@@ -144,47 +144,57 @@ private void CrawlPage(string url)
144
144
Console . WriteLine ( linkParser . GoodUrls ) ;
145
145
Console . WriteLine ( _externalUrlRepository . List ) ;
146
146
//Crawl all the links found on the page.
147
+ int loopBreak1 = 0 ;
147
148
foreach ( string link in _externalUrlRepository . List )
148
149
{
149
150
string formattedLink = link ;
151
+ loopBreak1 ++ ;
150
152
try
151
153
{
152
154
formattedLink = FixPath ( url , formattedLink ) ;
153
155
154
156
if ( formattedLink != String . Empty )
155
157
{
156
- links . Add ( new Link ( " " , "==============Crawling to new external page...==============" ) ) ;
158
+ links . Add ( new Link ( " " , "==============Crawling to new external page " + link + " ...==============") ) ;
157
159
CrawlPage ( formattedLink ) ;
158
160
}
159
161
}
160
162
catch ( Exception exc )
161
163
{
162
164
_failedUrlRepository . List . Add ( formattedLink + " (on page at url " + url + ") - " + exc . Message ) ;
163
165
}
166
+ if ( loopBreak1 >= 100 )
167
+ {
168
+ break ;
169
+ }
164
170
}
165
171
int loopBreak = 0 ;
166
- foreach ( string link in linkParser . GoodUrls )
172
+ foreach ( string iLink in linkParser . GoodUrls )
167
173
{
168
- string formattedLink = link ;
174
+ string formattediLink = iLink ;
169
175
loopBreak ++ ;
170
176
try
171
177
{
172
- formattedLink = FixPath ( url , formattedLink ) ;
173
-
174
- if ( formattedLink != String . Empty )
178
+ formattediLink = FixPath ( url , formattediLink ) ;
179
+
180
+ if ( formattediLink != String . Empty )
175
181
{
176
- links . Add ( new Link ( " " , "==============Crawling to new internal page...==============" ) ) ;
177
- CrawlPage ( formattedLink ) ;
182
+ links . Add ( new Link ( " " , "==============Crawling to new internal page from " + iLink + " ...==============") ) ;
183
+ CrawlPage ( formattediLink ) ;
178
184
}
179
185
}
180
186
catch ( Exception exc )
181
187
{
182
- _failedUrlRepository . List . Add ( formattedLink + " (on page at url " + url + ") - " + exc . Message ) ;
188
+ _failedUrlRepository . List . Add ( formattediLink + " (on page at url " + url + ") - " + exc . Message ) ;
183
189
}
184
- if ( loopBreak >= 100 )
190
+ if ( loopBreak >= 5 )
185
191
{
186
192
break ;
187
193
}
194
+ /*if (formattediLink == "/covid19")
195
+ {
196
+ break;
197
+ }*/
188
198
}
189
199
}
190
200
}
0 commit comments