compressor selection: add zlib, add zstd,8, descriptions, fixes #145 (#184)

setup.cfg

@@ -83,6 +83,7 @@ deps =
     pytest-mock
     pytest-xdist
     pytest-faulthandler
+    PyQt5==5.11.3
 commands=pytest
 passenv = DISPLAY
 

src/vorta/i18n/__init__.py

@@ -18,11 +18,13 @@
 
 
 class VortaTranslator(QTranslator):
+    """
+    Extends QTranslator to increase the length of strings for testing. Fallback for untranslated
+    strings to English doesn't work currently. So only use for testing.
+    """
     def translate(self, context, text, disambiguation=None, n=-1):
         translated = super().translate(context, text, disambiguation=disambiguation, n=n)
         scale = trans_scale
-        if scale == 100:  # normal, production usage
-            return translated
 
         # for UI layout debugging:
         has_placeholders = '%' in translated
@@ -44,9 +46,14 @@ def translate(self, context, text, disambiguation=None, n=-1):
 
 
 def init_translations(app):
+    """
+    Loads translations for a given input app. If a scaling factor is defined for testing, we use
+    our own subclass of QTranslator.
+    """
     global application, translator, locale  # if we don't keep a reference on these, it stops working. pyqt bug?
     application = app
-    translator = VortaTranslator()
+    translator = QTranslator() if trans_scale == 100 else VortaTranslator()
+
     locale = QLocale(os.environ.get('LANG', None))
     qm_path = os.path.abspath(os.path.join(os.path.dirname(__file__), 'qm'))
     ui_langs = locale.uiLanguages()

src/vorta/models.py

@@ -13,7 +13,7 @@
 from vorta.i18n import trans_late
 from vorta.utils import slugify, uses_dark_mode
 
-SCHEMA_VERSION = 10
+SCHEMA_VERSION = 11
 
 db = pw.Proxy()
 
@@ -315,3 +315,12 @@ def init_db(con):
             migrator.add_column(BackupProfileModel._meta.table_name, 'post_backup_cmd',
                                 pw.CharField(default='')),
         )
+
+    if current_schema.version < 11:
+        _apply_schema_update(current_schema, 11)
+        for profile in BackupProfileModel:
+            if profile.compression == 'zstd':
+                profile.compression = 'zstd,3'
+            if profile.compression == 'lzma,6':
+                profile.compression = 'auto,lzma,6'
+            profile.save()

src/vorta/views/repo_tab.py

@@ -31,9 +31,20 @@ def __init__(self, parent=None):
         self.repoSelector.currentIndexChanged.connect(self.repo_select_action)
         self.repoRemoveToolbutton.clicked.connect(self.repo_unlink_action)
 
-        self.repoCompression.addItem(self.tr('LZ4 (default)'), 'lz4')
-        self.repoCompression.addItem(self.tr('Zstandard (medium)'), 'zstd')
-        self.repoCompression.addItem(self.tr('LZMA (high)'), 'lzma,6')
+        # note: it is hard to describe these algorithms with attributes like low/medium/high
+        # compression or speed on a unified scale. this is not 1-dimensional and also depends
+        # on the input data. so we just tell what we know for sure.
+        # "auto" is used for some slower / older algorithms to avoid wasting a lot of time
+        # on uncompressible data.
+        self.repoCompression.addItem(self.tr('LZ4 (modern, default)'), 'lz4')
+        self.repoCompression.addItem(self.tr('Zstandard Level 3 (modern)'), 'zstd,3')
+        self.repoCompression.addItem(self.tr('Zstandard Level 8 (modern)'), 'zstd,8')
+        # zlib and lzma come from python stdlib and are there (and in borg) since long.
+        # but maybe not much reason to start with these nowadays, considering zstd supports
+        # a very wide range of compression levels and has great speed. if speed is more
+        # important than compression, lz4 is even a little better.
+        self.repoCompression.addItem(self.tr('ZLIB Level 6 (auto, legacy)'), 'auto,zlib,6')
+        self.repoCompression.addItem(self.tr('LZMA Level 6 (auto, legacy)'), 'auto,lzma,6')
         self.repoCompression.addItem(self.tr('No Compression'), 'none')
         self.repoCompression.currentIndexChanged.connect(self.compression_select_action)