ruby · kddnewton · Jun 5, 2024 · Jun 4, 2024 · Jun 4, 2024 · Jun 4, 2024
diff --git a/Gemfile b/Gemfile
@@ -4,18 +4,16 @@ source "https://rubygems.org"
 
 gemspec
 
+gem "benchmark-ips"
 gem "rake"
 gem "rake-compiler"
 gem "test-unit"
 
-# Some gems we don't want to install on JRuby or TruffleRuby.
-platforms = %i[mri mswin mingw x64_mingw]
-
-gem "ffi", platform: platforms
-gem "parser", platform: platforms
-gem "ruby_parser", platform: platforms
-gem "benchmark-ips"
+platforms :mri, :mswin, :mingw, :x64_mingw do
+  gem "ffi"
+  gem "parser"
+  gem "ruby_memcheck"
+  gem "ruby_parser"
+end  
 
-group :memcheck do
-  gem "ruby_memcheck", platform: platforms
-end
+gem "onigmo", platforms: :ruby
diff --git a/Gemfile.lock b/Gemfile.lock
@@ -13,6 +13,7 @@ GEM
     nokogiri (1.16.5)
       mini_portile2 (~> 2.8.2)
       racc (~> 1.4)
+    onigmo (0.1.0)
     parser (3.3.1.0)
       ast (~> 2.4.1)
       racc
@@ -36,6 +37,7 @@ PLATFORMS
 DEPENDENCIES
   benchmark-ips
   ffi
+  onigmo
   parser
   prism!
   rake

diff --git a/config.yml b/config.yml
@@ -206,8 +206,8 @@ errors:
   - PARAMETER_SPLAT_MULTI
   - PARAMETER_STAR
   - PARAMETER_UNEXPECTED_FWD
-  - PARAMETER_WILD_LOOSE_COMMA
   - PARAMETER_UNEXPECTED_NO_KW
+  - PARAMETER_WILD_LOOSE_COMMA
   - PATTERN_CAPTURE_DUPLICATE
   - PATTERN_EXPRESSION_AFTER_BRACKET
   - PATTERN_EXPRESSION_AFTER_COMMA
@@ -236,6 +236,7 @@ errors:
   - REGEXP_INCOMPAT_CHAR_ENCODING
   - REGEXP_INVALID_UNICODE_RANGE
   - REGEXP_NON_ESCAPED_MBC
+  - REGEXP_PARSE_ERROR
   - REGEXP_TERM
   - REGEXP_UNKNOWN_OPTIONS
   - REGEXP_UTF8_CHAR_NON_UTF8_REGEXP

diff --git a/docs/fuzzing.md b/docs/fuzzing.md
@@ -26,7 +26,7 @@ fuzz
 There are currently three fuzzing targets
 
 - `pm_serialize_parse` (parse)
-- `pm_regexp_named_capture_group_names` (regexp)
+- `pm_regexp_parse` (regexp)
 
 Respectively, fuzzing can be performed with
 

diff --git a/fuzz/regexp.c b/fuzz/regexp.c
@@ -1,8 +1,21 @@
 #include <prism.h>
 
+void
+regexp_name_callback(const pm_string_t *name, void *data) {
+    // Do nothing
+}
+
+void
+regexp_error_callback(const uint8_t *start, const uint8_t *end, const char *message, void *data) {
+    // Do nothing
+}
+
 void
 harness(const uint8_t *input, size_t size) {
-    pm_string_list_t capture_list = { 0 };
-    pm_regexp_named_capture_group_names(input, size, &capture_list, false, PM_ENCODING_UTF_8_ENTRY);
-    pm_string_list_free(&capture_list);
+    pm_parser_t parser;
+    pm_parser_init(&parser, input, size, NULL);
+
+    pm_regexp_parse(&parser, input, size, regexp_name_callback, NULL, regexp_error_callback, NULL);
+
+    pm_parser_free(&parser);
 }
diff --git a/gemfiles/2.7/Gemfile b/gemfiles/2.7/Gemfile
@@ -7,8 +7,9 @@ ruby "~> 2.7.0"
 gemspec path: "../.."
 
 gem "ffi"
+gem "onigmo", platforms: :ruby
 gem "parser"
-gem "rake"
 gem "rake-compiler"
+gem "rake"
 gem "rbs"
 gem "test-unit"
diff --git a/gemfiles/2.7/Gemfile.lock b/gemfiles/2.7/Gemfile.lock
@@ -8,6 +8,7 @@ GEM
   specs:
     ast (2.4.2)
     ffi (1.16.3)
+    onigmo (0.1.0)
     parser (3.3.2.0)
       ast (~> 2.4.1)
       racc
@@ -25,6 +26,7 @@ PLATFORMS
 
 DEPENDENCIES
   ffi
+  onigmo
   parser
   prism!
   rake

diff --git a/gemfiles/3.0/Gemfile b/gemfiles/3.0/Gemfile
@@ -7,9 +7,10 @@ ruby "~> 3.0.0"
 gemspec path: "../.."
 
 gem "ffi"
+gem "onigmo", platforms: :ruby
 gem "parser"
-gem "rake"
 gem "rake-compiler"
+gem "rake"
 gem "rbs"
 gem "ruby_memcheck"
 gem "test-unit"
diff --git a/gemfiles/3.0/Gemfile.lock b/gemfiles/3.0/Gemfile.lock
@@ -13,6 +13,7 @@ GEM
     nokogiri (1.16.5)
       mini_portile2 (~> 2.8.2)
       racc (~> 1.4)
+    onigmo (0.1.0)
     parser (3.3.2.0)
       ast (~> 2.4.1)
       racc
@@ -33,6 +34,7 @@ PLATFORMS
 
 DEPENDENCIES
   ffi
+  onigmo
   parser
   prism!
   rake

diff --git a/gemfiles/3.1/Gemfile b/gemfiles/3.1/Gemfile
@@ -7,9 +7,10 @@ ruby "~> 3.1.0"
 gemspec path: "../.."
 
 gem "ffi"
+gem "onigmo", platforms: :ruby
 gem "parser"
-gem "rake"
 gem "rake-compiler"
+gem "rake"
 gem "rbs"
 gem "ruby_memcheck"
 gem "test-unit"
diff --git a/gemfiles/3.1/Gemfile.lock b/gemfiles/3.1/Gemfile.lock
@@ -13,6 +13,7 @@ GEM
     nokogiri (1.16.5)
       mini_portile2 (~> 2.8.2)
       racc (~> 1.4)
+    onigmo (0.1.0)
     parser (3.3.2.0)
       ast (~> 2.4.1)
       racc
@@ -33,6 +34,7 @@ PLATFORMS
 
 DEPENDENCIES
   ffi
+  onigmo
   parser
   prism!
   rake

diff --git a/gemfiles/3.2/Gemfile b/gemfiles/3.2/Gemfile
@@ -7,9 +7,10 @@ ruby "~> 3.2.0"
 gemspec path: "../.."
 
 gem "ffi"
+gem "onigmo", platforms: :ruby
 gem "parser"
-gem "rake"
 gem "rake-compiler"
+gem "rake"
 gem "rbs"
 gem "ruby_memcheck"
 gem "test-unit"
diff --git a/gemfiles/3.2/Gemfile.lock b/gemfiles/3.2/Gemfile.lock
@@ -13,6 +13,7 @@ GEM
     nokogiri (1.16.5)
       mini_portile2 (~> 2.8.2)
       racc (~> 1.4)
+    onigmo (0.1.0)
     parser (3.3.2.0)
       ast (~> 2.4.1)
       racc
@@ -33,6 +34,7 @@ PLATFORMS
 
 DEPENDENCIES
   ffi
+  onigmo
   parser
   prism!
   rake

diff --git a/gemfiles/3.3/Gemfile b/gemfiles/3.3/Gemfile
@@ -7,9 +7,10 @@ ruby "~> 3.3.0"
 gemspec path: "../.."
 
 gem "ffi"
+gem "onigmo", platforms: :ruby
 gem "parser"
-gem "rake"
 gem "rake-compiler"
+gem "rake"
 gem "rbs"
 gem "ruby_memcheck"
 gem "test-unit"
diff --git a/gemfiles/3.3/Gemfile.lock b/gemfiles/3.3/Gemfile.lock
@@ -13,6 +13,7 @@ GEM
     nokogiri (1.16.5)
       mini_portile2 (~> 2.8.2)
       racc (~> 1.4)
+    onigmo (0.1.0)
     parser (3.3.2.0)
       ast (~> 2.4.1)
       racc
@@ -33,6 +34,7 @@ PLATFORMS
 
 DEPENDENCIES
   ffi
+  onigmo
   parser
   prism!
   rake

diff --git a/gemfiles/3.4/Gemfile b/gemfiles/3.4/Gemfile
@@ -7,9 +7,10 @@ ruby "~> 3.4.0.dev"
 gemspec path: "../.."
 
 gem "ffi"
+gem "onigmo", platforms: :ruby
 gem "parser"
-gem "rake"
 gem "rake-compiler"
+gem "rake"
 gem "rbs"
 gem "ruby_memcheck"
 gem "test-unit"
diff --git a/gemfiles/3.4/Gemfile.lock b/gemfiles/3.4/Gemfile.lock
@@ -13,6 +13,7 @@ GEM
     nokogiri (1.16.5)
       mini_portile2 (~> 2.8.2)
       racc (~> 1.4)
+    onigmo (0.1.0)
     parser (3.3.2.0)
       ast (~> 2.4.1)
       racc
@@ -33,6 +34,7 @@ PLATFORMS
 
 DEPENDENCIES
   ffi
+  onigmo
   parser
   prism!
   rake

diff --git a/include/prism/regexp.h b/include/prism/regexp.h
@@ -10,24 +10,33 @@
 #include "prism/parser.h"
 #include "prism/encoding.h"
 #include "prism/util/pm_memchr.h"
-#include "prism/util/pm_string_list.h"
 #include "prism/util/pm_string.h"
 
 #include <stdbool.h>
 #include <stddef.h>
 #include <string.h>
 
 /**
- * Parse a regular expression and extract the names of all of the named capture
- * groups.
+ * This callback is called when a named capture group is found.
+ */
+typedef void (*pm_regexp_name_callback_t)(const pm_string_t *name, void *data);
+
+/**
+ * This callback is called when a parse error is found.
+ */
+typedef void (*pm_regexp_error_callback_t)(const uint8_t *start, const uint8_t *end, const char *message, void *data);
+
+/**
+ * Parse a regular expression.
  *
+ * @param parser The parser that is currently being used.
  * @param source The source code to parse.
  * @param size The size of the source code.
- * @param named_captures The list to add the names of the named capture groups.
- * @param encoding_changed Whether or not the encoding changed from the default.
- * @param encoding The encoding of the source code.
- * @return Whether or not the parsing was successful.
+ * @param name_callback The optional callback to call when a named capture group is found.
+ * @param name_data The optional data to pass to the name callback.
+ * @param error_callback The callback to call when a parse error is found.
+ * @param error_data The data to pass to the error callback.
  */
-PRISM_EXPORTED_FUNCTION bool pm_regexp_named_capture_group_names(const uint8_t *source, size_t size, pm_string_list_t *named_captures, bool encoding_changed, const pm_encoding_t *encoding);
+PRISM_EXPORTED_FUNCTION void pm_regexp_parse(pm_parser_t *parser, const uint8_t *source, size_t size, pm_regexp_name_callback_t name_callback, void *name_data, pm_regexp_error_callback_t error_callback, void *error_data);
 
 #endif
diff --git a/include/prism/util/pm_string_list.h b/include/prism/util/pm_string_list.h
diff --git a/prism.gemspec b/prism.gemspec
@@ -65,7 +65,6 @@ Gem::Specification.new do |spec|
     "include/prism/util/pm_newline_list.h",
     "include/prism/util/pm_strncasecmp.h",
     "include/prism/util/pm_string.h",
-    "include/prism/util/pm_string_list.h",
     "include/prism/util/pm_strpbrk.h",
     "include/prism/version.h",
     "lib/prism.rb",
@@ -148,7 +147,6 @@ Gem::Specification.new do |spec|
     "src/util/pm_list.c",
     "src/util/pm_memchr.c",
     "src/util/pm_newline_list.c",
-    "src/util/pm_string_list.c",
     "src/util/pm_string.c",
     "src/util/pm_strncasecmp.c",
     "src/util/pm_strpbrk.c"