Keep the progress bar estimate when the last file count is exceeded

When a backup has more files than the last one, estimate the total from the files rsync has found so far,
capped at 99% until the scan ends, instead of falling back to "(scanning)".

man/simple_backup.1

@@ -117,8 +117,9 @@ Don't use systemd journal for logging.
 Don't show a progress bar while rsync copies the files. The progress bar is only shown when
 the output is a terminal, so it never appears in the systemd journal or in redirected output.
 While rsync is still looking for files, the percentage and estimated time (marked with
-\(aq~\(aq) are based on the number of files in the last backup, or the bar shows
-\(aq(scanning)\(aq if that is unknown or already exceeded.
+\(aq~\(aq) are based on the number of files in the last backup (or on the files found so far,
+once this backup has more), and stay below 100%. Without a previous count the bar shows
+\(aq(scanning)\(aq instead.
 .TP
 .B \-\-rsync\-options OPTIONS [OPTION...]
 By default, the following rsync options are used:

simple_backup/simple_backup.py

@@ -135,13 +135,16 @@ class _ProgressBar:
         percent: int | None
 
         if self.total is None:
-            # Estimate from the last backup, unless this one already has more files
-            if 0 < self._checked < self._estimate:
-                percent = self._checked * 100 // self._estimate
-                text = f'{f"~{percent}":>3}%  {self._checked:,}/~{self._estimate:,} files  {size} copied'
+            # Estimate from the last backup; past it, from the files found so far, as the scan is then nearly done
+            estimate = max(self._estimate, self._total)
+
+            if self._estimate > 0 and self._checked > 0:
+                # The real total is still unknown, so never claim 100%
+                percent = min(self._checked * 100 // estimate, 99)
+                text = f'{f"~{percent}":>3}%  {self._checked:,}/~{estimate:,} files  {size} copied'
 
                 if percent > 0:
-                    text += f'  ETA ~{self._eta(self._estimate)}'
+                    text += f'  ETA ~{self._eta(estimate)}'
             else:
                 text = f'{self._checked:,} files checked  {size} copied  (scanning)'
                 percent = None

tests/test_local.py

@@ -303,12 +303,31 @@ class TestProgressBar:
         assert capsys.readouterr().err == \
             '\r[#######-----------------------]  ~25%  1,000/~4,000 files  2.00M copied  ETA ~0:01:30\033[K'
 
-    def test_estimate_exceeded(self, capsys: pytest.CaptureFixture[str]) -> None:
+    def test_estimate_exceeded(self, monkeypatch: pytest.MonkeyPatch, capsys: pytest.CaptureFixture[str]) -> None:
+        monkeypatch.setattr(sb, 'get_terminal_size', lambda: os.terminal_size((100, 24)))
         bar = sb._ProgressBar(estimate=1000)
 
-        # More files than in the last backup: back to plain scanning
+        # More files than in the last backup: estimate from the files found so far
         assert bar.update('          2.00M  25%    1.47MB/s    0:00:02 (xfr#1, ir-chk=500/2000)')
-        assert capsys.readouterr().err == '\r1,500 files checked  2.00M copied  (scanning)\033[K'
+        assert capsys.readouterr().err == \
+            '\r[######################--------]  ~75%  1,500/~2,000 files  2.00M copied  ETA ~0:00:10\033[K'
+
+    def test_estimate_capped_while_scanning(self, monkeypatch: pytest.MonkeyPatch,
+                                            capsys: pytest.CaptureFixture[str]) -> None:
+        monkeypatch.setattr(sb, 'get_terminal_size', lambda: os.terminal_size((100, 24)))
+        bar = sb._ProgressBar(estimate=1000)
+
+        # All files found so far are checked, but rsync may still find more
+        assert bar.update('          2.00M  25%    1.47MB/s    0:00:02 (xfr#1, ir-chk=0/2000)')
+        assert bar.total is None
+        assert '  ~99%  2,000/~2,000 files  2.00M copied  ETA ~0:00:00\033[K' in capsys.readouterr().err
+
+    def test_no_estimate_without_last_count(self, capsys: pytest.CaptureFixture[str]) -> None:
+        bar = sb._ProgressBar()
+
+        # The files found so far are only a guess at the total near the end of the scan
+        assert bar.update('          2.00M  25%    1.47MB/s    0:00:02 (xfr#1, ir-chk=0/2000)')
+        assert capsys.readouterr().err == '\r2,000 files checked  2.00M copied  (scanning)\033[K'
 
     def test_real_total_replaces_estimate(self, capsys: pytest.CaptureFixture[str]) -> None:
         bar = sb._ProgressBar(estimate=4000)