New feat : Disk usage summary (#318)

* Disk usage summary

* Explain why there is more files than the max specified

* Update advanced.php

* Update advanced.php

* Few improvements

* Improve alerts

* Define HOME

* Use single alert

* Shorten message for single alert

* Add execute permission to scripts/disk_species_count.sh

* Removed MAX_FILE_SPECIES

* Update advanced.php

* Improve message

* Update scripts/advanced.php

Co-authored-by: Nachtzuster <Nachtzuster@users.noreply.github.com>

* Clean

* Update scripts/advanced.php

Co-authored-by: Nachtzuster <Nachtzuster@users.noreply.github.com>

* Typo

---------

Co-authored-by: root <root@db21ed7f-webtop.local.hass.io>
Co-authored-by: Nachtzuster <Nachtzuster@users.noreply.github.com>

scripts/advanced.php

@@ -5,9 +5,18 @@ error_reporting(E_ERROR);
 require_once "scripts/common.php";
 $home = get_home();
 $config = get_config();
+$user = get_user();
 
 ensure_authenticated();
 
+if (isset($_GET['run_species_count'])) {
+   echo "<script>";
+   $output = shell_exec("sudo -u $user ".$home."/BirdNET-Pi/scripts/disk_species_count.sh 2>&1");
+   $escaped_output = htmlspecialchars($output, ENT_QUOTES | ENT_SUBSTITUTE);
+   echo "alert(`$escaped_output`);";
+   echo "</script>";
+ }
+
 if(isset($_GET['submit'])) {
   $contents = file_get_contents('/etc/birdnet/birdnet.conf');
   $restart_livestream = false;
@@ -307,14 +316,15 @@ $newconfig = get_config();
       <label for="purge_threshold">Purge Threshold (Disk Used %):</label>
       <input name="purge_threshold" type="number" style="width:6em;" min="20" max="99" step="1" value="<?php print($newconfig['PURGE_THRESHOLD']);?>"/>
       <p>Defines how full the disk should be before the purge operations occur.<br>Note: This variable is still active if Keep is set. This means that the servies will be stopped at the purge threshold.</p><br>
-      <label for="max_files_species">Amount of files to keep for each species :</label>
+      <label for="max_files_species">Number of files to keep for each species :</label>
       <input name="max_files_species" type="number" style="width:6em;" min="0" step="1" value="<?php print($newconfig['MAX_FILES_SPECIES']);?>"/>
       </td></tr><tr><td>
-      If different than 0 (keep all), defines the maximum number of files to be kept for each species, with priority give to files with highest confidence. 
-      This value does not take into account the last 7 days (protected by default).
+      If different than 0 (keep all), defines the number of files to keep for each species, with priority given to files with higher confidence. This value does not include files from the last 7 days, these new files are protected against auto-deletion.
       </td></tr><tr><td>
       Note only the spectrogram and audio files are deleted, the obsevation data remains in the database.
       The files protected through the "lock" icon are also not affected.
+      <br>
+      <button type="submit" name="run_species_count" value="1" onclick="{this.innerHTML = 'Loading ... please wait.';this.classList.add('disabled')}"><i>[Click here for disk usage summary]</i></button>
       </td></tr></table><br>
       <table class="settingstable"><tr><td>
 

scripts/disk_species_count.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+# Get default values
+source /etc/birdnet/birdnet.conf
+base_dir="$(readlink -f "$HOME/BirdSongs/Extracted/By_Date")"
+cd "$base_dir" || exit 1
+
+# Function to format numbers to k if ≥1000
+format_k() {
+    local value=$1
+    if [ "$value" -ge 1000 ]; then
+        awk -v v="$value" 'BEGIN { printf "%.1fk", v/1000 }'
+    else
+        echo "$value"
+    fi
+}
+
+# Get bird names from the database
+bird_names=$(sqlite3 -readonly "$HOME"/BirdNET-Pi/scripts/birds.db <<EOF
+.headers off
+.mode list
+SELECT DISTINCT Com_Name FROM detections;
+EOF
+)
+
+# Sanitize names for folder matching
+sanitized_names="$(echo "$bird_names" | tr ' ' '_' | tr -d "'" | grep '[[:alnum:]]')"
+sanitized_names=$(echo "$sanitized_names" | sed 's/_*$//')
+
+# Count species
+species_count=$(echo "$sanitized_names" | wc -l)
+total_file_count=0
+
+# Temp files
+data_file=$(mktemp)
+output_file=$(mktemp)
+
+# Loop through each species
+while read -r species; do
+    # Count total files
+    total=$(find */"$species" -type f -name "*[0-9][0-9][0-9][0-9]-[0-9][0-9]-[0-9][0-9]*.*" \
+        -not -iname "*.png" 2>/dev/null | wc -l)
+    total_file_count=$((total_file_count + total))
+
+    # Format total
+    total_display=$(format_k "$total")
+
+    # Clean species name for display
+    species_display=$(echo "$species" | tr '_' ' ')
+
+    # Save padded sort key + display line
+    printf "%05d %s : %s\n" "$total" "$total_display" "$species_display" >> "$data_file"
+done <<<"$sanitized_names"
+
+# Avoid TERM error if not running in a terminal
+[ -t 1 ] && clear
+
+# Build final output
+{
+    echo "BirdSongs stored on your drive"
+    echo " "
+    echo "Location : $base_dir: "
+    echo "Free space    : $(df -h "$base_dir" | awk 'NR==2 {print $4}' | sed 's/G/ GB/; s/M/ MB/; s/K/ KB/')"
+    echo "Total species : $species_count"
+    echo "Total files   : $(format_k "$total_file_count")"
+    echo "Total size    : $(du -sh . | sed 's/G/ GB/; s/M/ MB/; s/K/ KB/' | cut -f1)"
+    echo " "
+    sort -r "$data_file" | sed 's/^[0-9]* //'
+} > "$output_file"
+
+# Show results
+cat "$output_file"
+
+# Clean up
+rm -f "$data_file" "$output_file"