[ruby/yarp] Use common fields for yp_string_t

* Otherwise it is undefined behavior to access the field of another `.as`.
* Accessing the right `.as` field according mode would be extra overhead.

https://github.com/ruby/yarp/commit/7dc41ee803

yarp/unescape.c

@@ -461,10 +461,8 @@ yp_unescape_manipulate_string(yp_parser_t *parser, const char *value, size_t len
         return;
     }
 
-    yp_string_owned_init(string, allocated, length);
-
     // This is the memory address where we're putting the unescaped string.
-    char *dest = string->as.owned.source;
+    char *dest = allocated;
     size_t dest_length = 0;
 
     // This is the current position in the source string that we're looking at.
@@ -525,7 +523,7 @@ yp_unescape_manipulate_string(yp_parser_t *parser, const char *value, size_t len
     // We also need to update the length at the end. This is because every escape
     // reduces the length of the final string, and we don't want garbage at the
     // end.
-    string->as.owned.length = dest_length + ((size_t) (end - cursor));
+    yp_string_owned_init(string, allocated, dest_length + ((size_t) (end - cursor)));
 }
 
 YP_EXPORTED_FUNCTION bool

yarp/util/yp_string.c

@@ -13,12 +13,11 @@
 // Initialize a shared string that is based on initial input.
 void
 yp_string_shared_init(yp_string_t *string, const char *start, const char *end) {
+    assert(start <= end);
     *string = (yp_string_t) {
         .type = YP_STRING_SHARED,
-        .as.shared = {
+        .source = (char*) start,
-            .start = start,
+        .length = (size_t) (end - start)
-            .end = end
-        }
     };
 }
 
@@ -27,10 +26,8 @@ void
 yp_string_owned_init(yp_string_t *string, char *source, size_t length) {
     *string = (yp_string_t) {
         .type = YP_STRING_OWNED,
-        .as.owned = {
+        .source = source,
-            .source = source,
+        .length = length
-            .length = length
-        }
     };
 }
 
@@ -39,10 +36,8 @@ void
 yp_string_constant_init(yp_string_t *string, const char *source, size_t length) {
     *string = (yp_string_t) {
         .type = YP_STRING_CONSTANT,
-        .as.constant = {
+        .source = (char*) source,
-            .source = source,
+        .length = length
-            .length = length
-        }
     };
 }
 
@@ -50,10 +45,8 @@ static void
 yp_string_mapped_init_internal(yp_string_t *string, char *source, size_t length) {
     *string = (yp_string_t) {
         .type = YP_STRING_MAPPED,
-        .as.mapped = {
+        .source = source,
-            .source = source,
+        .length = length
-            .length = length
-        }
     };
 }
 
@@ -62,7 +55,7 @@ size_t
 yp_string_memsize(const yp_string_t *string) {
     size_t size = sizeof(yp_string_t);
     if (string->type == YP_STRING_OWNED) {
-        size += string->as.owned.length;
+        size += string->length;
     }
     return size;
 }
@@ -77,40 +70,32 @@ yp_string_ensure_owned(yp_string_t *string) {
     const char *source = yp_string_source(string);
 
     yp_string_owned_init(string, malloc(length), length);
-    memcpy(string->as.owned.source, source, length);
+    memcpy(string->source, source, length);
 }
 
 // Returns the length associated with the string.
 YP_EXPORTED_FUNCTION size_t
 yp_string_length(const yp_string_t *string) {
-    if (string->type == YP_STRING_SHARED) {
+    return string->length;
-        return (size_t) (string->as.shared.end - string->as.shared.start);
-    } else {
-        return string->as.owned.length;
-    }
 }
 
 // Returns the start pointer associated with the string.
 YP_EXPORTED_FUNCTION const char *
 yp_string_source(const yp_string_t *string) {
-    if (string->type == YP_STRING_SHARED) {
+    return string->source;
-        return string->as.shared.start;
-    } else {
-        return string->as.owned.source;
-    }
 }
 
 // Free the associated memory of the given string.
 YP_EXPORTED_FUNCTION void
 yp_string_free(yp_string_t *string) {
     if (string->type == YP_STRING_OWNED) {
-        free(string->as.owned.source);
+        free(string->source);
-    } else if (string->type == YP_STRING_MAPPED && string->as.mapped.length) {
+    } else if (string->type == YP_STRING_MAPPED && string->length) {
-        void *memory = (void *) string->as.mapped.source;
+        void *memory = (void *) string->source;
         #if defined(_WIN32)
         UnmapViewOfFile(memory);
         #elif defined(HAVE_MMAP)
-        munmap(memory, string->as.mapped.length);
+        munmap(memory, string->length);
         #else
         free(memory);
         #endif

yarp/util/yp_string.h

@@ -3,6 +3,7 @@
 
 #include "yarp/defines.h"
 
+#include <assert.h>
 #include <stdbool.h>
 #include <stddef.h>
 #include <stdlib.h>
@@ -11,30 +12,12 @@
 // This struct represents a string value.
 typedef struct {
     enum { YP_STRING_SHARED, YP_STRING_OWNED, YP_STRING_CONSTANT, YP_STRING_MAPPED } type;
-
+    char *source;
-    union {
+    size_t length;
-        struct {
-            const char *start;
-            const char *end;
-        } shared;
-
-        struct {
-            char *source;
-            size_t length;
-        } owned;
-
-        struct {
-            const char *source;
-            size_t length;
-        } constant;
-
-        struct {
-            char *source;
-            size_t length;
-        } mapped;
-    } as;
 } yp_string_t;
 
+#define YP_EMPTY_STRING ((yp_string_t) { .type = YP_STRING_CONSTANT, .source = NULL, .length = 0 })
+
 // Initialize a shared string that is based on initial input.
 void yp_string_shared_init(yp_string_t *string, const char *start, const char *end);
 

yarp/yarp.c

@@ -445,7 +445,6 @@ not_provided(yp_parser_t *parser) {
     return (yp_token_t) { .type = YP_TOKEN_NOT_PROVIDED, .start = parser->start, .end = parser->start };
 }
 
-#define YP_EMPTY_STRING ((yp_string_t) { .type = YP_STRING_SHARED, .as.shared.start = NULL, .as.shared.end = NULL })
 #define YP_LOCATION_NULL_VALUE(parser) ((yp_location_t) { .start = parser->start, .end = parser->start })
 #define YP_LOCATION_TOKEN_VALUE(token) ((yp_location_t) { .start = (token)->start, .end = (token)->end })
 #define YP_LOCATION_NODE_VALUE(node) ((yp_location_t) { .start = (node)->location.start, .end = (node)->location.end })
@@ -9345,8 +9344,8 @@ parse_heredoc_dedent(yp_parser_t *parser, yp_node_t *node, yp_heredoc_quote_t qu
         // Now get the bounds of the existing string. We'll use this as a
         // destination to move bytes into. We'll also use it for bounds checking
         // since we don't require that these strings be null terminated.
-        size_t dest_length = string->as.owned.length;
+        size_t dest_length = yp_string_length(string);
-        char *source_start = string->as.owned.source;
+        char *source_start = string->source;
 
         const char *source_cursor = source_start;
         const char *source_end = source_cursor + dest_length;
@@ -9400,7 +9399,7 @@ parse_heredoc_dedent(yp_parser_t *parser, yp_node_t *node, yp_heredoc_quote_t qu
             dedent_next = true;
         }
 
-        string->as.owned.length = dest_length;
+        string->length = dest_length;
     }
 }
 
@@ -12450,7 +12449,7 @@ parse_expression_infix(yp_parser_t *parser, yp_node_t *node, yp_binding_power_t
                         yp_string_t *name = &named_captures.strings[index];
                         assert(name->type == YP_STRING_SHARED);
 
-                        yp_parser_local_add_location(parser, name->as.shared.start, name->as.shared.end);
+                        yp_parser_local_add_location(parser, name->source, name->source + name->length);
                     }
                 }