2 * Copyright 2005 the original author or authors.
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
8 * http://www.apache.org/licenses/LICENSE-2.0
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
17 package org.wamblee.crawler.impl;
19 import java.io.InputStream;
20 import java.io.PrintStream;
21 import java.util.ArrayList;
22 import java.util.Iterator;
23 import java.util.List;
25 import org.apache.commons.httpclient.NameValuePair;
26 import org.dom4j.Document;
27 import org.dom4j.DocumentException;
28 import org.dom4j.Element;
29 import org.dom4j.io.SAXReader;
30 import org.wamblee.crawler.Configuration;
31 import org.wamblee.crawler.GetPageRequest;
32 import org.wamblee.crawler.PageRequest;
33 import org.wamblee.crawler.PostPageRequest;
36 * Parsing of the configuration from an XML file.
38 public class ConfigurationParser {
40 private static final String ELEM_URL = "url";
42 private static final String ELEM_TYPE = "type";
44 private static final String ELEM_PATTERN = "pattern";
46 private static final String ELEM_METHOD = "method";
48 private static final String ELEM_XSLT = "xslt";
50 private static final String ELEM_PARAM = "param";
52 private static final String AT_NAME = "name";
54 private static final String AT_VALUE = "value";
56 private static final String METHOD_POST = "post";
58 private static final String METHOD_GET = "get";
60 private static final int MAX_TRIES = 3;
62 private static final int MAX_DELAY = 100;
64 private PrintStream _os;
67 * Constructs the configuration parser.
68 * @param aOs The stream for logging requests
70 public ConfigurationParser(PrintStream aOs) {
75 * Parses the configuration from an input stream.
76 * @param aStream Input file.
77 * @return Configuration.
79 public Configuration parse(InputStream aStream) {
81 SAXReader reader = new SAXReader();
82 Document document = reader.read(aStream);
84 Element root = document.getRootElement();
85 List<UrlConfig> urlConfigs = parseUrlConfigs(root);
86 List<PageTypeConfig> pageTypeConfigs = parsePageTypeConfigs(root);
87 return new ConfigurationImpl(urlConfigs, pageTypeConfigs);
88 } catch (DocumentException e) {
89 throw new RuntimeException("Problem parsing config file", e);
94 * Parses the URL-based configuration.
95 * @param aRoot Root of the configuration file document.
96 * @return List of URL-based configurations.
98 private List<UrlConfig> parseUrlConfigs(Element aRoot) {
99 List<UrlConfig> configs = new ArrayList<UrlConfig>();
100 for (Iterator i = aRoot.elementIterator(ELEM_URL); i.hasNext();) {
101 Element url = (Element) i.next();
102 UrlConfig config = parseUrlConfig(url);
109 * Parses the page type based configurations.
110 * @param aRoot Root of the configuration file document.
111 * @return LIst of page type based configurations.
113 private List<PageTypeConfig> parsePageTypeConfigs(Element aRoot) {
114 List<PageTypeConfig> configs = new ArrayList<PageTypeConfig>();
115 for (Iterator i = aRoot.elementIterator(ELEM_TYPE); i.hasNext();) {
116 Element url = (Element) i.next();
117 PageTypeConfig config = parsePageTypeConfig(url);
124 * Parses a URL-based configuration.
125 * @param aUrlElem Configuration element.
126 * @return Configuration.
128 private UrlConfig parseUrlConfig(Element aUrlElem) {
129 String pattern = aUrlElem.elementText(ELEM_PATTERN);
130 PageRequest request = parseRequestConfig(aUrlElem);
131 return new UrlConfig(pattern, request);
135 * Parses a page type based configuration.
136 * @param aTypeElem Configuration element.
137 * @return Configuration.
139 private PageTypeConfig parsePageTypeConfig(Element aTypeElem) {
140 String pattern = aTypeElem.elementText(ELEM_PATTERN);
141 PageRequest request = parseRequestConfig(aTypeElem);
142 return new PageTypeConfig(pattern, request);
146 * Parses a request configuration describing how to execute requests.
147 * @param aElem Configuration element.
148 * @return Page request.
150 private PageRequest parseRequestConfig(Element aElem) {
151 String method = aElem.elementText(ELEM_METHOD);
152 String xslt = aElem.elementText(ELEM_XSLT);
153 List<NameValuePair> params = new ArrayList<NameValuePair>();
154 for (Iterator i = aElem.elementIterator(ELEM_PARAM); i.hasNext();) {
155 Element paramElem = (Element) i.next();
156 NameValuePair param = parseParameter(paramElem);
160 NameValuePair[] paramsArray = params.toArray(new NameValuePair[0]);
162 if (METHOD_POST.equals(method)) {
163 request = new PostPageRequest(MAX_TRIES, MAX_DELAY, paramsArray,
165 } else if (METHOD_GET.equals(method) || method == null) {
166 request = new GetPageRequest(MAX_TRIES, MAX_DELAY, paramsArray,
169 throw new RuntimeException("Unknown request method '" + method
170 + "'. Only " + METHOD_GET + " and " + METHOD_POST
177 * Parses a parameter definition.
178 * @param aParam Parameter.
179 * @return Name value pair describing a parameter.
181 private NameValuePair parseParameter(Element aParam) {
182 String name = aParam.attributeValue(AT_NAME);
183 String value = aParam.attributeValue(AT_VALUE);
184 return new NameValuePair(name, value);