support higheredjobs in jobs_to_org

diff --git a/jobs_to_org b/jobs_to_org
index fcdfd35345aab2e61ffc34d3591c0414c6961ad8..03777c568300eb17ae03aa29002db009d742e7b1 100755 (executable)
--- a/jobs_to_org
+++ b/jobs_to_org
@@ -115,6 +115,36 @@ my %sites =
                                      'page-title page-title--two-col'},
                            ],
               },
+     higheredjobs => {url => 'https://www.higheredjobs.com/search/advanced_action.cfm?JobCat=113&JobCat=259&JobCat=99&JobCat=100&JobCat=108&JobCat=107&PosType=1&InstType=1&InstType=2&InstType=3&Keyword=&Remote=1&Remote=2&Region=&Submit=Search+Jobs',
+                      next_selector_tree => [class => 'js-click-submit',
+                                             href => qr{advanced_action\.cfm},
+                                             sub {my @c = $_[0]->content_list();
+                                                  return 0 unless @c;
+                                                  return (defined $c[0]->attr('href') and
+                                                          defined $c[0]->attr('src') =~ /active-right\.gif/);
+                                              },
+                                            ],
+                      job_selector => [url_regex => qr{^details.cfm\?JobCode=\d+},
+                                      ],
+                      university => [class => qr/field-value/,
+                                     sub {my $p = $_[0]->parent();
+                                    my $c = $p->look_down(class => qr/field-label/,
+                                                         );
+                                    defined $c and $c->as_text() =~ qr/institution/i;
+                                }
+                              ],
+                      date => [class => qr/field-value/,
+                               sub {my $p = $_[0]->parent();
+                                    my $c = $p->look_down(class => qr/field-label/,
+                                                         );
+                                    defined $c and $c->as_text() =~ qr/posted/i;
+                                }
+                              ],
+                      position => [id => 'jobtitle-header',
+                                   _tag => 'h1',
+                                  ],
+                      description => [id => 'jobDesc'],
+                     },
     );
 
 binmode STDOUT,":utf8";
@@ -163,7 +193,14 @@ sub get_jobs {
             print format_job($university,$position,$j_u->URI->abs(),$description,$date);
             $m->back();
         }
-        $m->follow_link(@{$s->{next_selector}}) or die "Unable to find next link";
+        if (exists $s->{next_selector}) {
+            $m->follow_link(@{$s->{next_selector}}) or die "Unable to find next link";
+        } elsif (exists $s->{next_selector_tree}) {
+            my $link = $m->tree->look_down(@{$s->{next_selector_tree}}) or
+                die "Unable to find next link";
+            $m->get($link->attr('href')) or
+                die "Unable to get next page";
+        }
     }
 }