@@ -79,9 +79,9 @@ public void InitializeCrawl(string key, string url)
79
79
{
80
80
//var crawlUrl = ConfigurationManager.AppSettings["url"];
81
81
Configuration configuration = ConfigurationManager . OpenExeConfiguration ( ConfigurationUserLevel . None ) ;
82
+
82
83
configuration . AppSettings . Settings [ "url" ] . Value = url ;
83
84
configuration . Save ( ) ;
84
-
85
85
ConfigurationManager . RefreshSection ( "appSettings" ) ;
86
86
CrawlPage ( ConfigurationManager . AppSettings [ "url" ] ) ;
87
87
}
@@ -130,13 +130,19 @@ private void CrawlPage(string url)
130
130
{
131
131
links . Add ( new Link ( "Page Title" , urlToParse ) ) ;
132
132
log . Add ( new Log ( $ "New Entry: { url } ", DateTime . Now ) ) ;
133
+
134
+ }
135
+ if ( linkParser . GoodUrls . Count == 0 )
136
+ {
137
+ links . Add ( new Link ( "Page Title" , "==============No New URLs From Page==============" ) ) ;
133
138
}
134
139
135
140
foreach ( string exception in linkParser . Exceptions )
136
141
_exceptions . Add ( exception ) ;
137
142
138
143
isCurrentPage = false ;
139
-
144
+ Console . WriteLine ( linkParser . GoodUrls ) ;
145
+ Console . WriteLine ( _externalUrlRepository . List ) ;
140
146
//Crawl all the links found on the page.
141
147
foreach ( string link in _externalUrlRepository . List )
142
148
{
@@ -147,6 +153,7 @@ private void CrawlPage(string url)
147
153
148
154
if ( formattedLink != String . Empty )
149
155
{
156
+ links . Add ( new Link ( " " , "==============Crawling to new external page...==============" ) ) ;
150
157
CrawlPage ( formattedLink ) ;
151
158
}
152
159
}
@@ -155,6 +162,30 @@ private void CrawlPage(string url)
155
162
_failedUrlRepository . List . Add ( formattedLink + " (on page at url " + url + ") - " + exc . Message ) ;
156
163
}
157
164
}
165
+ int loopBreak = 0 ;
166
+ foreach ( string link in linkParser . GoodUrls )
167
+ {
168
+ string formattedLink = link ;
169
+ loopBreak ++ ;
170
+ try
171
+ {
172
+ formattedLink = FixPath ( url , formattedLink ) ;
173
+
174
+ if ( formattedLink != String . Empty )
175
+ {
176
+ links . Add ( new Link ( " " , "==============Crawling to new internal page...==============" ) ) ;
177
+ CrawlPage ( formattedLink ) ;
178
+ }
179
+ }
180
+ catch ( Exception exc )
181
+ {
182
+ _failedUrlRepository . List . Add ( formattedLink + " (on page at url " + url + ") - " + exc . Message ) ;
183
+ }
184
+ if ( loopBreak >= 100 )
185
+ {
186
+ break ;
187
+ }
188
+ }
158
189
}
159
190
}
160
191
0 commit comments