| author | Christian Urban <christian.urban@kcl.ac.uk> |
| Sun, 27 Sep 2020 09:15:32 +0100 | |
| changeset 764 | 6718ef6143b8 |
| parent 330 | 0806e45d873c |
| permissions | -rw-r--r-- |
|
254
dcd4688690ce
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
97
diff
changeset
|
1 |
val http_pattern = """"https?://[^"]*"""".r |
| 7 | 2 |
val email_pattern = |
3 |
"""([a-z0-9_\.-]+)@([\da-z\.-]+)\.([a-z\.]{2,6})""".r
|
|
4 |
||
|
254
dcd4688690ce
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
97
diff
changeset
|
5 |
def print_str(s: String) = |
|
dcd4688690ce
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
97
diff
changeset
|
6 |
if (s == "") () else println(s) |
|
dcd4688690ce
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
97
diff
changeset
|
7 |
|
| 7 | 8 |
def crawl(url: String, n: Int) : Unit = {
|
9 |
if (n == 0) () |
|
10 |
else {
|
|
|
96
9fcd3de53c06
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
93
diff
changeset
|
11 |
println(s"Visiting: $n $url") |
| 7 | 12 |
val page = get_page(url) |
|
254
dcd4688690ce
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
97
diff
changeset
|
13 |
print_str(email_pattern.findAllIn(page).mkString("\n"))
|
|
dcd4688690ce
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
97
diff
changeset
|
14 |
for (u <- get_all_URLs(page).par) crawl(u, n - 1) |
| 7 | 15 |
} |
16 |
} |