Improve size column readability in archives tab (#1598)

Use one size unit for all archives. The unit is selected by a simple algorithm that picks the largest unit that can represent the smallest size with a given precision. Align all sizes in archive and source tab to the left.

* src/vorta/utils.py : Implement `find_best_size_formatting`. Add missing sizes to `sort_sizes`. Simplify `pretty_bytes` and add `fixed_unit` option for use with `find_best_size_formatting`. Implement `clamp` utility function. Add type hints and docstrings.

* src/vorta/views/archive_tab.py (ArchiveTab.populate_from_profile): Use `find_best_sizes_formatting`.

* src/vorta/views/source_tab.py (SizeItem.__init__): Set alignment to left.

* tests/test_utils.py : Add comprehensive tests for `pretty_bytes` and `find_best_sizes_formatting`.

Co-authored-by: yfprojects <62463991+real-yfprojects@users.noreply.github.com>

src/vorta/utils.py

@@ -2,6 +2,7 @@
 import errno
 import fnmatch
 import getpass
+import math
 import os
 import platform
 import re
@@ -9,7 +10,7 @@
 import unicodedata
 from datetime import datetime as dt
 from functools import reduce
-from typing import Any, Callable, Iterable, Tuple
+from typing import Any, Callable, Iterable, Optional, Tuple, TypeVar
 import psutil
 from paramiko import SSHException
 from paramiko.ecdsakey import ECDSAKey
@@ -226,7 +227,7 @@ def get_private_keys():
 def sort_sizes(size_list):
     """Sorts sizes with extensions. Assumes that size is already in largest unit possible"""
     final_list = []
-    for suffix in [" B", " KB", " MB", " GB", " TB"]:
+    for suffix in [" B", " KB", " MB", " GB", " TB", " PB", " EB", " ZB", " YB"]:
         sub_list = [
             float(size[: -len(suffix)])
             for size in size_list
@@ -240,7 +241,43 @@ def sort_sizes(size_list):
     return final_list
 
 
-def pretty_bytes(size, metric=True, sign=False, precision=1):
+Number = TypeVar("Number", int, float)
+
+
+def clamp(n: Number, min_: Number, max_: Number) -> Number:
+    """Restrict the number n inside a range"""
+    return min(max_, max(n, min_))
+
+
+def find_best_unit_for_sizes(sizes: Iterable[int], metric: bool = True, precision: int = 1) -> int:
+    """
+    Selects the index of the biggest unit (see the lists in the pretty_bytes function) capable of
+    representing the smallest size in the sizes iterable.
+    """
+    min_size = min((s for s in sizes if isinstance(s, int)), default=None)
+    return find_best_unit_for_size(min_size, metric=metric, precision=precision)
+
+
+def find_best_unit_for_size(size: Optional[int], metric: bool = True, precision: int = 1) -> int:
+    """
+    Selects the index of the biggest unit (see the lists in the pretty_bytes function) capable of
+    representing the passed size.
+    """
+    if not isinstance(size, int) or size == 0:  # this will also take care of the None case
+        return 0
+    power = 10**3 if metric else 2**10
+    n = math.floor(math.log(size * 10**precision, power))
+    return n
+
+
+def pretty_bytes(
+    size: int, metric: bool = True, sign: bool = False, precision: int = 1, fixed_unit: Optional[int] = None
+) -> str:
+    """
+    Formats the size with the requested unit and precision. The find_best_size_unit function
+    can be used to find the correct unit for a list of sizes. If no fixed_unit is passed it will
+    find the biggest unit to represent the size
+    """
     if not isinstance(size, int):
         return ''
     prefix = '+' if sign and size > 0 else ''
@@ -249,15 +286,18 @@ def pretty_bytes(size, metric=True, sign=False, precision=1):
         if metric
         else (2**10, ['', 'Ki', 'Mi', 'Gi', 'Ti', 'Pi', 'Ei', 'Zi', 'Yi'])
     )
-    n = 0
-    while abs(round(size, precision)) >= power and n + 1 < len(units):
-        size /= power
-        n += 1
+    if fixed_unit is None:
+        n = find_best_unit_for_size(size, metric=metric, precision=precision)
+    else:
+        n = fixed_unit
+    n = clamp(n, 0, len(units) - 1)
+    size /= power**n
     try:
         unit = units[n]
-        return f'{prefix}{round(size, precision)} {unit}B'
-    except KeyError as e:
-        logger.error(e)
+        digits = f'%.{precision}f' % (round(size, precision))
+        return f'{prefix}{digits} {unit}B'
+    except KeyError as error:
+        logger.error(error)
         return "NaN"
 
 

src/vorta/views/archive_tab.py

@@ -32,7 +32,14 @@
 from vorta.borg.umount import BorgUmountJob
 from vorta.i18n import translate
 from vorta.store.models import ArchiveModel, BackupProfileMixin
-from vorta.utils import choose_file_dialog, format_archive_name, get_asset, get_mount_points, pretty_bytes
+from vorta.utils import (
+    choose_file_dialog,
+    find_best_unit_for_sizes,
+    format_archive_name,
+    get_asset,
+    get_mount_points,
+    pretty_bytes,
+)
 from vorta.views import diff_result, extract_dialog
 from vorta.views.diff_result import DiffResultDialog, DiffTree
 from vorta.views.extract_dialog import ExtractDialog, ExtractTree
@@ -45,6 +52,10 @@
 logger = logging.getLogger(__name__)
 
 
+#: The number of decimal digits to show in the size column
+SIZE_DECIMAL_DIGITS = 1
+
+
 class ArchiveTab(ArchiveTabBase, ArchiveTabUI, BackupProfileMixin):
     prune_intervals = ['hour', 'day', 'week', 'month', 'year']
 
@@ -243,12 +254,15 @@ def populate_from_profile(self):
 
             sorting = self.archiveTable.isSortingEnabled()
             self.archiveTable.setSortingEnabled(False)
+            best_unit = find_best_unit_for_sizes((a.size for a in archives), precision=SIZE_DECIMAL_DIGITS)
             for row, archive in enumerate(archives):
                 self.archiveTable.insertRow(row)
 
                 formatted_time = archive.time.strftime('%Y-%m-%d %H:%M')
                 self.archiveTable.setItem(row, 0, QTableWidgetItem(formatted_time))
-                self.archiveTable.setItem(row, 1, SizeItem(pretty_bytes(archive.size)))
+                self.archiveTable.setItem(
+                    row, 1, SizeItem(pretty_bytes(archive.size, fixed_unit=best_unit, precision=SIZE_DECIMAL_DIGITS))
+                )
                 if archive.duration is not None:
                     formatted_duration = str(timedelta(seconds=round(archive.duration)))
                 else:

src/vorta/views/source_tab.py

@@ -21,6 +21,10 @@ class SourceColumn:
 
 
 class SizeItem(QTableWidgetItem):
+    def __init__(self, s):
+        super().__init__(s)
+        self.setTextAlignment(Qt.AlignVCenter + Qt.AlignRight)
+
     def __lt__(self, other):
         if other.text() == '':
             return False

tests/test_utils.py

@@ -1,5 +1,6 @@
 import uuid
 from vorta.keyring.abc import VortaKeyring
+from vorta.utils import find_best_unit_for_sizes, pretty_bytes
 
 
 def test_keyring():
@@ -9,3 +10,72 @@ def test_keyring():
     keyring = VortaKeyring.get_keyring()
     keyring.set_password('vorta-repo', REPO, UNICODE_PW)
     assert keyring.get_password("vorta-repo", REPO) == UNICODE_PW
+
+
+def test_best_size_unit_precision0():
+    MB = 1000000
+    sizes = [int(0.1 * MB), 100 * MB, 2000 * MB]
+    unit = find_best_unit_for_sizes(sizes, metric=True, precision=0)
+    assert unit == 1  # KB, min=100KB
+
+
+def test_best_size_unit_precision1():
+    MB = 1000000
+    sizes = [int(0.1 * MB), 100 * MB, 2000 * MB]
+    unit = find_best_unit_for_sizes(sizes, metric=True, precision=1)
+    assert unit == 2  # MB, min=0.1MB
+
+
+def test_best_size_unit_empty():
+    sizes = []
+    unit = find_best_unit_for_sizes(sizes, metric=True, precision=1)
+    assert unit == 0  # bytes
+
+
+def test_best_size_unit_precision3():
+    MB = 1000000
+    sizes = [1 * MB, 100 * MB, 2000 * MB]
+    unit = find_best_unit_for_sizes(sizes, metric=True, precision=3)
+    assert unit == 3  # GB, min=0.001 GB
+
+
+def test_best_size_unit_nonmetric1():
+    sizes = [102]
+    unit = find_best_unit_for_sizes(sizes, metric=False, precision=1)
+    assert unit == 0  # 102 < 0.1KB
+
+
+def test_best_size_unit_nonmetric2():
+    sizes = [103]
+    unit = find_best_unit_for_sizes(sizes, metric=False, precision=1)
+    assert unit == 1  # 103bytes == 0.1KB
+
+
+def test_pretty_bytes_metric_fixed1():
+    s = pretty_bytes(1000000, metric=True, precision=0, fixed_unit=2)
+    assert s == "1 MB"
+
+
+def test_pretty_bytes_metric_fixed2():
+    s = pretty_bytes(1000000, metric=True, precision=1, fixed_unit=2)
+    assert s == "1.0 MB"
+
+
+def test_pretty_bytes_metric_fixed3():
+    s = pretty_bytes(100000, metric=True, precision=1, fixed_unit=2)
+    assert s == "0.1 MB"
+
+
+def test_pretty_bytes_nonmetric_fixed1():
+    s = pretty_bytes(1024 * 1024, metric=False, precision=1, fixed_unit=2)
+    assert s == "1.0 MiB"
+
+
+def test_pretty_bytes_metric_nonfixed2():
+    s = pretty_bytes(1000000, metric=True, precision=1)
+    assert s == "1.0 MB"
+
+
+def test_pretty_bytes_metric_large():
+    s = pretty_bytes(10**30, metric=True, precision=1)
+    assert s == "1000000.0 YB"