From 841d82421646408ca4cc3a8b13ab48427d6836c6 Mon Sep 17 00:00:00 2001
From: Alexander Medvednikov <alexander@medvednikov.com>
Date: Tue, 1 Oct 2019 02:14:12 +0300
Subject: [PATCH] builtin: a very early version of the hashmap

---
 compiler/main.v             |  1 +
 vlib/builtin/hashmap.v      | 80 +++++++++++++++++++++++++++++++++++++
 vlib/builtin/hashmap_test.v | 33 +++++++++++++++
 vlib/strings/strings.v      |  1 +
 4 files changed, 115 insertions(+)
 create mode 100644 vlib/builtin/hashmap.v
 create mode 100644 vlib/builtin/hashmap_test.v

diff --git a/compiler/main.v b/compiler/main.v
index d5a2f4fe62..56e92e176a 100644
--- a/compiler/main.v
+++ b/compiler/main.v
@@ -875,6 +875,7 @@ fn new_v(args[]string) &V {
 	'int.v',
 	'utf8.v',
 	'map.v',
+	'hashmap.v',
 	'option.v',
 	]
 	//println(builtins)
diff --git a/vlib/builtin/hashmap.v b/vlib/builtin/hashmap.v
new file mode 100644
index 0000000000..c0ebb5a1ab
--- /dev/null
+++ b/vlib/builtin/hashmap.v
@@ -0,0 +1,80 @@
+// Copyright (c) 2019 Alexander Medvednikov. All rights reserved.
+// Use of this source code is governed by an MIT license
+// that can be found in the LICENSE file.
+
+module builtin
+
+/*
+	This is work in progress.
+	A very early test version of the hashmap with a fixed size.
+	Only works with string keys and int values for now.
+*/
+
+import math
+
+struct hashmap {
+	cap          int
+	keys         []string
+	table        []hashmapentry
+	elm_size int
+pub:
+	nr_collisions int
+}
+
+struct hashmapentry {
+        key string
+        val int
+        next &hashmapentry  // linked list for collisions
+}
+
+const (
+	min_cap = 2 << 10
+	max_cap = 2 << 20
+)
+
+fn new_hashmap(planned_nr_items int) hashmap {
+	mut cap := planned_nr_items * 3
+	if cap < min_cap {
+		cap = min_cap
+	}	
+	if cap > max_cap {
+		cap = max_cap
+	}	
+	return hashmap{
+		cap: cap
+		elm_size: 4
+		table: _make(cap, cap, sizeof(hashmapentry))
+	}	
+}	
+
+fn (m mut hashmap) set(key string, val int) {
+	hash := int(math.abs(key.hash()))
+	idx := hash % m.cap
+	if m.table[idx].key.len != 0 {
+		//println('\nset() idx=$idx key="$key" hash="$hash" val=$val')
+		m.nr_collisions++
+		//println('collision:' + m.table[idx].key)
+		mut e := &m.table[idx]
+		for e.next != 0 {
+			e = e.next
+		}	
+		e.next = &hashmapentry{key, val, 0}
+	} else {
+		m.table[idx] = hashmapentry{key, val, 0}
+	}
+}	
+
+fn (m mut hashmap) get(key string) int {
+	hash := int(math.abs(key.hash()))
+	idx := hash % m.cap
+	mut e := &m.table[idx]
+	for e.next != 0 { // todo unsafe {
+		if e.key == key {
+			return e.val
+		}	
+		e = e.next
+	}	
+	return e.val
+}
+
+
diff --git a/vlib/builtin/hashmap_test.v b/vlib/builtin/hashmap_test.v
new file mode 100644
index 0000000000..4bb588342d
--- /dev/null
+++ b/vlib/builtin/hashmap_test.v
@@ -0,0 +1,33 @@
+import rand
+import strings
+
+fn test_random_strings() {
+	mut m := new_hashmap(1000)
+	for i in 0..1000 {
+		mut buf := []byte
+		for j in 0..10 {
+			buf << byte(rand.next(int(`z`) - int(`a`)) + `a`)
+		}
+		s := string(buf)
+		//println(s)
+		m.set(s, i)
+		assert m.get(s) == i
+	}
+	m.set('foo', 12)
+	val := m.get('foo')
+	assert val == 12
+}	
+
+fn test_large_hashmap() {
+	N := 300 * 1000
+	mut nums := new_hashmap(N)
+	for i := 0; i < N; i++ {
+	        key := i.str()
+	        nums.set(key, i)
+	}
+	println('nr collisions: $nums.nr_collisions')
+	for i := 0; i < N; i++ {
+		key := i.str()
+		assert nums.get(key) == i
+	}
+}	
diff --git a/vlib/strings/strings.v b/vlib/strings/strings.v
index fcb8044f9b..d0507142ea 100644
--- a/vlib/strings/strings.v
+++ b/vlib/strings/strings.v
@@ -8,3 +8,4 @@ pub fn repeat(c byte, n int) string {
 	arr[n] = `\0`
 	return string(arr, n)
 }
+