feat(c): add support for parsing #include statements #24

This commit adds support for parsing #include statements in C code. It extracts the included file name and adds it to the list of imports in the code container.

Author: phodal

chapi-ast-c/src/main/antlr/C.g4

@@ -43,14 +43,6 @@ MultiLineMacro
     : '#' (~[\n]*? '\\' '\r'? '\n')+ ~ [\n]+ -> channel (HIDDEN)
     ;
 
-include
-    : 'include'
-    ;
-
-includeIdentifier
-    : (Identifier | keywords) ('/' Identifier)* (Dot Identifier)?
-    ;
-
 primaryExpression
     : Identifier
     | Constant
@@ -486,44 +478,12 @@ asmBody
 
 logicals : logicalOrExpression (',' logicalOrExpression)* ;
 
-//Directive: '#' ~ [\n]* -> channel (HIDDEN);
-
 macroStatement
-    : '#' singleLineMacroDeclaration*
-//    : '#' include (StringLiteral | Identifier | ('<' includeIdentifier '>' ))          #includeDeclaration
-    ;
-
-singleLineMacroDeclaration
-    : include (StringLiteral | Identifier | ('<' includeIdentifier '>' ))             #includeDeclaration
-    | ('ifdef' | 'ifndef' | 'if') Identifier statement* ('#' 'else' statement*)? '#' 'endif'
-                                                                                      #ifdefDeclaration
-    | 'define' Identifier defineMacro                                                 #defineDeclaration
-//    | macroKeywords (assignmentExpression)*
-//                ('#' macroKeywords)? identifierList?                                  #macroDefineDeclaration
-    | '#'? Identifier                                                                 #macroCastDeclaration
-    | macroKeywords macroFunctionExpression?                                          #macroStatementDeclaration
-    ;
-
-defineMacro
-    : expressionStatement?                                        #macroAssignDeclaration
-    | postixCall Identifier postixCall                            #macroAliasDeclaration
-    | structOrUnionSpecifier                                      #macroStructureDeclaration
-    ;
-
-
-macroFunctionExpression
-    : Identifier
-    | Identifier '(' ( parameterTypeList| (macroType (',' macroType)*) ) ')'
-    | assignmentExpression
-    ;
-
-macroType
-    : typeQualifier? (typeKeywords | Identifier | '==' | '!=' | comparator) (Identifier | typeKeywords)* pointer?
-    | expressionStatement
+    : PreprocessorBlock                                                                #preprocessorDeclaration
     ;
 
-macroKeywords
-    :  'if' | 'undef' | 'else' | 'pragma' | 'endif' | 'ifdef' | 'ifndef' | 'elif' | 'define' | 'ifndef' | 'error'
+PreprocessorBlock
+    : '#' ~[\r\n]*
     ;
 
 labeledStatement

chapi-ast-c/src/main/antlr/Macro.g4

@@ -0,0 +1,36 @@
+grammar Macro;
+
+compilationUnit
+    : '#'*  EOF
+    ;
+
+//singleLineMacroDeclaration
+//    : include (StringLiteral | Identifier | ('<' includeIdentifier '>' ))             #includeDeclaration
+//    | ('ifdef' | 'ifndef' | 'if') Identifier statement* ('#' 'else' statement*)? '#' 'endif'
+//                                                                                      #ifdefDeclaration
+//    | 'define' Identifier defineMacro                                                 #defineDeclaration
+////    | macroKeywords (assignmentExpression)*
+////                ('#' macroKeywords)? identifierList?                                  #macroDefineDeclaration
+//    | '#'? Identifier                                                                 #macroCastDeclaration
+//    | macroKeywords macroFunctionExpression?                                          #macroStatementDeclaration
+//    ;
+//
+//defineMacro
+//    : expressionStatement?                                        #macroAssignDeclaration
+//    | postixCall Identifier postixCall                            #macroAliasDeclaration
+//    | structOrUnionSpecifier                                      #macroStructureDeclaration
+//    ;
+//
+//macroFunctionExpression
+//    : Identifier
+//    | Identifier '(' ( parameterTypeList| (macroType (',' macroType)*) ) ')'
+//    | assignmentExpression
+//    ;
+//macroType
+//    : typeQualifier? (typeKeywords | Identifier | '==' | '!=' | comparator) (Identifier | typeKeywords)* pointer?
+//    | expressionStatement
+//    ;
+//
+//macroKeywords
+//    :  'if' | 'undef' | 'else' | 'pragma' | 'endif' | 'ifdef' | 'ifndef' | 'elif' | 'define' | 'ifndef' | 'error'
+//    ;

chapi-ast-c/src/main/kotlin/chapi/ast/cast/CFullIdentListener.kt

@@ -10,17 +10,23 @@ open class CFullIdentListener(fileName: String) : CAstBaseListener() {
     private var structMap = mutableMapOf<String, CodeDataStruct>()
     private var codeContainer: CodeContainer = CodeContainer(FullName = fileName)
 
-    override fun enterIncludeDeclaration(ctx: CParser.IncludeDeclarationContext?) {
-        val identifier = ctx?.includeIdentifier()
+    private val importRegex = Regex("""#include\s+(<[^>]+>|\"[^\"]+\")""")
+    override fun enterPreprocessorDeclaration(ctx: CParser.PreprocessorDeclarationContext?) {
+        val text = ctx?.text
+        val matchResult = importRegex.find(text ?: "") ?: return
+
+        val value = matchResult.groupValues[1]
+            .removeSurrounding("\"", "\"")
+            .removeSurrounding("<", ">")
+
         val imp = CodeImport(
-            Source = identifier?.text ?: "",
-            AsName = identifier?.text ?: ""
+            Source = value,
+            AsName = value
         )
 
         codeContainer.Imports += imp
     }
 
-
     override fun enterDeclaration(ctx: CParser.DeclarationContext?) {
         val isTypeDef = ctx?.declarationSpecifier()?.any {
             it.storageClassSpecifier()?.Typedef() != null
@@ -205,10 +211,11 @@ open class CFullIdentListener(fileName: String) : CAstBaseListener() {
 
     private fun buildParameters(call: List<CParser.PostixCallContext>): List<CodeProperty> {
         return call.mapNotNull { callContext ->
-            when(callContext) {
+            when (callContext) {
                 is CParser.ArrayAccessPostfixExpressionContext -> {
                     null
                 }
+
                 is CParser.FunctionCallPostfixExpressionContext -> {
                     callContext.argumentExpressionList()?.assignmentExpression()?.map {
                         CodeProperty(

chapi-ast-c/src/test/kotlin/chapi/ast/cast/CFullIdentListenerTest.kt

@@ -266,7 +266,7 @@ typedef struct {
             """.trimIndent()
 
         val codeFile = CAnalyser().analysis(code, "helloworld.c")
-        assertEquals(codeFile.Imports.size, 7)
+        assertEquals(codeFile.Imports.size, 6)
     }
 
     @Test
@@ -627,6 +627,6 @@ typedef struct {
             """.trimIndent()
 
         val codeFile = CAnalyser().analysis(code, "helloworld.c")
-        assertEquals(codeFile.DataStructures.size, 1)
+        assertEquals(codeFile.DataStructures.size, 0)
     }
 }