From 1450a1030a65fa3ef0ce77639b1a875130f1847b Mon Sep 17 00:00:00 2001
From: Aaron Burrow <burrows@charstarstar.com>
Date: Sun, 16 Jun 2019 03:51:00 -0400
Subject: [PATCH] Add ClosestMatch.id

This field keeps IDs unique across multiple invocations of
`ClosestMatch::Add()`.
---
 closestmatch.go      | 25 +++++++++++++++++++------
 closestmatch_test.go | 23 +++++++++++++++++++++++
 2 files changed, 42 insertions(+), 6 deletions(-)

diff --git a/closestmatch.go b/closestmatch.go
index e933625..87def95 100755
--- a/closestmatch.go
+++ b/closestmatch.go
@@ -3,6 +3,7 @@ package closestmatch
 import (
 	"compress/gzip"
 	"encoding/json"
+	"math"
 	"math/rand"
 	"os"
 	"sort"
@@ -17,6 +18,7 @@ type ClosestMatch struct {
 	SubstringSizes []int
 	SubstringToID  map[string]map[uint32]struct{}
 	ID             map[uint32]IDInfo
+	IDx            uint32
 	mux            sync.Mutex
 }
 
@@ -28,19 +30,25 @@ type IDInfo struct {
 
 // New returns a new structure for performing closest matches
 func New(possible []string, subsetSize []int) *ClosestMatch {
+	if len(possible)-1 > math.MaxUint32 {
+		panic("List is too long. ClosestMatch::id will overflow.")
+	}
+
 	cm := new(ClosestMatch)
 	cm.SubstringSizes = subsetSize
 	cm.SubstringToID = make(map[string]map[uint32]struct{})
 	cm.ID = make(map[uint32]IDInfo)
-	for i, s := range possible {
+	cm.IDx = 0
+	for _, s := range possible {
 		substrings := cm.splitWord(strings.ToLower(s))
-		cm.ID[uint32(i)] = IDInfo{Key: s, NumSubstrings: len(substrings)}
+		cm.ID[cm.IDx] = IDInfo{Key: s, NumSubstrings: len(substrings)}
 		for substring := range substrings {
 			if _, ok := cm.SubstringToID[substring]; !ok {
 				cm.SubstringToID[substring] = make(map[uint32]struct{})
 			}
-			cm.SubstringToID[substring][uint32(i)] = struct{}{}
+			cm.SubstringToID[substring][cm.IDx] = struct{}{}
 		}
+		cm.IDx++
 	}
 
 	return cm
@@ -67,16 +75,21 @@ func Load(filename string) (*ClosestMatch, error) {
 
 // Add more words to ClosestMatch structure
 func (cm *ClosestMatch) Add(possible []string) {
+	if len(possible)-1 > math.MaxUint32 || uint32(len(possible)-1) > math.MaxUint32-cm.IDx {
+		panic("List is too long. ClosestMatch::id will overflow.")
+	}
+
 	cm.mux.Lock()
-	for i, s := range possible {
+	for _, s := range possible {
 		substrings := cm.splitWord(strings.ToLower(s))
-		cm.ID[uint32(i)] = IDInfo{Key: s, NumSubstrings: len(substrings)}
+		cm.ID[cm.IDx] = IDInfo{Key: s, NumSubstrings: len(substrings)}
 		for substring := range substrings {
 			if _, ok := cm.SubstringToID[substring]; !ok {
 				cm.SubstringToID[substring] = make(map[uint32]struct{})
 			}
-			cm.SubstringToID[substring][uint32(i)] = struct{}{}
+			cm.SubstringToID[substring][cm.IDx] = struct{}{}
 		}
+		cm.IDx++
 	}
 	cm.mux.Unlock()
 }
diff --git a/closestmatch_test.go b/closestmatch_test.go
index 1a19789..e77518a 100755
--- a/closestmatch_test.go
+++ b/closestmatch_test.go
@@ -197,3 +197,26 @@ func TestSaveLoad(t *testing.T) {
 		t.Errorf("Differing answers: '%s' '%s'", answer1, answer2)
 	}
 }
+
+func TestMultipleAddInvocations(t *testing.T) {
+	cm := New([]string{}, []int{2})
+	for _, x := range []string{"uppermost", "up", "uppity"} {
+		cm.Add([]string{x})
+	}
+	if cm.Closest("uppermost") != "uppermost" {
+		t.Errorf("Should have been an exact match.")
+	}
+}
+
+func TestAddAfterLoad(t *testing.T) {
+	cm := New([]string{"Darth", "Vader", "loves", "Doritos"}, []int{2})
+	cm.Save("test/vader.cm.gz")
+	cm, err := Load("test/vader.cm.gz")
+	if err != nil {
+		t.Errorf("Load should succeed")
+	}
+	cm.Add([]string{"Elephant"})
+	if cm.Closest("Darth") != "Darth" {
+		t.Errorf("Should have been an exact match.")
+	}
+}