質問編集履歴
1
全文の追加
test
CHANGED
|
File without changes
|
test
CHANGED
|
@@ -4,9 +4,107 @@
|
|
|
4
4
|
|
|
5
5
|
```R
|
|
6
6
|
|
|
7
|
+
library(RSelenium)
|
|
8
|
+
|
|
9
|
+
library(rvest)
|
|
10
|
+
|
|
11
|
+
library(XML)
|
|
12
|
+
|
|
13
|
+
|
|
14
|
+
|
|
15
|
+
remDr = remoteDriver(remoteServerAddr = "localhost", port = 4444, browserName = "chrome")
|
|
16
|
+
|
|
17
|
+
remDr$open()
|
|
18
|
+
|
|
19
|
+
|
|
20
|
+
|
|
21
|
+
iterater <- 1
|
|
22
|
+
|
|
23
|
+
max_page <- 3
|
|
24
|
+
|
|
25
|
+
patent_linkDF <-data.frame('', '')
|
|
26
|
+
|
|
27
|
+
sleep <- 3
|
|
28
|
+
|
|
29
|
+
|
|
30
|
+
|
|
31
|
+
url <- 'https://jglobal.jst.go.jp/search/patents#{"category":"3","keyword":"農薬"}'
|
|
32
|
+
|
|
33
|
+
remDr$navigate(url)
|
|
34
|
+
|
|
35
|
+
planeHtmlList <- remDr$getPageSource()
|
|
36
|
+
|
|
37
|
+
fileNmae <- paste(formatC(iterater, width = 5, flag = 0), ".html", sep = "")
|
|
38
|
+
|
|
39
|
+
write(unlist(planeHtmlList), fileNmae)
|
|
40
|
+
|
|
41
|
+
|
|
42
|
+
|
|
43
|
+
path <- paste(getwd(), fileNmae, sep = "/")
|
|
44
|
+
|
|
45
|
+
html <- read_html(path)
|
|
46
|
+
|
|
47
|
+
parsed_doc <- htmlParse(html)
|
|
48
|
+
|
|
49
|
+
title <- xpathSApply(doc = parsed_doc , path = "//a[@href]", xmlValue)
|
|
50
|
+
|
|
51
|
+
link <- xpathSApply(doc = parsed_doc , path = "//a[@href]", xmlGetAttr, "href")
|
|
52
|
+
|
|
53
|
+
tempDF <- data.frame(title, link)
|
|
54
|
+
|
|
55
|
+
patent_linkDF <- tempDF[-c(1,2, nrow(tempDF)), ]
|
|
56
|
+
|
|
57
|
+
iterater <- iterater + 1
|
|
58
|
+
|
|
59
|
+
Sys.sleep(sleep)
|
|
60
|
+
|
|
61
|
+
|
|
62
|
+
|
|
7
63
|
while(iterater <= max_page){
|
|
8
64
|
|
|
9
|
-
url <- paste('https://
|
|
65
|
+
url <- paste('https://jglobal.jst.go.jp/search/patents#{"category":"3","keyword":"農薬","page":', iterater, "}", sep='') remDr$navigate(url)
|
|
66
|
+
|
|
67
|
+
planeHtmlList <- remDr$getPageSource()
|
|
68
|
+
|
|
69
|
+
fileNmae <- paste(formatC(iterater, width = 5, flag = 0), ".html", sep = "")
|
|
70
|
+
|
|
71
|
+
write(unlist(planeHtmlList), fileNmae)
|
|
72
|
+
|
|
73
|
+
|
|
74
|
+
|
|
75
|
+
path <- paste(getwd(), fileNmae, sep = "/")
|
|
76
|
+
|
|
77
|
+
html <- read_html(path)
|
|
78
|
+
|
|
79
|
+
parsed_doc <- htmlParse(html)
|
|
80
|
+
|
|
81
|
+
title <- xpathSApply(doc = parsed_doc , path = "//a[@href]", xmlValue)
|
|
82
|
+
|
|
83
|
+
link <- xpathSApply(doc = parsed_doc , path = "//a[@href]", xmlGetAttr, "href")
|
|
84
|
+
|
|
85
|
+
tempDF <- data.frame(title, link)
|
|
86
|
+
|
|
87
|
+
patent_page_linkDF <- tempDF[-c(1,2, nrow(tempDF)), ]
|
|
88
|
+
|
|
89
|
+
patent_linkDF <- rbind(patent_linkDF, tempDF);
|
|
90
|
+
|
|
91
|
+
iterater <- iterater + 1
|
|
92
|
+
|
|
93
|
+
Sys.sleep(sleep)
|
|
94
|
+
|
|
95
|
+
}
|
|
96
|
+
|
|
97
|
+
|
|
98
|
+
|
|
99
|
+
|
|
100
|
+
|
|
101
|
+
write.csv(patent_linkDF, "patent_link.csv")
|
|
102
|
+
|
|
103
|
+
|
|
104
|
+
|
|
105
|
+
|
|
106
|
+
|
|
107
|
+
|
|
10
108
|
|
|
11
109
|
```
|
|
12
110
|
|