tree-sitter-sed 0.3.0 → 0.5.0

package/CMakeLists.txt

@@ -0,0 +1,121 @@
+cmake_minimum_required(VERSION 3.13)
+
+project(
+  tree-sitter-sed
+  VERSION 0.5.0
+  DESCRIPTION "Tree-sitter grammars for POSIX sed."
+  HOMEPAGE_URL "https://github.com/konomanoasa/tree-sitter-sed"
+  LANGUAGES C
+)
+
+option(BUILD_SHARED_LIBS "Build using shared libraries" ON)
+option(TREE_SITTER_REUSE_ALLOCATOR "Reuse the Tree-sitter allocator" OFF)
+
+file(
+  STRINGS src/parser.c
+  TREE_SITTER_ABI_DEFINITION
+  REGEX "^#define LANGUAGE_VERSION [0-9]+$"
+)
+file(
+  STRINGS sed_ere/src/parser.c
+  TREE_SITTER_ERE_ABI_DEFINITION
+  REGEX "^#define LANGUAGE_VERSION [0-9]+$"
+)
+if(
+  NOT TREE_SITTER_ABI_DEFINITION MATCHES "^#define LANGUAGE_VERSION [0-9]+$"
+  OR NOT TREE_SITTER_ERE_ABI_DEFINITION MATCHES
+    "^#define LANGUAGE_VERSION [0-9]+$"
+)
+  message(FATAL_ERROR "Both parsers must define one LANGUAGE_VERSION")
+endif()
+if(NOT TREE_SITTER_ABI_DEFINITION STREQUAL TREE_SITTER_ERE_ABI_DEFINITION)
+  message(FATAL_ERROR "The sed and sed_ere parsers must use the same ABI")
+endif()
+string(
+  REGEX REPLACE "^#define LANGUAGE_VERSION " ""
+  TREE_SITTER_ABI_VERSION
+  "${TREE_SITTER_ABI_DEFINITION}"
+)
+
+include(GNUInstallDirs)
+if(CMAKE_CURRENT_SOURCE_DIR STREQUAL CMAKE_SOURCE_DIR)
+  include(CTest)
+endif()
+
+add_library(
+  tree-sitter-sed
+  src/parser.c
+  src/scanner.c
+  sed_ere/src/parser.c
+  sed_ere/src/scanner.c
+)
+set_source_files_properties(
+  src/scanner.c
+  PROPERTIES INCLUDE_DIRECTORIES "${CMAKE_CURRENT_SOURCE_DIR}/src"
+)
+set_source_files_properties(
+  sed_ere/src/scanner.c
+  PROPERTIES INCLUDE_DIRECTORIES "${CMAKE_CURRENT_SOURCE_DIR}/sed_ere/src"
+)
+target_include_directories(
+  tree-sitter-sed
+  INTERFACE
+    $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/bindings/c>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}>
+)
+target_compile_definitions(
+  tree-sitter-sed
+  PRIVATE
+    $<$<BOOL:${TREE_SITTER_REUSE_ALLOCATOR}>:TREE_SITTER_REUSE_ALLOCATOR>
+    $<$<CONFIG:Debug>:TREE_SITTER_DEBUG>
+)
+set_target_properties(
+  tree-sitter-sed
+  PROPERTIES
+    C_STANDARD 11
+    C_STANDARD_REQUIRED ON
+    DEFINE_SYMBOL ""
+    POSITION_INDEPENDENT_CODE ON
+    SOVERSION "${TREE_SITTER_ABI_VERSION}.${PROJECT_VERSION_MAJOR}"
+)
+
+if(BUILD_TESTING AND CMAKE_CURRENT_SOURCE_DIR STREQUAL CMAKE_SOURCE_DIR)
+  add_executable(tree-sitter-sed-binding-test test/binding.test.c)
+  target_link_libraries(tree-sitter-sed-binding-test PRIVATE tree-sitter-sed)
+  add_test(
+    NAME tree-sitter-sed-binding
+    COMMAND tree-sitter-sed-binding-test
+  )
+endif()
+
+configure_file(
+  bindings/c/tree-sitter-sed.pc.in
+  "${CMAKE_CURRENT_BINARY_DIR}/tree-sitter-sed.pc"
+  @ONLY
+)
+
+install(
+  TARGETS tree-sitter-sed
+  ARCHIVE DESTINATION "${CMAKE_INSTALL_LIBDIR}"
+  LIBRARY DESTINATION "${CMAKE_INSTALL_LIBDIR}"
+  RUNTIME DESTINATION "${CMAKE_INSTALL_BINDIR}"
+)
+install(
+  DIRECTORY bindings/c/tree_sitter
+  DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}"
+  FILES_MATCHING PATTERN "*.h"
+)
+install(
+  FILES "${CMAKE_CURRENT_BINARY_DIR}/tree-sitter-sed.pc"
+  DESTINATION "${CMAKE_INSTALL_LIBDIR}/pkgconfig"
+)
+install(
+  DIRECTORY queries/
+  DESTINATION "${CMAKE_INSTALL_DATADIR}/tree-sitter/queries/sed"
+  FILES_MATCHING PATTERN "*.scm"
+)
+install(
+  DIRECTORY sed_ere/queries/
+  DESTINATION "${CMAKE_INSTALL_DATADIR}/tree-sitter/queries/sed_ere"
+  FILES_MATCHING PATTERN "*.scm"
+)

package/Makefile

@@ -0,0 +1,145 @@
+LANGUAGE_NAME := tree-sitter-sed
+HOMEPAGE_URL := https://github.com/konomanoasa/tree-sitter-sed
+VERSION := 0.5.0
+DESCRIPTION := Tree-sitter grammars for POSIX sed.
+
+PREFIX ?= /usr/local
+DATADIR ?= $(PREFIX)/share
+INCLUDEDIR ?= $(PREFIX)/include
+LIBDIR ?= $(PREFIX)/lib
+BINDIR ?= $(PREFIX)/bin
+PCLIBDIR ?= $(LIBDIR)/pkgconfig
+
+SOURCES := src/parser.c src/scanner.c \
+  sed_ere/src/parser.c sed_ere/src/scanner.c
+OBJECTS := $(SOURCES:.c=.o)
+MAKE_BUILD_DIR := build/make
+STATIC_LIBRARY := lib$(LANGUAGE_NAME).a
+PKG_CONFIG := $(LANGUAGE_NAME).pc
+TEST := $(MAKE_BUILD_DIR)/binding.test
+
+ARFLAGS := rcs
+override CFLAGS += -std=c11 -fPIC
+
+SONAME_MAJOR := $(shell sed -n 's/\#define LANGUAGE_VERSION //p' src/parser.c)
+SED_ERE_SONAME_MAJOR := $(shell sed -n 's/\#define LANGUAGE_VERSION //p' sed_ere/src/parser.c)
+SONAME_MINOR := $(word 1,$(subst ., ,$(VERSION)))
+
+ifeq ($(SONAME_MAJOR),)
+$(error Both parsers must define LANGUAGE_VERSION)
+endif
+ifneq ($(SONAME_MAJOR),$(SED_ERE_SONAME_MAJOR))
+$(error The sed and sed_ere parsers must use the same ABI)
+endif
+
+MACHINE := $(shell $(CC) -dumpmachine)
+
+ifneq ($(findstring darwin,$(MACHINE)),)
+SOEXT := dylib
+SOEXTVER_MAJOR := $(SONAME_MAJOR).$(SOEXT)
+SOEXTVER := $(SONAME_MAJOR).$(SONAME_MINOR).$(SOEXT)
+LINKSHARED := -dynamiclib -Wl,-install_name,$(LIBDIR)/lib$(LANGUAGE_NAME).$(SOEXTVER)
+else ifneq ($(findstring mingw32,$(MACHINE)),)
+SOEXT := dll
+LINKSHARED := -s -shared -Wl,--out-implib,lib$(LANGUAGE_NAME).dll.a
+TEST := $(MAKE_BUILD_DIR)/binding.test.exe
+else
+SOEXT := so
+SOEXTVER_MAJOR := $(SOEXT).$(SONAME_MAJOR)
+SOEXTVER := $(SOEXT).$(SONAME_MAJOR).$(SONAME_MINOR)
+LINKSHARED := -shared -Wl,-soname,lib$(LANGUAGE_NAME).$(SOEXTVER)
+ifneq ($(filter $(shell uname),FreeBSD NetBSD DragonFly),)
+PCLIBDIR := $(PREFIX)/libdata/pkgconfig
+endif
+endif
+
+SHARED_LIBRARY := lib$(LANGUAGE_NAME).$(SOEXT)
+
+all: $(STATIC_LIBRARY) $(SHARED_LIBRARY) $(PKG_CONFIG)
+
+$(STATIC_LIBRARY): $(OBJECTS)
+	$(AR) $(ARFLAGS) $@ $^
+
+$(SHARED_LIBRARY): $(OBJECTS)
+	$(CC) $(LDFLAGS) $(LINKSHARED) $^ $(LDLIBS) -o $@
+ifneq ($(STRIP),)
+	$(STRIP) $@
+endif
+
+$(PKG_CONFIG): bindings/c/$(LANGUAGE_NAME).pc.in FORCE
+	sed -e 's|@PROJECT_VERSION@|$(VERSION)|' \
+		-e 's|@CMAKE_INSTALL_LIBDIR@|$(LIBDIR:$(PREFIX)/%=%)|' \
+		-e 's|@CMAKE_INSTALL_INCLUDEDIR@|$(INCLUDEDIR:$(PREFIX)/%=%)|' \
+		-e 's|@PROJECT_DESCRIPTION@|$(DESCRIPTION)|' \
+		-e 's|@PROJECT_HOMEPAGE_URL@|$(HOMEPAGE_URL)|' \
+		-e 's|@CMAKE_INSTALL_PREFIX@|$(PREFIX)|' $< > $@
+
+src/parser.o: src/tree_sitter/parser.h
+src/scanner.o: common/scanner.h src/tree_sitter/alloc.h src/tree_sitter/parser.h
+src/scanner.o: CPPFLAGS += -Isrc
+sed_ere/src/parser.o: sed_ere/src/tree_sitter/parser.h
+sed_ere/src/scanner.o: common/scanner.h sed_ere/src/tree_sitter/alloc.h \
+  sed_ere/src/tree_sitter/parser.h
+sed_ere/src/scanner.o: CPPFLAGS += -Ised_ere/src
+
+$(MAKE_BUILD_DIR):
+	mkdir -p $@
+
+$(TEST): test/binding.test.c $(STATIC_LIBRARY) | $(MAKE_BUILD_DIR)
+	$(CC) $(CPPFLAGS) $(CFLAGS) -Ibindings/c $< $(STATIC_LIBRARY) $(LDFLAGS) $(LDLIBS) -o $@
+
+test: $(TEST)
+	./$(TEST)
+
+install: all
+	install -d '$(DESTDIR)$(DATADIR)'/tree-sitter/queries/sed \
+		'$(DESTDIR)$(DATADIR)'/tree-sitter/queries/sed_ere \
+		'$(DESTDIR)$(INCLUDEDIR)'/tree_sitter \
+		'$(DESTDIR)$(PCLIBDIR)' \
+		'$(DESTDIR)$(LIBDIR)'
+	install -m644 bindings/c/tree_sitter/$(LANGUAGE_NAME).h \
+		'$(DESTDIR)$(INCLUDEDIR)'/tree_sitter/$(LANGUAGE_NAME).h
+	install -m644 $(PKG_CONFIG) '$(DESTDIR)$(PCLIBDIR)'/$(PKG_CONFIG)
+	install -m644 $(STATIC_LIBRARY) '$(DESTDIR)$(LIBDIR)'/$(STATIC_LIBRARY)
+ifneq ($(findstring mingw32,$(MACHINE)),)
+	install -d '$(DESTDIR)$(BINDIR)'
+	install -m755 $(SHARED_LIBRARY) '$(DESTDIR)$(BINDIR)'/$(SHARED_LIBRARY)
+	install -m644 lib$(LANGUAGE_NAME).dll.a \
+		'$(DESTDIR)$(LIBDIR)'/lib$(LANGUAGE_NAME).dll.a
+else
+	install -m755 $(SHARED_LIBRARY) \
+		'$(DESTDIR)$(LIBDIR)'/lib$(LANGUAGE_NAME).$(SOEXTVER)
+	cd '$(DESTDIR)$(LIBDIR)' && \
+		ln -sf lib$(LANGUAGE_NAME).$(SOEXTVER) \
+			lib$(LANGUAGE_NAME).$(SOEXTVER_MAJOR)
+	cd '$(DESTDIR)$(LIBDIR)' && \
+		ln -sf lib$(LANGUAGE_NAME).$(SOEXTVER_MAJOR) $(SHARED_LIBRARY)
+endif
+	install -m644 queries/*.scm \
+		'$(DESTDIR)$(DATADIR)'/tree-sitter/queries/sed
+	install -m644 sed_ere/queries/*.scm \
+		'$(DESTDIR)$(DATADIR)'/tree-sitter/queries/sed_ere
+
+uninstall:
+	$(RM) '$(DESTDIR)$(LIBDIR)'/$(STATIC_LIBRARY) \
+		'$(DESTDIR)$(INCLUDEDIR)'/tree_sitter/$(LANGUAGE_NAME).h \
+		'$(DESTDIR)$(PCLIBDIR)'/$(PKG_CONFIG)
+ifneq ($(findstring mingw32,$(MACHINE)),)
+	$(RM) '$(DESTDIR)$(BINDIR)'/$(SHARED_LIBRARY) \
+		'$(DESTDIR)$(LIBDIR)'/lib$(LANGUAGE_NAME).dll.a
+else
+	$(RM) '$(DESTDIR)$(LIBDIR)'/lib$(LANGUAGE_NAME).$(SOEXTVER) \
+		'$(DESTDIR)$(LIBDIR)'/lib$(LANGUAGE_NAME).$(SOEXTVER_MAJOR) \
+		'$(DESTDIR)$(LIBDIR)'/$(SHARED_LIBRARY)
+endif
+	$(RM) -r '$(DESTDIR)$(DATADIR)'/tree-sitter/queries/sed \
+		'$(DESTDIR)$(DATADIR)'/tree-sitter/queries/sed_ere
+
+clean:
+	$(RM) $(OBJECTS) $(STATIC_LIBRARY) $(SHARED_LIBRARY) \
+		lib$(LANGUAGE_NAME).dll.a $(PKG_CONFIG)
+	$(RM) -r $(MAKE_BUILD_DIR)
+
+FORCE:
+
+.PHONY: all clean FORCE install test uninstall

package/README.md

@@ -14,7 +14,12 @@ npm install tree-sitter-sed
 
 ## Grammars
 
-This repository contains two grammars. You can find [`sed_ere`](sed_ere) here.
+This repository contains the following two grammars.
+
+| Grammar   | Regexp | C function              |
+| --------- | ------ | ----------------------- |
+| `sed`     | BRE    | `tree_sitter_sed()`     |
+| `sed_ere` | ERE    | `tree_sitter_sed_ere()` |
 
 ## Specifications
 

package/bindings/c/tree-sitter-sed.pc.in

@@ -0,0 +1,10 @@
+prefix=@CMAKE_INSTALL_PREFIX@
+libdir=${prefix}/@CMAKE_INSTALL_LIBDIR@
+includedir=${prefix}/@CMAKE_INSTALL_INCLUDEDIR@
+
+Name: tree-sitter-sed
+Description: @PROJECT_DESCRIPTION@
+URL: @PROJECT_HOMEPAGE_URL@
+Version: @PROJECT_VERSION@
+Libs: -L${libdir} -ltree-sitter-sed
+Cflags: -I${includedir}

package/bindings/c/tree_sitter/tree-sitter-sed.h

@@ -0,0 +1,17 @@
+#ifndef TREE_SITTER_SED_H_
+#define TREE_SITTER_SED_H_
+
+typedef struct TSLanguage TSLanguage;
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+const TSLanguage *tree_sitter_sed(void);
+const TSLanguage *tree_sitter_sed_ere(void);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

package/common/dsl.js

@@ -2,6 +2,28 @@ function issueRuleName(id) {
   return `_${id}_issue`;
 }
 
+function outcomeRuleName(id) {
+  return `_${id}_outcome`;
+}
+
+function reasonRuleName(id) {
+  return `_${id}_reason`;
+}
+
+function defineIssueRules(definitions) {
+  const rules = {};
+
+  for (const definition of definitions) {
+    const { outcome, reason, rule } = definition;
+    const id = definition.id ?? reason;
+    rules[reasonRuleName(id)] = rule;
+    rules[outcomeRuleName(id)] = ($) => alias($[reasonRuleName(id)], $[reason]);
+    rules[issueRuleName(id)] = ($) => alias($[outcomeRuleName(id)], $[outcome]);
+  }
+
+  return rules;
+}
+
 function namedExternal($, external, name) {
   return alias(external, $[name]);
 }
@@ -14,4 +36,4 @@ function issueNode($, id) {
   return alias($[issueRuleName(id)], $.syntax_issue);
 }
 
-module.exports = { issueField, issueNode, issueRuleName, namedExternal };
+module.exports = { defineIssueRules, issueField, issueNode, namedExternal };

package/common/grammar.js

@@ -1,33 +1,11 @@
 const {
+  defineIssueRules,
   issueField,
   issueNode,
-  issueRuleName,
   namedExternal,
 } = require("./dsl");
 const regularExpressionRules = require("./regex");
 
-function outcomeRuleName(id) {
-  return `_${id}_outcome`;
-}
-
-function reasonRuleName(id) {
-  return `_${id}_reason`;
-}
-
-function defineIssueRules(definitions) {
-  const rules = {};
-
-  for (const definition of definitions) {
-    const { outcome, reason, rule } = definition;
-    const id = definition.id ?? reason;
-    rules[reasonRuleName(id)] = rule;
-    rules[outcomeRuleName(id)] = ($) => alias($[reasonRuleName(id)], $[reason]);
-    rules[issueRuleName(id)] = ($) => alias($[outcomeRuleName(id)], $[outcome]);
-  }
-
-  return rules;
-}
-
 function missingAtEndOrBoundary($, reason) {
   return choice(
     issueField($, reason),
@@ -573,29 +551,17 @@ function operandRules(mode) {
   }
 
   return {
-    _substitute_function_without_write: ($) =>
+    _substitute_function: ($) =>
       choice(
         seq(
           $._complete_substitute_function,
           optional(
-            field(
-              "flags",
-              alias($._substitution_flags_without_write, $.substitution_flags),
-            ),
+            field("flags", alias($._substitution_flags, $.substitution_flags)),
           ),
         ),
         $._incomplete_substitute_function,
       ),
 
-    _substitute_function_with_write: ($) =>
-      seq(
-        $._complete_substitute_function,
-        field(
-          "flags",
-          alias($._substitution_flags_with_write, $.substitution_flags),
-        ),
-      ),
-
     _complete_substitute_function: ($) =>
       seq(
         functionVerb($, "s"),
@@ -681,6 +647,12 @@ function operandRules(mode) {
     escaped_newline: ($) =>
       namedExternal($, $._replacement_escaped_newline, "escaped_newline_token"),
 
+    _substitution_flags: ($) =>
+      choice(
+        $._substitution_flags_without_write,
+        seq(optional($._substitution_flags_without_write), $.write_flag),
+      ),
+
     _substitution_flags_without_write: ($) =>
       choice(
         substitutionFlagChoice($),
@@ -689,9 +661,6 @@ function operandRules(mode) {
         ),
       ),
 
-    _substitution_flags_with_write: ($) =>
-      seq(optional($._substitution_flags_without_write), $.write_flag),
-
     occurrence_flag: () => /[[:digit:]]+/,
 
     global_flag: () => "g",
@@ -706,8 +675,7 @@ function operandRules(mode) {
         choice(
           seq(
             $._flag_after_write_marker,
-            issueField($, "flag_after_write_flag"),
-            repeat(postWriteSubstitutionFlagChoice($)),
+            repeat1(postWriteSubstitutionFlagChoice($)),
             $._blanks,
             choice(
               field("wfile", alias($._substitution_wfile, $.wfile)),
@@ -977,39 +945,19 @@ function editingCommandRules() {
 
     _line_terminated_editing_command: ($) =>
       seq(
-        choice(
-          $._line_terminated_regular_editing_command_body,
-          $._line_terminated_substitute_editing_command_body,
-        ),
+        addressedForms($, lineTerminated, "line_terminated"),
         optional(issueField($, "unexpected_command_text")),
       ),
 
     _recovered_line_terminated_editing_command: ($) =>
-      choice(
-        prec.right(
-          seq(
-            $._line_terminated_regular_editing_command_body,
-            optional(issueField($, "unexpected_command_text")),
-            issueField($, "forbidden_command_separator"),
-            optional($._blanks),
-          ),
-        ),
-        prec.right(
-          seq(
-            $._line_terminated_substitute_editing_command_body,
-            optional(issueField($, "unexpected_command_text")),
-            issueField($, "command_after_write_flag"),
-            optional($._blanks),
-          ),
+      prec.right(
+        seq(
+          $._line_terminated_editing_command,
+          issueField($, "forbidden_command_separator"),
+          optional($._blanks),
         ),
       ),
 
-    _line_terminated_regular_editing_command_body: ($) =>
-      addressedForms($, lineTerminated, "line_terminated"),
-
-    _line_terminated_substitute_editing_command_body: ($) =>
-      addressedFunction($, $._line_terminated_substitute_function),
-
     _recovered_editing_command: ($) =>
       choice(
         addressedFunction($, $._unknown_function),
@@ -1056,10 +1004,7 @@ function editingCommandRules() {
     _missing_function: ($) => missingAtEndOrBoundary($, "missing_function"),
 
     _chainable_substitute_function: ($) =>
-      alias($._substitute_function_without_write, $.substitute_function),
-
-    _line_terminated_substitute_function: ($) =>
-      alias($._substitute_function_with_write, $.substitute_function),
+      alias($._substitute_function, $.substitute_function),
 
     negation: ($) =>
       seq(
@@ -1272,16 +1217,6 @@ function issueDefinitions(mode) {
       outcome: "undefined_syntax",
       rule: ($) => $._translate_nonportable_escape,
     },
-    {
-      reason: "command_after_write_flag",
-      outcome: "undefined_syntax",
-      rule: () => token.immediate(";"),
-    },
-    {
-      reason: "flag_after_write_flag",
-      outcome: "undefined_syntax",
-      rule: ($) => postWriteSubstitutionFlagChoice($),
-    },
     {
       reason: "equivalence_class_range_start",
       outcome: "unspecified_syntax",

package/common/regex.js

@@ -340,6 +340,14 @@ function breDuplicationSymbol($) {
   );
 }
 
+function misplacedBreDuplSymbol($, reason) {
+  return choice(
+    seq(issueField($, reason), field("operator", $.bre_dupl_symbol)),
+    issueField($, "malformed_interval"),
+    issueField($, "incomplete_interval"),
+  );
+}
+
 function breRules() {
   return {
     bre_extension_escape: ($) =>
@@ -483,24 +491,10 @@ function breRules() {
       intervalExpression($, "back_open_brace", "back_close_brace"),
 
     leading_bre_dupl_symbol: ($) =>
-      choice(
-        seq(
-          issueField($, "leading_duplication_symbol"),
-          field("operator", $.bre_dupl_symbol),
-        ),
-        issueField($, "malformed_interval"),
-        issueField($, "incomplete_interval"),
-      ),
+      misplacedBreDuplSymbol($, "leading_duplication_symbol"),
 
     adjacent_bre_dupl_symbol: ($) =>
-      choice(
-        seq(
-          issueField($, "adjacent_duplication_symbol"),
-          field("operator", $.bre_dupl_symbol),
-        ),
-        issueField($, "malformed_interval"),
-        issueField($, "incomplete_interval"),
-      ),
+      misplacedBreDuplSymbol($, "adjacent_duplication_symbol"),
   };
 }