app4.scala
author Christian Urban <urbanc@in.tum.de>
Sat, 29 Sep 2012 12:46:42 +0100
changeset 11 5353d405e4e6
parent 7 73cf4406b773
child 13 6d554a18b81e
permissions -rw-r--r--
added slides 2
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
7
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
     1
val http_pattern = """\"https?://[^\"]*\"""".r
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
     2
val my_urls = """urbanc""".r
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
     3
val email_pattern = 
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
     4
  """([a-z0-9_\.-]+)@([\da-z\.-]+)\.([a-z\.]{2,6})""".r
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
     5
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
     6
def crawl(url: String, n: Int) : Unit = {
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
     7
  if (n == 0) ()
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
     8
  //else if (my_urls.findFirstIn(url) == None) ()
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
     9
  else {
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    10
    println("Visiting: " + n + " " + url)
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    11
    val page = get_page(url)
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    12
    println(email_pattern.findAllIn(page).mkString("\n"))
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    13
    for (u <- get_all_URLs(page)) crawl(u, n - 1)
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    14
  }
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    15
}