aboutsummaryrefslogtreecommitdiff
path: root/kbbi_client.go
diff options
context:
space:
mode:
Diffstat (limited to 'kbbi_client.go')
-rw-r--r--kbbi_client.go417
1 files changed, 417 insertions, 0 deletions
diff --git a/kbbi_client.go b/kbbi_client.go
new file mode 100644
index 0000000..23767b4
--- /dev/null
+++ b/kbbi_client.go
@@ -0,0 +1,417 @@
+// Copyright 2020, Shulhan <ms@kilabit.info>. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package kamusku
+
+import (
+ "bytes"
+ "encoding/gob"
+ "errors"
+ "fmt"
+ "io/ioutil"
+ "log"
+ "net/http"
+ "net/http/cookiejar"
+ "net/url"
+ "os"
+ "path/filepath"
+ "strconv"
+ "strings"
+ "time"
+
+ "github.com/shuLhan/share/lib/debug"
+ libhttp "github.com/shuLhan/share/lib/http"
+ "github.com/shuLhan/share/lib/net/html"
+ "golang.org/x/net/publicsuffix"
+)
+
+const (
+ kbbiUrlBase = "https://kbbi.kemdikbud.go.id"
+ kbbiUrlLogin = kbbiUrlBase + "/Account/Login"
+ kbbiPathEntri = "/entri/"
+
+ attrNameClass = "class"
+ attrNameHref = "href"
+ attrNameTitle = "title"
+ attrNameValue = "value"
+
+ attrValueRootWord = "rootword"
+
+ paramNameMasukan = "masukan"
+ paramNameMasukanLengkap = "masukanLengkap"
+ paramNameIngatSaya = "IngatSaya"
+ paramNameKataSandi = "KataSandi"
+ paramNamePage = "page"
+ paramNamePosel = "Posel"
+ paramNameRequestVerificationToken = "__RequestVerificationToken" //nolint: gosec
+
+ paramValueDasar = "dasar"
+ paramValueFalse = "false"
+
+ tagNameAnchor = "a"
+ tagNameFont = "font"
+ tagNameHeader2 = "h2"
+ tagNameInput = "input"
+ tagNameItalic = "i"
+ tagNameOrderedList = "ol"
+ tagNameSpan = "span"
+ tagNameUnorderedList = "ul"
+
+ cookieFile = "cookie"
+ configDir = "kamusku"
+ defTimeout = 20 * time.Second
+ maxPageNumber = 501
+)
+
+//
+// KbbiClient client for official KBBI web using HTTP.
+//
+type KbbiClient struct {
+ baseDir string
+ cookieURL *url.URL
+ cookies []*http.Cookie
+ httpc *http.Client
+}
+
+//
+// NewKbbiClient create and initialize new client that connect directly to
+// KBBI official website.
+//
+func NewKbbiClient() (cl *KbbiClient, err error) {
+ cookieURL, err := url.Parse(kbbiUrlBase)
+ if err != nil {
+ return nil, fmt.Errorf("New: %w", err)
+ }
+
+ jarOpt := &cookiejar.Options{
+ PublicSuffixList: publicsuffix.List,
+ }
+
+ jar, err := cookiejar.New(jarOpt)
+ if err != nil {
+ return nil, fmt.Errorf("New: %w", err)
+ }
+
+ cl = &KbbiClient{
+ cookieURL: cookieURL,
+ httpc: &http.Client{
+ Jar: jar,
+ Timeout: defTimeout,
+ },
+ }
+
+ err = cl.loadCookies()
+ if err != nil {
+ return nil, fmt.Errorf("New: %w", err)
+ }
+
+ if cl.cookies != nil {
+ jar.SetCookies(cookieURL, cl.cookies)
+ }
+
+ return cl, nil
+}
+
+//
+// Lookup lookup definition of one or more words.
+//
+func (cl *KbbiClient) Lookup(ins []string) (res LookupResponse, err error) {
+ res = make(LookupResponse, len(ins))
+
+ for _, in := range ins {
+ _, ok := res[in]
+ if ok {
+ continue
+ }
+
+ kata := &Word{}
+ res[in] = kata
+
+ entriURL := kbbiUrlBase + kbbiPathEntri + in
+ httpRes, err := cl.httpc.Get(entriURL)
+ if err != nil {
+ kata.err = err
+ continue
+ }
+
+ defer httpRes.Body.Close()
+
+ body, err := ioutil.ReadAll(httpRes.Body)
+ if err != nil {
+ kata.err = err
+ continue
+ }
+
+ if debug.Value >= 3 {
+ fmt.Printf(">>> HTML body for %s:\n%s", entriURL, body)
+ }
+
+ err = kata.parseHTMLEntri(in, body)
+ if err != nil {
+ kata.err = err
+ }
+
+ if len(kata.Definition) == 0 && len(kata.Message) == 0 {
+ kata.Message = "Entri tidak ditemukan"
+ }
+ }
+
+ return res, nil
+}
+
+//
+// ListRootWords list all of the root words in dictionary.
+//
+func (cl *KbbiClient) ListRootWords() (rootWords Words, err error) {
+ params := url.Values{
+ paramNameMasukan: []string{paramValueDasar},
+ paramNameMasukanLengkap: []string{paramValueDasar},
+ }
+
+ urlPage := kbbiUrlBase + "/Cari/Jenis?"
+
+ rootWords = make(Words)
+
+ for pageNumber := 1; pageNumber <= maxPageNumber; pageNumber++ {
+ params.Set(paramNamePage, strconv.Itoa(pageNumber))
+
+ req, err := http.NewRequest(http.MethodGet, urlPage+params.Encode(), nil)
+ if err != nil {
+ return rootWords, err
+ }
+
+ res, err := cl.httpc.Do(req)
+ if err != nil {
+ return rootWords, fmt.Errorf("ListRootWords: page %d: %w",
+ pageNumber, err)
+ }
+
+ defer res.Body.Close()
+
+ body, err := ioutil.ReadAll(res.Body)
+ if err != nil {
+ return rootWords, fmt.Errorf("ListRootWords: page %d: %w",
+ pageNumber, err)
+ }
+
+ got, err := cl.parseHTMLRootWords(body)
+ if err != nil {
+ return rootWords, fmt.Errorf("ListRootWords: page %d: %w",
+ pageNumber, err)
+ }
+ if len(got) == 0 {
+ break
+ }
+
+ rootWords.merge(got)
+
+ log.Printf("ListRootWords: halaman %d, jumlah kata %d, total kata %d",
+ pageNumber, len(got), len(rootWords))
+ }
+
+ return rootWords, nil
+}
+
+//
+// IsAuthenticated will return true if the client already login; otherwise it
+// will return false.
+//
+func (cl *KbbiClient) IsAuthenticated() bool {
+ return len(cl.cookies) > 0
+}
+
+//
+// Login authenticate the client using user email and password.
+//
+func (cl *KbbiClient) Login(email, pass string) (err error) {
+ tokenLogin, err := cl.preLogin()
+ if err != nil {
+ return fmt.Errorf("Login: %w", err)
+ }
+
+ params := url.Values{
+ paramNameRequestVerificationToken: []string{tokenLogin},
+ paramNamePosel: []string{email},
+ paramNameKataSandi: []string{pass},
+ paramNameIngatSaya: []string{paramValueFalse},
+ }
+
+ reqBody := strings.NewReader(params.Encode())
+
+ req, err := http.NewRequest(http.MethodPost, kbbiUrlLogin, reqBody)
+ if err != nil {
+ return fmt.Errorf("Login: %w", err)
+ }
+
+ req.Header.Set(libhttp.HeaderContentType, libhttp.ContentTypeForm)
+
+ res, err := cl.httpc.Do(req)
+ if err != nil {
+ return fmt.Errorf("Login: %w", err)
+ }
+
+ defer res.Body.Close()
+
+ resBody, err := ioutil.ReadAll(res.Body)
+ if err != nil {
+ return fmt.Errorf("Login: %w", err)
+ }
+
+ if res.StatusCode >= http.StatusBadRequest {
+ return fmt.Errorf("login: %d %s", res.StatusCode, resBody)
+ }
+
+ cl.cookies = cl.httpc.Jar.Cookies(cl.cookieURL)
+ cl.setCookies()
+ cl.saveCookies()
+
+ return nil
+}
+
+//
+// setCookies for HTTP request that need an authentication.
+//
+func (cl *KbbiClient) setCookies() {
+ cl.httpc.Jar.SetCookies(cl.cookieURL, cl.cookies)
+}
+
+func (cl *KbbiClient) parseHTMLRootWords(htmlBody []byte) (
+ rootWords Words, err error,
+) {
+ iter, err := html.Parse(bytes.NewReader(htmlBody))
+ if err != nil {
+ return nil, err
+ }
+
+ rootWords = make(Words)
+
+ for node := iter.Next(); node != nil; node = iter.Next() {
+ if !node.IsElement() {
+ continue
+ }
+ if node.Data != tagNameAnchor {
+ continue
+ }
+ hrefValue := node.GetAttrValue(attrNameHref)
+ if !strings.HasPrefix(hrefValue, kbbiPathEntri) {
+ continue
+ }
+ k := strings.TrimSpace(node.FirstChild.Data)
+ rootWords[k] = struct{}{}
+ }
+
+ return rootWords, nil
+}
+
+//
+// parseHTMLLogin get the token at the form login.
+//
+func (cl *KbbiClient) parseHTMLLogin(htmlBody []byte) (
+ token string, err error,
+) {
+ iter, err := html.Parse(bytes.NewReader(htmlBody))
+ if err != nil {
+ return "", err
+ }
+
+ for node := iter.Next(); node != nil; node = iter.Next() {
+ if !node.IsElement() {
+ continue
+ }
+ if node.Data != tagNameInput {
+ continue
+ }
+
+ token := node.GetAttrValue(attrNameValue)
+ if len(token) > 0 {
+ return token, nil
+ }
+ }
+
+ return "", fmt.Errorf("token login not found")
+}
+
+//
+// preLogin initialize the client to get the first cookie.
+//
+func (cl *KbbiClient) preLogin() (token string, err error) {
+ req, err := http.NewRequest(http.MethodGet, kbbiUrlLogin, nil)
+ if err != nil {
+ return "", err
+ }
+
+ res, err := cl.httpc.Do(req)
+ if err != nil {
+ return "", err
+ }
+
+ defer res.Body.Close()
+
+ body, err := ioutil.ReadAll(res.Body)
+ if err != nil {
+ return "", err
+ }
+
+ token, err = cl.parseHTMLLogin(body)
+ if err != nil {
+ return "", err
+ }
+
+ return token, nil
+}
+
+//
+// loadCookies load the KBBI cookies from file.
+//
+func (cl *KbbiClient) loadCookies() (err error) {
+ cl.baseDir, err = os.UserConfigDir()
+ if err != nil {
+ return fmt.Errorf("loadCookies: %w", err)
+ }
+
+ f := filepath.Join(cl.baseDir, configDir, cookieFile)
+
+ _, err = os.Stat(f)
+ if errors.Is(err, os.ErrNotExist) {
+ return nil
+ }
+
+ body, err := ioutil.ReadFile(f)
+ if err != nil {
+ return fmt.Errorf("loadCookies: %w", err)
+ }
+
+ dec := gob.NewDecoder(bytes.NewReader(body))
+
+ err = dec.Decode(&cl.cookies)
+ if err != nil {
+ return fmt.Errorf("loadCookies: %w", err)
+ }
+
+ return nil
+}
+
+//
+// saveCookies store the client cookies to the file for future use.
+//
+func (cl *KbbiClient) saveCookies() {
+ err := os.MkdirAll(filepath.Join(cl.baseDir, configDir), 0700)
+ if err != nil {
+ log.Println("saveCookies:", err)
+ }
+
+ f := filepath.Join(cl.baseDir, configDir, cookieFile)
+
+ var buf bytes.Buffer
+ enc := gob.NewEncoder(&buf)
+ err = enc.Encode(cl.cookies)
+ if err != nil {
+ log.Println("saveCookies: ", err)
+ }
+
+ err = ioutil.WriteFile(f, buf.Bytes(), 0600)
+ if err != nil {
+ log.Println("saveCookies: ", err)
+ }
+}
generated by cgit v1.3 (git 2.53.0) at 2026-04-15 04:09:43 +0000