink_dupes(files)
             logging.info('Scanning %d files' % len(files))
-            self.results.groups = self.scanner.get_dupe_groups(files, self.ignore_list, j)
+            self.results.groups = scanner.get_dupe_groups(files, self.ignore_list, j)
+            self.discarded_file_count = scanner.discarded_file_count
 
-        if not self.directories.has_any_file():
-            self.view.show_message(tr("The selected directories contain no scannable file."))
-            return
-        self.results.groups = []
-        self._results_changed()
         self._start_job(JobType.Scan, do)
 
     def toggle_selected_mark_state(self):
@@ -783,7 +794,7 @@ class DupeGuru(Broadcaster):
     @property
     def stat_line(self):
         result = self.results.stat_line
-        if self.scanner.discarded_file_count:
+        if self.discarded_file_count:
-            result = tr("%s (%d discarded)") % (result, self.scanner.discarded_file_count)
+            result = tr("%s (%d discarded)") % (result, self.discarded_file_count)
         return result
 

core/scanner.py

@@ -124,7 +124,8 @@ class Scanner:
             return True
         return len(dupe.path) > len(ref.path)
 
-    def get_scan_options(self):
+    @staticmethod
+    def get_scan_options():
         """Returns a list of scanning options for this scanner.
 
         Returns a list of ``ScanOption``.

core/tests/app_test.py

@@ -105,7 +105,7 @@ class TestCaseDupeGuru:
         os.link(str(tmppath['myfile']), str(tmppath['hardlink']))
         app = TestApp().app
         app.directories.add_path(tmppath)
-        app.scanner.scan_type = ScanType.Contents
+        app.options['scan_type'] = ScanType.Contents
         app.options['ignore_hardlink_matches'] = True
         app.start_scanning()
         eq_(len(app.results.groups), 0)

core_me/scanner.py

@@ -13,7 +13,8 @@ class ScannerME(ScannerBase):
     def _key_func(dupe):
         return (-dupe.bitrate, -dupe.size)
 
-    def get_scan_options(self):
+    @staticmethod
+    def get_scan_options():
         return [
             ScanOption(ScanType.Filename, tr("Filename")),
             ScanOption(ScanType.Fields, tr("Filename - Fields")),

core_pe/scanner.py

@@ -16,7 +16,8 @@ class ScannerPE(Scanner):
     match_scaled = False
     threshold = 75
 
-    def get_scan_options(self):
+    @staticmethod
+    def get_scan_options():
         return [
             ScanOption(ScanType.FuzzyBlock, tr("Contents")),
             ScanOption(ScanType.ExifTimestamp, tr("EXIF Timestamp")),

core_se/scanner.py

@@ -9,7 +9,8 @@ from hscommon.trans import tr
 from core.scanner import Scanner as ScannerBase, ScanOption, ScanType
 
 class ScannerSE(ScannerBase):
-    def get_scan_options(self):
+    @staticmethod
+    def get_scan_options():
         return [
             ScanOption(ScanType.Filename, tr("Filename")),
             ScanOption(ScanType.Contents, tr("Contents")),

qt/base/app.py

@@ -89,7 +89,7 @@ class DupeGuru(QObject):
         createActions(ACTIONS, self)
 
     def _update_options(self):
-        self.model.scanner.mix_file_kind = self.prefs.mix_file_kind
+        self.model.options['mix_file_kind'] = self.prefs.mix_file_kind
         self.model.options['escape_filter_regexp'] = not self.prefs.use_regexp
         self.model.options['clean_empty_dirs'] = self.prefs.remove_empty_folders
         self.model.options['ignore_hardlink_matches'] = self.prefs.ignore_hardlink_matches

qt/base/directories_dialog.py

@@ -28,7 +28,7 @@ class DirectoriesDialog(QMainWindow):
         self.lastAddedFolder = platform.INITIAL_FOLDER_IN_DIALOGS
         self.recentFolders = Recent(self.app, 'recentFolders')
         self._setupUi()
-        SCAN_TYPE_ORDER = [so.scan_type for so in self.app.model.scanner.get_scan_options()]
+        SCAN_TYPE_ORDER = [so.scan_type for so in self.app.model.SCANNER_CLASS.get_scan_options()]
         scan_type_index = SCAN_TYPE_ORDER.index(self.app.prefs.scan_type)
         self.scanTypeComboBox.setCurrentIndex(scan_type_index)
         self.directoriesModel = DirectoriesModel(self.app.model.directory_tree, view=self.treeView)
@@ -130,7 +130,7 @@ class DirectoriesDialog(QMainWindow):
         self.scanTypeComboBox = QComboBox(self)
         self.scanTypeComboBox.setSizePolicy(QSizePolicy(QSizePolicy.Expanding, QSizePolicy.Fixed))
         self.scanTypeComboBox.setMaximumWidth(400)
-        for scan_option in self.app.model.scanner.get_scan_options():
+        for scan_option in self.app.model.SCANNER_CLASS.get_scan_options():
             self.scanTypeComboBox.addItem(scan_option.label)
         hl.addWidget(self.scanTypeComboBox)
         self.showPreferencesButton = QPushButton(tr("Options"), self.centralwidget)
@@ -240,7 +240,7 @@ class DirectoriesDialog(QMainWindow):
         self.app.model.start_scanning()
 
     def scanTypeChanged(self, index):
-        scan_options = self.app.model.scanner.get_scan_options()
+        scan_options = self.app.model.SCANNER_CLASS.get_scan_options()
         self.app.prefs.scan_type = scan_options[index].scan_type
         self.app._update_options()
 

qt/me/app.py

@@ -1,6 +1,4 @@
-# Created By: Virgil Dupras
+# Copyright 2016 Hardcoded Software (http://www.hardcoded.net)
-# Created On: 2009-05-21
-# Copyright 2015 Hardcoded Software (http://www.hardcoded.net)
 #
 # This software is licensed under the "GPLv3" License as described in the "LICENSE" file,
 # which should be included with this package. The terms are also available at
@@ -28,10 +26,10 @@ class DupeGuru(DupeGuruBase):
 
     def _update_options(self):
         DupeGuruBase._update_options(self)
-        self.model.scanner.min_match_percentage = self.prefs.filter_hardness
+        self.model.options['min_match_percentage'] = self.prefs.filter_hardness
-        self.model.scanner.scan_type = self.prefs.scan_type
+        self.model.options['scan_type'] = self.prefs.scan_type
-        self.model.scanner.word_weighting = self.prefs.word_weighting
+        self.model.options['word_weighting'] = self.prefs.word_weighting
-        self.model.scanner.match_similar_words = self.prefs.match_similar
+        self.model.options['match_similar_words'] = self.prefs.match_similar
         scanned_tags = set()
         if self.prefs.scan_tag_track:
             scanned_tags.add('track')
@@ -45,5 +43,5 @@ class DupeGuru(DupeGuruBase):
             scanned_tags.add('genre')
         if self.prefs.scan_tag_year:
             scanned_tags.add('year')
-        self.model.scanner.scanned_tags = scanned_tags
+        self.model.options['scanned_tags'] = scanned_tags
 

qt/pe/app.py

@@ -1,6 +1,4 @@
-# Created By: Virgil Dupras
+# Copyright 2016 Hardcoded Software (http://www.hardcoded.net)
-# Created On: 2009-04-25
-# Copyright 2015 Hardcoded Software (http://www.hardcoded.net)
 #
 # This software is licensed under the "GPLv3" License as described in the "LICENSE" file,
 # which should be included with this package. The terms are also available at
@@ -86,7 +84,7 @@ class DupeGuru(DupeGuruBase):
 
     def _update_options(self):
         DupeGuruBase._update_options(self)
-        self.model.scanner.scan_type = self.prefs.scan_type
+        self.model.options['scan_type'] = self.prefs.scan_type
-        self.model.scanner.match_scaled = self.prefs.match_scaled
+        self.model.options['match_scaled'] = self.prefs.match_scaled
-        self.model.scanner.threshold = self.prefs.filter_hardness
+        self.model.options['threshold'] = self.prefs.filter_hardness
 

qt/se/app.py

@@ -1,6 +1,4 @@
-# Created By: Virgil Dupras
+# Copyright 2016 Hardcoded Software (http://www.hardcoded.net)
-# Created On: 2009-05-24
-# Copyright 2015 Hardcoded Software (http://www.hardcoded.net)
 #
 # This software is licensed under the "GPLv3" License as described in the "LICENSE" file,
 # which should be included with this package. The terms are also available at
@@ -43,10 +41,10 @@ class DupeGuru(DupeGuruBase):
 
     def _update_options(self):
         DupeGuruBase._update_options(self)
-        self.model.scanner.min_match_percentage = self.prefs.filter_hardness
+        self.model.options['min_match_percentage'] = self.prefs.filter_hardness
-        self.model.scanner.scan_type = self.prefs.scan_type
+        self.model.options['scan_type'] = self.prefs.scan_type
-        self.model.scanner.word_weighting = self.prefs.word_weighting
+        self.model.options['word_weighting'] = self.prefs.word_weighting
-        self.model.scanner.match_similar_words = self.prefs.match_similar
+        self.model.options['match_similar_words'] = self.prefs.match_similar
         threshold = self.prefs.small_file_threshold if self.prefs.ignore_small_files else 0
-        self.model.scanner.size_threshold = threshold * 1024 # threshold is in KB. the scanner wants bytes
+        self.model.options['size_threshold'] = threshold * 1024 # threshold is in KB. the scanner wants bytes