update: milestone! hello world stdout!

park671 · Nov 25, 2024 · 5d677ea · 5d677ea
1 parent 2ec474e
commit 5d677ea
Show file tree

Hide file tree

Showing 20 changed files with 329 additions and 72 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -74,3 +74,11 @@ add_executable(pcc
         generator/arm64/internal_func_arm64.cpp
         generator/arm64/internal_func_arm64.h
 )
+
+add_custom_command(
+        TARGET pcc POST_BUILD
+        COMMAND ${CMAKE_COMMAND} -E copy_directory
+        ${CMAKE_SOURCE_DIR}/include
+        ${CMAKE_BINARY_DIR}/include
+        COMMENT "Copying include directory to build output"
+)
diff --git a/compiler/ast.h b/compiler/ast.h
@@ -307,13 +307,19 @@ struct AstStatementBlock {
     //}
 };
 
+enum MethodDefineType {
+    METHOD_IMPL,
+    METHOD_EXTERN,
+};
+
 struct AstMethodDefine {
+    MethodDefineType defineType;
     AstType *type;
     AstIdentity *identity;
     // (
     AstParamList *paramList;
     //)
-    AstStatementBlock *statementBlock;
+    AstStatementBlock *statementBlock;//nullable
 };
 
 struct AstMethodSeq {

diff --git a/compiler/lexer.cpp b/compiler/lexer.cpp
@@ -10,14 +10,11 @@
 
 using namespace std;
 
-//16KB
-#define BUFFER_SIZE 16384
-
 static const char *LEXER_TAG = "lexer";
 static const char *VAR_TAG = "var";
 
-static const char *keywords[] = {"if", "else", "for", "while", "return"};
-static const size_t keywordSize = 5;
+static const char *keywords[] = {"if", "else", "for", "while", "return", "extern"};
+static const size_t keywordSize = 6;
 
 static const char *types[] = {"void", "char", "int", "short", "long", "float", "double"};
 static const size_t typeSize = 7;
@@ -60,26 +57,26 @@ const char *getTokenTypeName(TokenType tokenType) {
     return "unknown";
 }
 
-static void trimString(char *buffer) {
+static void trimString(ProcessedSource *source) {
     int startIndex = -1;
     int endIndex = -1;
-    for (int i = 0; i < BUFFER_SIZE; i++) {
+    for (int i = 0; i < source->length; i++) {
         if (startIndex == -1) {
-            if (buffer[i] == ' ') {
+            if (source->line[i] == ' ') {
                 continue;
             }
             startIndex = i;
         }
-        if (buffer[i] == '\n' || buffer[i] == '\0') {
+        if (source->line[i] == '\n' || source->line[i] == '\0') {
             endIndex = i;
             break;
         }
     }
     int size = endIndex - startIndex;
     for (int i = 0; i < size; i++) {
-        buffer[i] = buffer[i + startIndex];
+        source->line[i] = source->line[i + startIndex];
     }
-    buffer[size] = '\0';
+    source->line[size] = '\0';
 }
 
 inline static bool isBoolOperator(char a) {
@@ -186,10 +183,21 @@ static Token *lexer(Token *tail, const char *buffer) {
     bool nextCharEscape = false;
     bool inCharsSession = false;
     for (int i = 0; i < length; i++) {
-
         if (inCharsSession && isEscape(buffer[i])) {
             nextCharEscape = true;
             continue;
+        } else if (nextCharEscape) {
+            if (buffer[i] == 'n') {
+                snprintf(tmp, 256, "%s\n", tmp);
+            } else if (buffer[i] == 'r') {
+                snprintf(tmp, 256, "%s\r", tmp);
+            } else if (buffer[i] == 't') {
+                snprintf(tmp, 256, "%s\t", tmp);
+            } else if (buffer[i] == 'b') {
+                snprintf(tmp, 256, "%s\b", tmp);
+            }
+            nextCharEscape = false;
+            continue;
         } else if (isQuotation(buffer[i]) && !nextCharEscape) {
             if (!inCharsSession) {
                 tail = processCompletedString(tail, tmp);
@@ -293,18 +301,16 @@ static Token *lexer(Token *tail, const char *buffer) {
     return tail;
 }
 
-Token *buildTokens(const char *sourceFilePath) {
+Token *buildTokens(ProcessedSource *source) {
     logd(LEXER_TAG, "lexical analysis...");
-    FILE *inputFile = fopen(sourceFilePath, "r");
-    char buffer[BUFFER_SIZE];
     Token *tokenHead = (Token *) (pccMalloc(LEXER_TAG, sizeof(Token)));
     tokenHead->tokenType = TOKEN_HEAD;
     Token *tokenTail = tokenHead;
-    while (fgets(buffer, BUFFER_SIZE, inputFile) != nullptr) {
-        trimString(buffer);
-        tokenTail = lexer(tokenTail, buffer);
+    while (source != nullptr) {
+        trimString(source);
+        tokenTail = lexer(tokenTail, source->line);
+        source = source->next;
     }
-    fclose(inputFile);
     return tokenHead;
 }
 

diff --git a/compiler/lexer.h b/compiler/lexer.h
@@ -6,9 +6,12 @@
 #define PCC_CC_LEXER_H
 
 #include "token.h"
+#include "preprocessor.h"
+
+extern Token *buildTokens(ProcessedSource *source);
 
-extern Token *buildTokens(const char *sourceFilePath);
 extern void printTokenStack(Token *tokens);
+
 extern void releaseLexerMemory();
 
 #endif //PCC_CC_LEXER_H
diff --git a/compiler/mir.cpp b/compiler/mir.cpp
@@ -1301,6 +1301,9 @@ void generateParam(AstParamList *astParamList, MirMethod *mirMethod) {
         //real travel ast
         mirMethodParam->paramName = curAstParamList->paramDefine->identity->name;
         mirMethodParam->sign = true;
+        if (curAstParamList->paramDefine->type->isPointer) {
+            mirMethodParam->pointer = true;
+        }
         switch (curAstParamList->paramDefine->type->primitiveType) {
             case TYPE_CHAR: {
                 mirMethodParam->integer = true;
@@ -1368,6 +1371,13 @@ void generateMethod(AstMethodDefine *astMethodDefine, MirMethod *mirMethod) {
     addMethodInfo(
             astMethodDefine->identity->name,
             type);
+    if (astMethodDefine->defineType == METHOD_EXTERN) {
+        logd(MIR_TAG, "skip extern method mir generation: %s", astMethodDefine->identity->name);
+        mirMethod->isExtern = true;
+        resetTempValIndex();
+        return;
+    }
+    mirMethod->isExtern = false;
     AstStatementSeq *astStatementSeq = astMethodDefine->statementBlock->statementSeq;
     //start mir code session
     startMirCodeSession();
@@ -1396,17 +1406,21 @@ Mir *generateMir(AstProgram *program) {
     while (astMethodSeq != nullptr) {
         //maintain mir linked-list
         mirMethod = (MirMethod *) pccMalloc(MIR_TAG, sizeof(MirMethod));
-        mirMethod->next = nullptr;
-        if (firstMirMethod == nullptr) {
-            firstMirMethod = mirMethod;
-        }
-        if (lastMirMethod != nullptr) {
-            lastMirMethod->next = mirMethod;
-        }
-        lastMirMethod = mirMethod;
-        //real travel ast
-        methodSize++;
         generateMethod(astMethodSeq->methodDefine, mirMethod);
+        if (mirMethod->isExtern) {
+            pccFree(MIR_TAG, mirMethod);
+        } else {
+            mirMethod->next = nullptr;
+            if (firstMirMethod == nullptr) {
+                firstMirMethod = mirMethod;
+            }
+            if (lastMirMethod != nullptr) {
+                lastMirMethod->next = mirMethod;
+            }
+            lastMirMethod = mirMethod;
+            //real travel ast
+            methodSize++;
+        }
         astMethodSeq = astMethodSeq->nextAstMethodSeq;
     }
     mir->mirMethod = firstMirMethod;

diff --git a/compiler/mir.h b/compiler/mir.h
@@ -36,6 +36,7 @@ enum MirType {
 struct MirCode;
 
 struct MirMethod {
+    bool isExtern = false;
     const char *label;//identity, in another word: method name.
     MirMethodParam *param;//nullable
     MirCode *code;
@@ -45,6 +46,7 @@ struct MirMethod {
 
 struct MirMethodParam {
     const char *paramName;
+    bool pointer;
     bool integer;//true-int, false-float
     bool sign;//true-signed, false-unsigned
     int byte;//1-int8, 2-int16, 4-int32, 8-int64

diff --git a/compiler/optimization.cpp b/compiler/optimization.cpp
@@ -58,7 +58,8 @@ bool replaceFutureUsedMir2(MirCode *currentMirCode, const char *identity, MirOpe
 
         if (mirCode->mirType == MIR_2) {
             Mir2 *mir2 = mirCode->mir2;
-            if (mir2->fromValue.type.primitiveType == OPERAND_IDENTITY && strcmp(mir2->fromValue.identity, identity) == 0) {
+            if (mir2->fromValue.type.primitiveType == OPERAND_IDENTITY &&
+                strcmp(mir2->fromValue.identity, identity) == 0) {
                 mir2->fromValue = *replaceWith;
                 result = true;
             }
@@ -81,18 +82,21 @@ bool replaceFutureUsedMir2(MirCode *currentMirCode, const char *identity, MirOpe
         } else if (mirCode->mirType == MIR_RET) {
             MirRet *mirRet = mirCode->mirRet;
             if (mirRet->value != nullptr) {
-                if (mirRet->value->type.primitiveType == OPERAND_IDENTITY && strcmp(mirRet->value->identity, identity) == 0) {
+                if (mirRet->value->type.primitiveType == OPERAND_IDENTITY &&
+                    strcmp(mirRet->value->identity, identity) == 0) {
                     mirRet->value = replaceWith;
                     result = true;
                 }
             }
         } else if (mirCode->mirType == MIR_CMP) {
             MirCmp *mirCmp = mirCode->mirCmp;
-            if (mirCmp->value1.type.primitiveType == OPERAND_IDENTITY && strcmp(mirCmp->value1.identity, identity) == 0) {
+            if (mirCmp->value1.type.primitiveType == OPERAND_IDENTITY &&
+                strcmp(mirCmp->value1.identity, identity) == 0) {
                 mirCmp->value1 = *replaceWith;
                 result = true;
             }
-            if (mirCmp->value2.type.primitiveType == OPERAND_IDENTITY && strcmp(mirCmp->value2.identity, identity) == 0) {
+            if (mirCmp->value2.type.primitiveType == OPERAND_IDENTITY &&
+                strcmp(mirCmp->value2.identity, identity) == 0) {
                 mirCmp->value2 = *replaceWith;
                 result = true;
             }
@@ -144,7 +148,10 @@ void printMirCode(MirMethod *mirMethod);
 void foldMir2(MirMethod *mirMethod) {
     MirCode *mirCode = mirMethod->code;
     while (mirCode != nullptr) {
-        if (mirCode->mirType == MIR_2) {
+        if (mirCode->mirType == MIR_2
+            && mirCode->mir2->op == OP_ASSIGNMENT
+            && !mirCode->mir2->fromValue.type.isPointer) {
+            //todo this is a BIG shit, will fix in future
             Mir2 *mir2 = mirCode->mir2;
             if (replaceFutureUsedMir2(mirCode, mir2->distIdentity, &mir2->fromValue)) {
                 mirMethod->code = removeMirCode(mirMethod, mirCode);