@konomanoasa/tree-sitter-awk 0.2.0 → 0.4.0

package/CMakeLists.txt

@@ -0,0 +1,90 @@
+cmake_minimum_required(VERSION 3.13)
+
+project(
+  tree-sitter-awk
+  VERSION 0.4.0
+  DESCRIPTION "Tree-sitter grammar for POSIX awk."
+  HOMEPAGE_URL "https://github.com/konomanoasa/tree-sitter-awk"
+  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]+$"
+)
+if(NOT TREE_SITTER_ABI_DEFINITION MATCHES "^#define LANGUAGE_VERSION [0-9]+$")
+  message(FATAL_ERROR "The parser must define one LANGUAGE_VERSION")
+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-awk src/parser.c src/scanner.c)
+target_include_directories(
+  tree-sitter-awk
+  INTERFACE
+    $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/bindings/c>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}>
+)
+target_compile_definitions(
+  tree-sitter-awk
+  PRIVATE
+    $<$<BOOL:${TREE_SITTER_REUSE_ALLOCATOR}>:TREE_SITTER_REUSE_ALLOCATOR>
+    $<$<CONFIG:Debug>:TREE_SITTER_DEBUG>
+)
+set_target_properties(
+  tree-sitter-awk
+  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-awk-binding-test test/binding.test.c)
+  target_link_libraries(tree-sitter-awk-binding-test PRIVATE tree-sitter-awk)
+  add_test(
+    NAME tree-sitter-awk-binding
+    COMMAND tree-sitter-awk-binding-test
+  )
+endif()
+
+configure_file(
+  bindings/c/tree-sitter-awk.pc.in
+  "${CMAKE_CURRENT_BINARY_DIR}/tree-sitter-awk.pc"
+  @ONLY
+)
+
+install(
+  TARGETS tree-sitter-awk
+  ARCHIVE DESTINATION "${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-awk.pc"
+  DESTINATION "${CMAKE_INSTALL_LIBDIR}/pkgconfig"
+)
+install(
+  DIRECTORY queries/
+  DESTINATION "${CMAKE_INSTALL_DATADIR}/tree-sitter/queries/awk"
+  FILES_MATCHING PATTERN "*.scm"
+)

package/Makefile

@@ -0,0 +1,131 @@
+LANGUAGE_NAME := tree-sitter-awk
+HOMEPAGE_URL := https://github.com/konomanoasa/tree-sitter-awk
+VERSION := 0.4.0
+DESCRIPTION := Tree-sitter grammar for POSIX awk.
+
+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
+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)
+SONAME_MINOR := $(word 1,$(subst ., ,$(VERSION)))
+
+ifeq ($(SONAME_MAJOR),)
+$(error The parser must define LANGUAGE_VERSION)
+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)|' $< > $@
+
+src/parser.o: src/tree_sitter/parser.h
+src/scanner.o: src/tree_sitter/alloc.h src/tree_sitter/parser.h
+
+$(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/awk \
+		'$(DESTDIR)$(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/awk
+
+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/awk
+
+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/bindings/c/tree-sitter-awk.pc.in

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

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

@@ -0,0 +1,16 @@
+#ifndef TREE_SITTER_AWK_H_
+#define TREE_SITTER_AWK_H_
+
+typedef struct TSLanguage TSLanguage;
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+const TSLanguage *tree_sitter_awk(void);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

package/grammar.js

@@ -30,9 +30,6 @@ const optionalNewlineLayout = ($) =>
     optional(seq($.newline_opt, repeat($.line_continuation))),
   );
 
-// A statement owns only the line continuations interior to its own trailing
-// newline_opt. Continuations after that gap belong to the surrounding list,
-// action, or item, keeping every continuation a single owner.
 const statementEndLayout = ($) =>
   optional(seq(repeat($.line_continuation), $.newline_opt));
 
@@ -232,10 +229,7 @@ const actionBoundaryBody = ($) =>
   actionBody(
     $,
     alias(
-      statementListWithTail(
-        $,
-        alias($._action_body_boundary_control, $.unterminated_statement),
-      ),
+      statementListWithTail($, actionBoundaryControlBody($)),
       $.unterminated_statement_list,
     ),
   );
@@ -286,9 +280,6 @@ const aliasedAnyTier = ($, context, tier) =>
 const expressionTargetGuard = ($, context) =>
   context.input ? $._expression_target_guard : $._print_expression_target_guard;
 
-const continuedPresentTierExpression = ($, context, name, tier) =>
-  continuedExpression($, name, aliasedAnyTier($, context, tier));
-
 const nonUnaryAtom = ($, context) => {
   const atoms = [
     $._parenthesized_expression,
@@ -306,15 +297,6 @@ const nonUnaryAtom = ($, context) => {
   return choice(...atoms);
 };
 
-const prefixUpdateExpression = ($) =>
-  prec.right(
-    PRECEDENCE.prefixUpdate,
-    seq(
-      field("operator", choice($.incr, $.decr)),
-      continuedExpression($, "operand", $.lvalue),
-    ),
-  );
-
 const tieredExpressionRules = (context) => {
   const rules = {};
   const unary = (tier) => classTierName(context, "unary", tier);
@@ -370,6 +352,7 @@ const tieredExpressionRules = (context) => {
   };
 
   const addLeftAssociativeTier = (tier, nextTier, operator, precedence) => {
+    addAnyTier(nextTier);
     const tail = `_${context.prefix}_${tier}_tail`;
     rules[tail] = ($) => seq(operator($), requiredTier($, "right", nextTier));
     for (const classification of ["unary", "non_unary"]) {
@@ -461,7 +444,7 @@ const tieredExpressionRules = (context) => {
         requiredAfterOptionalNewline(
           $,
           expressionTargetGuard($, context),
-          continuedPresentTierExpression($, context, "right", nextTier),
+          continuedExpression($, "right", aliasedAnyTier($, context, nextTier)),
           requiredTier($, "right", nextTier),
         ),
       );
@@ -574,7 +557,6 @@ const tieredExpressionRules = (context) => {
       );
   }
 
-  addAnyTier("multiplicative");
   addLeftAssociativeTier(
     "additive",
     "multiplicative",
@@ -582,7 +564,6 @@ const tieredExpressionRules = (context) => {
     PRECEDENCE.additive,
   );
 
-  addAnyTier("unary");
   addLeftAssociativeTier(
     "multiplicative",
     "unary",
@@ -1231,7 +1212,14 @@ module.exports = grammar({
 
     non_unary_expr: ($) => $._normal_non_unary_assignment_expr,
 
-    _prefix_update_expr: ($) => prefixUpdateExpression($),
+    _prefix_update_expr: ($) =>
+      prec.right(
+        PRECEDENCE.prefixUpdate,
+        seq(
+          field("operator", choice($.incr, $.decr)),
+          continuedExpression($, "operand", $.lvalue),
+        ),
+      ),
 
     ...normalExpressionRules,
 
@@ -1677,6 +1665,8 @@ module.exports = grammar({
 
     // Tree-sitter rejects the POSIX bracket spelling for these delimiter
     // characters, so the exclusion sets use its hexadecimal regex escape.
+    // The comment extra competes in ERE lexical states, so every ERE token
+    // whose set contains "#" carries a lexical precedence above the comment.
     _ordinary_character: () =>
       token.immediate(prec(1, /[^.\x5B\x5C*^$+?{|}()/\n]/)),
 
@@ -1700,19 +1690,19 @@ module.exports = grammar({
     _ere_bracket_close_character: () => token.immediate(prec(1, "]")),
 
     _ere_compound_nonmeta_character: () =>
-      token.immediate(/[^\x2D\x2F\x5C\x5D\x5E\n]/),
+      token.immediate(prec(1, /[^\x2D\x2F\x5C\x5D\x5E\n]/)),
 
     _ere_compound_meta_character: () => token.immediate(/[\x2D\x5D\x5E]/),
 
     _ere_class_name_spelling: () => token.immediate(/[A-Za-z][A-Za-z0-9]*/),
 
-    _ere_named_escape_character: () => token.immediate(/[abfnrtv]/),
+    _ere_named_escape_character: () => token.immediate(prec(2, /[abfnrtv]/)),
 
     _ere_quoted_escape_character: () =>
-      token.immediate(/[().*+?{}|^$\x5B\x5C\x5D]/),
+      token.immediate(prec(2, /[().*+?{}|^$\x5B\x5C\x5D]/)),
 
     _ere_undefined_escape_character: () =>
-      token.immediate(prec(-1, /[^0-7\x2F\x5C\n]/)),
+      token.immediate(prec(1, /[^0-7\x2F\x5C\n]/)),
 
     newline_opt: ($) => rawNewlines($),
 

package/package.json

@@ -1,31 +1,35 @@
 {
   "name": "@konomanoasa/tree-sitter-awk",
-  "version": "0.2.0",
+  "version": "0.4.0",
   "description": "Tree-sitter grammar for POSIX awk.",
   "license": "MIT",
   "repository": "git+https://github.com/konomanoasa/tree-sitter-awk.git",
   "files": [
+    "bindings/c",
+    "CMakeLists.txt",
     "grammar.js",
+    "Makefile",
     "queries/*.scm",
     "src/*.c",
     "src/*.json",
     "src/tree_sitter/*.h",
+    "test/binding.test.c",
     "tree-sitter.json"
   ],
   "publishConfig": {
     "access": "public"
   },
   "scripts": {
-    "check": "biome check . && node scripts/check-scanner.js && node scripts/check-generated.js && npm test",
-    "format": "biome check --write . && node scripts/check-scanner.js --write",
+    "check": "biome check . && node scripts/check-c.js && node scripts/check-generated.js && npm test",
+    "format": "biome check --write . && node scripts/check-c.js --write",
     "generate": "npm run tree-sitter -- generate",
     "test": "npm run tree-sitter -- test --rebuild && node --test",
     "test:fuzz": "npm run tree-sitter -- fuzz --rebuild --iterations 100 --edits 5",
     "tree-sitter": "node scripts/tree-sitter.js"
   },
   "devDependencies": {
-    "@biomejs/biome": "2.5.5",
-    "tree-sitter-cli": "0.26.11"
+    "@biomejs/biome": "2.5.11",
+    "tree-sitter-cli": "0.26.13"
   },
   "allowScripts": {
     "tree-sitter-cli": true

package/src/grammar.json

@@ -10122,8 +10122,12 @@
     "_ere_compound_nonmeta_character": {
       "type": "IMMEDIATE_TOKEN",
       "content": {
-        "type": "PATTERN",
-        "value": "[^\\x2D\\x2F\\x5C\\x5D\\x5E\\n]"
+        "type": "PREC",
+        "value": 1,
+        "content": {
+          "type": "PATTERN",
+          "value": "[^\\x2D\\x2F\\x5C\\x5D\\x5E\\n]"
+        }
       }
     },
     "_ere_compound_meta_character": {
@@ -10143,22 +10147,30 @@
     "_ere_named_escape_character": {
       "type": "IMMEDIATE_TOKEN",
       "content": {
-        "type": "PATTERN",
-        "value": "[abfnrtv]"
+        "type": "PREC",
+        "value": 2,
+        "content": {
+          "type": "PATTERN",
+          "value": "[abfnrtv]"
+        }
       }
     },
     "_ere_quoted_escape_character": {
       "type": "IMMEDIATE_TOKEN",
       "content": {
-        "type": "PATTERN",
-        "value": "[().*+?{}|^$\\x5B\\x5C\\x5D]"
+        "type": "PREC",
+        "value": 2,
+        "content": {
+          "type": "PATTERN",
+          "value": "[().*+?{}|^$\\x5B\\x5C\\x5D]"
+        }
       }
     },
     "_ere_undefined_escape_character": {
       "type": "IMMEDIATE_TOKEN",
       "content": {
         "type": "PREC",
-        "value": -1,
+        "value": 1,
         "content": {
           "type": "PATTERN",
           "value": "[^0-7\\x2F\\x5C\\n]"