// id doesn't contain the prefix "SAMPLE"
public String get_sample_attributes(String id) throws HttpException, IOException {
String prefix = prefixMap.get("SAMPLE");
String postfix = id + "&format=text";
String url = HelpFunctions.getUrl(prefix, postfix);
String webpage = getWebPage(url);
String[] lines = webpage.split("\n");
String result = "";
String comma = ",";
boolean start = false;
boolean descriptionTag = false;
for (int i = 0; i < lines.length; i++) {
String line = lines[i].trim();
if (line.length() == 0) continue;
if (line.equals("</pre>")) break;
if (line.equals("Attributes:") || line.equals("Additional attributes:")) {
start = true;
continue;
} else if (line.equals("Description:")) {
result += "description=\"";
// read the next line
result += (lines[i + 1].trim() + "\"");
break;
} else if (start) {
if (line.charAt(0) == '/') line = line.substring(1);
result += (line + comma);
}
} // for
// remove the comma
if (result.charAt(result.length() - 1) == ',')
result = result.substring(0, result.length() - 1);
return result;
}
// read xml File
void readXmlFiles() throws ParserConfigurationException, SAXException, IOException {
// prefix2url.xml
Document document = HelpFunctions.getDocument(prefixFilePath);
Element element = document.getDocumentElement();
// System.out.println("根元素为:" + element.getTagName());
NodeList childList = element.getChildNodes();
for (int i = 0; i < childList.getLength(); i++) {
Node node = childList.item(i);
if (!(node instanceof Element)) continue;
String prefix = node.getNodeName();
String url = node.getTextContent();
prefixMap.put(prefix, url);
}
// accessError.xml
document = HelpFunctions.getDocument(errorInfoPath);
element = document.getDocumentElement();
// System.out.println("根元素为:" + element.getTagName());
childList = element.getChildNodes();
for (int i = 0; i < childList.getLength(); i++) {
Node node = childList.item(i);
if (node instanceof Element) {
NodeList childList2 = node.getChildNodes();
String url = null;
String errorInfo = null;
for (int j = 0; j < childList2.getLength(); j++) {
Node node2 = childList2.item(j);
if (node2 instanceof Element) {
if (url == null) url = node2.getTextContent();
else {
errorInfo = node2.getTextContent();
break;
}
}
}
errorInfoMap.put(url, errorInfo);
}
}
}
public String get_pmid(String doi) throws HttpException, IOException {
String prefix = prefixMap.get("pmid");
String postfix = "?term=" + doi + "&presentation=xml";
String url = HelpFunctions.getUrl(prefix, postfix);
String webpage = getWebPage(url);
// System.out.println(webpage);
String beginMark = "<ArticleId IdType=\"pubmed\">";
String endMark = "<";
int beginIndex = webpage.indexOf(beginMark) + beginMark.length();
if (beginIndex < beginMark.length()) return null;
int endIndex = webpage.indexOf(endMark, beginIndex);
String pmid = webpage.substring(beginIndex, endIndex).trim();
if (pmid.equals("")) return null;
return pmid;
}
public String get_doi(String pmid) throws HttpException, IOException {
String prefix = prefixMap.get("pmid");
String postfix = "?term=" + pmid + "&presentation=xml";
String url = HelpFunctions.getUrl(prefix, postfix);
// String url = prefixMap.get("pmid") + "?term=" + pmid
// + "&presentation=xml";
String webpage = getWebPage(url);
// System.out.println(webpage);
String beginMark = "<ArticleId IdType=\"doi\">";
String endMark = "<";
int beginIndex = webpage.indexOf(beginMark) + beginMark.length();
if (beginIndex < beginMark.length()) return null;
int endIndex = webpage.indexOf(endMark, beginIndex);
String doi = webpage.substring(beginIndex, endIndex);
return doi;
}
// postfix should not be null or ""
boolean accessTest(String postfix, String prefix, String field, String log)
throws HttpException, IOException {
if (postfix == null || postfix.equals("") || postfix.equals("null")) return true;
// String logon_site
// int LOGON_PORT=80;
String url = HelpFunctions.getUrl(prefix, postfix);
// if(ur)
// System.out.println("**"+url+"** "+field);
String urlRegex = Validation.urlRegex;
if (!Validation.regexTest(urlRegex, url, field, log)) return false;
// client.getHostConfiguration().setHost(url, LOGON_PORT);
try {
GetMethod getMethod = new GetMethod(url);
int status = client.executeMethod(getMethod);
// not found the webpage
if (status == 404) {
Excel2Database.excel2DBLog.writeLine(
"ValidTest for " + field + " : '" + postfix + "' cann't be resolved." + log);
return false;
} else if (status == 200) {
if (errorInfoMap.keySet().contains(prefix)) {
String pageContent = getMethod.getResponseBodyAsString();
String errorInfo = errorInfoMap.get(prefix);
if (pageContent.contains(errorInfo)) {
Excel2Database.excel2DBLog.writeLine(
"ValidTest for " + field + " : '" + postfix + "' can't be resolved." + log);
return false;
}
}
}
getMethod.releaseConnection();
} catch (Exception e) {
// TODO: handle exception
System.out.println(url);
}
return true;
}
