Sync

Task/Yahoo--search-interface/D/yahoo--search-interface.d

@@ -0,0 +1,59 @@
+import std.stdio, std.exception, std.regex, std.algorithm, std.string,
+       std.net.curl;
+
+struct YahooResult {
+    string url, title, content;
+
+    string toString() const {
+        return "\nTitle: %s\nLink:  %s\nText:  %s"
+               .format(title, url, content);
+    }
+}
+
+struct YahooSearch {
+    private string query, content;
+    private uint page;
+
+    this(in string query_, in uint page_ = 0) {
+        this.query = query_;
+        this.page = page_;
+        this.content = "http://search.yahoo.com/search?p=%s&b=%d"
+                       .format(query, page * 10 + 1).get.assumeUnique;
+    }
+
+    @property results() const {
+        immutable re = `<li>
+                          <div \s class="res">
+                            <div>
+                              <h3>
+                                <a \s (?P<linkAttributes> [^>]+)>
+                                  (?P<linkText> .*?)
+                                </a>
+                              </h3>
+                            </div>
+                            <div \s class="abstr">
+                              (?P<abstract> .*?)
+                            </div>
+                            .*?
+                          </div>
+                        </li>`;
+
+        const clean = (string s) => s.replace("<[^>]*>".regex("g"),"");
+
+        return content.match(re.regex("gx")).map!(m => YahooResult(
+            clean(m.captures["linkAttributes"]
+                  .findSplitAfter(`href="`)[1]
+                  .findSplitBefore(`"`)[0]),
+            clean(m.captures["linkText"]),
+            clean(m.captures["abstract"])
+        ));
+    }
+
+    YahooSearch nextPage() const {
+        return YahooSearch(query, page + 1);
+    }
+}
+
+void main() {
+    writefln("%(%s\n%)", "test".YahooSearch.results);
+}

Task/Yahoo--search-interface/Racket/yahoo--search-interface.rkt

@@ -0,0 +1,29 @@
+#lang racket
+(require net/url)
+(define *yaho-url* "http://search.yahoo.com/search?p=~a&b=~a")
+(define *current-page* 0)
+(define *current-query* "")
+(define request (compose port->string get-pure-port string->url))
+
+;;strip html tags
+(define (remove-tags text)
+  (regexp-replace* #px"<[^<]+?>" text ""))
+
+;;search, parse and print
+(define (search-yahoo query)
+  (unless (string=? *current-query* query) ;different query, go back to page 1
+    (set! *current-query* query)
+    (set! *current-page* 0))
+  (let* ([current-page (number->string (add1 (* 10 *current-page*)))]
+         [html (request (format *yaho-url* query current-page))]
+         [results (regexp-match* #px"lass=\"yschttl spt\" href=\".+?\">(.+?)<span class=url>(.+?)</span>.+?<div class=\"abstr\">(.+?)</div>" html #:match-select cdr)])
+    (for ([result (in-list results)])
+      (printf "Title: ~a \n Link: ~a \n Text: ~a \n\n"
+              (remove-tags (first result))
+              (remove-tags (second result) )
+              (remove-tags (third result))))))
+
+;;search nexxt page
+(define (next-page)
+  (set! *current-page* (add1 *current-page*))
+  (search-yahoo *current-query*))