Skip to content

Commit d09116c

Browse files
author
spenney
committed
added some concats
1 parent 0cf80bf commit d09116c

File tree

1 file changed

+20
-10
lines changed

1 file changed

+20
-10
lines changed

WebCrawler/Model/Crawler.cs

Lines changed: 20 additions & 10 deletions
Original file line numberDiff line numberDiff line change
@@ -144,47 +144,57 @@ private void CrawlPage(string url)
144144
Console.WriteLine(linkParser.GoodUrls);
145145
Console.WriteLine(_externalUrlRepository.List);
146146
//Crawl all the links found on the page.
147+
int loopBreak1 = 0;
147148
foreach (string link in _externalUrlRepository.List)
148149
{
149150
string formattedLink = link;
151+
loopBreak1++;
150152
try
151153
{
152154
formattedLink = FixPath(url, formattedLink);
153155

154156
if (formattedLink != String.Empty)
155157
{
156-
links.Add(new Link(" ", "==============Crawling to new external page...=============="));
158+
links.Add(new Link(" ", "==============Crawling to new external page " + link + "...=============="));
157159
CrawlPage(formattedLink);
158160
}
159161
}
160162
catch (Exception exc)
161163
{
162164
_failedUrlRepository.List.Add(formattedLink + " (on page at url " + url + ") - " + exc.Message);
163165
}
166+
if (loopBreak1 >= 100)
167+
{
168+
break;
169+
}
164170
}
165171
int loopBreak = 0;
166-
foreach (string link in linkParser.GoodUrls)
172+
foreach (string iLink in linkParser.GoodUrls)
167173
{
168-
string formattedLink = link;
174+
string formattediLink = iLink;
169175
loopBreak++;
170176
try
171177
{
172-
formattedLink = FixPath(url, formattedLink);
173-
174-
if (formattedLink != String.Empty)
178+
formattediLink = FixPath(url, formattediLink);
179+
180+
if (formattediLink != String.Empty)
175181
{
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);
178184
}
179185
}
180186
catch (Exception exc)
181187
{
182-
_failedUrlRepository.List.Add(formattedLink + " (on page at url " + url + ") - " + exc.Message);
188+
_failedUrlRepository.List.Add(formattediLink + " (on page at url " + url + ") - " + exc.Message);
183189
}
184-
if (loopBreak >= 100)
190+
if (loopBreak >= 5)
185191
{
186192
break;
187193
}
194+
/*if (formattediLink == "/covid19")
195+
{
196+
break;
197+
}*/
188198
}
189199
}
190200
}

0 commit comments

Comments
 (0)