From dee0950f74376ae89d19cdf5220578335386faa8 Mon Sep 17 00:00:00 2001
From: ed <s@ocv.me>
Date: Sat, 6 Jan 2024 01:15:43 +0000
Subject: [PATCH] misc; * scripts: add log repacker * bench/filehash: msys
 support + add more stats

---
 scripts/bench/filehash.sh | 29 ++++++++++++----
 scripts/logpack.sh        | 73 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 96 insertions(+), 6 deletions(-)
 create mode 100755 scripts/logpack.sh

diff --git a/scripts/bench/filehash.sh b/scripts/bench/filehash.sh
index f729c071..5a5f4c98 100755
--- a/scripts/bench/filehash.sh
+++ b/scripts/bench/filehash.sh
@@ -12,6 +12,11 @@ set -euo pipefail
 #
 # can be adjusted with --hash-mt (but alpine caps out at 5)
 
+fsize=256
+nfiles=128
+pybin=$(command -v python3 || command -v python)
+#pybin=~/.pyenv/versions/nogil-3.9.10-2/bin/python3
+
 [ $# -ge 1 ] || {
 	echo 'need arg 1: path to copyparty-sfx.py'
 	echo ' (remaining args will be passed on to copyparty,'
@@ -22,6 +27,8 @@ sfx="$1"
 shift
 sfx="$(realpath "$sfx" || readlink -e "$sfx" || echo "$sfx")"
 awk=$(command -v gawk || command -v awk)
+uname -s | grep -E MSYS && win=1 || win=
+totalsize=$((fsize*nfiles))
 
 # try to use /dev/shm to avoid hitting filesystems at all,
 # otherwise fallback to mktemp which probably uses /tmp
@@ -30,20 +37,24 @@ mkdir $td || td=$(mktemp -d)
 trap "rm -rf $td" INT TERM EXIT
 cd $td
 
-echo creating 256 MiB testfile in $td
-head -c $((1024*1024*256)) /dev/urandom > 1
+echo creating $fsize MiB testfile in $td
+sz=$((1024*1024*fsize))
+head -c $sz /dev/zero | openssl enc -aes-256-ctr -iter 1 -pass pass:k -nosalt 2>/dev/null >1 || true
+wc -c 1 | awk '$1=='$sz'{r=1}END{exit 1-r}' || head -c $sz /dev/urandom >1
 
-echo creating 127 symlinks to it
-for n in $(seq 2 128); do ln -s 1 $n; done
+echo creating $((nfiles-1)) symlinks to it
+for n in $(seq 2 $nfiles); do MSYS=winsymlinks:nativestrict ln -s 1 $n; done
 
 echo warming up cache
 cat 1 >/dev/null
 
 echo ok lets go
-python3 "$sfx" -p39204 -e2dsa --dbd=yolo --exit=idx -lo=t -q "$@"
+$pybin "$sfx" -p39204 -e2dsa --dbd=yolo --exit=idx -lo=t -q "$@" && err= || err=$?
+[ $win ] && [ $err = 15 ] && err=  # sigterm doesn't hook on windows, ah whatever
+[ $err ] && echo ERROR $err && exit $err
 
 echo and the results are...
-LC_ALL=C $awk '/1 volumes in / {s=$(NF-1); printf "speed: %.1f MiB/s  (time=%.2fs)\n", 256*128/s, s}' <t
+LC_ALL=C $awk '/1 volumes in / {s=$(NF-1); printf "speed: %.1f MiB/s  (time=%.2fs)\n", '$totalsize'/s, s}' <t
 
 echo deleting $td and exiting
 
@@ -52,6 +63,8 @@ echo deleting $td and exiting
 
 # MiB/s @ cpu or device  (copyparty, pythonver, distro/os)  // comment
 
+#  3887 @ Ryzen 5 4500U  (cpp 1.9.5, nogil 3.9, fedora 39)  // --hash-mt=6; laptop
+#  3732 @ Ryzen 5 4500U  (cpp 1.9.5, py 3.12.1, fedora 39)  // --hash-mt=6; laptop
 #  3608 @ Ryzen 5 4500U  (cpp 1.9.5, py 3.11.5, fedora 38)  // --hash-mt=6; laptop
 #  2726 @ Ryzen 5 4500U  (cpp 1.9.5, py 3.11.5, fedora 38)  // --hash-mt=4 (old-default)
 #  2202 @ Ryzen 5 4500U  (cpp 1.9.5, py 3.11.5, docker-alpine 3.18.3) ??? alpine slow
@@ -62,6 +75,10 @@ echo deleting $td and exiting
 #  5544 @ Intel i5-12500 (cpp 1.9.5, py 3.11.2, debian 12.0)  // --hash-mt=12; desktop
 #  5197 @ Ryzen 7 3700X  (cpp 1.9.5, py 3.9.18, freebsd 13.2)  // --hash-mt=8; 2u server
 #  4551 @ mbp 2020 m1    (cpp 1.9.5, py 3.11.7, macos 14.2.1)
+#  4190 @ Ryzen 7 5800X  (cpp 1.9.5, py 3.11.6, fedora 37)  // --hash-mt=8 (vbox-VM on win10-17763.4974)
+#  3028 @ Ryzen 7 5800X  (cpp 1.9.5, py 3.11.6, fedora 37)  // --hash-mt=5 (vbox-VM on win10-17763.4974)
+#  2629 @ Ryzen 7 5800X  (cpp 1.9.5, py 3.11.7, win10-ltsc-1809-17763.4974)  // --hash-mt=5 (default)
+#  2576 @ Ryzen 7 5800X  (cpp 1.9.5, py 3.11.7, win10-ltsc-1809-17763.4974)  // --hash-mt=8 (hello??)
 #  2606 @ Ryzen 7 3700X  (cpp 1.9.5, py 3.9.18, freebsd 13.2)  // --hash-mt=4 (old-default)
 #  1436 @ Ryzen 5 5500U  (cpp 1.9.5, py 3.11.4, alpine 3.18.3)  // nuc
 #  1065 @ Pixel 7        (cpp 1.9.5, py 3.11.5, termux 2023-09)
diff --git a/scripts/logpack.sh b/scripts/logpack.sh
new file mode 100755
index 00000000..b122e9b1
--- /dev/null
+++ b/scripts/logpack.sh
@@ -0,0 +1,73 @@
+#!/bin/bash
+set -e
+
+# recompress logs so they decompress faster + save some space;
+# * will not recurse into subfolders
+# * each file in current folder gets recompressed to zstd; input file is DELETED
+# * any xz-compressed logfiles are decompressed before converting to zstd
+# * SHOULD ignore and skip files which are currently open; SHOULD be safe to run while copyparty is running
+
+
+# for files larger than $cutoff, compress with `zstd -T0`
+# (otherwise do several files in parallel (scales better))
+cutoff=400M
+
+
+# osx support:
+# port install findutils gsed coreutils
+command -v gfind >/dev/null &&
+command -v gsed  >/dev/null &&
+command -v gsort >/dev/null && {
+    find() { gfind "$@"; }
+    sed()  { gsed  "$@"; }
+    sort() { gsort "$@"; }
+}
+
+packfun() {
+    local jobs=$1 fn="$2"
+    printf '%s\n' "$fn" | grep -qF .zst && return
+
+    local of="$(printf '%s\n' "$fn" | sed -r 's/\.(xz|txt)/.zst/')"
+    [ "$fn" = "$of" ] &&
+        of="$of.zst"
+
+    [ -e "$of" ] &&
+        echo "SKIP: output file exists: $of" &&
+        return
+
+    lsof -- "$fn" 2>/dev/null | grep -E .. &&
+        printf "SKIP: file in use: %s\n\n" $fn &&
+        return
+
+    # determine by header; old copyparty versions would produce xz output without .xz names
+    head -c3 "$fn" | grep -qF 7z &&
+        cmd="xz -dkc" || cmd="cat"
+
+    printf '<%s> T%d: %s\n' "$cmd" $jobs "$of"
+
+    $cmd <"$fn" >/dev/null || {
+        echo "ERROR: uncompress failed: $fn"
+        return
+    }
+
+    $cmd <"$fn" | zstd --long -19 -T$jobs >"$of"
+    touch -r "$fn" -- "$of"
+
+    cmp <($cmd <"$fn") <(zstd -d <"$of") || {
+        echo "ERROR: data mismatch: $of"
+        mv "$of"{,.BAD}
+        return
+    }
+    rm -- "$fn"
+}
+
+# do small files in parallel first (in descending size);
+# each file can use 4 threads in case the cutoff is poor
+export -f packfun
+export -f sed 2>/dev/null || true
+find -maxdepth 1 -type f -size -$cutoff -printf '%s %p\n' |
+sort -nr | sed -r 's`[^ ]+ ``; s`^\./``' | tr '\n' '\0' |
+xargs "$@" -0i -P$(nproc) bash -c 'packfun 4 "$@"' _ {}
+
+# then the big ones, letting each file use the whole cpu
+for f in *; do packfun 0 "$f"; done