Data commit

Task/Web-scraping/ALGOL-68/web-scraping.alg

@@ -0,0 +1,42 @@
+STRING
+   domain="tycho.usno.navy.mil",
+   page="cgi-bin/timer.pl";
+
+STRING # search for the needle in the haystack #
+   needle = "UTC",
+   hay stack = "http://"+domain+"/"+page,
+
+   re success="^HTTP/[0-9.]* 200",
+   re result description="^HTTP/[0-9.]* [0-9]+ [a-zA-Z ]*",
+   re doctype ="\s\s<![Dd][Oo][Cc][Tt][Yy][Pp][Ee] [^>]+>\s+";
+
+PROC raise error = (STRING msg)VOID: ( put(stand error, (msg, new line)); stop);
+
+PROC is html page = (REF STRING page) BOOL: (
+     BOOL out=grep in string(re success, page, NIL, NIL) = 0;
+     IF INT start, end;
+        grep in string(re result description, page, start, end) = 0
+     THEN
+        page:=page[end+1:];
+        IF grep in string(re doctype, page, start, end) = 0
+        THEN page:=page[start+2:]
+        ELSE raise error("unknown format retrieving page")
+        FI
+     ELSE raise error("unknown error retrieving page")
+     FI;
+     out
+);
+
+STRING reply;
+INT rc = http content (reply, domain, haystack, 0);
+IF rc = 0 AND is html page (reply)
+THEN
+  STRING line; FILE freply; associate(freply, reply);
+  on logical file end(freply, (REF FILE freply)BOOL: (done; SKIP));
+  DO
+    get(freply,(line, new line));
+    IF string in string(needle, NIL, line) THEN print((line, new line)) FI
+  OD;
+  done: SKIP
+ELSE raise error (strerror (rc))
+FI