converter__org.bibliome.alvisnlp.modules.pattern.ElementPattern

#ElementPattern

Synopsis

Converts into an annotation pattern to be used by PatternMatcher.

String conversion

A pattern consists in a sequence of predicates. A predicate is a condition on an annotation, thus the pattern matches subsequences of annotations that yield true for each predicate.

Available predicates:

• any: yields true for any annotation;

• featureKey: yields true iff the annotation has the feature featureKey;

• featureKey == "featureValue": yields true iff the annotation has the feature featureKey and its value is equal to featureValue;

• "featureValue": yields true iff the feature form of the annotation is equal to featureValue, this is a short for form == "featureValue";

• /regexp/: yields true iff the feature form of the annotation matches the regular expression regexp (Java syntax);

• featureKey =~ "regexp" or featureKey =~ /regexp/: yields true iff the annotation has the feature featureKey and its value matches the regular expression regexp (Java syntax);

• featureKey op integer: yields true iff the annotation has the feature featureKey, and its value can be converted into an integer, and this value satisfies the comparison according to the specified operator and integer, op can be either ==, !=, <, >, <= or >=;

• pred1 or pred2: yields true iff one of predicates pred1 or pred2 yield true for the annotation;

• pred1 and pred2: yields true iff predicates pred1 and pred2 both yield true for the annotation;

• not pred: yields true iff the predicate pred yields false.

  The operator precedence is the usual one (*or* > *and* > *not*). The precedence can be overriden using parentheses.
  

Additionally the following two pseudo-predicates are available:

• start: matches the start of the annotation sequence, meaning that the following predicate must match the first annotation in the sequence;
• end: matches the end of the annotation sequence, meaning that the preceding predicate must match the last annotation in the sequence.

Predicates can be quantified. The following quantifiers are available:

• ?: matches the preceding predicate or group zero or once;

• ***: matches the preceding predicate or group zero to several times (unlimited);

• +: matches the preceding predicate or group at least once (unlimited);

• {n}: matches the preceding predicate or group exactly n times;

• {n,}: matches the preceding predicate or group at least n times (unlimited);

• {n,m}: matches the preceding predicate or group at least n times but no more than m times.

  These are greedy quantifiers. If there is a question mark (?) after the quantifier, then this quantifier is reluctant. See Java Pattern documentation for a complete description of greedy and reluctant quantifiers.

Predicate sequences may be grouped between square brackets. This allows the quantification of sub-patterns. Groups can be named by specifying the name after the opening bracket and

Examples:

• posTag =~ /^N/ +: matches a single annotation or an unlimited number of consecutive annotations whose feature posTag starts with a N;
• posTag =~ /^J/ * posTag =~ /^N/ +: matches nothing or an unlimited number of consecutive annotations whose feature posTag starts with a J, then followed by a single annotation or an unlimited number of consecutive annotations whose feature posTag starts with a N;
• posTag =~ /^J/ * posTag =~ /^N/ * [head: posTag =~ /^N/ ]: like the preceding one, but captures the last annotation of the match in a sub-pattern named head, this name can be referenced in PatternMatcher#actions.

XML conversion

<param value="PATTERN"/>```


	or
	```xml
<param pattern="PATTERN"/>```


	or
	```xml
<param>PATTERN</param>```

*PATTERN* will be converted as a string.