public String extractLunchNameFromTitle(final String title) throws ParseException {
final Pattern pattern =
Pattern.compile("\\d{4}-\\d{2}-\\d{2}\\s(Bastians|Mittagessen|\\s|-)*(.*?)");
final Matcher matcher = pattern.matcher(title);
if (!matcher.matches()) {
return null;
}
final String result = stringUtil.trim(matcher.group(2));
if (result.isEmpty()) {
return null;
}
return result;
}
public String extractLunchNameFromTitleOrContent(final String title, final String htmlContent)
throws ParseException {
{
try {
final String name = stringUtil.trim(extractLunchNameFromContent(htmlContent));
if (name != null && name.length() > 1) {
logger.debug("lunch name found in content: '" + name + "'");
return name;
}
} catch (ParseException e) {
logger.trace("extract lunch title form content failed", e);
}
}
{
final String name = stringUtil.trim(extractLunchNameFromTitle(title));
if (name != null && name.length() > 1) {
logger.debug("lunch name found in title: '" + name + "'");
return name;
}
}
logger.debug("no lunch name found");
return null;
}
public String extractLunchNameFromContent(final String htmlContent) throws ParseException {
final Document document = Jsoup.parse(htmlContent);
{
final Elements elements = document.getElementsByClass("tipMacro");
for (final Element element : elements) {
for (final Element td : element.getElementsByTag("p")) {
final String innerHtml = td.html();
final String result = stringUtil.trim(htmlUtil.filterHtmlTages(innerHtml));
if (result != null && result.length() > 0) {
logger.debug("found lunch lame " + result);
return result;
}
}
}
}
{
int pos = 0;
pos = parseUtil.indexOf(htmlContent, "ac:name=\"tip\"", pos);
try {
pos = parseUtil.indexOf(htmlContent, "INLINE", pos);
} catch (final ParseException e) {
// nop
}
final int pstart = parseUtil.indexOf(htmlContent, "<ac:rich-text-body>", pos);
final int pend = parseUtil.indexOf(htmlContent, "</ac:rich-text-body>", pstart);
final String result =
stringUtil.trim(htmlUtil.filterHtmlTages(htmlContent.substring(pstart, pend)));
if (result != null && result.length() > 0) {
logger.debug("found lunch name " + result);
return result;
}
}
logger.debug("extractLunchNameFromContent failed " + htmlContent);
return null;
}