progs/app4.scala
author Christian Urban <christian dot urban at kcl dot ac dot uk>
Tue, 24 Sep 2013 23:16:00 +0100
changeset 97 60a3ba90dd53
parent 96 9fcd3de53c06
child 254 dcd4688690ce
permissions -rw-r--r--
added
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 {
96
9fcd3de53c06 updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 93
diff changeset
     9
    println(s"Visiting: $n $url")
7
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    10
    val page = get_page(url)
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    11
    println(email_pattern.findAllIn(page).mkString("\n"))
97
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 96
diff changeset
    12
    for (u <- get_all_URLs(page)) 
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents: 96
diff changeset
    13
      crawl(u, n - 1)
7
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    14
  }
73cf4406b773 updated
Christian Urban <urbanc@in.tum.de>
parents:
diff changeset
    15
}