A comprehensive filtering system.
is a strategy to decide whether a certain object should be accepted or not; the type of objects
a filter considers is called the base type of the filter. In most cases, the base type
is going to be a URL or a fetched page. More precisely, a prefetch filter is one that
BURL as its base type (typically: to decide whether a
URL should be scheduled for later visit,
or should be fetched); a postfetch filter is one that has FetchedResponse as base type
and decides whether to do something with that response (typically: to parse
it, to store it, etc.).
Various kinds of filters are available, and moreover they can be composed with boolean operators
using the static methods specified in the
Filters class. Additionally, a filter parser is
provided in the it.unimi.dsi.law.ubi.filters.parser package; since the parser itself is written
using JavaCC, we provide a description of it here.
Two filters are called homogeneous if they filter the same kind of objects, heterogeneous otherwise.
A filter parser is instantiated on the basis of the kind of filters it will actually return; more precisely a
FilterParser<T> is a filter parser that will return a
Filter<T>; for technical
reasons, the class
T must be provided as unique parameter when the parser is constructed. A parser
can be used many times. Every time a filter is sought, the
parse(String x) method of the parser
is called, which returns a filter of the correct kind, or throws a parse exception.
The syntax used by the filter parser is available. Basically, it is a propositional calculus, with and (denoted by infix and or &), or (denoted by infix or or |) and not (denoted by prefix not or !), whose ground terms have the same form as returned by the toString() method of the Filter class.
Here are some examples:
- (HostEndsWith(foo.bar) and not ForbiddenHost(http://xxx.yyy.zzz/list-of-forbidden-hosts)) or NoMoreSlashThan(10)
Usually, an expression should only contain references to homogeneous filters of type
T is the type used to instantiate the parser. Nonetheless, if some ground term refers to a
filter of some other type
D, the parser will try to find a static method in the
having the following signature:
public static Filter<T> adaptFilterD2T( Filter<D> f )
that adapts the given filter
f to a filter of the correct type. If this method is missing,
the parser will itself throw an exception.
Interface Summary Interface Description Filter<T>A filter is a strategy to decide whether to accept a given object or not.
Class Summary Class Description AbstractFilter<T> ContentTypeStartsWithAccepts only fetched response whose content type starts with a given string. DigestEqualsAccepts only records of given digest, specified as a hexadecimal string. DuplicateSegmentsLessThanAccepts only URIs whose path does not contain too many duplicate segments. FiltersA collection of static methods to deal with
HostEndsWithAccepts only URIs whose host ends with (case-insensitively) a certain suffix. HostEqualsAccepts only URIs whose host equals (case-insensitively) a certain string. IsHttpResponseAccepts only records that are http/https responses. IsProbablyBinaryAccepts only http responses whose content stream appears to be binary. PathEndsWithOneOfAccepts only URIs whose path ends (case-insensitively) with one of a given set of suffixes. SchemeEqualsAccepts only URIs whose scheme equals a certain string (typically,
StatusCategoryAccepts only fetched response whose status category (status/100) has a certain value. URLEqualsAccepts only a given URIs. URLMatchesRegexAccepts only URIs that match a certain regular expression. URLShorterThanAccepts only URIs whose overall length is below a given threshold.