Implement dictionary data serialization to improve startup times

dict/.gitignore

@@ -1 +1,2 @@
 JMdict.xml
+dict.bin

dict/main.go

@@ -1,6 +1,7 @@
 package main
 
 import (
+	"encoding/gob"
 	"encoding/json"
 	"fmt"
 	"log"
@@ -13,18 +14,51 @@ import (
 	"github.com/gorilla/mux"
 )
 
-var dict jmdict.Jmdict
+var dict map[string]Entry
 
 func LoadDict() error {
+	// Loading from binary
+	const binaryFile = "dict.bin"
+	file, err := os.Open(binaryFile)
+	if err != nil {
+		if !os.IsNotExist(err) {
+			return err
+		}
+	} else {
+		defer file.Close()
+		decoder := gob.NewDecoder(file)
+		err = decoder.Decode(&dict)
+		return err
+	}
+
+	// Loading from JMdict
 	const jmdictFile = "JMdict.xml"
 	reader, err := os.Open(jmdictFile)
 	if err != nil {
 		return err
 	}
-	dict, _, err = jmdict.LoadJmdict(reader)
+	jmdict, _, err := jmdict.LoadJmdict(reader)
 	if err != nil {
 		return err
 	}
+	dict = make(map[string]Entry)
+	for _, jmdictEntry := range jmdict.Entries {
+		entry := ParseEntry(&jmdictEntry)
+		dict[entry.Kanji] = entry
+	}
+
+	// Encoding to binary
+	file, err = os.Create(binaryFile)
+	if err != nil {
+		return err
+	}
+	defer file.Close()
+	encoder := gob.NewEncoder(file)
+	err = encoder.Encode(&dict)
+	if err != nil {
+		return err
+	}
+
 	return nil
 }
 
@@ -39,7 +73,7 @@ type Definition struct {
 	PartOfSpeech []string
 }
 
-func ParseEntry(entry jmdict.JmdictEntry) Entry {
+func ParseEntry(entry *jmdict.JmdictEntry) Entry {
 	kanji := ""
 	if len(entry.Kanji) > 0 {
 		kanji = entry.Kanji[0].Expression
@@ -75,26 +109,22 @@ func Search(query string) queryResult {
 	otherResults := make([]Entry, 0)
 	truncated := false
 	count := 0
-	for _, jmdictEntry := range dict.Entries {
+	for kanji := range dict {
 		exactMatch := false
-		for _, kanji := range jmdictEntry.Kanji {
+		entry := dict[kanji]
-			if kanji.Expression == query {
+		if kanji == query {
-				exactMatch = true
+			exactMatch = true
-				goto match
+			goto match
-			}
+		}
-			if strings.Contains(kanji.Expression, query) {
+		if strings.Contains(kanji, query) {
-				goto match
+			goto match
-			}
 		}
 		// TODO: Skip if query contains kanji
-		for _, reading := range jmdictEntry.Readings {
+		if strings.Contains(entry.Reading, query) {
-			if strings.Contains(reading.Reading, query) {
+			goto match
-				goto match
-			}
 		}
 		continue
 	match:
-		entry := ParseEntry(jmdictEntry)
 		if exactMatch {
 			exactResults = append(exactResults, entry)
 		} else {
@@ -116,9 +146,8 @@ func Search(query string) queryResult {
 }
 
 func Lookup(word string) *Entry {
-	for _, jmdictEntry := range dict.Entries {
+	for kanji, entry := range dict {
-		entry := ParseEntry(jmdictEntry)
+		if kanji == word {
-		if entry.Kanji == word {
 			return &entry
 		}
 	}