chore(release): v3.0.0

jedld · jedld · commit 0c52e5eeb87f · 2025-09-11T08:10:26.000+08:00
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
@@ -2,35 +2,27 @@ name: CI
 
 on:
   push:
+    branches: [ master ]
   pull_request:
-  workflow_dispatch:
-#  schedule:
-#    - cron: '42 5 * * *'
+    branches: [ master ]
 
 jobs:
   test:
+    runs-on: ubuntu-latest
     strategy:
       fail-fast: false
       matrix:
-        ruby: [ '3.0' ]
-
-    runs-on: ubuntu-latest
-    name: Ruby ${{matrix.ruby}}
-    container: ruby:${{matrix.ruby}}
-
+        ruby: [ '3.0', '3.1', '3.2', '3.3', '3.4' ]
     steps:
-    - uses: actions/checkout@v3
-
-    - name: Show ruby Version
-      run: |
-        ruby -v
-
-    - name: Install Modules
-      run: ./bin/setup
-
-    - name: Compile
-      run: rake compile
-
-    - name: Run tests
-      run: rake spec
+      - uses: actions/checkout@v4
+      - uses: ruby/setup-ruby@v1
+        with:
+          ruby-version: ${{ matrix.ruby }}
+          bundler-cache: true
+      - name: Install dependencies
+        run: bundle install --jobs 4 --retry 3
+      - name: Compile native extension
+        run: rake compile
+      - name: Run specs
+        run: rake spec
 
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,4 +1,27 @@
 # 2.0.2 (August 10, 2023)
+# 2.0.2 (August 10, 2023)
+
+# 3.0.0 (September 11, 2025)
+
+## Features
+
+- New `MultiStringReplace::Automaton` API to compile patterns once and reuse for faster repeated match/replace.
+
+## Performance
+
+- Faster replace path: cached Ruby method IDs, fewer API calls, preallocated buffers, and deferred string allocation for no-match fast path.
+- Optimized trie setup/clear and memory copies.
+
+## Stability and build
+
+- Fixed Bundler conflicts on Ruby 3.4+; Rakefile avoids dual-Bundler loading.
+- Added GC marking to prevent crashes with the new Automaton.
+- Safer behavior when replacement is missing; falls back to original substring.
+
+## Tooling
+
+- Benchmark script now supports `-f/--file`, `-n/--iters`, and `-A/--automaton` flags.
+- Added GitHub Actions CI across Ruby 3.0–3.4.
 
 ## Bug fixes:
 
diff --git a/Gemfile.lock b/Gemfile.lock
@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    multi_string_replace (2.0.2)
+    multi_string_replace (3.0.0)
 
 GEM
   remote: https://rubygems.org/
@@ -36,4 +36,4 @@ DEPENDENCIES
   rspec (~> 3.0)
 
 BUNDLED WITH
-   2.3.27
+   2.6.9
diff --git a/ext/multi_string_replace/ahocorasick.c b/ext/multi_string_replace/ahocorasick.c
@@ -194,9 +194,8 @@ VALUE aho_replace_text(struct ahocorasick * restrict aho, const char* data,
     struct aho_trie_node* travasal_node = NULL;
 
     travasal_node = &(aho->trie.root);
-    // Preallocate with input size; Ruby will grow if needed
-    VALUE main_result = rb_str_buf_new((long)data_len);
-
+    // Defer allocation until first match; handle no-match fast path
+    VALUE main_result = Qnil;
     unsigned long long last_concat_pos = 0;
 
     for (i = 0; i < data_len; i++)
@@ -212,8 +211,11 @@ VALUE aho_replace_text(struct ahocorasick * restrict aho, const char* data,
     const int rlen = result->len;
     unsigned long long pos = i - rlen + 1;
 
-        // concatenate from last_concat_pos
-        if (pos > last_concat_pos) {
+        // On first match, allocate result and copy prefix if any
+        if (NIL_P(main_result)) {
+            main_result = rb_str_buf_new((long)data_len);
+            if (pos > 0) rb_str_cat(main_result, &data[0], pos);
+        } else if (pos > last_concat_pos) {
             rb_str_cat(main_result, &data[last_concat_pos], pos - last_concat_pos);
         }
         
@@ -249,11 +251,15 @@ VALUE aho_replace_text(struct ahocorasick * restrict aho, const char* data,
         last_concat_pos = i + 1;
     }
 
-    if (last_concat_pos < data_len) {
-        rb_str_cat(main_result, &data[last_concat_pos], (long)(data_len - last_concat_pos));
+    if (NIL_P(main_result)) {
+        // No matches; return a copy of input (preserves previous API behavior of returning a new String)
+        return rb_str_new(data, (long)data_len);
+    } else {
+        if (last_concat_pos < data_len) {
+            rb_str_cat(main_result, &data[last_concat_pos], (long)(data_len - last_concat_pos));
+        }
+        return main_result;
     }
-
-    return main_result;
 }
 
 inline void aho_register_match_callback(VALUE rb_result_container, struct ahocorasick * restrict aho,
diff --git a/ext/multi_string_replace/extconf.rb b/ext/multi_string_replace/extconf.rb
@@ -1,3 +1,13 @@
 require 'mkmf'
-create_header
-create_makefile 'multi_string_replace/multi_string_replace'
+
+# Optional: allow users to tweak optimization flags
+optflags = ENV['MSR_OPTFLAGS'] || '-O3 -fno-strict-aliasing'
+warnflags = ENV['MSR_WARNFLAGS']
+
+with_cflags(optflags) do
+	# avoid treating warnings as errors across diverse compilers
+	with_werror(false) do
+		create_header
+		create_makefile 'multi_string_replace/multi_string_replace'
+	end
+end
diff --git a/ext/multi_string_replace/multi_string_replace.c b/ext/multi_string_replace/multi_string_replace.c
@@ -186,6 +186,11 @@ VALUE multi_string_replace(VALUE self, VALUE body, VALUE replace)
   VALUE replace_values = rb_funcall(replace, id_values, 0);
   long size =  RARRAY_LEN(keys);
 
+  if (size == 0) {
+    // Nothing to replace; return a copy of input (preserves API semantics)
+    return rb_str_dup(body);
+  }
+
   long value_sizes[size];
   char *values[size];
   VALUE ruby_val[size];
diff --git a/lib/multi_string_replace/version.rb b/lib/multi_string_replace/version.rb
@@ -1,3 +1,3 @@
 module MultiStringReplace
-  VERSION = "2.0.2"
+  VERSION = "3.0.0"
 end
diff --git a/spec/automaton_spec.rb b/spec/automaton_spec.rb
@@ -0,0 +1,29 @@
+require 'spec_helper'
+RSpec.describe MultiStringReplace::Automaton do
+  it 'matches with compiled automaton' do
+    ac = described_class.new(['brown', 'fox'])
+    res = ac.match('The quick brown fox')
+    expect(res).to be_a(Hash)
+    # ids 0 and 1 should exist with positions
+    expect(res[0]).to include(10) # 'brown'
+    expect(res[1]).to include(16) # 'fox'
+  end
+
+  it 'replaces with compiled automaton' do
+    ac = described_class.new(['brown', 'fox'])
+    out = ac.replace('The quick brown fox', { 'brown' => 'black', 'fox' => 'wolf' })
+    expect(out).to eq('The quick black wolf')
+  end
+
+  it 'falls back to original when no replacement provided' do
+    ac = described_class.new(['brown'])
+    out = ac.replace('brown', {})
+    expect(out).to eq('brown')
+  end
+
+  it 'supports proc values' do
+    ac = described_class.new(['x'])
+    out = ac.replace('xx', { 'x' => ->(s, e) { 'y' } })
+    expect(out).to eq('yy')
+  end
+end
