Implements post query

This commit is contained in:
Mikhail Klementyev 2016-11-20 13:22:00 +03:00
parent 03756cafb3
commit df10b44fe7
3 changed files with 287 additions and 50 deletions

View file

@ -9,73 +9,88 @@
package commands
import (
"bytes"
"database/sql"
"strings"
"fmt"
"io/ioutil"
"log"
"net/http"
"net/url"
"github.com/jollheef/wi/storage"
"github.com/PuerkitoBio/goquery"
"github.com/jaytaylor/html2text"
"golang.org/x/net/html"
"golang.org/x/net/html/charset"
)
func parseLink(db *sql.DB, oldPage, value string, lastUrl *url.URL) (htmlPage string, err error) {
linkUrl, err := lastUrl.Parse(value)
if err != nil {
return
}
func fixLinks(db *sql.DB, doc *goquery.Document, pageUrl *url.URL) (err error) {
linkNo, err := storage.GetLinkID(db, linkUrl.String())
if err != nil {
linkNo, err = storage.AddLink(db, linkUrl.String())
doc.Find("a").Each(func(i int, s *goquery.Selection) {
url, exists := s.Attr("href")
if !exists {
return
}
linkUrl, err := pageUrl.Parse(url)
if err != nil {
return
}
}
htmlPage = oldPage
linkNo, err := storage.GetLinkID(db, linkUrl.String())
if err != nil {
linkNo, err = storage.AddLink(db, linkUrl.String())
if err != nil {
log.Fatalln("Add link:", err)
}
}
for _, s := range []string{value, html.EscapeString(value)} {
htmlPage = strings.Replace(htmlPage, "\""+s+"\"",
"\""+fmt.Sprintf("%d", linkNo)+"\"", -1)
}
s.SetAttr("href", fmt.Sprintf("%d", linkNo))
})
return
}
func parseLinks(db *sql.DB, body []byte, lastUrl *url.URL) (htmlPage string, err error) {
htmlPage = string(body)
func fixForms(db *sql.DB, doc *goquery.Document, pageUrl *url.URL) (err error) {
z := html.NewTokenizer(bytes.NewReader(body))
for {
tt := z.Next()
if tt == html.ErrorToken {
break
}
for {
key, value, moreAttr := z.TagAttr()
if string(key) == "href" {
htmlPage, err = parseLink(db, htmlPage, string(value), lastUrl)
if err != nil {
doc.Find("form").Each(func(i int, s *goquery.Selection) {
var fields []storage.Field
s.Find("input").Map(
func(i int, s *goquery.Selection) (str string) {
f := storage.Field{}
var exists bool
f.Name, exists = s.Attr("name")
if !exists {
return
}
}
f.Value, _ = s.Attr("value")
hidden, _ := s.Attr("type")
if hidden == "hidden" {
f.Hidden = true
}
fields = append(fields, f)
return
})
if !moreAttr {
break
action, _ := s.Attr("action")
actionUrl, err := pageUrl.Parse(action)
if err != nil {
return
}
method, _ := s.Attr("method")
formNo, err := storage.GetFormID(db, fields, actionUrl.String(), method)
if err != nil {
formNo, err = storage.AddForm(db, fields, actionUrl.String(), method)
if err != nil {
log.Fatalln(err)
}
}
}
s.AppendHtml(fmt.Sprintf("(%d %s)", formNo, strings.ToUpper(method)))
})
return
}
@ -94,8 +109,12 @@ func Get(db *sql.DB, linkUrl string) {
linkUrl = "https://" + linkUrl
}
// TODO Full url encoding
req, err := http.NewRequest("GET", strings.Replace(linkUrl, " ", "%20", -1), nil)
u, err := url.Parse(linkUrl)
if err != nil {
log.Fatalln(err)
}
req, err := http.NewRequest("GET", u.String(), nil)
if err != nil {
log.Fatalln(err)
}
@ -120,14 +139,24 @@ func Get(db *sql.DB, linkUrl string) {
log.Fatalln("Encoding error:", err)
}
body, err := ioutil.ReadAll(utf8)
doc, err := goquery.NewDocumentFromReader(utf8)
if err != nil {
log.Fatalln("IO error:", err)
log.Fatalln("Create document error:", err)
}
htmlPage, err := parseLinks(db, body, lastUrl)
err = fixLinks(db, doc, lastUrl)
if err != nil {
log.Fatalln("Parse links error:", err)
log.Fatalln("Fix links error:", err)
}
err = fixForms(db, doc, lastUrl)
if err != nil {
log.Fatalln("Fix forms error", err)
}
htmlPage, err := doc.Html()
if err != nil {
log.Fatalln("Convert to html error:", err)
}
text, err := html2text.FromString(htmlPage)
@ -139,8 +168,72 @@ func Get(db *sql.DB, linkUrl string) {
fmt.Println(text)
}
func Post(db *sql.DB, formID int64, formArgs []string) {
fmt.Println("Not implemented")
func Form(db *sql.DB, formID int64, formArgs []string) {
fields, formUrl, post, err := storage.GetForm(db, formID)
if err != nil {
log.Fatalln("Get form:", err)
}
if len(formArgs) == 0 {
if post {
fmt.Print("POST ")
}
fmt.Println(formUrl)
fmt.Print("Values: ")
for i, f := range fields {
if i != 0 {
fmt.Print("\n\t")
}
fmt.Printf(`%s="%s"`, f.Name, f.Value)
}
fmt.Println()
return
}
urlData := url.Values{}
for _, f := range fields {
urlData.Set(f.Name, f.Value)
}
for _, fa := range formArgs {
nameAndValue := strings.Split(fa, " ")
if len(nameAndValue) != 2 {
continue
}
name := nameAndValue[0]
value := nameAndValue[1]
urlData.Set(name, value)
}
client := &http.Client{}
var lastUrl *url.URL
client.CheckRedirect = func(r *http.Request, via []*http.Request) (err error) {
lastUrl = r.URL
return
}
resp, err := client.PostForm(formUrl, urlData)
if err != nil {
fmt.Println(err)
}
if lastUrl == nil {
lastUrl, _ = resp.Location()
}
log.Println(resp.Status)
var status int64
fmt.Sscanf(resp.Status, "%d", &status)
if status >= 300 && status < 400 {
Get(db, lastUrl.String())
}
}
func Link(db *sql.DB, linkID int64, fromHistory bool) {