Try implemente benchmarks

.gitignore

@@ -3,3 +3,5 @@
 target
 tarpaulin-report.html
 cobertura.xml
+svejk.txt
+rur.txt

Justfile

@@ -9,7 +9,7 @@ dev: format build clippy test docs
 dev-docs:
     #!/usr/bin/env -S sh -eux
     mdbook serve --open &
-    cargo watch ---ignore docs -- just docs &
+    cargo watch --ignore '*.{md,css,sh,txt}' -- just docs &
     trap 'kill $(jobs -pr)' EXIT
     wait
 
@@ -51,6 +51,10 @@ test:
 docs:
     cargo run --package xtask -- docs
 
+# Run benchmarks
+bench:
+    ./bench.sh >./docs/guide/benchmarks.md
+
 # Run fuzzer
 fuzz:
     cargo +nightly fuzz run --jobs {{num_cpus()}} pattern

bench.sh

@@ -0,0 +1,161 @@
+#!/usr/bin/env sh
+
+set -eu
+
+echo "# Benchmarks"
+echo
+echo 'This page contains `rew` benchmarks against various other tools:'
+echo
+echo "- [GNU coreutils](https://www.gnu.org/software/coreutils/)"
+echo "- [uutils coreutils](https://github.com/uutils/coreutils?tab=readme-ov-file)"
+
+specs() {
+  echo "<details>"
+  echo "<summary><code>$1</code></summary>"
+  echo
+  echo '```'
+  sh -c "$1"
+  echo '```'
+  echo
+  echo "</details>"
+}
+
+echo
+echo "## Environment"
+echo
+echo "All benchmarks were run within the following environment."
+echo
+echo "System:"
+echo
+specs 'uname -a'
+specs 'lscpu'
+specs 'lsmem'
+echo
+echo "Software:"
+echo
+specs 'hyperfine --version'
+specs 'rew --version'
+specs 'coreutils --help | head -n1'
+specs 'seq --version'
+specs 'head --version'
+
+BASE_DATA_URL=https://jpikl.github.io/data
+
+SVEJK_FILE=svejk.txt
+SVEJK_REPEAT=100
+SVEJK_INIT="curl $BASE_DATA_URL/$SVEJK_FILE | rew loop $SVEJK_REPEAT >$SVEJK_FILE"
+
+RUR_FILE=rur.txt
+RUR_REPEAT=1000
+RUR_INIT="curl $BASE_DATA_URL/$RUR_FILE | rew loop $RUR_REPEAT >$RUR_FILE"
+
+[ -f "$SVEJK_FILE" ] || sh -c "$SVEJK_INIT"
+[ -f "$RUR_FILE" ] || sh -c "$RUR_INIT"
+
+echo
+echo "## Data"
+echo
+echo "The following text files were used as input data for benchmarking."
+echo
+echo '```shell'
+echo "$SVEJK_INIT"
+echo '```'
+echo
+echo "- UTF-8, non-ascii (czech diacritics)"
+echo "- 5913 lines (repeated $SVEJK_REPEAT times)"
+echo "- very long lines (up to 4952 chars)"
+echo "- trimmed whitespaces"
+echo
+echo '```shell'
+echo "$RUR_INIT"
+echo '```'
+echo
+echo "- UTF-8, non-ascii (czech diacritics)"
+echo "- 4477 lines (repeated $RUR_REPEAT times)"
+echo "- short lines (up to 81 chars)"
+echo "- untrimmed whitespaces"
+
+BENCH_OUT=.bench.md
+
+bench_group() {
+  echo
+  echo "### $1"
+}
+
+bench() {
+  set -- hyperfine --warmup 5 "$@"
+  "$@" --sort mean-time --export-markdown "$BENCH_OUT" 1>&2
+
+  echo
+  echo "Command:"
+  echo
+  echo '```shell'
+
+  printf "$1"
+  shift
+
+  while [ $# -gt 0 ]; do
+    printf ' \\\n    '
+
+    case "$1" in
+    --warmup)
+      printf "%s %s" "$1" "$2"
+      shift 2
+      ;;
+    --setup | --cleanup)
+      printf "%s '%s'" "$1" "$2"
+      shift 2
+      ;;
+    *)
+      printf "'%s'" "$1"
+      shift
+      ;;
+    esac
+  done
+
+  echo
+  echo '```'
+  echo
+  echo "Results:"
+  echo
+
+  cat "$BENCH_OUT"
+  rm "$BENCH_OUT"
+}
+
+echo
+echo "## Runs"
+
+bench_group "rew seq"
+for MULT in '' 0 00; do
+  bench \
+    "seq 1 1$MULT 1000000$MULT" \
+    "rew seq 1..1000000$MULT 1$MULT"
+done
+exit
+bench_group "rew first"
+for FILE in "$SVEJK_FILE" "$RUR_FILE"; do
+  bench \
+    "head -n 100000 <$FILE" \
+    "coreutils head -n 100000 <$FILE" \
+    "rew first 100000 <$FILE"
+done
+
+bench_group "rew last"
+for FILE in "$SVEJK_FILE" "$RUR_FILE"; do
+  bench \
+    "tail -n 100000 <$FILE" \
+    "coreutils tail -n 100000 <$FILE" \
+    "rew last 100000 <$FILE"
+done
+
+bench_group "rew cat"
+for FILE in "$SVEJK_FILE" "$RUR_FILE"; do
+  bench \
+    "cat <$FILE" \
+    "coreutils cat <$FILE" \
+    "rew cat <$FILE" \
+    "rew cat --bytes <$FILE" \
+    "rew cat --chars <$FILE" \
+    "rew cat --lines <$FILE"
+done

docs/SUMMARY.md

@@ -6,6 +6,7 @@
 
 - [Installation](./guide/installation.md)
 - [FAQ](./guide/faq.md)
+- [Benchmarks](./guide/benchmarks.md)
 
 # Reference