X-Git-Url: http://wamblee.org/gitweb/?a=blobdiff_plain;f=crawler%2Fbasic%2Fsrc%2Forg%2Fwamblee%2Fcrawler%2Fimpl%2FConfigurationParser.java;h=792d0d34ad0f7ac1281e1c56a0afe24975420a83;hb=0119c69858055c84592338a202e9d1b18510a29b;hp=2f2e5f5bf95cdf9cc866c182551da757de820769;hpb=9d2a49c65872cd9330670a3cace19faf493df04d;p=utils diff --git a/crawler/basic/src/org/wamblee/crawler/impl/ConfigurationParser.java b/crawler/basic/src/org/wamblee/crawler/impl/ConfigurationParser.java index 2f2e5f5b..792d0d34 100644 --- a/crawler/basic/src/org/wamblee/crawler/impl/ConfigurationParser.java +++ b/crawler/basic/src/org/wamblee/crawler/impl/ConfigurationParser.java @@ -17,7 +17,6 @@ package org.wamblee.crawler.impl; import java.io.InputStream; -import java.io.PrintStream; import java.util.ArrayList; import java.util.Iterator; import java.util.List; @@ -59,14 +58,20 @@ public class ConfigurationParser { private static final int MAX_TRIES = 3; - private static final int MAX_DELAY = 5000; + private static final int MAX_DELAY = 100; - private PrintStream _os; - - public ConfigurationParser(PrintStream aOs) { - _os = aOs; + /** + * Constructs the configuration parser. + */ + public ConfigurationParser() { + // Empty } + /** + * Parses the configuration from an input stream. + * @param aStream Input file. + * @return Configuration. + */ public Configuration parse(InputStream aStream) { try { SAXReader reader = new SAXReader(); @@ -82,8 +87,9 @@ public class ConfigurationParser { } /** - * @param aRoot - * @return + * Parses the URL-based configuration. + * @param aRoot Root of the configuration file document. + * @return List of URL-based configurations. */ private List parseUrlConfigs(Element aRoot) { List configs = new ArrayList(); @@ -95,6 +101,11 @@ public class ConfigurationParser { return configs; } + /** + * Parses the page type based configurations. + * @param aRoot Root of the configuration file document. + * @return LIst of page type based configurations. + */ private List parsePageTypeConfigs(Element aRoot) { List configs = new ArrayList(); for (Iterator i = aRoot.elementIterator(ELEM_TYPE); i.hasNext();) { @@ -105,12 +116,22 @@ public class ConfigurationParser { return configs; } + /** + * Parses a URL-based configuration. + * @param aUrlElem Configuration element. + * @return Configuration. + */ private UrlConfig parseUrlConfig(Element aUrlElem) { String pattern = aUrlElem.elementText(ELEM_PATTERN); PageRequest request = parseRequestConfig(aUrlElem); return new UrlConfig(pattern, request); } + /** + * Parses a page type based configuration. + * @param aTypeElem Configuration element. + * @return Configuration. + */ private PageTypeConfig parsePageTypeConfig(Element aTypeElem) { String pattern = aTypeElem.elementText(ELEM_PATTERN); PageRequest request = parseRequestConfig(aTypeElem); @@ -118,14 +139,15 @@ public class ConfigurationParser { } /** - * @param aUrlElem - * @return + * Parses a request configuration describing how to execute requests. + * @param aElem Configuration element. + * @return Page request. */ - private PageRequest parseRequestConfig(Element aUrlElem) { - String method = aUrlElem.elementText(ELEM_METHOD); - String xslt = aUrlElem.elementText(ELEM_XSLT); + private PageRequest parseRequestConfig(Element aElem) { + String method = aElem.elementText(ELEM_METHOD); + String xslt = aElem.elementText(ELEM_XSLT); List params = new ArrayList(); - for (Iterator i = aUrlElem.elementIterator(ELEM_PARAM); i.hasNext();) { + for (Iterator i = aElem.elementIterator(ELEM_PARAM); i.hasNext();) { Element paramElem = (Element) i.next(); NameValuePair param = parseParameter(paramElem); params.add(param); @@ -135,10 +157,10 @@ public class ConfigurationParser { PageRequest request; if (METHOD_POST.equals(method)) { request = new PostPageRequest(MAX_TRIES, MAX_DELAY, paramsArray, - xslt, _os); + xslt); } else if (METHOD_GET.equals(method) || method == null) { request = new GetPageRequest(MAX_TRIES, MAX_DELAY, paramsArray, - xslt, _os); + xslt); } else { throw new RuntimeException("Unknown request method '" + method + "'. Only " + METHOD_GET + " and " + METHOD_POST @@ -147,6 +169,11 @@ public class ConfigurationParser { return request; } + /** + * Parses a parameter definition. + * @param aParam Parameter. + * @return Name value pair describing a parameter. + */ private NameValuePair parseParameter(Element aParam) { String name = aParam.attributeValue(AT_NAME); String value = aParam.attributeValue(AT_VALUE);