Use an arena for parser metadata

Author: kddnewton

include/prism/parser.h

@@ -639,6 +639,9 @@ struct pm_parser {
     /** The arena used for all AST-lifetime allocations. Caller-owned. */
     pm_arena_t *arena;
 
+    /** The arena used for parser metadata (comments, diagnostics, etc.). */
+    pm_arena_t metadata_arena;
+
     /**
      * The next node identifier that will be assigned. This is a unique
      * identifier used to track nodes such that the syntax tree can be dropped

include/prism/util/pm_char.h

@@ -36,7 +36,7 @@ size_t pm_strspn_whitespace(const uint8_t *string, ptrdiff_t length);
  * @return The number of characters at the start of the string that are
  *     whitespace.
  */
-size_t pm_strspn_whitespace_newlines(const uint8_t *string, ptrdiff_t length, pm_line_offset_list_t *line_offsets, uint32_t start_offset);
+size_t pm_strspn_whitespace_newlines(const uint8_t *string, ptrdiff_t length, pm_arena_t *arena, pm_line_offset_list_t *line_offsets, uint32_t start_offset);
 
 /**
  * Returns the number of characters at the start of the string that are inline

include/prism/util/pm_line_offset_list.h

@@ -15,6 +15,7 @@
 #define PRISM_LINE_OFFSET_LIST_H
 
 #include "prism/defines.h"
+#include "prism/util/pm_arena.h"
 
 #include <assert.h>
 #include <stdbool.h>
@@ -48,14 +49,13 @@ typedef struct {
 } pm_line_column_t;
 
 /**
- * Initialize a new line offset list with the given capacity. Returns true if
- * the allocation of the offsets succeeds, otherwise returns false.
+ * Initialize a new line offset list with the given capacity.
  *
+ * @param arena The arena to allocate from.
  * @param list The list to initialize.
  * @param capacity The initial capacity of the list.
- * @return True if the allocation of the offsets succeeds, otherwise false.
  */
-bool pm_line_offset_list_init(pm_line_offset_list_t *list, size_t capacity);
+void pm_line_offset_list_init(pm_arena_t *arena, pm_line_offset_list_t *list, size_t capacity);
 
 /**
  * Clear out the offsets that have been appended to the list.
@@ -65,15 +65,13 @@ bool pm_line_offset_list_init(pm_line_offset_list_t *list, size_t capacity);
 void pm_line_offset_list_clear(pm_line_offset_list_t *list);
 
 /**
- * Append a new offset to the list. Returns true if the reallocation of the
- * offsets succeeds (if one was necessary), otherwise returns false.
+ * Append a new offset to the list.
  *
+ * @param arena The arena to allocate from.
  * @param list The list to append to.
  * @param cursor The offset to append.
- * @return True if the reallocation of the offsets succeeds (if one was
- *     necessary), otherwise false.
  */
-bool pm_line_offset_list_append(pm_line_offset_list_t *list, uint32_t cursor);
+void pm_line_offset_list_append(pm_arena_t *arena, pm_line_offset_list_t *list, uint32_t cursor);
 
 /**
  * Returns the line of the given offset. If the offset is not in the list, the
@@ -98,11 +96,4 @@ int32_t pm_line_offset_list_line(const pm_line_offset_list_t *list, uint32_t cur
  */
 PRISM_EXPORTED_FUNCTION pm_line_column_t pm_line_offset_list_line_column(const pm_line_offset_list_t *list, uint32_t cursor, int32_t start_line);
 
-/**
- * Free the internal memory allocated for the list.
- *
- * @param list The list to free.
- */
-void pm_line_offset_list_free(pm_line_offset_list_t *list);
-
 #endif

src/prism.c

@@ -83,15 +83,15 @@ pm_strspn_whitespace(const uint8_t *string, ptrdiff_t length) {
  * searching past the given maximum number of characters.
  */
 size_t
-pm_strspn_whitespace_newlines(const uint8_t *string, ptrdiff_t length, pm_line_offset_list_t *line_offsets, uint32_t start_offset) {
+pm_strspn_whitespace_newlines(const uint8_t *string, ptrdiff_t length, pm_arena_t *arena, pm_line_offset_list_t *line_offsets, uint32_t start_offset) {
     if (length <= 0) return 0;
 
     uint32_t size = 0;
     uint32_t maximum = (uint32_t) length;
 
     while (size < maximum && (pm_byte_table[string[size]] & PRISM_CHAR_BIT_WHITESPACE)) {
         if (string[size] == '\n') {
-            pm_line_offset_list_append(line_offsets, start_offset + size + 1);
+            pm_line_offset_list_append(arena, line_offsets, start_offset + size + 1);
         }
 
         size++;

src/util/pm_char.c

@@ -1,20 +1,16 @@
 #include "prism/util/pm_line_offset_list.h"
 
 /**
- * Initialize a new newline list with the given capacity. Returns true if the
- * allocation of the offsets succeeds, otherwise returns false.
+ * Initialize a new line offset list with the given capacity.
  */
-bool
-pm_line_offset_list_init(pm_line_offset_list_t *list, size_t capacity) {
-    list->offsets = (uint32_t *) xcalloc(capacity, sizeof(uint32_t));
-    if (list->offsets == NULL) return false;
+void
+pm_line_offset_list_init(pm_arena_t *arena, pm_line_offset_list_t *list, size_t capacity) {
+    list->offsets = (uint32_t *) pm_arena_zalloc(arena, capacity * sizeof(uint32_t), PRISM_ALIGNOF(uint32_t));
 
     // This is 1 instead of 0 because we want to include the first line of the
-    // file as having offset 0, which is set because of calloc.
+    // file as having offset 0, which is set because of the zero-initialization.
     list->size = 1;
     list->capacity = capacity;
-
-    return true;
 }
 
 /**
@@ -26,26 +22,22 @@ pm_line_offset_list_clear(pm_line_offset_list_t *list) {
 }
 
 /**
- * Append a new offset to the newline list. Returns true if the reallocation of
- * the offsets succeeds (if one was necessary), otherwise returns false.
+ * Append a new offset to the newline list.
  */
-bool
-pm_line_offset_list_append(pm_line_offset_list_t *list, uint32_t cursor) {
+void
+pm_line_offset_list_append(pm_arena_t *arena, pm_line_offset_list_t *list, uint32_t cursor) {
     if (list->size == list->capacity) {
-        uint32_t *original_offsets = list->offsets;
+        size_t new_capacity = (list->capacity * 3) / 2;
+        uint32_t *new_offsets = (uint32_t *) pm_arena_alloc(arena, new_capacity * sizeof(uint32_t), PRISM_ALIGNOF(uint32_t));
 
-        list->capacity = (list->capacity * 3) / 2;
-        list->offsets = (uint32_t *) xcalloc(list->capacity, sizeof(uint32_t));
-        if (list->offsets == NULL) return false;
+        memcpy(new_offsets, list->offsets, list->size * sizeof(uint32_t));
 
-        memcpy(list->offsets, original_offsets, list->size * sizeof(uint32_t));
-        xfree_sized(original_offsets, list->size * sizeof(uint32_t));
+        list->offsets = new_offsets;
+        list->capacity = new_capacity;
     }
 
     assert(list->size == 0 || cursor > list->offsets[list->size - 1]);
     list->offsets[list->size++] = cursor;
-
-    return true;
 }
 
 /**
@@ -103,11 +95,3 @@ pm_line_offset_list_line_column(const pm_line_offset_list_t *list, uint32_t curs
         .column = cursor - list->offsets[left - 1]
     });
 }
-
-/**
- * Free the internal memory allocated for the newline list.
- */
-void
-pm_line_offset_list_free(pm_line_offset_list_t *list) {
-    xfree_sized(list->offsets, list->capacity * sizeof(uint32_t));
-}

src/util/pm_line_offset_list.c

@@ -5,7 +5,7 @@
  */
 static inline void
 pm_strpbrk_invalid_multibyte_character(pm_parser_t *parser, uint32_t start, uint32_t length) {
-    pm_diagnostic_list_append_format(&parser->error_list, start, length, PM_ERR_INVALID_MULTIBYTE_CHARACTER, parser->start[start]);
+    pm_diagnostic_list_append_format(&parser->metadata_arena, &parser->error_list, start, length, PM_ERR_INVALID_MULTIBYTE_CHARACTER, parser->start[start]);
 }
 
 /**
@@ -19,7 +19,7 @@ pm_strpbrk_explicit_encoding_set(pm_parser_t *parser, uint32_t start, uint32_t l
         } else if (parser->explicit_encoding == PM_ENCODING_UTF_8_ENTRY) {
             // Not okay, we already found a Unicode escape sequence and this
             // conflicts.
-            pm_diagnostic_list_append_format(&parser->error_list, start, length, PM_ERR_MIXED_ENCODING, parser->encoding->name);
+            pm_diagnostic_list_append_format(&parser->metadata_arena, &parser->error_list, start, length, PM_ERR_MIXED_ENCODING, parser->encoding->name);
         } else {
             // Should not be anything else.
             assert(false && "unreachable");

src/util/pm_strpbrk.c

@@ -8,6 +8,7 @@
 
 #include "prism/ast.h"
 #include "prism/defines.h"
+#include "prism/util/pm_arena.h"
 #include "prism/util/pm_list.h"
 
 #include <stdbool.h>
@@ -48,13 +49,6 @@ typedef struct {
     /** The message associated with the diagnostic. */
     const char *message;
 
-    /**
-     * Whether or not the memory related to the message of this diagnostic is
-     * owned by this diagnostic. If it is, it needs to be freed when the
-     * diagnostic is freed.
-     */
-    bool owned;
-
     /**
      * The level of the diagnostic, see `pm_error_level_t` and
      * `pm_warning_level_t` for possible values.
@@ -99,32 +93,25 @@ const char * pm_diagnostic_id_human(pm_diagnostic_id_t diag_id);
  * Append a diagnostic to the given list of diagnostics that is using shared
  * memory for its message.
  *
+ * @param arena The arena to allocate from.
  * @param list The list to append to.
  * @param start The source offset of the start of the diagnostic.
  * @param length The length of the diagnostic.
  * @param diag_id The diagnostic ID.
- * @return Whether the diagnostic was successfully appended.
  */
-bool pm_diagnostic_list_append(pm_list_t *list, uint32_t start, uint32_t length, pm_diagnostic_id_t diag_id);
+void pm_diagnostic_list_append(pm_arena_t *arena, pm_list_t *list, uint32_t start, uint32_t length, pm_diagnostic_id_t diag_id);
 
 /**
  * Append a diagnostic to the given list of diagnostics that is using a format
  * string for its message.
  *
+ * @param arena The arena to allocate from.
  * @param list The list to append to.
  * @param start The source offset of the start of the diagnostic.
  * @param length The length of the diagnostic.
  * @param diag_id The diagnostic ID.
  * @param ... The arguments to the format string for the message.
- * @return Whether the diagnostic was successfully appended.
- */
-bool pm_diagnostic_list_append_format(pm_list_t *list, uint32_t start, uint32_t length, pm_diagnostic_id_t diag_id, ...);
-
-/**
- * Deallocate the internal state of the given diagnostic list.
- *
- * @param list The list to deallocate.
  */
-void pm_diagnostic_list_free(pm_list_t *list);
+void pm_diagnostic_list_append_format(pm_arena_t *arena, pm_list_t *list, uint32_t start, uint32_t length, pm_diagnostic_id_t diag_id, ...);
 
 #endif

templates/include/prism/diagnostic.h.erb

@@ -1,4 +1,5 @@
 #include "prism/diagnostic.h"
+#include "prism/util/pm_arena.h"
 
 #define PM_DIAGNOSTIC_ID_MAX <%= errors.length + warnings.length %>
 
@@ -451,29 +452,26 @@ pm_diagnostic_level(pm_diagnostic_id_t diag_id) {
 /**
  * Append an error to the given list of diagnostic.
  */
-bool
-pm_diagnostic_list_append(pm_list_t *list, uint32_t start, uint32_t length, pm_diagnostic_id_t diag_id) {
-    pm_diagnostic_t *diagnostic = (pm_diagnostic_t *) xcalloc(1, sizeof(pm_diagnostic_t));
-    if (diagnostic == NULL) return false;
+void
+pm_diagnostic_list_append(pm_arena_t *arena, pm_list_t *list, uint32_t start, uint32_t length, pm_diagnostic_id_t diag_id) {
+    pm_diagnostic_t *diagnostic = (pm_diagnostic_t *) pm_arena_zalloc(arena, sizeof(pm_diagnostic_t), PRISM_ALIGNOF(pm_diagnostic_t));
 
     *diagnostic = (pm_diagnostic_t) {
         .location = { .start = start, .length = length },
         .diag_id = diag_id,
         .message = pm_diagnostic_message(diag_id),
-        .owned = false,
         .level = pm_diagnostic_level(diag_id)
     };
 
     pm_list_append(list, (pm_list_node_t *) diagnostic);
-    return true;
 }
 
 /**
  * Append a diagnostic to the given list of diagnostics that is using a format
  * string for its message.
  */
-bool
-pm_diagnostic_list_append_format(pm_list_t *list, uint32_t start, uint32_t length, pm_diagnostic_id_t diag_id, ...) {
+void
+pm_diagnostic_list_append_format(pm_arena_t *arena, pm_list_t *list, uint32_t start, uint32_t length, pm_diagnostic_id_t diag_id, ...) {
     va_list arguments;
     va_start(arguments, diag_id);
 
@@ -482,20 +480,13 @@ pm_diagnostic_list_append_format(pm_list_t *list, uint32_t start, uint32_t lengt
     va_end(arguments);
 
     if (result < 0) {
-        return false;
+        return;
     }
 
-    pm_diagnostic_t *diagnostic = (pm_diagnostic_t *) xcalloc(1, sizeof(pm_diagnostic_t));
-    if (diagnostic == NULL) {
-        return false;
-    }
+    pm_diagnostic_t *diagnostic = (pm_diagnostic_t *) pm_arena_zalloc(arena, sizeof(pm_diagnostic_t), PRISM_ALIGNOF(pm_diagnostic_t));
 
     size_t message_length = (size_t) (result + 1);
-    char *message = (char *) xmalloc(message_length);
-    if (message == NULL) {
-        xfree_sized(diagnostic, sizeof(pm_diagnostic_t));
-        return false;
-    }
+    char *message = (char *) pm_arena_alloc(arena, message_length, 1);
 
     va_start(arguments, diag_id);
     vsnprintf(message, message_length, format, arguments);
@@ -505,27 +496,9 @@ pm_diagnostic_list_append_format(pm_list_t *list, uint32_t start, uint32_t lengt
         .location = { .start = start, .length = length },
         .diag_id = diag_id,
         .message = message,
-        .owned = true,
         .level = pm_diagnostic_level(diag_id)
     };
 
     pm_list_append(list, (pm_list_node_t *) diagnostic);
-    return true;
 }
 
-/**
- * Deallocate the internal state of the given diagnostic list.
- */
-void
-pm_diagnostic_list_free(pm_list_t *list) {
-    pm_diagnostic_t *node = (pm_diagnostic_t *) list->head;
-
-    while (node != NULL) {
-        pm_diagnostic_t *next = (pm_diagnostic_t *) node->node.next;
-
-        if (node->owned) xfree((void *) node->message);
-        xfree_sized(node, sizeof(pm_diagnostic_t));
-
-        node = next;
-    }
-}