From 7a895f5717fada3ebcf7ac5ca2a81ec5bd51276f Mon Sep 17 00:00:00 2001
From: Josef Haupt <josef.haupt@informatik.tu-chemnitz.de>
Date: Mon, 6 Jul 2026 12:03:18 +0200
Subject: [PATCH] Faster look up for input selection in multi file tab

---
 birdnet_analyzer/gui/multi_file.py | 17 +++++++++++++----
 birdnet_analyzer/gui/utils.py      | 12 ++++++++++++
 birdnet_analyzer/utils.py          | 25 +++++++++++++++++++++++++
 3 files changed, 50 insertions(+), 4 deletions(-)

diff --git a/birdnet_analyzer/gui/multi_file.py b/birdnet_analyzer/gui/multi_file.py
index 6029d2bb9..d5ce5d270 100644
--- a/birdnet_analyzer/gui/multi_file.py
+++ b/birdnet_analyzer/gui/multi_file.py
@@ -143,19 +143,28 @@ def build_multi_analysis_tab() -> gu.TAB_BUILDER_RESULT:
             ],
         )
 
+        preview_limit = 100
+
         def select_directory_on_empty():
             folder = gu.select_folder(state_key="batch-analysis-data-dir")
 
             if folder:
-                files_and_durations = gu.get_audio_files_and_durations(folder)
-                if len(files_and_durations) > 100:
+                # Only load durations for the first files shown in the preview.
+                # Fetch one extra to detect whether more files exist without
+                # walking (and probing durations for) the whole directory.
+                files_and_durations = gu.get_audio_files_and_durations(
+                    folder, max_files=preview_limit + 1
+                )
+                if len(files_and_durations) > preview_limit:
+                    # Count the remaining files with a fast walk that skips durations.
+                    total = gu.count_audio_files(folder)
                     return [
                         folder,
                         folder,
                         [
-                            *files_and_durations[:100],
+                            *files_and_durations[:preview_limit],
                             (
-                                f"{len(files_and_durations) - 100} "
+                                f"{total - preview_limit} "
                                 f"{loc.localize('multi-tab-more-files-label')}",
                                 "...",
                             ),
diff --git a/birdnet_analyzer/gui/utils.py b/birdnet_analyzer/gui/utils.py
index e7422c26f..b678177d1 100644
--- a/birdnet_analyzer/gui/utils.py
+++ b/birdnet_analyzer/gui/utils.py
@@ -150,6 +150,18 @@ def get_audio_files_and_durations(folder, max_files=None):
     return files_and_durations
 
 
+def count_audio_files(folder):
+    """Counts the audio files in a folder without collecting their paths or durations.
+
+    Args:
+        folder (str): The path to the folder containing audio files.
+
+    Returns:
+        int: The number of audio files in the folder (recursively).
+    """
+    return utils.count_audio_files(folder)
+
+
 def set_window(window):
     """
     Sets the global _WINDOW variable to the provided window object.
diff --git a/birdnet_analyzer/utils.py b/birdnet_analyzer/utils.py
index 880cf47a4..042f9829f 100644
--- a/birdnet_analyzer/utils.py
+++ b/birdnet_analyzer/utils.py
@@ -145,6 +145,31 @@ def collect_audio_files(path: str, max_files: int | None = None):
     return sorted(files)
 
 
+def count_audio_files(path: str) -> int:
+    """Counts all audio files in the given directory.
+
+    Faster than ``collect_audio_files`` when only the number of files is needed,
+    as it neither stores nor sorts the paths.
+
+    Args:
+        path: The directory to be searched.
+
+    Returns:
+        The number of audio files in the directory (recursively).
+    """
+    count = 0
+
+    for _, _, flist in os.walk(path):
+        for f in flist:
+            if (
+                not f.startswith(".")
+                and f.rsplit(".", 1)[-1].lower() in ALLOWED_FILETYPES
+            ):
+                count += 1
+
+    return count
+
+
 def read_lines(
     path: str | Path | None, trim: bool = False, fail_on_blank_lines: bool = False
 ) -> list[str]: