git://wamblee.org
/
utils
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
timeout is now 5 seconds max.
[utils]
/
crawler
/
basic
/
src
/
org
/
wamblee
/
crawler
/
Crawler.java
diff --git
a/crawler/basic/src/org/wamblee/crawler/Crawler.java
b/crawler/basic/src/org/wamblee/crawler/Crawler.java
index f55eebb30afc1bed9b86c73d0e29eb6dddd80cbb..3615d9bc3a1e612e620725140a383fd17ac8e41b 100644
(file)
--- a/
crawler/basic/src/org/wamblee/crawler/Crawler.java
+++ b/
crawler/basic/src/org/wamblee/crawler/Crawler.java
@@
-16,24
+16,30
@@
package org.wamblee.crawler;
package org.wamblee.crawler;
+import org.apache.commons.httpclient.NameValuePair;
+
/**
/**
- *
Represents a crawled page
.
+ *
The object that actually obtains pages based on URL
.
*/
public interface Crawler {
/**
* Gets the content for a specific page.
* @param aUrl Url of page.
*/
public interface Crawler {
/**
* Gets the content for a specific page.
* @param aUrl Url of page.
+ * @param aParameters Paremeters to supply.
* @return Page to retrieve.
* @return Page to retrieve.
+ * @throws PageException In case of problems retrieving the page.
*/
*/
- Page getPage(String aUrl
)
;
+ Page getPage(String aUrl
, NameValuePair[] aParameters) throws PageException
;
/**
* Gets the content for a specific page.
/**
* Gets the content for a specific page.
- * @param aUrl Url of page.
+ * @param aUrl Url of page.
+ * @param aParameters Parameters to supply.
* @param aType Type of page.
* @param aType Type of page.
- * @return Page.
+ * @return Page.
+ * @throws PageException In case of problems retrieving the page.
*/
*/
- Page getPage(String aUrl,
PageType aType)
;
+ Page getPage(String aUrl,
NameValuePair[] aParameters, PageType aType) throws PageException
;
}
}