|
1
|
|
|
package main |
|
2
|
|
|
|
|
3
|
|
|
import ( |
|
4
|
|
|
"encoding/xml" |
|
5
|
|
|
"fmt" |
|
6
|
|
|
"net/http" |
|
7
|
|
|
"os" |
|
8
|
|
|
"sync" |
|
9
|
|
|
"time" |
|
10
|
|
|
) |
|
11
|
|
|
|
|
12
|
|
|
// URLSet is root for site mite |
|
13
|
|
|
type URLSet struct { |
|
14
|
|
|
XMLName xml.Name `xml:"urlset"` |
|
15
|
|
|
XMLNs string `xml:"xmlns,attr"` |
|
16
|
|
|
URL []URL `xml:"url"` |
|
17
|
|
|
} |
|
18
|
|
|
|
|
19
|
|
|
// URL is for every single location url |
|
20
|
|
|
type URL struct { |
|
21
|
|
|
Loc string `xml:"loc"` |
|
22
|
|
|
LastMod string `xml:"lastmod,omitempty"` |
|
23
|
|
|
ChangeFreq string `xml:"changefreq,omitempty"` |
|
24
|
|
|
Priority float32 `xml:"priority,omitempty"` |
|
25
|
|
|
} |
|
26
|
|
|
|
|
27
|
|
|
type ValidURL struct { |
|
28
|
|
|
IsValid bool |
|
29
|
|
|
URL URL |
|
30
|
|
|
StatusCode int |
|
31
|
|
|
} |
|
32
|
|
|
|
|
33
|
|
|
func (us *URLSet) saveToFile(filename string) error { |
|
34
|
|
|
m, err := xml.MarshalIndent((*us), "\r\n", " ") |
|
35
|
|
|
if err != nil { |
|
36
|
|
|
return err |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
file, err := os.OpenFile(filename, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0777) |
|
40
|
|
|
file.Write([]byte(xml.Header)) |
|
41
|
|
|
file.Write(m) |
|
42
|
|
|
file.Close() |
|
43
|
|
|
return err |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
func (us *URLSet) validate() URLSet { |
|
47
|
|
|
// Create an HTTP client with a timeout |
|
48
|
|
|
client := &http.Client{ |
|
49
|
|
|
Timeout: 10 * time.Second, |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
// Initialize the new URLSet |
|
53
|
|
|
newURLSet := URLSet{ |
|
54
|
|
|
XMLNs: us.XMLNs, |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
// Create a semaphore to limit the number of concurrent requests |
|
58
|
|
|
maxConcurrentRequests := 10 |
|
59
|
|
|
sem := make(chan struct{}, maxConcurrentRequests) |
|
60
|
|
|
|
|
61
|
|
|
// Use a WaitGroup to wait for all goroutines |
|
62
|
|
|
var wg sync.WaitGroup |
|
63
|
|
|
var mu sync.Mutex |
|
64
|
|
|
|
|
65
|
|
|
n := len(us.URL) |
|
66
|
|
|
for i, url := range us.URL { |
|
67
|
|
|
wg.Add(1) |
|
68
|
|
|
sem <- struct{}{} // Acquire a semaphore slot |
|
69
|
|
|
|
|
70
|
|
|
go func(i int, url URL) { |
|
71
|
|
|
defer wg.Done() |
|
72
|
|
|
defer func() { <-sem }() // Release the semaphore slot |
|
73
|
|
|
|
|
74
|
|
|
resp, err := client.Get(url.Loc) |
|
75
|
|
|
if err != nil { |
|
76
|
|
|
fmt.Printf("Url %d/%d error: %s\n", i, n, url.Loc) |
|
77
|
|
|
return |
|
78
|
|
|
} |
|
79
|
|
|
defer resp.Body.Close() |
|
80
|
|
|
|
|
81
|
|
|
if resp.StatusCode == 200 { |
|
82
|
|
|
fmt.Printf("Url %d/%d check (200): %s\n", i, n, url.Loc) |
|
83
|
|
|
mu.Lock() |
|
84
|
|
|
newURLSet.URL = append(newURLSet.URL, url) |
|
85
|
|
|
mu.Unlock() |
|
86
|
|
|
} else { |
|
87
|
|
|
fmt.Printf("Url %d/%d dead (%d): %s\n", i, n, resp.StatusCode, url.Loc) |
|
88
|
|
|
} |
|
89
|
|
|
}(i, url) |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
wg.Wait() // Wait for all requests to complete |
|
93
|
|
|
|
|
94
|
|
|
return newURLSet |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
// i will use first parameter to determine sitemapIndex or not. |
|
98
|
|
|
func newURLSetFromXML(rawXMLData []byte) (bool, URLSet) { |
|
99
|
|
|
us := URLSet{} |
|
100
|
|
|
|
|
101
|
|
|
err := xml.Unmarshal(rawXMLData, &us) |
|
102
|
|
|
|
|
103
|
|
|
if err != nil { //some kind of goto |
|
104
|
|
|
sitemapIndex := newSitemapIndexFromXML(rawXMLData) |
|
105
|
|
|
sitemapIndexValidate(sitemapIndex) |
|
106
|
|
|
return true, URLSet{} |
|
107
|
|
|
} |
|
108
|
|
|
return false, us |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
func singleProcess(uri string, filename string) { |
|
112
|
|
|
client := &http.Client{ |
|
113
|
|
|
Timeout: 100 * time.Second, |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
resp, err := client.Get(uri) |
|
117
|
|
|
if err != nil { |
|
118
|
|
|
fmt.Printf("Url cannot fetched: %s\n", uri) |
|
119
|
|
|
fmt.Println(err) |
|
120
|
|
|
os.Exit(1) |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
rawXMLData := readXMLFromResponse(resp) |
|
124
|
|
|
|
|
125
|
|
|
isJumped, urlSet := newURLSetFromXML(rawXMLData) |
|
126
|
|
|
if !isJumped { |
|
127
|
|
|
|
|
128
|
|
|
newURLSet := urlSet.validate() |
|
129
|
|
|
|
|
130
|
|
|
err = newURLSet.saveToFile(filename) |
|
131
|
|
|
|
|
132
|
|
|
if err != nil { |
|
133
|
|
|
fmt.Println(err) |
|
134
|
|
|
os.Exit(1) |
|
135
|
|
|
} |
|
136
|
|
|
} |
|
137
|
|
|
} |
|
138
|
|
|
|
workouse /
sitemap-checker
