mittens 0.3.3 → 0.3.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a293e52f3e6ecf7669615bbf24aae34a9fcef78305b7324f7f739fc97e7eb0fc
-  data.tar.gz: e5f1ba1b94ac634316599a25d969c1d71d4a0ff591ef043011c2481fe641814d
+  metadata.gz: cc9a43dbea4ac1069238dd003f295c3fec249d658f1aa911e7da3e8abeea246f
+  data.tar.gz: d2c745ba5dea06757c857787cf89891a38f118cbecbd12720b577216605d09cf
 SHA512:
-  metadata.gz: 4c392905c194b9eb723f44e7e97fd040395757751a43a034c34dfef51f399f24e2be4e390414e6e34e79bd26410987d696cd23fa197396ff9510f60a5503730e
-  data.tar.gz: 9dbff4611a7d1b43d42d19fa0e873a3c5055b2358b5b0d75e3080e9686ccd315be564dbd93af62e28a63c3e1b50706968a2b6dbcd01f9dc7cd41420bf782ac09
+  metadata.gz: 326722dc85ee85717aa395d7bef8d9cab953172f598fdf2fe26622fc3ab1b2e96bba1005c773a0f2ac4ae0b684b55e116f84e611c1643cc3941f30577e220fcc
+  data.tar.gz: 0113f43ea4d07fe4f5f77ffc5a7b9b6f1a5ad6b85208876878c5a0cd378686c8b344d095d93957455757b33bf9c73aae63e72288effc8bf0c8c26b5b868fa3d0

data/CHANGELOG.md

@@ -1,3 +1,11 @@
+## 0.3.5 (2026-10-28)
+
+- Updated Snowball to 3.2.0
+
+## 0.3.4 (2026-08-06)
+
+- Fixed `dup` and `clone` methods
+
 ## 0.3.3 (2026-08-05)
 
 - Fixed issue with GC compaction

data/LICENSE.txt

@@ -1,7 +1,7 @@
 Copyright (c) 2001, Dr Martin Porter
 Copyright (c) 2004,2005, Richard Boulton
 Copyright (c) 2013, Yoshiki Shibukawa
-Copyright (c) 2006,2007,2009,2010,2011,2014-2019, Olly Betts
+Copyright (c) 2006-2025, Olly Betts
 Copyright (c) 2022-2025, Andrew Kane
 All rights reserved.
 

data/README.md

@@ -36,7 +36,7 @@ Specify the language
 stemmer = Mittens::Stemmer.new(language: "french")
 ```
 
-Supports `arabic`, `armenian`, `basque`, `catalan`, `czech`, `danish`, `dutch`, `dutch_porter`, `english`, `esperanto`, `estonian`, `finnish`, `french`, `german`, `greek`, `hindi`, `hungarian`, `indonesian`, `irish`, `italian`, `lithuanian`, `nepali`, `norwegian`, `persian`, `polish`, `porter`, `portuguese`, `romanian`, `russian`, `serbian`, `sesotho`, `spanish`, `swedish`, `tamil`, `turkish`, and `yiddish`
+Supports `arabic`, `armenian`, `basque`, `catalan`, `czech`, `danish`, `dutch`, `dutch_porter`, `earlymodernenglish`, `english`, `esperanto`, `estonian`, `finnish`, `french`, `german`, `greek`, `hindi`, `hungarian`, `indonesian`, `irish`, `italian`, `lithuanian`, `nepali`, `norwegian`, `persian`, `polish`, `porter`, `portuguese`, `romanian`, `russian`, `serbian`, `sesotho`, `spanish`, `swedish`, `tamil`, `turkish`, `ukrainian`, and `yiddish`
 
 ## History
 

data/ext/mittens/ext.c

@@ -3,21 +3,35 @@
 #include "libstemmer.h"
 #include "ruby/ruby.h"
 
+#define GetAlgorithm(language) (NIL_P(language) ? "english" : StringValueCStr(language))
+#define GetStemmer(obj, ptr) TypedData_Get_Struct((obj), stemmer_t, &stemmer_data_type, (ptr))
+
 typedef struct stemmer {
     struct sb_stemmer* stemmer;
+    VALUE language;
 } stemmer_t;
 
+static void stemmer_mark(void* ptr)
+{
+    stemmer_t* stemmer = (stemmer_t*) ptr;
+    rb_gc_mark(stemmer->language);
+}
+
 static void stemmer_free(void* ptr)
 {
     stemmer_t* stemmer = (stemmer_t*) ptr;
-    // safe to pass null pointer according to docs
-    sb_stemmer_delete(stemmer->stemmer);
+
+    // safe to pass null pointer, but check anyways
+    if (stemmer->stemmer != NULL)
+        sb_stemmer_delete(stemmer->stemmer);
+
     xfree(ptr);
 }
 
 const rb_data_type_t stemmer_data_type = {
-    .wrap_struct_name = "stemmer",
+    .wrap_struct_name = "Mittens::Stemmer",
     .function = {
+        .dmark = stemmer_mark,
         .dfree = stemmer_free,
     },
     .flags = RUBY_TYPED_FREE_IMMEDIATELY
@@ -28,38 +42,37 @@ static VALUE stemmer_allocate(VALUE klass)
     stemmer_t* stemmer;
     VALUE obj = TypedData_Make_Struct(klass, stemmer_t, &stemmer_data_type, stemmer);
     stemmer->stemmer = NULL;
+    stemmer->language = Qnil;
     return obj;
 }
 
 static VALUE stemmer_initialize(int argc, VALUE* argv, VALUE self)
 {
+    stemmer_t* stemmer;
+    GetStemmer(self, stemmer);
+
+    if (stemmer->stemmer != NULL)
+        rb_raise(rb_eRuntimeError, "stemmer already initialized");
+
     VALUE opts;
     rb_scan_args(argc, argv, ":", &opts);
 
     VALUE language = Qnil;
-    const char* algorithm = "english";
-    if (!NIL_P(opts)) {
+    if (!NIL_P(opts))
         language = rb_hash_aref(opts, ID2SYM(rb_intern("language")));
-        if (!NIL_P(language)) {
-            Check_Type(language, T_STRING);
-            algorithm = RSTRING_PTR(language);
-        }
-    }
-
-    stemmer_t* stemmer;
-    TypedData_Get_Struct(self, stemmer_t, &stemmer_data_type, stemmer);
-
-    // in case called multiple times
-    sb_stemmer_delete(stemmer->stemmer);
 
     // if adding support for encoding, may want to change encoding returned from stem
-    stemmer->stemmer = sb_stemmer_new(algorithm, NULL);
-    if (stemmer->stemmer == NULL) {
-        rb_raise(rb_eArgError, "unknown language: %s", algorithm);
+    stemmer->stemmer = sb_stemmer_new(GetAlgorithm(language), NULL);
+    if (stemmer->stemmer == NULL)
+    {
+        if (!NIL_P(language))
+            rb_raise(rb_eArgError, "unknown language: %" PRIsVALUE, language);
+
+        rb_raise(rb_eNoMemError, "failed to allocate memory");
     }
 
-    // must be placed after last use of algorithm
-    RB_GC_GUARD(language);
+    // copy (after GetAlgorithm) in case string is modified in future
+    stemmer->language = rb_obj_dup(language);
 
     return self;
 }
@@ -67,17 +80,43 @@ static VALUE stemmer_initialize(int argc, VALUE* argv, VALUE self)
 static VALUE stemmer_stem(VALUE self, VALUE value)
 {
     stemmer_t* stemmer;
-    TypedData_Get_Struct(self, stemmer_t, &stemmer_data_type, stemmer);
+    GetStemmer(self, stemmer);
 
-    Check_Type(value, T_STRING);
+    if (stemmer->stemmer == NULL)
+        rb_raise(rb_eRuntimeError, "stemmer not initialized");
 
-    const sb_symbol* word = (const sb_symbol*) RSTRING_PTR(value);
+    const char* word = StringValuePtr(value);
     long size = RSTRING_LEN(value);
     if (size > INT_MAX)
         rb_raise(rb_eArgError, "string exceeds max length");
-    const sb_symbol* pointer_out = sb_stemmer_stem(stemmer->stemmer, word, (int) size);
 
-    return rb_utf8_str_new_cstr((char*) pointer_out);
+    const sb_symbol* pointer_out = sb_stemmer_stem(stemmer->stemmer, (const sb_symbol*) word, (int) size);
+    int length_out = sb_stemmer_length(stemmer->stemmer);
+
+    // place after last use of word
+    RB_GC_GUARD(value);
+
+    return rb_utf8_str_new((char*) pointer_out, (long) length_out);
+}
+
+static VALUE stemmer_initialize_copy(VALUE self, VALUE orig)
+{
+    stemmer_t* stemmer;
+    GetStemmer(self, stemmer);
+
+    stemmer_t* orig_stemmer;
+    GetStemmer(orig, orig_stemmer);
+
+    if (stemmer->stemmer != NULL)
+        rb_raise(rb_eRuntimeError, "stemmer already initialized");
+
+    stemmer->stemmer = sb_stemmer_new(GetAlgorithm(orig_stemmer->language), NULL);
+    if (stemmer->stemmer == NULL)
+        rb_raise(rb_eNoMemError, "failed to allocate memory");
+
+    stemmer->language = orig_stemmer->language;
+
+    return self;
 }
 
 static VALUE stemmer_languages(VALUE klass)
@@ -85,10 +124,8 @@ static VALUE stemmer_languages(VALUE klass)
     VALUE out = rb_ary_new();
 
     const char** language = sb_stemmer_list();
-    while (*language != NULL) {
-        rb_ary_push(out, rb_utf8_str_new_cstr(*language));
-        language++;
-    }
+    while (*language != NULL)
+        rb_ary_push(out, rb_utf8_str_new_cstr(*language++));
 
     return out;
 }
@@ -100,5 +137,6 @@ void Init_ext(void)
     rb_define_alloc_func(cStemmer, stemmer_allocate);
     rb_define_method(cStemmer, "initialize", stemmer_initialize, -1);
     rb_define_method(cStemmer, "stem", stemmer_stem, 1);
+    rb_define_method(cStemmer, "initialize_copy", stemmer_initialize_copy, 1);
     rb_define_singleton_method(cStemmer, "languages", stemmer_languages, 0);
 }

data/ext/mittens/extconf.rb

@@ -10,6 +10,8 @@ output, status = Open3.capture2(make_cmd, "CFLAGS=#{cflags}", chdir: vendor)
 puts output
 raise "Command failed" unless status.success?
 
+append_cflags("-Wall -Wextra -Wpedantic -Wconversion")
+
 $INCFLAGS += " -I$(srcdir)/../../vendor/snowball/include"
 $LDFLAGS += " $(srcdir)/../../vendor/snowball/libstemmer.a"
 

data/lib/mittens/version.rb

@@ -1,3 +1,3 @@
 module Mittens
-  VERSION = "0.3.3"
+  VERSION = "0.3.5"
 end

data/vendor/snowball/.github/workflows/ci.yml

@@ -48,6 +48,10 @@ jobs:
         - name: "C++"
           CXX: g++
           WERROR: '-pedantic -Werror'
+        - name: "C++ widechars"
+          CXX: g++
+          MAKE: 'make SNOWBALL_WIDE=1'
+          WERROR: '-pedantic -Werror'
         - name: java
           JAVA: java
           JAVAC: javac
@@ -112,6 +116,11 @@ jobs:
           os: 'ubuntu-24.04'
           apt_packages: 'python3.12'
           THIN_FACTOR: 1
+        - name: Python 3.14
+          PYTHON: python3.14
+          os: 'ubuntu-26.04'
+          apt_packages: 'python3.14'
+          THIN_FACTOR: 1
         - name: Python (pypy3)
           PYTHON: pypy3
           apt_packages: 'pypy3'
@@ -128,6 +137,32 @@ jobs:
           MAKE: mingw32-make
           mingw64_packages: 'mingw-w64-ucrt-x86_64-go'
           ccache: sccache
+        - name: Windows MSVC (C)
+          os: windows-latest
+          c_tests: check
+          ccache: sccache
+          CC: cl
+          CFLAGS: -nologo -std:c11 -W3 -WX
+        - name: Windows MSVC clatest (C)
+          os: windows-latest
+          c_tests: check
+          ccache: sccache
+          CC: cl
+          CFLAGS: -nologo -std:clatest -W3 -WX
+        - name: Windows MSVC2026 arm (C)
+          os: windows-11-vs2026-arm
+          msvc_arch: arm64
+          c_tests: check
+          ccache: sccache
+          CC: cl
+          CFLAGS: -nologo -std:c11 -W3 -WX
+        - name: Windows MSVC2026 clatest arm (C)
+          os: windows-11-vs2026-arm
+          msvc_arch: arm64
+          c_tests: check
+          ccache: sccache
+          CC: cl
+          CFLAGS: -nologo -std:clatest -W3 -WX
 
       fail-fast: false
 
@@ -188,6 +223,11 @@ jobs:
             git clone "$UPSTREAM_REPO_URL"
           fi
         fi
+    - name: Set up Visual Studio shell
+      if: ${{ endsWith(matrix.CC, 'cl') }}
+      uses: egor-tensin/vs-shell@v2
+      with:
+        arch: ${{ matrix.msvc_arch || 'x64' }}
     - name: Hook up runtime testing
       if: ${{ inputs.runtime-tests }}
       run: |
@@ -229,7 +269,7 @@ jobs:
         msystem: ucrt64
         install: base-devel ${{ matrix.mingw64_packages }}
     - name: Build
-      run: $MAKE CC="${{ matrix.ccache || 'ccache' }} $CC"
+      run: $MAKE CC="${{ matrix.ccache || 'ccache' }} $CC" ${{ format('{0}{1}{2}', matrix.CFLAGS && 'CFLAGS="', matrix.CFLAGS, matrix.CFLAGS && '"') }}
     - name: Test C dist
       if: matrix.CFLAGS_DIST_BUILD
       run: |
@@ -249,7 +289,7 @@ jobs:
         case '${{ matrix.c_tests }}' in
           *generate*) mkdir ada/bin && ln -s /bin/true ada/bin/generate ;;
         esac
-        $MAKE ${{ matrix.c_tests }} CC="$CC"
+        $MAKE ${{ matrix.c_tests }} CC="$CC" ${{ format('{0}{1}{2}', matrix.CFLAGS && 'CFLAGS="', matrix.CFLAGS, matrix.CFLAGS && '"') }}
     - name: Test C++
       if: matrix.CXX
       run: |
@@ -277,7 +317,7 @@ jobs:
       if: ${{ startsWith(matrix.JSRUN, 'deno') }}
       run: |
         deno lint javascript/*.js js_out/*.js
-        deno check javascript/*.js js_out/*.js
+        deno check --check-js javascript/*.js js_out/*.js
     - name: Test Rust
       if: matrix.RUST
       run: $MAKE check_rust RUST="${{ matrix.RUST }}"

data/vendor/snowball/.github/workflows/coverage.yml

@@ -90,7 +90,8 @@ jobs:
           ls -1 algorithms|sed 's,\.sbl$,,p;d'
         elif [ -n "$GITHUB_HEAD_REF" ] ; then
           # Pull-request.
-          git diff --numstat "$GITHUB_BASE_REF"... algorithms|sed 's,^[0-9].*/\([^ ]*\)\.sbl$,\1,p;d'
+          git fetch --depth=1 origin "$GITHUB_BASE_REF":"$GITHUB_BASE_REF"
+          git diff --numstat "$GITHUB_BASE_REF" HEAD -- algorithms|sed 's,^[0-9].*/\([^ ]*\)\.sbl$,\1,p;d'
         else
           # Push.
           #
@@ -109,9 +110,11 @@ jobs:
           # That size means generating a coverage report for it takes something
           # like 20-25 minutes.
           if [ -f "${STEMMING_DATA}/$a/voc.txt" ] ; then
-            $MAKE -s "check_utf8_$a" 2> coverage.log
-            echo ; echo "=== Coverage report for $a ===" ; echo
-            sort -k2,2 -k3,3n -k3,3 -k5,5g coverage.log |
-              uniq -c |awk '{$1=sprintf("% 8d",$1-1);print}'
+            if grep -q "^$a " libstemmer/modules.txt ; then
+              $MAKE -s "check_utf8_$a" 2> coverage.log
+              echo ; echo "=== Coverage report for $a ===" ; echo
+              sort -k2,2 -k3,3n -k3,3 -k5,5g coverage.log |
+                uniq -c |awk '{$1=sprintf("% 8d",$1-1);print}'
+            fi
           fi
         done

data/vendor/snowball/GNUmakefile

@@ -2,7 +2,7 @@
 
 # After changing this, run `make update_version` to update various sources
 # which hard-code it.
-SNOWBALL_VERSION = 3.1.1
+SNOWBALL_VERSION = 3.2.0
 
 ifeq ($(OS),Windows_NT)
 EXEEXT = .exe
@@ -34,6 +34,20 @@ ada_src_dir = $(ada_src_main_dir)/algorithms
 
 ARFLAGS = -cr
 c_src_dir = src_c
+ifeq '$(filter %cl,$(CC))' ''
+o_for_obj = -o
+o_for_exe = -o
+ARFLAGS_ = $(ARFLAGS) # Includes a trailing space.
+else
+o_for_obj = -Fo:
+o_for_exe = -Fe:
+AR = lib -nologo
+# Frustratingly, LIB.EXE's `-out:` option doesn't allow a space to follow
+# (unlike CL.EXE's `-Fo:` and `-Fe:` options) so ARFLAGS_ has to be used
+# without a space after it.  For more sensible compilers ARFLAGS_ is set
+# to include a trailing space.
+ARFLAGS_ = -out:
+endif
 
 # C++
 
@@ -41,6 +55,13 @@ CXX ?= c++
 CXXFLAGS=-g -O2 -W -Wall -Wcast-qual -Wmissing-declarations -Wshadow $(WERROR)
 cxx_src_dir = cxx
 
+ifdef SNOWBALL_WIDE
+CXXFLAGS += -DSNOWBALL_WIDE
+cxx_encoding_opt := -w
+else
+cxx_encoding_opt := -u
+endif
+
 # C#
 
 MONO ?= mono
@@ -182,7 +203,7 @@ COMPILER_HEADERS = compiler/header.h \
 # C
 
 RUNTIME_SOURCES  = runtime/api.c \
-		   runtime/utilities.c
+		   runtime/snowball_runtime.c
 
 RUNTIME_HEADERS  = runtime/api.h \
 		   runtime/snowball_runtime.h
@@ -198,7 +219,7 @@ STEMTEST_SOURCES = tests/stemtest.c
 # C++
 
 CXX_STEMWORDS_SOURCES = $(cxx_src_dir)/stemwords.cxx
-CXX_RUNTIME_SOURCES = $(cxx_src_dir)/stemmer.cxx $(cxx_src_dir)/utilities.cxx
+CXX_RUNTIME_SOURCES = $(cxx_src_dir)/stemmer.cxx $(cxx_src_dir)/snowball_runtime.cxx
 CXX_SOURCES = $(libstemmer_algorithms:%=$(cxx_src_dir)/%_stemmer.cxx)
 CXX_HEADERS = $(libstemmer_algorithms:%=$(cxx_src_dir)/%_stemmer.h)
 
@@ -369,7 +390,7 @@ $(STEMMING_DATA)/% $(STEMMING_DATA_ABS)/%:
 	@[ -f '$@' ] || { echo '$@: Test data not found'; echo 'Checkout the snowball-data repo as "$(STEMMING_DATA_ABS)"'; exit 1; }
 
 snowball$(EXEEXT): $(COMPILER_OBJECTS)
-	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $^
+	$(CC) $(CFLAGS) $(LDFLAGS) $(o_for_exe) $@ $^
 
 $(COMPILER_OBJECTS): $(COMPILER_HEADERS)
 
@@ -424,19 +445,19 @@ libstemmer/modules_utf8.h libstemmer/mkinc_utf8.mak: libstemmer/mkmodules.pl $(M
 libstemmer/libstemmer.o: libstemmer/modules.h $(C_LIB_HEADERS)
 
 libstemmer.a: libstemmer/libstemmer.o $(RUNTIME_OBJECTS) $(C_LIB_OBJECTS)
-	$(AR) $(ARFLAGS) $@ $^
+	$(AR) $(ARFLAGS_)$@ $^
 
 examples/%.o: examples/%.c
-	$(CC) $(CFLAGS) $(INCLUDES) $(CPPFLAGS) -c -o $@ $<
+	$(CC) $(CFLAGS) $(INCLUDES) $(CPPFLAGS) -c $(o_for_obj) $@ $<
 
 stemwords$(EXEEXT): $(STEMWORDS_OBJECTS) libstemmer.a
-	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $^
+	$(CC) $(CFLAGS) $(LDFLAGS) $(o_for_exe) $@ $^
 
 tests/%.o: tests/%.c
-	$(CC) $(CFLAGS) $(INCLUDES) $(CPPFLAGS) -c -o $@ $<
+	$(CC) $(CFLAGS) $(INCLUDES) $(CPPFLAGS) -c $(o_for_obj) $@ $<
 
 stemtest$(EXEEXT): $(STEMTEST_OBJECTS) libstemmer.a
-	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $^
+	$(CC) $(CFLAGS) $(LDFLAGS) $(o_for_exe) $@ $^
 
 $(c_src_dir)/stem_UTF_8_%.c $(c_src_dir)/stem_UTF_8_%.h: $(ALGORITHMS)/%.sbl snowball$(EXEEXT)
 	@mkdir -p $(c_src_dir)
@@ -455,7 +476,7 @@ $(c_src_dir)/stem_ISO_8859_2_%.c $(c_src_dir)/stem_ISO_8859_2_%.h: $(ALGORITHMS)
 	$(SNOWBALL_COMPILE) charsets/ISO-8859-2.sbl $< -o $@ -eprefix $*_ISO_8859_2_ -r ../runtime
 
 $(c_src_dir)/stem_%.o: $(c_src_dir)/stem_%.c $(c_src_dir)/stem_%.h
-	$(CC) $(CFLAGS) $(INCLUDES) $(CPPFLAGS) -c -o $@ $<
+	$(CC) $(CFLAGS) $(INCLUDES) $(CPPFLAGS) -c $(o_for_obj) $@ $<
 
 # C++
 
@@ -467,7 +488,7 @@ $(cxx_src_dir)/stemwords$(EXEEXT): $(CXX_STEMWORDS_OBJECTS) $(CXX_RUNTIME_OBJECT
 
 $(cxx_src_dir)/%_stemmer.cxx $(cxx_src_dir)/%_stemmer.h: $(ALGORITHMS)/%.sbl snowball$(EXEEXT)
 	@mkdir -p $(cxx_src_dir)
-	$(SNOWBALL_COMPILE) -c++ -cheader '"stemmer.h"' $< -o $@ -r ../runtime -u
+	$(SNOWBALL_COMPILE) -c++ -cheader '"stemmer.h"' $< -o $@ -r ../runtime $(cxx_encoding_opt)
 
 $(cxx_src_dir)/%stemmer.o: $(cxx_src_dir)/%stemmer.h
 
@@ -477,6 +498,8 @@ $(cxx_src_dir)/%.o: $(cxx_src_dir)/%.cxx
 $(cxx_src_dir)/stemmer.cxx: GNUmakefile $(cxx_src_dir)/generate_algorithms.pl $(MODULES)
 	$(cxx_src_dir)/generate_algorithms.pl $@ $(MODULES)
 
+$(cxx_src_dir)/snowball_runtime.cxx: runtime/snowball_runtime.c
+
 # C#
 
 csharp_stemwords$(EXEEXT): $(CSHARP_STEMWORDS_SOURCES) $(CSHARP_RUNTIME_SOURCES) $(CSHARP_SOURCES)
@@ -818,7 +841,13 @@ CLEANDIRS += $(ada_src_dir) ada/bin ada/obj
 
 .PHONY: check check_compilertest check_stemtest check_utf8 check_iso_8859_1 check_iso_8859_2 check_koi8r
 
-check: check_compilertest check_utf8 check_iso_8859_1 check_iso_8859_2 check_koi8r
+# We don't run these when runtime tests are enabled (for stemtest, because it
+# contains tests of the shipped algorithms; for compilertest, because it is
+# orthogonal to the algorithms and so redundant to run it in both
+# configuations).
+EXTRA_TESTS ?= check_compilertest check_stemtest
+
+check: $(EXTRA_TESTS) check_utf8 check_iso_8859_1 check_iso_8859_2 check_koi8r
 
 check_compilertest: tests/compilertest
 	cd tests && ./compilertest
@@ -1199,14 +1228,14 @@ CLEANFILES += $(ZIG_SOURCES) zig/stemwords$(EXEEXT)
 # Runtime test integration is currently a bit clunky, and you need to switch
 # your tree to a different state to run runtime tests.
 #
-# make clean setup_runtime_tests
+# make setup_runtime_tests
 #
 # Then targets like `check_utf8`, `check_python`, etc will run the runtime
 # tests for a particular target language.
 #
 # Once you're done, switch the tree back to the normal state:
 #
-# make clean clean_runtime_tests
+# make clean_runtime_tests
 
 .PHONY: setup_runtime_tests clean_runtime_tests
 
@@ -1219,12 +1248,13 @@ setup_runtime_tests: clean_runtime_tests
 	  d=`echo "$$t"|sed 's/\.sbl$$//'` ;\
 	  mkdir $$r/$$d ;\
 	  echo ok > $$r/$$d/voc.txt ;\
-	  echo ok > $$r/$$d/output.txt ;\
+	  if [ -f $$d.out ] ; then cp $$d.out $$r/$$d/output.txt ; else echo ok > $$r/$$d/output.txt ; fi ;\
 	  echo "$$d UTF_8,ISO_8859_1 $$d" >> $$r/modules.txt ;\
 	done
 	printf '%s:=%s\n' \
 	  ALGORITHMS 'tests/runtime' \
 	  BASELINE 'rbaseline' \
+	  EXTRA_TESTS '' \
 	  MODULES '$(RUNTIME_DATA_DIR)/modules.txt' \
 	  other_algorithms '' \
 	  SNOWBALL_FLAGS '-comments' \
@@ -1234,5 +1264,5 @@ setup_runtime_tests: clean_runtime_tests
 	rm -f algorithms.mk
 	$(MAKE) algorithms.mk
 
-clean_runtime_tests:
+clean_runtime_tests: clean
 	rm -rf $(RUNTIME_DATA_DIR) overrides.mk