1
|
|
|
package main |
2
|
|
|
|
3
|
|
|
import ( |
4
|
|
|
"encoding/xml" |
5
|
|
|
"fmt" |
6
|
|
|
"time" |
7
|
|
|
"net/http" |
8
|
|
|
"os" |
9
|
|
|
) |
10
|
|
|
|
11
|
|
|
// URLSet is root for site mite |
12
|
|
|
type URLSet struct { |
13
|
|
|
XMLName xml.Name `xml:"urlset"` |
14
|
|
|
XMLNs string `xml:"xmlns,attr"` |
15
|
|
|
URL []URL `xml:"url"` |
16
|
|
|
} |
17
|
|
|
|
18
|
|
|
// URL is for every single location url |
19
|
|
|
type URL struct { |
20
|
|
|
Loc string `xml:"loc"` |
21
|
|
|
LastMod string `xml:"lastmod,omitempty"` |
22
|
|
|
ChangeFreq string `xml:"changefreq,omitempty"` |
23
|
|
|
Priority float32 `xml:"priority,omitempty"` |
24
|
|
|
} |
25
|
|
|
|
26
|
|
|
type ValidURL struct{ |
|
|
|
|
27
|
|
|
IsValid bool |
28
|
|
|
URL URL |
29
|
|
|
StatusCode int |
30
|
|
|
} |
31
|
|
|
|
32
|
|
|
func (us *URLSet) saveToFile(filename string) error { |
33
|
|
|
m, err := xml.Marshal((*us)) |
34
|
|
|
if err != nil { |
35
|
|
|
return err |
36
|
|
|
} |
37
|
|
|
|
38
|
|
|
file, err := os.OpenFile(filename, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0777) |
39
|
|
|
file.Write([]byte(xml.Header)) |
40
|
|
|
file.Write(m) |
41
|
|
|
file.Close() |
42
|
|
|
return err |
43
|
|
|
} |
44
|
|
|
|
45
|
|
|
func (us *URLSet) validate() URLSet { |
46
|
|
|
client := &http.Client{ |
47
|
|
|
Timeout: 10*time.Second, |
48
|
|
|
} |
49
|
|
|
validURLChannel := make(chan ValidURL) |
50
|
|
|
|
51
|
|
|
|
52
|
|
|
for _, url := range (*us).URL { |
53
|
|
|
go func(url URL, validURLChannel chan ValidURL) { |
54
|
|
|
resp, err := client.Get(url.Loc) |
55
|
|
|
statusCode := (*resp).StatusCode |
56
|
|
|
validURL := ValidURL { |
57
|
|
|
IsValid: err == nil && statusCode == 200, |
58
|
|
|
URL: url, |
59
|
|
|
StatusCode: statusCode, |
60
|
|
|
} |
61
|
|
|
validURLChannel <- validURL |
62
|
|
|
}(url, validURLChannel) |
63
|
|
|
} |
64
|
|
|
|
65
|
|
|
newURLSet := URLSet{ |
66
|
|
|
XMLNs: us.XMLNs, |
67
|
|
|
} |
68
|
|
|
|
69
|
|
|
for range us.URL { |
70
|
|
|
validURL:= <-validURLChannel |
71
|
|
|
if validURL.IsValid { |
72
|
|
|
newURLSet.URL = append(newURLSet.URL, validURL.URL) |
73
|
|
|
}else{ |
74
|
|
|
fmt.Printf("Url is dead (%s): %s \n",validURL.StatusCode,validURL.URL.Loc) |
|
|
|
|
75
|
|
|
} |
76
|
|
|
} |
77
|
|
|
close(validURLChannel) |
78
|
|
|
|
79
|
|
|
return newURLSet |
80
|
|
|
} |
81
|
|
|
//i will use first parameter to determine sitemapIndex or not. |
82
|
|
|
func newURLSetFromXML(rawXMLData []byte) (bool,URLSet) { |
83
|
|
|
us := URLSet{} |
84
|
|
|
|
85
|
|
|
err := xml.Unmarshal(rawXMLData, &us) |
86
|
|
|
|
87
|
|
|
if err != nil { //some kind of goto |
88
|
|
|
sitemapIndex := newSitemapIndexFromXML(rawXMLData) |
89
|
|
|
sitemapIndexValidate(sitemapIndex) |
90
|
|
|
return true, URLSet{} |
91
|
|
|
} |
92
|
|
|
return false,us |
93
|
|
|
} |
94
|
|
|
|
95
|
|
|
func singleProcess(uri string, filename string) { |
96
|
|
|
client := &http.Client{ |
97
|
|
|
Timeout: 10*time.Second, |
98
|
|
|
} |
99
|
|
|
|
100
|
|
|
resp, err := client.Get(uri) |
101
|
|
|
if err != nil { |
102
|
|
|
fmt.Printf("Url cannot fetched: %s\n", uri) |
103
|
|
|
fmt.Println(err) |
104
|
|
|
os.Exit(1) |
105
|
|
|
} |
106
|
|
|
|
107
|
|
|
rawXMLData := readXMLFromResponse(resp) |
108
|
|
|
|
109
|
|
|
isJumped, urlSet := newURLSetFromXML(rawXMLData) |
110
|
|
|
if !isJumped { |
111
|
|
|
|
112
|
|
|
newURLSet := urlSet.validate() |
113
|
|
|
|
114
|
|
|
err = newURLSet.saveToFile(filename) |
115
|
|
|
|
116
|
|
|
if err != nil { |
117
|
|
|
fmt.Println(err) |
118
|
|
|
os.Exit(1) |
119
|
|
|
} |
120
|
|
|
} |
121
|
|
|
} |
122
|
|
|
|