feat(safeclean): add regenerable cache categories + manual-review report for large dirs

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

safeclean.sh

@@ -8,16 +8,27 @@
 # What it targets (all regenerable / safe to lose):
 #   core dumps, Python caches, virtualenvs, node_modules, build targets,
 #   ~/.m2, gradle/pip/npm/yarn/go caches, trash, thumbnails, Docker cruft,
-#   and (with --system) apt cache, journald logs, old snap revisions.
+#   browser & IDE caches (only under ~/.cache — never profiles), tool
+#   download caches (playwright/puppeteer/electron), ML model caches
+#   (huggingface/torch — models re-download on demand!), cargo registry
+#   cache, unused rustup toolchains, Flatpak app caches, PlatformIO's
+#   .cache — and, only with the explicit --platformio flag, PlatformIO
+#   packages/platforms (regenerable via pio, but slow to re-fetch).
+#   With --system: apt cache, journald logs, rotated logs, old snap
+#   revisions, systemd coredumps.
 #
-# What it NEVER touches: your documents, media, and especially cloud-sync
+# What it NEVER touches: your documents, media, ~/projects,
-# folders (Dropbox / sync / CosmoseGDrive) — those are pruned from every scan.
+# ~/homelab-preload, and especially cloud-sync folders (Dropbox / sync /
+# CosmoseGDrive) — those are pruned from every scan. Big mixed dirs
+# (~/.local/share, ~/.config, ~/.cache as a whole) are only REPORTED
+# for manual review, never deleted.
 #
 # Usage:
 #   ./safeclean.sh                 # report only (safe, default)
 #   ./safeclean.sh --apply         # delete, asking before each category
 #   ./safeclean.sh --apply -y      # delete without per-category prompts
 #   ./safeclean.sh --apply --system   # also clean system stuff (uses sudo)
+#   ./safeclean.sh --platformio    # also offer PlatformIO packages/platforms
 #   ./safeclean.sh --root /path    # scan a different tree (default: $HOME)
 #   ./safeclean.sh --days 30       # only items untouched for 30+ days (default 7)
 #   ./safeclean.sh --no-age        # ignore age, consider everything eligible
@@ -33,6 +44,8 @@ ASSUME_YES=0
 DO_SYSTEM=0
 VERBOSE=0
 AGGRESSIVE_DOCKER=0    # set via --docker-all to drop ALL unused images
+CLEAN_PLATFORMIO=0     # set via --platformio to also offer ~/.platformio
+                       # packages/platforms (regenerable via pio, but slow)
 AGE_DAYS=7             # only touch items NOT modified within this many days;
                        # set to 0 (or --no-age) to ignore age entirely
 
@@ -48,6 +61,8 @@ PRUNE_DIRS=(
     "$HOME/snap"            # per-app snap data
     "$HOME/.steam"
     "$HOME/Steam"
+    "$HOME/projects"        # real project data — hands off
+    "$HOME/homelab-preload" # preloaded data, expensive to re-fetch
 )
 
 # ------------------------------- arg parsing --------------------------------
@@ -58,12 +73,15 @@ while [ $# -gt 0 ]; do
         -y|--yes)       ASSUME_YES=1 ;;
         --system)       DO_SYSTEM=1 ;;
         --docker-all)   AGGRESSIVE_DOCKER=1 ;;
+        --platformio)   CLEAN_PLATFORMIO=1 ;;
         --days)         shift; AGE_DAYS="${1:?--days needs a number}" ;;
         --no-age)       AGE_DAYS=0 ;;
         -v|--verbose)   VERBOSE=1 ;;
         --root)         shift; SCAN_ROOT="${1:?--root needs a path}" ;;
         -h|--help)
-            sed -n '2,33p' "$0" | sed 's/^# \{0,1\}//'
+            # print the header comment block (everything up to the first
+            # non-comment line), so help never drifts out of sync
+            awk 'NR > 1 && !/^#/ { exit } NR > 1 { sub(/^# ?/, ""); print }' "$0"
             exit 0 ;;
         *) echo "unknown option: $1 (try --help)" >&2; exit 2 ;;
     esac
@@ -249,6 +267,123 @@ scan_user_caches() {
     report_and_clean "Dev & app caches"
 }
 
+# Browser & IDE caches: only the ~/.cache subtrees — profiles (~/.mozilla,
+# ~/.config/google-chrome, ~/.config/JetBrains) are never touched.
+scan_browser_ide_caches() {
+    reset_targets
+    local d
+    for d in \
+        "$HOME/.cache/JetBrains" \
+        "$HOME/.cache/google-chrome" \
+        "$HOME/.cache/chromium" \
+        "$HOME/.cache/mozilla"
+    do
+        [ -e "$d" ] && TARGETS+=("$d")
+    done
+    report_and_clean "Browser & IDE caches"
+}
+
+# Downloaded browser/runtime bundles; re-fetched on demand by the tool.
+scan_tool_download_caches() {
+    reset_targets
+    local d
+    for d in \
+        "$HOME/.cache/ms-playwright" \
+        "$HOME/.cache/puppeteer" \
+        "$HOME/.cache/electron" \
+        "$HOME/.cache/electron-builder"
+    do
+        [ -e "$d" ] && TARGETS+=("$d")
+    done
+    report_and_clean "Tool download caches"
+}
+
+# ML model caches: fully regenerable, but re-downloading can mean many GB.
+scan_ml_caches() {
+    reset_targets
+    local d
+    for d in "$HOME/.cache/huggingface" "$HOME/.cache/torch"; do
+        [ -e "$d" ] && TARGETS+=("$d")
+    done
+    if [ ${#TARGETS[@]} -gt 0 ]; then
+        printf '  %sNote: ML model caches — models re-download on demand (can be many GB).%s\n' \
+            "$c_yel" "$c_rst"
+    fi
+    report_and_clean "ML model caches"
+}
+
+# Downloaded .crate archives; cargo re-fetches them on demand.
+# (pip wheels live under ~/.cache/pip, already covered by "Dev & app caches".)
+scan_cargo_registry_cache() {
+    reset_targets
+    [ -e "$HOME/.cargo/registry/cache" ] && TARGETS+=("$HOME/.cargo/registry/cache")
+    report_and_clean "Cargo registry cache"
+}
+
+# Old rustup toolchains: regenerable via `rustup toolchain install`, but the
+# default toolchain must survive. If we can't tell which one that is, we skip
+# the whole category rather than guess.
+scan_rustup_toolchains() {
+    reset_targets
+    local tdir="$HOME/.rustup/toolchains" def="" d
+    if [ -d "$tdir" ]; then
+        if command -v rustup >/dev/null 2>&1; then
+            def=$(rustup show active-toolchain 2>/dev/null | awk 'NR==1{print $1}')
+        fi
+        if [ -z "$def" ] && [ -f "$HOME/.rustup/settings.toml" ]; then
+            def=$(awk -F'"' '/^default_toolchain/{print $2}' "$HOME/.rustup/settings.toml")
+        fi
+        if [ -z "$def" ]; then
+            printf '  %-26s %scannot determine default toolchain — skipped%s\n' \
+                "Rust toolchains (unused)" "$c_dim" "$c_rst"
+            return
+        fi
+        for d in "$tdir"/*/; do
+            d="${d%/}"
+            [ -d "$d" ] || continue
+            [ "$(basename "$d")" = "$def" ] && continue
+            TARGETS+=("$d")
+        done
+    fi
+    report_and_clean "Rust toolchains (unused)"
+}
+
+scan_flatpak_caches() {
+    reset_targets
+    local d
+    for d in "$HOME"/.var/app/*/cache; do
+        [ -d "$d" ] && TARGETS+=("$d")
+    done
+    report_and_clean "Flatpak app caches"
+}
+
+# PlatformIO packages/platforms: regenerable (pio re-downloads on the next
+# build) but slow to re-fetch, so only offered behind the explicit
+# --platformio flag. ~/.platformio/.cache is always covered ("Dev & app
+# caches" above).
+scan_platformio_packages() {
+    local pkgs="$HOME/.platformio/packages" plats="$HOME/.platformio/platforms"
+    if [ "$CLEAN_PLATFORMIO" = 0 ]; then
+        if [ -d "$pkgs" ] || [ -d "$plats" ]; then
+            local sz
+            sz=$(du -sch "$pkgs" "$plats" 2>/dev/null | tail -n1 | awk '{print $1}')
+            printf '  %-26s %s%s — skipped (opt in with --platformio)%s\n' \
+                "PlatformIO pkgs/platforms" "$c_dim" "${sz:-?}" "$c_rst"
+        fi
+        return
+    fi
+    reset_targets
+    local d
+    for d in "$pkgs" "$plats"; do
+        [ -e "$d" ] && TARGETS+=("$d")
+    done
+    if [ ${#TARGETS[@]} -gt 0 ]; then
+        printf '  %sWARNING: pio will re-download packages/platforms on the next build (slow).%s\n' \
+            "$c_yel" "$c_rst"
+    fi
+    report_and_clean "PlatformIO pkgs/platforms"
+}
+
 # Go's caches need their own tool: module cache files are read-only, so
 # `go clean` is the correct way to drop them (with a chmod+rm fallback).
 clean_go_cache() {
@@ -328,9 +463,10 @@ clean_system() {
     echo
     echo "${c_bold}System (sudo) ─────────────────────────────────────────────${c_rst}"
     if [ "$DRY_RUN" = 1 ]; then
-        echo "  ${c_dim}dry-run: would offer apt clean, journald vacuum, rotated log cleanup, old snap removal${c_rst}"
+        echo "  ${c_dim}dry-run: would offer apt clean, journald vacuum, rotated log cleanup, old snap removal, systemd coredump cleanup${c_rst}"
         echo "  apt archive cache:"; sudo -n du -sh /var/cache/apt/archives 2>/dev/null | sed 's/^/      /' || true
         echo "  journal logs:";       journalctl --disk-usage 2>/dev/null | sed 's/^/      /' || true
+        echo "  systemd coredumps:";  sudo -n du -sh /var/lib/systemd/coredump 2>/dev/null | sed 's/^/      /' || true
         echo "  rotated logs (/var/log/*.gz, *.1, *.2 …):"
         local rot_total
         rot_total=$(find /var/log -maxdepth 1 -type f \
@@ -367,6 +503,21 @@ clean_system() {
         fi
     fi
 
+    if [ -d /var/lib/systemd/coredump ]; then
+        local cd_size
+        cd_size=$(sudo du -sh /var/lib/systemd/coredump 2>/dev/null | awk '{print $1}')
+        if confirm "systemd coredumps: delete dumps older than ${AGE_DAYS} day(s) (dir total: ${cd_size:-?})?"; then
+            if [ "$AGE_DAYS" -gt 0 ]; then
+                sudo find /var/lib/systemd/coredump -type f -mtime "+$AGE_DAYS" -delete
+            else
+                sudo find /var/lib/systemd/coredump -type f -delete
+            fi
+            printf '      %sremoved.%s\n' "$c_grn" "$c_rst"
+        else
+            echo "      skipped."
+        fi
+    fi
+
     if confirm "snap: remove disabled (old) revisions?"; then
         snap list --all 2>/dev/null | awk '/disabled/{print $1, $3}' | \
         while read -r name rev; do
@@ -375,6 +526,23 @@ clean_system() {
     fi
 }
 
+# Report-only: the biggest subdirectories of trees that MIX regenerable junk
+# with real data (app state, settings, histories). This function never
+# deletes anything and never feeds TARGETS — the owner reviews by hand.
+report_large_dirs() {
+    echo
+    echo "${c_bold}Large dirs for MANUAL review (NEVER deleted by this script) ─${c_rst}"
+    echo "  ${c_dim}Top 15 subdirectories of ~/.local/share, ~/.config and ~/.cache by size:${c_rst}"
+    local r
+    {
+        for r in "$HOME/.local/share" "$HOME/.config" "$HOME/.cache"; do
+            # per-subdir sizes; drop du's trailing total line for the root itself
+            [ -d "$r" ] && du -h --max-depth=1 "$r" 2>/dev/null | sed '$d'
+        done
+    } | sort -rh | head -n 15 | sed 's/^/  /'
+    echo "  ${c_dim}Review by hand: du -sh <path> — these may hold real data, so decide yourself.${c_rst}"
+}
+
 # --------------------------------- run --------------------------------------
 
 echo "${c_bold}safeclean${c_rst} — root: $SCAN_ROOT"
@@ -407,11 +575,20 @@ scan_virtualenvs
 scan_node_modules
 scan_build_targets
 scan_user_caches
+scan_browser_ide_caches
+scan_tool_download_caches
+scan_ml_caches
+scan_cargo_registry_cache
+scan_rustup_toolchains
+scan_flatpak_caches
+scan_platformio_packages
 clean_go_cache
 clean_docker
 
 [ "$DO_SYSTEM" = 1 ] && clean_system
 
+report_large_dirs
+
 echo
 if [ "$DRY_RUN" = 1 ]; then
     echo "Done.  (this was a dry-run — add --apply to delete)"