camel/tests/misc/url-scan.c

   1 /* -*- Mode: C; tab-width: 8; indent-tabs-mode: t; c-basic-offset: 8 -*- */
   2 /*
   3  *  Authors: Jeffrey Stedfast <fejj@ximian.com>
   4  *
   5  *  Copyright 2004 Ximian, Inc. (www.ximian.com)
   6  *
   7  *  This program is free software; you can redistribute it and/or modify
   8  *  it under the terms of the GNU Lesser General Public License as published by
   9  *  the Free Software Foundation; either version 2 of the License, or
  10  *  (at your option) any later version.
  11  *
  12  *  This program is distributed in the hope that it will be useful,
  13  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  *  GNU Lesser General Public License for more details.
  16  *
  17  *  You should have received a copy of the GNU Lesser General Public License
  18  *  along with this program; if not, write to the Free Software
  19  *  Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  20  *
  21  */
  22
  23
  24 #ifdef HAVE_CONFIG_H
  25 #include <config.h>
  26 #endif
  27
  28 #include <stdio.h>
  29 #include <stdlib.h>
  30 #include <string.h>
  31
  32 #include <camel/camel-mime-filter-tohtml.h>
  33
  34 #include "camel-test.h"
  35
  36 struct {
  37         char *text, *url;
  38 } url_tests[] = {
  39         { "bob@foo.com", "mailto:bob@foo.com" },
  40         { "Ends with bob@foo.com", "mailto:bob@foo.com" },
  41         { "bob@foo.com at start", "mailto:bob@foo.com" },
  42         { "bob@foo.com.", "mailto:bob@foo.com" },
  43         { "\"bob@foo.com\"", "mailto:bob@foo.com" },
  44         { "<bob@foo.com>", "mailto:bob@foo.com" },
  45         { "(bob@foo.com)", "mailto:bob@foo.com" },
  46         { "bob@foo.com, 555-9999", "mailto:bob@foo.com" },
  47         { "|bob@foo.com|555-9999|", "mailto:bob@foo.com" },
  48         { "bob@ no match bob@", NULL },
  49         { "@foo.com no match @foo.com", NULL },
  50         { "\"bob\"@foo.com", NULL },
  51         { "M@ke money fast!", NULL },
  52         { "ASCII art @_@ @>->-", NULL },
  53
  54         { "http://www.foo.com", "http://www.foo.com" },
  55         { "Ends with http://www.foo.com", "http://www.foo.com" },
  56         { "http://www.foo.com at start", "http://www.foo.com" },
  57         { "http://www.foo.com.", "http://www.foo.com" },
  58         { "http://www.foo.com/.", "http://www.foo.com/" },
  59         { "<http://www.foo.com>", "http://www.foo.com" },
  60         { "(http://www.foo.com)", "http://www.foo.com" },
  61         { "http://www.foo.com, 555-9999", "http://www.foo.com" },
  62         { "|http://www.foo.com|555-9999|", "http://www.foo.com" },
  63         { "foo http://www.foo.com/ bar", "http://www.foo.com/" },
  64         { "foo http://www.foo.com/index.html bar", "http://www.foo.com/index.html" },
  65         { "foo http://www.foo.com/q?99 bar", "http://www.foo.com/q?99" },
  66         { "foo http://www.foo.com/;foo=bar&baz=quux bar", "http://www.foo.com/;foo=bar&baz=quux" },
  67         { "foo http://www.foo.com/index.html#anchor bar", "http://www.foo.com/index.html#anchor" },
  68         { "http://www.foo.com/index.html; foo", "http://www.foo.com/index.html" },
  69         { "http://www.foo.com/index.html: foo", "http://www.foo.com/index.html" },
  70         { "http://www.foo.com/index.html-- foo", "http://www.foo.com/index.html" },
  71         { "http://www.foo.com/index.html?", "http://www.foo.com/index.html" },
  72         { "http://www.foo.com/index.html!", "http://www.foo.com/index.html" },
  73         { "\"http://www.foo.com/index.html\"", "http://www.foo.com/index.html" },
  74         { "'http://www.foo.com/index.html'", "http://www.foo.com/index.html" },
  75         { "http://bob@www.foo.com/bar/baz/", "http://bob@www.foo.com/bar/baz/" },
  76         { "http no match http", NULL },
  77         { "http: no match http:", NULL },
  78         { "http:// no match http://", NULL },
  79         { "unrecognized://bob@foo.com/path", "mailto:bob@foo.com" },
  80
  81         { "src/www.c", NULL },
  82         { "Ewwwwww.Gross.", NULL },
  83
  84 };
  85
  86 static int num_url_tests = G_N_ELEMENTS (url_tests);
  87
  88 int main (int argc, char **argv)
  89 {
  90         char *html, *url, *p;
  91         int i, errors = 0;
  92         guint32 flags;
  93
  94         camel_test_init (argc, argv);
  95
  96         camel_test_start ("URL scanning");
  97
  98         flags = CAMEL_MIME_FILTER_TOHTML_CONVERT_URLS | CAMEL_MIME_FILTER_TOHTML_CONVERT_ADDRESSES;
  99         for (i = 0; i < num_url_tests; i++) {
 100                 camel_test_push ("'%s' => '%s'", url_tests[i].text, url_tests[i].url ? url_tests[i].url : "None");
 101
 102                 html = camel_text_to_html (url_tests[i].text, flags, 0);
 103
 104                 url = strstr (html, "href=\"");
 105                 if (url) {
 106                         url += 6;
 107                         p = strchr (url, '"');
 108                         if (p)
 109                                 *p = '\0';
 110
 111                         while ((p = strstr (url, "&amp;")))
 112                                 memmove (p + 1, p + 5, strlen (p + 5) + 1);
 113                 }
 114
 115                 if ((url && (!url_tests[i].url || strcmp (url, url_tests[i].url) != 0)) ||
 116                     (!url && url_tests[i].url)) {
 117                         printf ("FAILED on \"%s\" -> %s\n  (got %s)\n\n",
 118                                 url_tests[i].text,
 119                                 url_tests[i].url ? url_tests[i].url : "(nothing)",
 120                                 url ? url : "(nothing)");
 121                         errors++;
 122                 }
 123
 124                 g_free (html);
 125         }
 126
 127         printf ("\n%d errors\n", errors);
 128
 129         camel_test_end ();
 130
 131         return errors;
 132 }