From f944271488817729701d936f0f04aa3346b661c7 Mon Sep 17 00:00:00 2001
From: Geoffray Levasseur-Brandin <geoffray.levasseur-brandin-ext@onera.fr>
Date: Thu, 19 Jun 2025 14:32:29 +0200
Subject: [PATCH] added file_stats function

---
 profile.d/filefct.sh | 125 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 125 insertions(+)

diff --git a/profile.d/filefct.sh b/profile.d/filefct.sh
index feaf279..47d8a5d 100644
--- a/profile.d/filefct.sh
+++ b/profile.d/filefct.sh
@@ -211,5 +211,130 @@ rmspc()
 }
 export -f rmspc
 
+
+# ------------------------------------------------------------------------------
+# Rename all files in current directory to replace spaces with _
+# ------------------------------------------------------------------------------
+file_stats()
+{
+    local human=0 details=0 only_avg=0 only_med=0 only_count=0 only_total=0
+    local path="." show_all=1 ext_filter="" ext_list="" min_size="" max_size=""
+    local OPTIND opt
+
+    # Analyse options
+    while [[ "$1" =~ ^- ]]; do
+        case "$1" in
+            -H) human=1 ;;
+            -d) details=1 ;;
+            -m) only_avg=1; show_all=0 ;;
+            -M) only_med=1; show_all=0 ;;
+            -c) only_count=1; show_all=0 ;;
+            -t) only_total=1; show_all=0 ;;
+            -a) human=1; details=1 ;;
+            -x) ext_filter="${2#.}"; shift ;;
+            -X) ext_list="${2}"; shift ;;
+            --min) min_size="$2"; shift ;;
+            --max) max_size="$2"; shift ;;
+            --) shift; break ;;
+            -*) echo "Usage: file_stats [-h] [-d] [-mMctaxX --min N --max N] [path]"; return 1 ;;
+        esac
+        shift
+    done
+
+    [ -n "$1" ] && path="$1"
+
+    # Prepare find filters
+    local find_cmd=(find "$path" -type f)
+
+    # Extension simple
+    if [[ -n "$ext_filter" ]]; then
+        find_cmd+=(-iname "*.$ext_filter")
+    fi
+
+    # Extension liste
+    if [[ -n "$ext_list" ]]; then
+        IFS=',' read -ra exts <<< "$ext_list"
+        find_cmd+=('(')
+        for i in "${!exts[@]}"; do
+            [[ $i -ne 0 ]] && find_cmd+=(-o)
+            find_cmd+=(-iname "*.${exts[$i]}")
+        done
+        find_cmd+=(')')
+    fi
+
+    # Taille min/max (à évaluer en octets)
+    if [[ -n "$min_size" ]]; then
+        find_cmd+=(-size +"$(numfmt --from=iec "$min_size")"c)
+    fi
+    if [[ -n "$max_size" ]]; then
+        find_cmd+=(-size -"$(( $(numfmt --from=iec "$max_size") + 1 ))"c)
+    fi
+
+    # Exécution
+    "${find_cmd[@]}" -printf "%s\n" 2>/dev/null | sort -n | awk -v human="$human" -v details="$details" -v only_avg="$only_avg" -v only_med="$only_med" -v only_count="$only_count" -v only_total="$only_total" -v show_all="$show_all" -v path="$path" '
+    function human_readable(x) {
+        split("B KiB MiB GiB TiB", units)
+        for (i=1; x>=1024 && i<5; i++) x /= 1024
+        return sprintf("%.2f %s", x, units[i])
+    }
+    {
+        sizes[NR] = $1
+        total += $1
+        if (min == "" || $1 < min) min = $1
+        if (max == "" || $1 > max) max = $1
+        if ($1 == 0) bucket[0]++
+        else {
+            b = int(log($1)/log(1024))
+            bucket[b]++
+        }
+    }
+    END {
+        count = NR
+        if (count == 0) {
+            print "Aucun fichier trouvé."; exit
+        }
+
+        moyenne = total / count
+        if (count % 2 == 1)
+            mediane = sizes[(count + 1) / 2]
+        else
+            mediane = (sizes[count / 2] + sizes[count / 2 + 1]) / 2
+
+        function out(label, val) {
+            if (human) val = human_readable(val)
+            printf "%-20s : %s\n", label, val
+        }
+
+        if (only_avg)      out("Taille moyenne", moyenne)
+        else if (only_med) out("Taille médiane", mediane)
+        else if (only_count) printf "Nombre de fichiers     : %d\n", count
+        else if (only_total) out("Taille totale", total)
+        else {
+            if (show_all || human || details) {
+                printf "Statistiques sur \"%s\"\n", path
+                printf "-------------------------\n"
+            }
+            out("Nombre de fichiers", count)
+            out("Taille totale", total)
+            out("Taille moyenne", moyenne)
+            out("Taille médiane", mediane)
+            out("Taille minimale", min)
+            out("Taille maximale", max)
+        }
+
+        if (details) {
+            print "\nHistogramme des tailles :"
+            for (i = 0; i in bucket; i++) {
+                low = 2^i
+                high = 2^(i+1)
+                if (i == 0)
+                    label = sprintf("%4s – %4s", "0", "1K")
+                else
+                    label = sprintf("%4s – %4s", human_readable(low), human_readable(high))
+                printf "%-20s : %5d fichiers\n", label, bucket[i]
+            }
+        }
+    }'
+}
 # ------------------------------------------------------------------------------
 # EOF