Merge pull request #20 from MisakaVan/dev-field-check

feat(parser): support identifier overlapping checking within struct/union's field scope

Author: MisakaVan

.clang-format

@@ -22,8 +22,8 @@ AlignConsecutiveDeclarations:
 #   AcrossComments:  false
 #   AlignCompound:   false
 #   PadOperators:    false
-# AlignConsecutiveMacros:
-#   Enabled:         false
+AlignConsecutiveMacros:
+  Enabled:         True
 #   AcrossEmptyLines: false
 #   AcrossComments:  false
 #   AlignCompound:   false

readme.md

@@ -11,6 +11,7 @@
 - [ ] 利用 `lib.h` 给的哈希表存放一些类型前置定义
   - [x] 检查在全局变量/Struct/Union/Enum/Typedef声明中出现的标识符是否已经在全局出现过
   - [x] 检查Struct/Union/Enum/Typedef被使用时是否已经有前置定义
-  - [ ] 检查Struct/Union的字段内是否有变量名重复
+  - [x] 检查Struct/Union的字段内是否有变量名重复
+  - [ ] 指出检查的完备性（没有遗漏检查）
 - [ ] 完善build、test、使用文档
 

resource/test/test-identifier-predeclared/testcase-inside-fields.jtl.h

@@ -0,0 +1,11 @@
+struct Foo {
+    int var1;
+    int var2;
+    char var2; // Warning: redeclaration of 'var2' field
+};
+
+union Bar {
+    char var1; // OK: 'var1' is not declared yet in this scope
+    int (*var2)(int); // OK: 'var2' is not declared yet in this scope
+    struct Foo var2; // Warning: redeclaration of 'var2' field
+};

src/lang.c

@@ -102,6 +102,7 @@ struct left_type* TNewStructType(char* name, struct type_list* fld) {
     res->d.NEW_STRUCT_TYPE.fld = fld;
 
     register_identifier_struct(name);
+    check_field_list(fld);
     return res;
 }
 
@@ -122,6 +123,7 @@ struct left_type* TNewUnionType(char* name, struct type_list* fld) {
     res->d.NEW_UNION_TYPE.fld = fld;
 
     register_identifier_union(name);
+    check_field_list(fld);
 
     return res;
 }
@@ -173,6 +175,7 @@ struct var_decl_expr* TOrigType(char* name) {
     struct var_decl_expr* res = new_var_decl_expr_ptr();
     res->t = T_ORIG_TYPE;
     res->d.ORIG_TYPE.name = name;
+    res->d.ORIG_TYPE.lineno = yylineno;
     return res;
 }
 
@@ -208,6 +211,7 @@ struct glob_item* TStructDef(char* name, struct type_list* fld) {
     res->d.STRUCT_DEF.fld = fld;
 
     register_identifier_struct(name);
+    check_field_list(fld);
 
     return res;
 }
@@ -229,6 +233,7 @@ struct glob_item* TUnionDef(char* name, struct type_list* fld) {
     res->d.UNION_DEF.fld = fld;
 
     register_identifier_union(name);
+    check_field_list(fld);
 
     return res;
 }
@@ -271,22 +276,7 @@ struct glob_item* TTypeDef(struct left_type* t, struct var_decl_expr* e) {
     res->d.TYPE_DEF.e = e;
 
     // get the core type name and register it
-    struct var_decl_expr* ptr = e;
-    while (ptr->t != T_ORIG_TYPE) {
-        switch (ptr->t) {
-            case T_PTR_TYPE:
-                ptr = ptr->d.PTR_TYPE.base;
-                break;
-            case T_ARRAY_TYPE:
-                ptr = ptr->d.ARRAY_TYPE.base;
-                break;
-            case T_FUNC_TYPE:
-                ptr = ptr->d.FUNC_TYPE.ret;
-                break;
-            case T_ORIG_TYPE:
-                break;
-        }
-    }
+    struct var_decl_expr* ptr = get_core_type(e);
     register_identifier_typedef(ptr->d.ORIG_TYPE.name);
 
     return res;
@@ -299,22 +289,7 @@ struct glob_item* TVarDef(struct left_type* t, struct var_decl_expr* e) {
     res->d.VAR_DEF.e = e;
 
     // get the core type name and register it
-    struct var_decl_expr* ptr = e;
-    while (ptr->t != T_ORIG_TYPE) {
-        switch (ptr->t) {
-            case T_PTR_TYPE:
-                ptr = ptr->d.PTR_TYPE.base;
-                break;
-            case T_ARRAY_TYPE:
-                ptr = ptr->d.ARRAY_TYPE.base;
-                break;
-            case T_FUNC_TYPE:
-                ptr = ptr->d.FUNC_TYPE.ret;
-                break;
-            case T_ORIG_TYPE:
-                break;
-        }
-    }
+    struct var_decl_expr* ptr = get_core_type(e);
     register_identifier_variable(ptr->d.ORIG_TYPE.name);
     return res;
 }

src/lang.h

@@ -1,7 +1,6 @@
 #ifndef LANG_H_INCLUDED
 #define LANG_H_INCLUDED
 
-#include "lib.h"
 #include <stdio.h>
 #include <stdlib.h>
 
@@ -104,6 +103,7 @@ struct var_decl_expr {
     union {
         struct {
             char* name;
+            int lineno;
         } ORIG_TYPE;
         struct {
             struct var_decl_expr* base;

src/lib.c

@@ -1,8 +1,11 @@
 #include "lib.h"
+#include "lang.h"
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
+int log_level = 0;
+
 unsigned int build_nat(char* c, int len) {
     int s = 0, i = 0;
     for (i = 0; i < len; ++i) {
@@ -110,6 +113,28 @@ void SLL_hash_delete(struct SLL_hash_table* t, char* key) {
 
 struct SLL_hash_table* identifier_table;
 
+
+
+struct var_decl_expr *get_core_type(struct var_decl_expr *ptr){
+    while (ptr->t != T_ORIG_TYPE) {
+        switch (ptr->t) {
+            case T_PTR_TYPE:
+                ptr = ptr->d.PTR_TYPE.base;
+                break;
+            case T_ARRAY_TYPE:
+                ptr = ptr->d.ARRAY_TYPE.base;
+                break;
+            case T_FUNC_TYPE:
+                ptr = ptr->d.FUNC_TYPE.ret;
+                break;
+            case T_ORIG_TYPE:
+                break;
+        }
+    }
+    return ptr;
+}
+
+
 struct IdentifierInfo* init_identifier_info() {
     struct IdentifierInfo* res =
         (struct IdentifierInfo*)malloc(sizeof(struct IdentifierInfo));
@@ -151,9 +176,18 @@ char* identifier_type_str[6] = {
 // use yylineno to record the line number of the first registration
 extern int yylineno;
 
-void register_identifier(char* name, enum IdentifierType type) {
-    pdebug("Line %d:\n", yylineno);
-    pdebug("Registering identifier %s as %s\n", name, identifier_type_str[type]);
+// lineno: use -1 to fallback to yylineno.
+void register_identifier_in_table(char* name, enum IdentifierType type, struct SLL_hash_table *table, int lineno, char *description) {
+    if (lineno == -1){
+        lineno = yylineno;
+    }
+    if (description == NULL){
+        // fallback to the default description
+        description = identifier_type_str[type];
+    }
+
+    pdebug("Line %d:\n", lineno);
+    pdebug("Registering identifier %s as %s\n", name, description);
 
     if (name==NULL){
         pdebug("Identifier name is NULL (Anonymous), skip registering\n");
@@ -162,11 +196,11 @@ void register_identifier(char* name, enum IdentifierType type) {
 
     struct IdentifierInfo *info = NULL;
 
-    info = (struct IdentifierInfo*)SLL_hash_get(identifier_table, name);
+    info = (struct IdentifierInfo*)SLL_hash_get(table, name);
     if ((long long)info == NONE) {
         pdebug("Identifier %s is not in the hashtable\n", name);
         info = init_identifier_info();
-        SLL_hash_set(identifier_table, name, (long long)info);
+        SLL_hash_set(table, name, (long long)info);
     }
     else{
         pdebug("Identifier %s is in the hashtable\n", name);
@@ -182,18 +216,22 @@ void register_identifier(char* name, enum IdentifierType type) {
         }
         if (info->flags & (1 << i)) {
             // identifier is already registered as a conflicting type.
-            printf("Warning: (Line %d) Identifier %s is already registered as %s at line %d\n", yylineno, name, identifier_type_str[i], info->lineno[i]);
+            printf("Warning: (Line %d) Identifier %s is already registered as %s at line %d\n", lineno, name, description, info->lineno[i]);
             conflict = 1;
         }
     }
 
     if (!conflict) {
         pdebug("No conflict. Now register %s as %s\n", name, identifier_type_str[type]);
         info->flags |= (1 << type);
-        info->lineno[type] = yylineno; // the line number of the first registration
+        info->lineno[type] = lineno; // the line number of the first registration
     }
 }
 
+void register_identifier(char* name, enum IdentifierType type) {
+    register_identifier_in_table(name, type, identifier_table, -1, NULL);
+}
+
 void register_identifier_variable(char* name) {
     register_identifier(name, IDENT_TYPE_VARIABLE);
 }
@@ -218,10 +256,10 @@ void register_identifier_typedef(char* name) {
     register_identifier(name, IDENT_TYPE_TYPEDEF);
 }
 
-void check_identifier(char *name, enum IdentifierType using_type){
+void check_identifier_in_table(char* name, enum IdentifierType using_type, struct SLL_hash_table *table){
     pdebug("Line %d:\n", yylineno);
     struct IdentifierInfo *info = NULL;
-    info = (struct IdentifierInfo*)SLL_hash_get(identifier_table, name);
+    info = (struct IdentifierInfo*)SLL_hash_get(table, name);
     if ((long long)info == NONE) {
         printf("Warning: (Line %d) Identifier %s has never been registered\n", yylineno, name);
         return;
@@ -241,6 +279,10 @@ void check_identifier(char *name, enum IdentifierType using_type){
     }
 }
 
+void check_identifier(char *name, enum IdentifierType using_type){
+    check_identifier_in_table(name, using_type, identifier_table);
+}
+
 void check_identifier_enum(char *name){
     check_identifier(name, IDENT_TYPE_ENUM);
 }
@@ -257,3 +299,20 @@ void check_identifier_typedef(char *name){
     check_identifier(name, IDENT_TYPE_TYPEDEF);
 }
 
+void check_field_list(struct type_list *field_list){
+    // build a new hash table for the field list
+    struct SLL_hash_table *field_table = init_SLL_hash();
+    while (field_list != NULL){
+        struct left_type *t = field_list->t;
+        struct var_decl_expr *e = field_list->e;
+        struct var_decl_expr *core_type = get_core_type(e);
+        register_identifier_in_table(
+            core_type->d.ORIG_TYPE.name, 
+            IDENT_TYPE_VARIABLE,
+            field_table,
+            core_type->d.ORIG_TYPE.lineno,
+            "field variable"
+        );
+        field_list = field_list->next;
+    }
+}

src/lib.h

@@ -1,15 +1,25 @@
 #ifndef LIB_H_INCLUDED
 #define LIB_H_INCLUDED
 
+#include "lang.h"
+
 #define NONE 4294967295
 
-#define VERBOSE 1
+// global variable to indicate logging level
 
-#ifdef VERBOSE
-    #define pdebug(fmt, ...) printf("[DEBUG] " fmt, ##__VA_ARGS__)
-#else
-    #define pdebug(fmt, ...)
-#endif
+#define LOG_LEVEL_ERROR 0
+#define LOG_LEVEL_VERBOSE 5
+
+
+extern int log_level;
+
+#define pdebug(fmt, ...) \
+do { \
+    if (log_level >= LOG_LEVEL_VERBOSE) \
+        printf("[DEBUG] " fmt, ##__VA_ARGS__); \
+} while (0)
+
+extern int yylineno;
 
 // clang-format off
 unsigned int build_nat(char * c, int len);
@@ -21,6 +31,11 @@ void SLL_hash_set(struct SLL_hash_table * t, char * key, long long value);
 void SLL_hash_delete(struct SLL_hash_table * t, char * key);
 // clang-format on
 
+// helper functions
+
+// return the innermost core type. gaurenteed that ret->t == T_ORIG_TYPE.
+struct var_decl_expr *get_core_type(struct var_decl_expr *ptr);
+
 
 enum IdentifierType {
     IDENT_TYPE_VARIABLE = 0,
@@ -50,6 +65,7 @@ struct IdentifierInfo {
 
 struct IdentifierInfo *init_identifier_info();
 
+void register_identifier_in_table(char* name, enum IdentifierType type, struct SLL_hash_table *table, int lineno, char *description);
 void register_identifier(char *name, enum IdentifierType type);
 
 void register_identifier_variable(char *name);
@@ -60,10 +76,15 @@ void register_identifier_enum(char *name);
 void register_identifier_typedef(char *name);
 
 // check if the identifier is already declared when using it
+void check_identifier_in_table(char* name, enum IdentifierType type, struct SLL_hash_table *table);
 void check_identifier(char *name, enum IdentifierType using_type);
+
 void check_identifier_enum(char *name);
 void check_identifier_struct(char *name);
 void check_identifier_union(char *name);
 void check_identifier_typedef(char *name);
 
+// check if identifiers overlap within a field list
+void check_field_list(struct type_list *field_list);
+
 #endif

src/main.c

@@ -4,6 +4,7 @@
 #include "astprint.h"
 #include "lib.h"
 #include <stdio.h>
+#include <stdlib.h>
 
 // use the new printer
 // #define ASTPRINT
@@ -24,6 +25,12 @@ int main(int argc, char** argv) {
         return 0;
     }
 
+    // read "PARSER_VERBOSE" environment variable
+    char *log_level_str = getenv("PARSER_VERBOSE");
+    if (log_level_str != NULL) {
+        log_level = LOG_LEVEL_VERBOSE;
+    }
+
     // initialize the identifier table
     identifier_table = init_SLL_hash();