Use xxh3 hashing for static literals

include/prism/internal/static_literals.h

@@ -1,91 +1,51 @@
 #ifndef PRISM_INTERNAL_STATIC_LITERALS_H
 #define PRISM_INTERNAL_STATIC_LITERALS_H
 
-#include "prism/internal/encoding.h"
-
 #include "prism/ast.h"
 #include "prism/buffer.h"
-#include "prism/line_offset_list.h"
-
-/*
- * An internal hash table for a set of nodes.
- */
-typedef struct {
-    /* The array of nodes in the hash table. */
-    pm_node_t **nodes;
-
-    /* The size of the hash table. */
-    uint32_t size;
-
-    /* The space that has been allocated in the hash table. */
-    uint32_t capacity;
-} pm_node_hash_t;
+#include "prism/parser.h"
 
 /*
  * Certain sets of nodes (hash keys and when clauses) check for duplicate nodes
  * to alert the user of potential issues. To do this, we keep a set of the nodes
  * that have been seen so far, and compare whenever we find a new node.
- *
- * We bucket the nodes based on their type to minimize the number of comparisons
- * that need to be performed.
  */
 typedef struct {
-    /*
-     * This is the set of IntegerNode and SourceLineNode instances.
-     */
-    pm_node_hash_t integer_nodes;
+    /* The node stored in this slot, or NULL if the slot is empty. */
+    pm_node_t *node;
 
-    /*
-     * This is the set of FloatNode instances.
-     */
-    pm_node_hash_t float_nodes;
+    /* The hash of the node, kept so that it is never recomputed. */
+    uint64_t hash;
+} pm_static_literals_entry_t;
 
-    /*
-     * This is the set of RationalNode and ImaginaryNode instances.
-     */
-    pm_node_hash_t number_nodes;
-
-    /*
-     * This is the set of StringNode and SourceFileNode instances.
-     */
-    pm_node_hash_t string_nodes;
-
-    /*
-     * This is the set of RegularExpressionNode instances.
-     */
-    pm_node_hash_t regexp_nodes;
+/*
+ * The number of nodes a set holds directly before it moves them into a hash
+ * table. Most sets are this small, so they are searched linearly without
+ * hashing their nodes or allocating any memory.
+ */
+#define PM_STATIC_LITERALS_INLINE_CAPACITY 4
 
+typedef struct {
     /*
-     * This is the set of SymbolNode instances.
+     * The nodes in the set while it has no hash table. Once the set outgrows
+     * this array, its nodes move into the hash table and this is unused.
      */
-    pm_node_hash_t symbol_nodes;
+    pm_node_t *nodes[PM_STATIC_LITERALS_INLINE_CAPACITY];
 
-    /*
-     * A pointer to the last TrueNode instance that was inserted, or NULL.
-     */
-    pm_node_t *true_node;
+    /* The array of entries in the hash table, or NULL if there is none. */
+    pm_static_literals_entry_t *entries;
 
-    /*
-     * A pointer to the last FalseNode instance that was inserted, or NULL.
-     */
-    pm_node_t *false_node;
-
-    /*
-     * A pointer to the last NilNode instance that was inserted, or NULL.
-     */
-    pm_node_t *nil_node;
+    /* The number of nodes in the set. */
+    uint32_t size;
 
-    /*
-     * A pointer to the last SourceEncodingNode instance that was inserted, or
-     * NULL.
-     */
-    pm_node_t *source_encoding_node;
+    /* The number of entries allocated in the hash table, or 0 if there is none. */
+    uint32_t capacity;
 } pm_static_literals_t;
 
 /*
  * Add a node to the set of static literals.
  */
-pm_node_t * pm_static_literals_add(const pm_line_offset_list_t *line_offsets, const uint8_t *start, int32_t start_line, const pm_encoding_t *encoding, pm_static_literals_t *literals, pm_node_t *node, bool replace);
+pm_node_t * pm_static_literals_add(const pm_parser_t *parser, pm_static_literals_t *literals, pm_node_t *node, bool replace);
 
 /*
  * Free the internal memory associated with the given static literals set.
@@ -95,6 +55,6 @@ void pm_static_literals_free(pm_static_literals_t *literals);
 /*
  * Create a string-based representation of the given static literal.
  */
-void pm_static_literal_inspect(pm_buffer_t *buffer, const pm_line_offset_list_t *line_offsets, const uint8_t *start, int32_t start_line, const pm_encoding_t *encoding, const pm_node_t *node);
+void pm_static_literal_inspect(pm_buffer_t *buffer, const pm_parser_t *parser, const pm_node_t *node);
 
 #endif

src/prism.c

@@ -13890,7 +13890,7 @@ parse_statements(pm_parser_t *parser, pm_context_t context, uint16_t depth) {
 static void
 pm_hash_key_duplicated_warn(pm_parser_t *parser, const pm_node_t *duplicated, const pm_node_t *node) {
     pm_buffer_t buffer = { 0 };
-    pm_static_literal_inspect(&buffer, &parser->line_offsets, parser->start, parser->start_line, parser->encoding, duplicated);
+    pm_static_literal_inspect(&buffer, parser, duplicated);
 
     pm_diagnostic_list_append_format(
         &parser->metadata_arena,
@@ -13912,7 +13912,7 @@ pm_hash_key_duplicated_warn(pm_parser_t *parser, const pm_node_t *duplicated, co
  */
 static void
 pm_hash_key_static_literals_add(pm_parser_t *parser, pm_static_literals_t *literals, pm_node_t *node) {
-    const pm_node_t *duplicated = pm_static_literals_add(&parser->line_offsets, parser->start, parser->start_line, parser->encoding, literals, node, true);
+    const pm_node_t *duplicated = pm_static_literals_add(parser, literals, node, true);
 
     if (duplicated != NULL) {
         pm_hash_key_duplicated_warn(parser, duplicated, node);
@@ -13936,7 +13936,7 @@ pm_hash_key_static_literals_merge(pm_parser_t *parser, pm_static_literals_t *lit
         switch (PM_NODE_TYPE(element)) {
             case PM_ASSOC_NODE: {
                 pm_node_t *key = ((pm_assoc_node_t *) element)->key;
-                const pm_node_t *duplicated = pm_static_literals_add(&parser->line_offsets, parser->start, parser->start_line, parser->encoding, literals, key, true);
+                const pm_node_t *duplicated = pm_static_literals_add(parser, literals, key, true);
 
                 if (duplicated != NULL && PM_NODE_START(duplicated) < boundary) {
                     pm_hash_key_duplicated_warn(parser, duplicated, key);
@@ -13967,7 +13967,7 @@ static void
 pm_when_clause_static_literals_add(pm_parser_t *parser, pm_static_literals_t *literals, pm_node_t *node) {
     pm_node_t *previous;
 
-    if ((previous = pm_static_literals_add(&parser->line_offsets, parser->start, parser->start_line, parser->encoding, literals, node, false)) != NULL) {
+    if ((previous = pm_static_literals_add(parser, literals, node, false)) != NULL) {
         pm_diagnostic_list_append_format(
             &parser->metadata_arena,
             &parser->warning_list,
@@ -17377,7 +17377,7 @@ parse_pattern_hash_implicit_value(pm_parser_t *parser, pm_constant_id_set_t *cap
  */
 static void
 parse_pattern_hash_key(pm_parser_t *parser, pm_static_literals_t *keys, pm_node_t *node) {
-    if (pm_static_literals_add(&parser->line_offsets, parser->start, parser->start_line, parser->encoding, keys, node, true) != NULL) {
+    if (pm_static_literals_add(parser, keys, node, true) != NULL) {
         pm_parser_err_node(parser, node, PM_ERR_PATTERN_HASH_KEY_DUPLICATE);
     }
 }