1
|
|
|
package main |
2
|
|
|
|
3
|
|
|
import ( |
4
|
|
|
"encoding/xml" |
5
|
|
|
"fmt" |
6
|
|
|
"net/http" |
7
|
|
|
"net/url" |
8
|
|
|
"os" |
9
|
|
|
"path" |
10
|
|
|
"time" |
11
|
|
|
) |
12
|
|
|
|
13
|
|
|
type SitemapIndex struct { |
|
|
|
|
14
|
|
|
XMLName xml.Name `xml:"sitemapindex"` |
15
|
|
|
XMLNs string `xml:"xmlns,attr"` |
16
|
|
|
Sitemap []Sitemap `xml:"sitemap"` |
17
|
|
|
} |
18
|
|
|
type Sitemap struct { |
|
|
|
|
19
|
|
|
Loc string `xml:"loc"` |
20
|
|
|
LastMod string `xml:"lastmod,omitempty"` |
21
|
|
|
} |
22
|
|
|
|
23
|
|
|
func (s Sitemap) findFileName() string { |
24
|
|
|
u, _ := url.Parse(s.Loc) |
25
|
|
|
dir := path.Dir(u.Path)[1:] |
26
|
|
|
filename := u.Path[len(dir)+1+1:] |
27
|
|
|
|
28
|
|
|
if _, err := os.Stat(dir); os.IsNotExist(err) != false { |
29
|
|
|
os.MkdirAll(dir, 0777) |
30
|
|
|
} |
31
|
|
|
filename = dir + string(os.PathSeparator) + filename |
32
|
|
|
return filename |
33
|
|
|
} |
34
|
|
|
func (si *SitemapIndex) validate() SitemapIndex { |
35
|
|
|
logChannel := make(chan string) |
36
|
|
|
validSitemapChannel := make(chan Sitemap) |
37
|
|
|
|
38
|
|
|
go func() { |
39
|
|
|
for _, sitemap := range (*si).Sitemap { |
40
|
|
|
sitemap.validate(logChannel,validSitemapChannel) |
41
|
|
|
} |
42
|
|
|
if Verbose {fmt.Println("Validation done")} |
43
|
|
|
close(logChannel) |
44
|
|
|
close(validSitemapChannel) |
45
|
|
|
}() |
46
|
|
|
|
47
|
|
|
go func() { |
48
|
|
|
for { |
49
|
|
|
logMsg,isLogChannelOpen := <-logChannel |
50
|
|
|
if !isLogChannelOpen { |
51
|
|
|
break |
52
|
|
|
} |
53
|
|
|
fmt.Println(logMsg) |
54
|
|
|
} |
55
|
|
|
}() |
56
|
|
|
|
57
|
|
|
|
58
|
|
|
newSitemapIndex := SitemapIndex{ |
59
|
|
|
XMLNs: si.XMLNs, |
60
|
|
|
} |
61
|
|
|
|
62
|
|
|
for { |
63
|
|
|
if Verbose { fmt.Println("Waits for sitemap data") } |
64
|
|
|
validSitemap, isValidSitemapChannelOpen := <-validSitemapChannel |
65
|
|
|
if !isValidSitemapChannelOpen { |
66
|
|
|
break |
67
|
|
|
} |
68
|
|
|
newSitemapIndex.Sitemap = append(newSitemapIndex.Sitemap, validSitemap) |
69
|
|
|
} |
70
|
|
|
|
71
|
|
|
return newSitemapIndex |
72
|
|
|
} |
73
|
|
|
|
74
|
|
|
func (s *Sitemap) validate(logChannel chan string,sitemapChannel chan Sitemap) { |
75
|
|
|
resp,err := http.Get((*s).Loc) |
76
|
|
|
if err!=nil { |
77
|
|
|
logChannel <- err.Error() |
78
|
|
|
return |
79
|
|
|
} |
80
|
|
|
logChannel <- fmt.Sprintf("Response code is %d for %s", resp.StatusCode, (*s).Loc) |
81
|
|
|
if resp.StatusCode == 200 { |
82
|
|
|
if Verbose { fmt.Println("Sitemap returning to channel") } |
83
|
|
|
sitemapChannel <- (*s) |
84
|
|
|
if Verbose { fmt.Println("Sitemap returned to channel") } |
85
|
|
|
} |
86
|
|
|
return |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
func (si *SitemapIndex) saveToFile(filename string) error { |
90
|
|
|
m, err := xml.Marshal((*si)) |
91
|
|
|
if err != nil { |
92
|
|
|
return err |
93
|
|
|
} |
94
|
|
|
|
95
|
|
|
file, err := os.OpenFile(filename, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0777) |
96
|
|
|
file.Write([]byte(xml.Header)) |
97
|
|
|
file.Write(m) |
98
|
|
|
file.Close() |
99
|
|
|
return err |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
func batchProcess(uri string) { |
103
|
|
|
resp, err := http.Get(uri) |
104
|
|
|
if err != nil { |
105
|
|
|
fmt.Printf("Url cannot fetched: %s\n", uri) |
106
|
|
|
fmt.Println(err) |
107
|
|
|
os.Exit(1) |
108
|
|
|
} |
109
|
|
|
|
110
|
|
|
rawXMLData := readXMLFromResponse(resp) |
111
|
|
|
if Verbose {fmt.Printf("XML readed from response\n")} |
112
|
|
|
|
113
|
|
|
sitemapIndex := newSitemapIndexFromXML(rawXMLData) |
114
|
|
|
if Verbose {fmt.Printf("New sitemap created\n")} |
115
|
|
|
newSitemapIndex := sitemapIndex.validate() |
116
|
|
|
if Verbose {fmt.Printf("Sitemap validated\n")} |
117
|
|
|
|
118
|
|
|
for _, sitemap := range newSitemapIndex.Sitemap { |
119
|
|
|
if Verbose {fmt.Printf("Wait for 2 sec.\n")} |
120
|
|
|
time.Sleep(time.Second * 2) |
121
|
|
|
filename := sitemap.findFileName() |
122
|
|
|
if Verbose {fmt.Printf("Filename is %s\n",filename)} |
123
|
|
|
singleProcess(sitemap.Loc, filename) |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
newSitemapIndex.saveToFile(OutputFileName) |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
func newSitemapIndexFromXML(rawXMLData []byte) SitemapIndex { |
130
|
|
|
sm := SitemapIndex{} |
131
|
|
|
err := xml.Unmarshal(rawXMLData, &sm) |
132
|
|
|
|
133
|
|
|
if err != nil { |
134
|
|
|
fmt.Printf("Sitemap index cannot parsed. Because: %s", err) |
135
|
|
|
return SitemapIndex{} |
136
|
|
|
} |
137
|
|
|
return sm |
138
|
|
|
} |
139
|
|
|
|