public NorwegianCompanyNameCleaner() { _sub = new LowerCaseNormalizeCleaner(); Add("\\s(a/s)(\\s|$)", "as"); Add("\\s(a\\\\s)(\\s|$)", "as"); Add("^(a/s)\\s", "as"); Add("^(a\\\\s)\\s", "as"); Add("\\s(a/l)(\\s|$)", "al"); Add("^(a/l)\\s", "al"); }
public PersonNameCleaner() { _sub = new LowerCaseNormalizeCleaner(); // load token translation _mapping (FIXME: move to static init?) try { _mapping = LoadMapping(); } catch (System.Exception ex) { logger.Error("Error initializing object: {0}", ex.Message); } }
public NorwegianAddressCleaner() { _sub = new LowerCaseNormalizeCleaner(); Add("^(co/ ?)", "c/o "); Add("^(c\\\\o)", "c/o"); Add("[A-Za-z]+(g\\.) [0-9]+", "gata"); Add("[A-Za-z]+ (gt?\\.?) [0-9]+", "gate"); Add("[A-Za-z]+(v\\.) [0-9]+", "veien"); Add("[A-Za-z]+ (v\\.?) [0-9]+", "vei"); Add("[A-Za-z]+(vn\\.?)[0-9]+", "veien "); Add("[A-Za-z]+(vn\\.?) [0-9]+", "veien"); Add("[A-Za-z]+(gt\\.?) [0-9]+", "gata"); Add("[A-Za-z]+(gaten) [0-9]+", "gata"); Add("(\\s|^)(pb\\.?) [0-9]+", "postboks", 2); Add("(\\s|^)(boks) [0-9]+", "postboks", 2); Add("[A-Za-z]+ [0-9]+(\\s+)[A-Za-z](\\s|$)", ""); Add("[A-Za-z]+(gata|veien)()[0-9]+[a-z]?(\\s|$)", " "); // FIXME: not sure about the following rules Add("postboks\\s+[0-9]+(\\s*-\\s*)", " "); }
public CapitalCleaner() { _sub = new LowerCaseNormalizeCleaner(); }
public void Init() { // Setup code goes here... cleaner = new LowerCaseNormalizeCleaner(); }
public void Cleanup() { // TearDown code goes here... cleaner = null; }
public CountryNameCleaner() { _sub = new LowerCaseNormalizeCleaner(); }