-
Notifications
You must be signed in to change notification settings - Fork 0
Cleaner
A cleaner's job is to make comparison easier by removing from data values all variations that are not likely to indicate genuine differences. For example, a cleaner might strip everything except digits from a zip code. Or, it might normalize and lowercase addresses. Or translate dates into a common format.
A number of cleaners are provided with Duke, but it's trivial to
implement your own by implementing the no.priv.garshol.duke.Cleaner
interface. If you want to use regular expressions to clean data you
can subclass the AbstractRuleBasedCleaner.
All cleaners are found in the package no.priv.garshol.duke.cleaners. The package javadoc has more details.
The most widely used cleaner for string values. It lowercases all letters, removes whitespce at beginning and end, and normalizes whitespace in between tokens. It also removes accents, turning é into e, and so on.
This cleaner removes everything which is not a digit, and could have been used to clean the zip code as described above.
Simply trims whitespace at the beginning and end of the input string.
This cleaner uses regular expressions to do the cleaning. Basically, it matches the input against the regular expression it's given, and returns the part matching group number 1 (by default), or the given group. If the expression does not match, it returns null.
regexp: The regular expression to use. Must contain at least one group.
group: The number of the group to extract. 1 by default.
discard-group: Changes the behaviour so the cleaner instead discards the matching group and keeps the rest of the string.
Uses a mapping file to map incoming values to new values.
mapping-file: The name of a CSV file containing the mapping. Column 1 has from values, column 2 to values.
Uses a sub-cleaner to clean values first, then removes common meaningless values. (For example, if many people have the phone number 999-999-9999, this can be used to remove it.)
generic: The generic value to remove.
sub-cleaner: Another cleaner, used to clean data before generic values are removed.
This cleaner understands the syntax of phone numbers and will normalize phone numbers to a standard format (+CC number, where CC = country code). It knows the codes and syntax for phone numbers for some countries, and can easily be extended to support more countries.
Experimental cleaner for English person names (translates "Sam" to "samuel" etc).
A rule-based cleaner which can normalize Norwegian addresses. It understands abbreviations like "gt" for "gate", "pb" for "postboks", and so on.
A rule-based cleaner which can normalize Norwegian company names. It normalizes "a/s" to "as", and performs other normalizations as well.