jichan.org-go/dict/main.go

package main

import (
	"bufio"
	"encoding/gob"
	"encoding/json"
	"fmt"
	"log"
	"net/http"
	"os"
	"strconv"
	"strings"

	"foosoft.net/projects/jmdict"
	"git.elnu.com/ElnuDev/jichanorg/httputils"
	"github.com/gorilla/mux"
)

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
	}
	jmdict, _, err := jmdict.LoadJmdict(reader)
	if err != nil {
		return err
	}

	const jmdictFuriganaFile = "JmdictFurigana.txt"
	reader, err = os.Open(jmdictFuriganaFile)
	if err != nil {
		return err
	}
	scanner := bufio.NewScanner(reader)
	furiganaData := make(map[string]string)
	for scanner.Scan() {
		params := strings.Split(scanner.Text(), "|")
		furiganaData[params[0]] = params[2]
	}

	dict = make(map[string]Entry)
	for _, jmdictEntry := range jmdict.Entries {
		// お願い致します|おねがいいたします|1:ねが;3:いた
		var furiganaInfo *string
		if len(jmdictEntry.Kanji) > 0 {
			data := furiganaData[jmdictEntry.Kanji[0].Expression]
			furiganaInfo = &data
		} else {
			furiganaInfo = nil
		}
		entry := ParseEntry(&jmdictEntry, furiganaInfo)
		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
}

type Furigana struct {
	Kanji    string
	Furigana string
}

type Entry struct {
	Kanji string
	// Mapping of character index to furigana
	Furigana    []Furigana
	Reading     string
	Definitions []Definition
}

type Definition struct {
	Definition   string
	PartOfSpeech []string
}

func ParseEntry(entry *jmdict.JmdictEntry, furiganaInfo *string) Entry {
	kanji := ""
	if len(entry.Kanji) > 0 {
		kanji = entry.Kanji[0].Expression
	} else {
		kanji = entry.Readings[0].Reading
	}
	reading := ""
	if kanji != "" {
		reading = entry.Readings[0].Reading
	}
	var definitions []Definition
	definitions = make([]Definition, len(entry.Sense))
	for i, sense := range entry.Sense {
		definition := sense.Glossary[0].Content
		if len(sense.Glossary) > 1 {
			for _, glossary := range sense.Glossary[1:] {
				definition += "; " + glossary.Content
			}
		}
		definitions[i] = Definition{
			Definition:   definition,
			PartOfSpeech: sense.PartsOfSpeech,
		}
	}
	// 1:ねが;3:いた
	var furiganaList []Furigana
	if reading == "" || furiganaInfo == nil || *furiganaInfo == "" {
		furiganaList = []Furigana{{Kanji: reading, Furigana: ""}}
	} else {
		furiganaEntries := strings.Split(*furiganaInfo, ";")
		// ["1:ねが", "3:いた"]
		type rawFurigana struct {
			from     int
			to       int
			furigana string
		}
		ruby := make([]rawFurigana, 0)
		for _, entry := range furiganaEntries {
			// 1:ねが
			// multiple: 0-1:きょう
			params := strings.Split(entry, ":")
			// ["1", "ねが"]
			// multiple: ["0-1", "きょう"]
			indexRange := strings.Split(params[0], "-")
			// [1]
			// multiple: [0, 1]
			var from, to int
			if len(indexRange) == 1 {
				index, _ := strconv.Atoi(indexRange[0])
				from, to = index, index
			} else {
				from, _ = strconv.Atoi(indexRange[0])
				to, _ = strconv.Atoi(indexRange[1])
			}
			ruby = append(ruby, rawFurigana{
				from:     from,
				to:       to,
				furigana: params[1],
			})
		}
		furiganaList = make([]Furigana, 0)
		slice := func(from, to int) string {
			return string([]rune(kanji)[from : to+1])
		}
		nextIndex := 0
		for _, raw := range ruby {
			if raw.from > nextIndex {
				furiganaList = append(furiganaList, Furigana{
					Kanji:    slice(nextIndex, raw.from-1),
					Furigana: "",
				})
			}
			furiganaList = append(furiganaList, Furigana{
				Kanji:    slice(raw.from, raw.to),
				Furigana: raw.furigana,
			})
			nextIndex = raw.to + 1
		}
		length := len([]rune(kanji))
		if nextIndex < length {
			furiganaList = append(furiganaList, Furigana{
				Kanji:    slice(nextIndex, length-1),
				Furigana: "",
			})
		}
	}
	return Entry{
		Kanji:       kanji,
		Furigana:    furiganaList,
		Reading:     reading,
		Definitions: definitions,
	}
}

func Search(query string) queryResult {
	query = strings.TrimSpace(query)
	exactResults := make([]Entry, 0)
	otherResults := make([]Entry, 0)
	truncated := false
	count := 0
	for kanji := range dict {
		exactMatch := false
		entry := dict[kanji]
		if kanji == query {
			exactMatch = true
			goto match
		}
		if strings.Contains(kanji, query) {
			goto match
		}
		// TODO: Skip if query contains kanji
		if strings.Contains(entry.Reading, query) {
			goto match
		}
		continue
	match:
		if exactMatch {
			exactResults = append(exactResults, entry)
		} else {
			otherResults = append(otherResults, entry)
		}
		count++
		if count >= 500 {
			truncated = true
			break
		}
	}
	return queryResult{
		Query:        query,
		ExactResults: exactResults,
		OtherResults: otherResults,
		Truncated:    truncated,
		Count:        len(exactResults) + len(otherResults),
	}
}

func Lookup(word string) *Entry {
	for kanji, entry := range dict {
		if kanji == word {
			return &entry
		}
	}
	return nil
}

type queryResult struct {
	// Fields must be capitalized
	// to be accessible in templates
	Query        string
	ExactResults []Entry
	OtherResults []Entry
	Truncated    bool
	Count        int
}

func main() {
	err := LoadDict()
	if err != nil {
		fmt.Println(err)
		return
	}
	fmt.Println("JMdict loaded!")
	r := mux.NewRouter()
	r.HandleFunc("/", httputils.GenerateHandler(
		func(w http.ResponseWriter, r *http.Request) bool { return true },
		httputils.NewTemplateSet("index.html"),
		func(w http.ResponseWriter, r *http.Request) (string, any) { return "index.html", nil },
		[]string{http.MethodGet},
	))
	rawSearchHandler := func(w http.ResponseWriter, r *http.Request) {
		r.ParseMultipartForm(0)
		q := r.FormValue("q")
		var redirect string
		if q == "" {
			redirect = "/"
		} else {
			redirect = "/search/" + q
		}
		http.Redirect(w, r, redirect, http.StatusMovedPermanently)
	}
	r.HandleFunc("/search", rawSearchHandler)
	r.HandleFunc("/search/", rawSearchHandler)
	r.HandleFunc("/search/{query}", httputils.GenerateHandler(
		// handler whether or not to use template
		func(w http.ResponseWriter, r *http.Request) bool {
			// If Accept: applicaiton/json we'll use the template
			if r.Header.Get("Accept") != "application/json" {
				return true
			}

			// Otherwise, let's send JSON
			query := mux.Vars(r)["query"]
			result := Search(query)
			jsonBytes, _ := json.Marshal(append(result.ExactResults, result.OtherResults...))

			w.Header().Set("Content-Type", "application/json; charset=utf-8")
			fmt.Fprint(w, string(jsonBytes))

			return false
		},
		httputils.NewTemplateSet("index.html", "search.html"),
		// template data
		func(w http.ResponseWriter, r *http.Request) (template string, data any) {
			if r.Header.Get("HX-Request") == "" {
				template = "search.html"
			} else {
				template = "search"
			}
			// Only runs if handler returns true
			query := mux.Vars(r)["query"]
			data = Search(query)
			return
		},
		[]string{http.MethodGet},
	))
	rawWordHandler := func(w http.ResponseWriter, r *http.Request) {
		fmt.Println("Redirecting raw word handler")
		http.Redirect(w, r, "/", http.StatusMovedPermanently)
	}
	r.HandleFunc("/word", rawWordHandler)
	r.HandleFunc("/word/", rawWordHandler)
	r.HandleFunc("/word/{word}", httputils.GenerateHandler(
		func(w http.ResponseWriter, r *http.Request) bool { return true },
		// Order matters
		// word.html overrided the results block in index.html
		// so should be loaded second
		httputils.NewTemplateSet("index.html", "word.html"),
		func(w http.ResponseWriter, r *http.Request) (template string, data any) {
			template = "word.html"
			query := mux.Vars(r)["word"]
			data = struct {
				Query any
				Entry *Entry
			}{
				Query: nil,
				Entry: Lookup(query),
			}
			return
		},
		[]string{http.MethodGet},
	))
	r.Handle("/", http.FileServer(http.Dir("static")))
	log.Fatal(http.ListenAndServe(":3334", r))
}