private void readMetaData(File f, FetchedDocument doc) {
try {
InputStreamReader is = new InputStreamReader(new FileInputStream(f), "UTF-8");
BufferedReader reader = new BufferedReader(is);
Map<String, String> metadata = new HashMap<String, String>();
String line = null;
while ((line = reader.readLine()) != null) {
if (line.length() == 0) {
continue;
}
String[] values = line.split(":", 2);
String key = values[0];
String value = values[1];
if ("url".equalsIgnoreCase(key)) {
doc.setDocumentURL(value);
} else if ("host".equalsIgnoreCase(key)) {
// skip, do nothing
} else if ("Content-Type".equalsIgnoreCase(key)) {
doc.setContentType(value);
} else if ("Charset".equalsIgnoreCase(key)) {
doc.setContentCharset(value);
} else {
metadata.put(key, value);
}
}
reader.close();
doc.setDocumentMetadata(metadata);
} catch (IOException e) {
throw new RuntimeException(
"Error while reading metadata from file: '" + f.getAbsolutePath() + "'", e);
}
}
