equal
deleted
inserted
replaced
1 val http_pattern = """\"https?://[^\"]*\"""".r |
1 val http_pattern = """"https?://[^"]*"""".r |
2 val my_urls = """urbanc""".r |
2 val my_urls = """urbanc""".r |
3 val email_pattern = |
3 val email_pattern = |
4 """([a-z0-9_\.-]+)@([\da-z\.-]+)\.([a-z\.]{2,6})""".r |
4 """([a-z0-9_\.-]+)@([\da-z\.-]+)\.([a-z\.]{2,6})""".r |
|
5 |
|
6 def print_str(s: String) = |
|
7 if (s == "") () else println(s) |
5 |
8 |
6 def crawl(url: String, n: Int) : Unit = { |
9 def crawl(url: String, n: Int) : Unit = { |
7 if (n == 0) () |
10 if (n == 0) () |
8 else { |
11 else { |
9 println(s"Visiting: $n $url") |
12 println(s"Visiting: $n $url") |
10 val page = get_page(url) |
13 val page = get_page(url) |
11 println(email_pattern.findAllIn(page).mkString("\n")) |
14 print_str(email_pattern.findAllIn(page).mkString("\n")) |
12 for (u <- get_all_URLs(page)) |
15 for (u <- get_all_URLs(page).par) crawl(u, n - 1) |
13 crawl(u, n - 1) |
|
14 } |
16 } |
15 } |
17 } |