From 1e2f3c263a924458a684cf124d8d074f6c6525b5 Mon Sep 17 00:00:00 2001
From: servostar <sven.vogel123@web.de>
Date: Wed, 5 Jun 2024 15:03:23 +0200
Subject: [PATCH] fixed: lexer memory leak due to strdup

---
 src/compiler.c  |  2 ++
 src/lex/lexer.l | 10 +++++-----
 src/lex/util.c  | 26 ++++++++++++++++++++++++++
 src/lex/util.h  |  4 ++++
 4 files changed, 37 insertions(+), 5 deletions(-)

diff --git a/src/compiler.c b/src/compiler.c
index 709592e..e9b5c8d 100644
--- a/src/compiler.c
+++ b/src/compiler.c
@@ -148,6 +148,8 @@ static void build_target(ModuleFileStack *unit, const TargetConfig *target) {
 
     AST_delete_node(ast);
 
+    lex_purge_str_cache();
+
     print_file_statistics(file);
 }
 
diff --git a/src/lex/lexer.l b/src/lex/lexer.l
index cb507f6..8f65ff7 100644
--- a/src/lex/lexer.l
+++ b/src/lex/lexer.l
@@ -89,20 +89,20 @@
 "lineno" {DEBUG("\"%s\" tokenized with \'FunLineno\'", yytext); return(FunLineno);};
 "extsupport" {DEBUG("\"%s\" tokenized with \'FunExtsupport\'", yytext); return(FunExtsupport);};
 
-[0-9]+    {DEBUG("\"%s\" tokenized with \'ValInt\'", yytext); yylval.string = strdup(yytext); return(ValInt); };
-[0-9]*\.[0-9]+ {DEBUG("\"%s\" tokenized with \'ValFloat\'", yytext); yylval.string = strdup(yytext); return(ValFloat);};
-[a-zA-Z_0-9]+ {DEBUG("\"%s\" tokenized with \'Ident\'", yytext); yylval.string = strdup(yytext); return(Ident); };
+[0-9]+    {DEBUG("\"%s\" tokenized with \'ValInt\'", yytext); yylval.string = lex_cached_strdup(yytext); return(ValInt); };
+[0-9]*\.[0-9]+ {DEBUG("\"%s\" tokenized with \'ValFloat\'", yytext); yylval.string = lex_cached_strdup(yytext); return(ValFloat);};
+[a-zA-Z_0-9]+ {DEBUG("\"%s\" tokenized with \'Ident\'", yytext); yylval.string = lex_cached_strdup(yytext); return(Ident); };
 
 \"([^\"\n])*\" {
     yytext = yytext +1;
     yytext[yyleng - 2] = 0;
     
-    DEBUG("\"%s\" tokenized with \'ValStr\'", yytext); yylval.string = strdup(yytext); return(ValStr);};
+    DEBUG("\"%s\" tokenized with \'ValStr\'", yytext); yylval.string = lex_cached_strdup(yytext); return(ValStr);};
 \"\"\"[^\"]*\"\"\" {
     yytext = yytext +3;
     yytext[yyleng - 4] = 0;
 
-    DEBUG("\"%s\" tokenized with \'ValMultistr\'", yytext); yylval.string = strdup(yytext); return(ValMultistr);};
+    DEBUG("\"%s\" tokenized with \'ValMultistr\'", yytext); yylval.string = lex_cached_strdup(yytext); return(ValMultistr);};
 [ \r\t] { /* ignore whitespace */ };
 . { return yytext[0]; /* passthrough unknown token, let parser handle the error */ };
 %%
diff --git a/src/lex/util.c b/src/lex/util.c
index 81f2855..0e01118 100644
--- a/src/lex/util.c
+++ b/src/lex/util.c
@@ -2,6 +2,7 @@
 #include <lex/util.h>
 #include <string.h>
 #include <stdlib.h>
+#include <glib.h>
 
 // implementation based on:
 // https://github.com/sunxfancy/flex-bison-examples/blob/master/error-handling/ccalc.c
@@ -16,12 +17,37 @@ static int nTokenStart = 0;
 static int nTokenLength = 0;
 static int nTokenNextStart = 0;
 
+static GArray* stringCache = NULL;
+
+char* lex_cached_strdup(char* string) {
+    char* dup = strdup(string);
+
+    g_array_append_val(stringCache, dup);
+
+    return dup;
+}
+
+void lex_purge_str_cache() {
+    DEBUG("purging string cache...");
+
+    const guint count = stringCache->len;
+
+    for (guint i = 0; i < count; i++) {
+        free(((char**) stringCache->data)[i]);
+    }
+
+    g_array_remove_range(stringCache, 0, count);
+}
+
 static void lex_deinit(void) {
+    lex_purge_str_cache();
+    g_array_free(stringCache, TRUE);
   free(buffer);
 }
 
 void lex_init(void) {
   buffer = malloc(MAX_READ_BUFFER_SIZE);
+    stringCache = g_array_new(FALSE, FALSE, sizeof(char*));
   atexit(lex_deinit);
 }
 
diff --git a/src/lex/util.h b/src/lex/util.h
index 6939526..6f96ef1 100644
--- a/src/lex/util.h
+++ b/src/lex/util.h
@@ -18,6 +18,10 @@ void lex_init(void);
 
 void lex_reset(void);
 
+char* lex_cached_strdup(char* string);
+
+void lex_purge_str_cache();
+
 /**
  * @brief Begin counting a new token. This will fill the global struct yylloc.
  * @param t the text of the token. Must be null terminated