| author | Christian Urban <urbanc@in.tum.de> | 
| Mon, 24 Oct 2016 14:37:46 +0100 | |
| changeset 464 | ce33d8b44fe2 | 
| parent 254 | dcd4688690ce | 
| child 559 | 893e756e18f5 | 
| permissions | -rw-r--r-- | 
| 7 | 1 | val my_urls = """urbanc""".r | 
| 2 | ||
| 3 | def crawl(url: String, n: Int) : Unit = {
 | |
| 4 | if (n == 0) () | |
| 254 
dcd4688690ce
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
97diff
changeset | 5 |   else if (my_urls.findFirstIn(url) == None) { 
 | 
| 
dcd4688690ce
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
97diff
changeset | 6 | println(s"Visiting: $n $url") | 
| 
dcd4688690ce
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
97diff
changeset | 7 | get_page(url); () | 
| 
dcd4688690ce
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
97diff
changeset | 8 | } | 
| 7 | 9 |   else {
 | 
| 96 
9fcd3de53c06
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
93diff
changeset | 10 | println(s"Visiting: $n $url") | 
| 254 
dcd4688690ce
updated
 Christian Urban <christian dot urban at kcl dot ac dot uk> parents: 
97diff
changeset | 11 | for (u <- get_all_URLs(get_page(url))) crawl(u, n - 1) | 
| 7 | 12 | } | 
| 13 | } |