targrep 0.6.1

Search inside compressed log archives (.tgz, .tar.zst, .tar.xz) with ripgrep's flags and per-member line numbers
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96

name: Benchmark

# bench/compare.sh on a GitHub-hosted runner, for architectures there is no
# other machine to measure on. Run by hand:
#
#     gh workflow run bench.yml                      # x86_64
#     gh workflow run bench.yml -f runner=ubuntu-24.04-arm
#
# The corpus is synthetic (bench/synth_logs.py): real logs never go to a public
# runner. Shared runner VMs are noisy, so read the ratios between rows, not the
# absolute times. The results land in the run summary and as an artifact.
on:
  workflow_dispatch:
    inputs:
      runner:
        description: "Runner label"
        default: ubuntu-latest
      reps:
        description: "Repetitions per row (the best is reported)"
        default: "3"
      suite:
        description: "compare (trg against zgrep and rg) or inflate-modes (--inflate A/B at three archive sizes)"
        default: compare

env:
  CARGO_TERM_COLOR: always

jobs:
  bench:
    runs-on: ${{ inputs.runner }}
    steps:
      - uses: actions/checkout@v4
      - uses: dtolnay/rust-toolchain@stable
      - uses: Swatinem/rust-cache@v2

      - name: Install the tools being compared
        run: sudo apt-get update -q && sudo apt-get install -y -q pigz ripgrep

      # Both backends: zlib-ng (the default) and the pure-Rust fallback.
      - name: Build
        run: |
          cargo build --release
          cargo build --release --no-default-features --target-dir target/miniz
          mkdir -p bin
          cp target/release/trg bin/trg-ng
          cp target/miniz/release/trg bin/trg-miniz

      - name: Build the corpus
        if: inputs.suite == 'compare'
        run: |
          mkdir -p synth
          python3 bench/synth_logs.py synth
          SRC=synth bench/build_corpus.sh /tmp/trgbench

      - name: Compare
        if: inputs.suite == 'compare'
        run: |
          {
            echo "## $(uname -m), $(nproc) cores"
            echo
            echo '```'
            lscpu | grep -E 'Model name|^CPU\(s\)|Thread|L2|L3' || true
            echo "SIMD: $(grep -owE 'avx2|avx512f|avx512vl|pclmulqdq|asimd|pmull|crc32' /proc/cpuinfo | sort -u | tr '\n' ' ')"
            echo "$(rg --version | head -n1), $(gzip --version | head -n1), $(pigz --version 2>&1)"
            echo
            REPS=${{ inputs.reps }} TRG=bin/trg-miniz TRG_NAME=miniz_oxide TRG_ZLIBNG=bin/trg-ng \
              bench/compare.sh /tmp/trgbench/corpus
            echo '```'
          } | tee results.md
          cat results.md >> "$GITHUB_STEP_SUMMARY"

      # Archives inflating to 4 MB and 42 MB (which `auto` buffers under the
      # default 64 MB budget) and 1 GB (which needs a 2G budget to buffer).
      - name: Inflate modes
        if: inputs.suite == 'inflate-modes'
        run: |
          {
            echo "## $(uname -m), $(nproc) cores: --inflate modes"
            echo
            echo '```'
            lscpu | grep -E 'Model name' || true
            for scale in 0.004 0.04 1; do
              mkdir -p "synth-$scale"
              python3 bench/synth_logs.py "synth-$scale" "$scale"
              SRC="synth-$scale" bench/build_corpus.sh "/tmp/trgbench-$scale" > /dev/null
              python3 bench/inflate_modes.py "/tmp/trgbench-$scale/corpus" bin/trg-ng ${{ inputs.reps }}
              rm -rf "/tmp/trgbench-$scale"
            done
            echo '```'
          } | tee results.md
          cat results.md >> "$GITHUB_STEP_SUMMARY"

      - uses: actions/upload-artifact@v4
        with:
          name: bench-${{ inputs.suite }}-${{ inputs.runner }}
          path: results.md