From 936d157785b04158ca1e03074ba45a5d410db7db Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Musab=20G=C3=BCltekin?= <gultekinmusab@gmail.com>
Date: Tue, 18 Jun 2019 13:03:00 +0300
Subject: [PATCH] Revert "Try parsing HTML even if content-type is empty."

This reverts commit f384fc2c
---
 middleware.go |  2 +-
 response.go   | 14 ++++++++++++++
 2 files changed, 15 insertions(+), 1 deletion(-)

diff --git a/middleware.go b/middleware.go
index ea502f9..432ed4b 100644
--- a/middleware.go
+++ b/middleware.go
@@ -43,7 +43,7 @@ func defaultHeadersMiddleware(g *Geziyor, r *Request) {
 
 // parseHTMLMiddleware parses response if response is HTML
 func parseHTMLMiddleware(g *Geziyor, r *Response) {
-	if !g.Opt.ParseHTMLDisabled {
+	if !g.Opt.ParseHTMLDisabled && r.isHTML() {
 		r.DocHTML, _ = goquery.NewDocumentFromReader(bytes.NewReader(r.Body))
 	}
 }
diff --git a/response.go b/response.go
index e216d95..1aa3ed3 100644
--- a/response.go
+++ b/response.go
@@ -4,6 +4,7 @@ import (
 	"github.com/PuerkitoBio/goquery"
 	"net/http"
 	"net/url"
+	"strings"
 )
 
 // Response type wraps http.Response
@@ -26,3 +27,16 @@ func (r *Response) JoinURL(relativeURL string) string {
 	joinedURL := r.Response.Request.URL.ResolveReference(parsedRelativeURL)
 	return joinedURL.String()
 }
+
+func (r *Response) isHTML() bool {
+	if r.Response == nil {
+		return len(r.Body) != 0
+	}
+	contentType := r.Header.Get("Content-Type")
+	for _, htmlContentType := range []string{"text/html", "application/xhtml+xml", "application/vnd.wap.xhtml+xml"} {
+		if strings.Contains(contentType, htmlContentType) {
+			return true
+		}
+	}
+	return false
+}