From 8043ca2183f83a68006d9c28f2f6894a1b01bb12 Mon Sep 17 00:00:00 2001
From: "K. Lange" <klange@toaruos.org>
Date: Tue, 13 Sep 2022 19:46:17 +0900
Subject: [PATCH] Pre-hash strings from str(),hex(),oct(),bin() conversions

---
 src/obj_bytes.c |  4 +++-
 src/obj_long.c  | 15 ++++++++++-----
 src/obj_str.c   |  4 ++--
 src/object.c    |  4 +++-
 src/private.h   |  5 +++++
 5 files changed, 23 insertions(+), 9 deletions(-)

diff --git a/src/obj_bytes.c b/src/obj_bytes.c
index b4cd0be..13a5469 100644
--- a/src/obj_bytes.c
+++ b/src/obj_bytes.c
@@ -4,6 +4,8 @@
 #include <kuroko/memory.h>
 #include <kuroko/util.h>
 
+#include "private.h"
+
 struct ByteArray {
 	KrkInstance inst;
 	KrkValue actual;
@@ -61,7 +63,7 @@ KRK_Method(bytes,__hash__) {
 	/* This is the so-called "sdbm" hash. It comes from a piece of
 	 * public domain code from a clone of ndbm. */
 	for (size_t i = 0; i < self->length; ++i) {
-		hash = (int)self->bytes[i] + (hash << 6) + (hash << 16) - hash;
+		krk_hash_advance(hash,self->bytes[i]);
 	}
 	return INTEGER_VAL(hash);
 }
diff --git a/src/obj_long.c b/src/obj_long.c
index 19e49cf..d522e11 100644
--- a/src/obj_long.c
+++ b/src/obj_long.c
@@ -20,6 +20,7 @@
 #include <kuroko/vm.h>
 #include <kuroko/value.h>
 #include <kuroko/util.h>
+#include "private.h"
 
 #define DIGIT_SHIFT 31
 #define DIGIT_MAX   0x7FFFFFFF
@@ -1006,7 +1007,7 @@ static char * _fast_conversion(const KrkLong * abs, unsigned int bits, char * wr
 /**
  * @brief Convert a long to a string in a given base.
  */
-static char * krk_long_to_str(const KrkLong * n, int _base, const char * prefix, size_t *size) {
+static char * krk_long_to_str(const KrkLong * n, int _base, const char * prefix, size_t *size, uint32_t *_hash) {
 	KrkLong abs;
 
 	krk_long_init_si(&abs, 0);
@@ -1039,11 +1040,14 @@ static char * krk_long_to_str(const KrkLong * n, int _base, const char * prefix,
 
 	char * rev = malloc(len);
 	char * out = rev;
+	uint32_t hash = 0;
 	while (writer != tmp) {
-		writer--;
-		*out++ = *writer;
+		*out = *--writer;
+		krk_hash_advance(hash,*out);
+		out++;
 	}
 	*out = '\0';
+	*_hash = hash;
 
 	free(tmp);
 
@@ -1377,8 +1381,9 @@ KRK_Method(long,__rtruediv__) {
 #define PRINTER(name,base,prefix) \
 	KRK_Method(long,__ ## name ## __) { \
 		size_t size; \
-		char * rev = krk_long_to_str(self->value, base, prefix, &size); \
-		return OBJECT_VAL(krk_takeString(rev,size)); \
+		uint32_t hash; \
+		char * rev = krk_long_to_str(self->value, base, prefix, &size, &hash); \
+		return OBJECT_VAL(krk_takeStringVetted(rev,size,size,KRK_OBJ_FLAGS_STRING_ASCII,hash)); \
 	}
 
 PRINTER(str,10,"")
diff --git a/src/obj_str.c b/src/obj_str.c
index 040a66b..9db5f8c 100644
--- a/src/obj_str.c
+++ b/src/obj_str.c
@@ -71,7 +71,7 @@ KRK_Method(str,__add__) {
 	/* Hashes can be extended, which saves us calculating the whole thing */
 	uint32_t hash = self->obj.hash;
 	for (size_t i = 0; i < bl; ++i) {
-		hash = (int)b[i] + (hash << 6) + (hash << 16) - hash;
+		krk_hash_advance(hash,b[i]);
 	}
 
 	KrkString * result = krk_takeStringVetted(chars, length, cpLength, type, hash);
@@ -390,7 +390,7 @@ KRK_Method(str,__mul__) {
 	for (krk_integer_type i = 0; i < howMany; ++i) {
 		for (size_t j = 0; j < self->length; ++j) {
 			*c = self->chars[j];
-			hash = (int)*c + (hash << 6) + (hash << 16) - hash;
+			krk_hash_advance(hash, *c);
 			c++;
 		}
 	}
diff --git a/src/object.c b/src/object.c
index 292fb1f..aedd961 100644
--- a/src/object.c
+++ b/src/object.c
@@ -8,6 +8,8 @@
 #include <kuroko/vm.h>
 #include <kuroko/table.h>
 
+#include "private.h"
+
 #define ALLOCATE_OBJECT(type, objectType) \
 	(type*)allocateObject(sizeof(type), objectType)
 
@@ -195,7 +197,7 @@ static uint32_t hashString(const char * key, size_t length) {
 	/* This is the so-called "sdbm" hash. It comes from a piece of
 	 * public domain code from a clone of ndbm. */
 	for (size_t i = 0; i < length; ++i) {
-		hash = (int)key[i] + (hash << 6) + (hash << 16) - hash;
+		krk_hash_advance(hash,key[i]);
 	}
 	return hash;
 }
diff --git a/src/private.h b/src/private.h
index ee08f27..4b2917d 100644
--- a/src/private.h
+++ b/src/private.h
@@ -62,3 +62,8 @@ struct ParsedFormatSpec {
 	int hasPrecision;
 	int fillSize;
 };
+
+/* We inline hashing in a few places, so it's nice to have this in one place.
+ * This is the "sdbm" hash. I've been using it in various places for many years,
+ * and this specific version apparently traces to gawk. */
+#define krk_hash_advance(hash,c) do { hash = (int)(c) + (hash << 6) + (hash << 16) - hash; } while (0)