This is a utility class implementing the 3 most common Perl5 operations
involving regular expressions:
- [m]/pattern/[i][m][s][x],
- s/pattern/replacement/[g][i][m][o][s][x],
- and split().
As with Perl, any non-alphanumeric character can be used in lieu of
the slashes.
The objective of the class is to minimize the amount of code a Java
programmer using Jakarta-ORO
has to write to achieve the same results as Perl by
transparently handling regular expression compilation, caching, and
matching. A second objective is to use the same Perl pattern matching
syntax to ease the task of Perl programmers transitioning to Java
(this also reduces the number of parameters to a method).
All the state affecting methods are synchronized to avoid
the maintenance of explicit locks in multithreaded programs. This
philosophy differs from the
org.apache.oro.text.regex
package, where
you are expected to either maintain explicit locks, or more preferably
create separate compiler and matcher instances for each thread.
To use this class, first create an instance using the default constructor
or initialize the instance with a PatternCache of your choosing using
the alternate constructor. The default cache used by Perl5Util is a
PatternCacheLRU of capacity GenericPatternCache.DEFAULT_CAPACITY. You may
want to create a cache with a different capacity, a different
cache replacement policy, or even devise your own PatternCache
implementation. The PatternCacheLRU is probably the best general purpose
pattern cache, but your specific application may be better served by
a different cache replacement policy. You should remember that you can
front-load a cache with all the patterns you will be using before
initializing a Perl5Util instance, or you can just let Perl5Util
fill the cache as you use it.
You might use the class as follows:
Perl5Util util = new Perl5Util();
String line;
DataInputStream input;
PrintStream output;
// Initialization of input and output omitted
while((line = input.readLine()) != null) {
// First find the line with the string we want to substitute because
// it is cheaper than blindly substituting each line.
if(util.match("/HREF=\"description1.html\"/")) {
line = util.substitute("s/description1\\.html/about1.html/", line);
}
output.println(line);
}
A couple of things to remember when using this class are that the
match()
methods have the same meaning as
Perl5Matcher.contains()
and
=~ m/pattern/
in Perl. The methods are named match
to more closely associate them with Perl and to differentiate them
from
Perl5Matcher.matches()
.
A further thing to keep in mind is that the
MalformedPerl5PatternException
class is derived from
RuntimeException which means you DON'T have to catch it. The reasoning
behind this is that you will detect your regular expression mistakes
as you write and debug your program when a MalformedPerl5PatternException
is thrown during a test run. However, we STRONGLY recommend that you
ALWAYS catch MalformedPerl5PatternException whenever you deal with a
DYNAMICALLY created pattern. Relying on a fatal
MalformedPerl5PatternException being thrown to detect errors while
debugging is only useful for dealing with static patterns, that is, actual
pregenerated strings present in your program. Patterns created from user
input or some other dynamic method CANNOT be relied upon to be correct
and MUST be handled by catching MalformedPerl5PatternException for your
programs to be robust.
Finally, as a convenience Perl5Util implements
the
MatchResult
interface.
The methods are merely wrappers which call the corresponding method of
the last
MatchResult
found (which can be accessed with
getMatch()
) by a match or
substitution (or even a split, but this isn't particularly useful).
At the moment, the
MatchResult
returned
by
getMatch()
is not stored in a thread-local variable. Therefore
concurrent calls to
getMatch()
will produce unpredictable
results. So if your concurrent program requires the match results,
you must protect the matching and the result retrieval in a critical
section. If you do not need match results, you don't need to do anything
special. If you feel the J2SE implementation of
getMatch()
should use a thread-local variable and obviate the need for a critical
section, please express your views on the oro-dev mailing list.
begin
public int begin(int group)
Returns the begin offset of the subgroup of the last match found
relative the beginning of the match.
- begin in interface MatchResult
group
- The pattern subgroup.
- The offset into group 0 of the first token in the indicated
pattern subgroup. If a group was never matched or does
not exist, returns -1. Be aware that a group that matches
the null string at the end of a match will have an offset
equal to the length of the string, so you shouldn't blindly
use the offset to index an array or String.
beginOffset
public int beginOffset(int group)
Returns an offset marking the beginning of the last pattern match
found relative to the beginning of the input from which the match
was extracted.
- beginOffset in interface MatchResult
group
- The pattern subgroup.
- The offset of the first token in the indicated
pattern subgroup. If a group was never matched or does
not exist, returns -1.
end
public int end(int group)
Returns the end offset of the subgroup of the last match found
relative the beginning of the match.
- end in interface MatchResult
group
- The pattern subgroup.
- Returns one plus the offset into group 0 of the last token in
the indicated pattern subgroup. If a group was never matched
or does not exist, returns -1. A group matching the null
string will return its start offset.
endOffset
public int endOffset(int group)
Returns an offset marking the end of the last pattern match found
relative to the beginning of the input from which the match was
extracted.
- endOffset in interface MatchResult
group
- The pattern subgroup.
- Returns one plus the offset of the last token in
the indicated pattern subgroup. If a group was never matched
or does not exist, returns -1. A group matching the null
string will return its start offset.
getMatch
public MatchResult getMatch()
Returns the last match found by a call to a match(), substitute(), or
split() method. This method is only intended for use to retrieve a match
found by the last match found by a match() method. This method should
be used when you want to save MatchResult instances. Otherwise, for
simply accessing match information, it is more convenient to use the
Perl5Util methods implementing the MatchResult interface.
- The org.apache.oro.text.regex.MatchResult instance containing the
last match found.
group
public String group(int group)
Returns the contents of the parenthesized subgroups of the last match
found according to the behavior dictated by the MatchResult interface.
- group in interface MatchResult
group
- The pattern subgroup to return.
- A string containing the indicated pattern subgroup. Group
0 always refers to the entire match. If a group was never
matched, it returns null. This is not to be confused with
a group matching the null string, which will return a String
of length 0.
groups
public int groups()
- groups in interface MatchResult
- The number of groups contained in the last match found.
This number includes the 0th group. In other words, the
result refers to the number of parenthesized subgroups plus
the entire match itself.
length
public int length()
Returns the length of the last match found.
- length in interface MatchResult
- The length of the last match found.
match
public boolean match(String pattern,
String input)
throws MalformedPerl5PatternException
Searches for the first pattern match in a String taking
a pattern specified in Perl5 native format:
[m]/pattern/[i][m][s][x]
The
m
prefix is optional and the meaning of the optional
trailing options are:
- i
- case insensitive match
- m
- treat the input as consisting of multiple lines
- s
- treat the input as consisting of a single line
- x
- enable extended expression syntax incorporating whitespace
and comments
As with Perl, any non-alphanumeric character can be used in lieu of
the slashes.
If the input contains the pattern, the
MatchResult
can be obtained by calling
getMatch()
.
However, Perl5Util implements the MatchResult interface as a wrapper
around the last MatchResult found, so you can call its methods to
access match information.
pattern
- The pattern to search for.input
- The String input to search.
- True if the input contains the pattern, false otherwise.
MalformedPerl5PatternException
- If there is an error in
the pattern. You are not forced to catch this exception
because it is derived from RuntimeException.
match
public boolean match(String pattern,
char[] input)
throws MalformedPerl5PatternException
Searches for the first pattern match somewhere in a character array
taking a pattern specified in Perl5 native format:
[m]/pattern/[i][m][s][x]
The
m
prefix is optional and the meaning of the optional
trailing options are:
- i
- case insensitive match
- m
- treat the input as consisting of multiple lines
- s
- treat the input as consisting of a single line
- x
- enable extended expression syntax incorporating whitespace
and comments
As with Perl, any non-alphanumeric character can be used in lieu of
the slashes.
If the input contains the pattern, the org.apache.oro.text.regex.MatchResult
can be obtained by calling
getMatch()
.
However, Perl5Util implements the MatchResult interface as a wrapper
around the last MatchResult found, so you can call its methods to
access match information.
pattern
- The pattern to search for.input
- The char[] input to search.
- True if the input contains the pattern, false otherwise.
MalformedPerl5PatternException
- If there is an error in
the pattern. You are not forced to catch this exception
because it is derived from RuntimeException.
match
public boolean match(String pattern,
PatternMatcherInput input)
throws MalformedPerl5PatternException
Searches for the next pattern match somewhere in a
org.apache.oro.text.regex.PatternMatcherInput instance, taking
a pattern specified in Perl5 native format:
[m]/pattern/[i][m][s][x]
The
m
prefix is optional and the meaning of the optional
trailing options are:
- i
- case insensitive match
- m
- treat the input as consisting of multiple lines
- s
- treat the input as consisting of a single line
- x
- enable extended expression syntax incorporating whitespace
and comments
As with Perl, any non-alphanumeric character can be used in lieu of
the slashes.
If the input contains the pattern, the
MatchResult
can be obtained by calling
getMatch()
.
However, Perl5Util implements the MatchResult interface as a wrapper
around the last MatchResult found, so you can call its methods to
access match information.
After the call to this method, the PatternMatcherInput current offset
is advanced to the end of the match, so you can use it to repeatedly
search for expressions in the entire input using a while loop as
explained in the
PatternMatcherInput
documentation.
pattern
- The pattern to search for.input
- The PatternMatcherInput to search.
- True if the input contains the pattern, false otherwise.
MalformedPerl5PatternException
- If there is an error in
the pattern. You are not forced to catch this exception
because it is derived from RuntimeException.
postMatch
public String postMatch()
Returns the part of the input following the last match found.
- The part of the input following the last match found.
postMatchCharArray
public char[] postMatchCharArray()
Returns the part of the input following the last match found as a char
array. This method eliminates the extra buffer copying caused by
preMatch().toCharArray().
- The part of the input following the last match found as a char[].
If the result is of zero length, returns null instead of a zero
length array.
preMatch
public String preMatch()
Returns the part of the input preceding the last match found.
- The part of the input following the last match found.
preMatchCharArray
public char[] preMatchCharArray()
Returns the part of the input preceding the last match found as a
char array. This method eliminates the extra
buffer copying caused by preMatch().toCharArray().
- The part of the input preceding the last match found as a char[].
If the result is of zero length, returns null instead of a zero
length array.
split
public void split(Collection results,
String input)
throws MalformedPerl5PatternException
Splits input in the default Perl manner, splitting on all whitespace.
This method is identical to calling:
split(results, "/\\s+/", input);
split
public void split(Collection results,
String pattern,
String input)
throws MalformedPerl5PatternException
This method is identical to calling:
split(results, pattern, input, SPLIT_ALL);
split
public void split(Collection results,
String pattern,
String input,
int limit)
throws MalformedPerl5PatternException
Splits a String into strings that are appended to a List, but no more
than a specified limit. The String is split using a regular expression
as the delimiter. The regular expression is a pattern specified
in Perl5 native format:
[m]/pattern/[i][m][s][x]
The
m
prefix is optional and the meaning of the optional
trailing options are:
- i
- case insensitive match
- m
- treat the input as consisting of multiple lines
- s
- treat the input as consisting of a single line
- x
- enable extended expression syntax incorporating whitespace
and comments
As with Perl, any non-alphanumeric character can be used in lieu of
the slashes.
The limit parameter causes the string to be split on at most the first
limit - 1 number of pattern occurences.
Of special note is that this split method performs EXACTLY the same
as the Perl split() function. In other words, if the split pattern
contains parentheses, additional Vector elements are created from
each of the matching subgroups in the pattern. Using an example
similar to the one from the Camel book:
split(list, "/([,-])/", "8-12,15,18")
produces the Vector containing:
{ "8", "-", "12", ",", "15", ",", "18" }
Furthermore, the following Perl behavior is observed: "leading empty
fields are preserved, and empty trailing one are deleted." This
has the effect that a split on a zero length string returns an empty
list.
The
Util.split()
method
does NOT implement these behaviors because it is intended to
be a general self-consistent and predictable split function usable
with Pattern instances other than Perl5Pattern.
results
- A Collection
to which the substrings of the input
that occur between the regular expression delimiter occurences
are appended. The input will not be split into any more substrings
than the specified
limit. A way of thinking of this is that only the first
limit - 1
matches of the delimiting regular expression will be used to split the
input. The Collection must support the
addAll(Collection)
operation.pattern
- The regular expression to use as a split delimiter.input
- The String to split.limit
- The limit on the size of the returned Vector
.
Values <= 0 produce the same behavior as the SPLIT_ALL constant which
causes the limit to be ignored and splits to be performed on all
occurrences of the pattern. You should use the SPLIT_ALL constant
to achieve this behavior instead of relying on the default behavior
associated with non-positive limit values.
MalformedPerl5PatternException
- If there is an error in
the expression. You are not forced to catch this exception
because it is derived from RuntimeException.
split
public Vector split(String input)
throws MalformedPerl5PatternException
Use
Splits input in the default Perl manner, splitting on all whitespace.
This method is identical to calling:
split("/\\s+/", input);
split(Collection results, String input)
instead.
split
public Vector split(String pattern,
String input)
throws MalformedPerl5PatternException
Use
This method is identical to calling:
split(pattern, input, SPLIT_ALL);
split(Collection results, String pattern, String input)
instead.
split
public Vector split(String pattern,
String input,
int limit)
throws MalformedPerl5PatternException
Use
Splits a String into strings contained in a Vector of size no greater
than a specified limit. The String is split using a regular expression
as the delimiter. The regular expression is a pattern specified
in Perl5 native format:
[m]/pattern/[i][m][s][x]
The
m
prefix is optional and the meaning of the optional
trailing options are:
- i
- case insensitive match
- m
- treat the input as consisting of multiple lines
- s
- treat the input as consisting of a single line
- x
- enable extended expression syntax incorporating whitespace
and comments
As with Perl, any non-alphanumeric character can be used in lieu of
the slashes.
The limit parameter causes the string to be split on at most the first
limit - 1 number of pattern occurences.
Of special note is that this split method performs EXACTLY the same
as the Perl split() function. In other words, if the split pattern
contains parentheses, additional Vector elements are created from
each of the matching subgroups in the pattern. Using an example
similar to the one from the Camel book:
split("/([,-])/", "8-12,15,18")
produces the Vector containing:
{ "8", "-", "12", ",", "15", ",", "18" }
The
Util.split()
method
does NOT implement this particular behavior because it is intended to
be usable with Pattern instances other than Perl5Pattern.
split(Collection results, String pattern, String input, int limit)
instead.
pattern
- The regular expression to use as a split delimiter.input
- The String to split.limit
- The limit on the size of the returned Vector
.
Values <= 0 produce the same behavior as the SPLIT_ALL constant which
causes the limit to be ignored and splits to be performed on all
occurrences of the pattern. You should use the SPLIT_ALL constant
to achieve this behavior instead of relying on the default behavior
associated with non-positive limit values.
- A
Vector
containing the substrings of the input
that occur between the regular expression delimiter occurences. The
input will not be split into any more substrings than the specified
limit. A way of thinking of this is that only the first
limit - 1
matches of the delimiting regular expression will be used to split the
input.
MalformedPerl5PatternException
- If there is an error in
the expression. You are not forced to catch this exception
because it is derived from RuntimeException.
substitute
public String substitute(String expression,
String input)
throws MalformedPerl5PatternException
Substitutes a pattern in a given input with a replacement string.
The substitution expression is specified in Perl5 native format.
- Calling this method is the same as:
-
String result;
StringBuffer buffer = new StringBuffer();
perl.substitute(buffer, expression, input);
result = buffer.toString();
expression
- The Perl5 substitution regular expression.input
- The input on which to perform substitutions.
- The input as a String after substitutions have been performed.
MalformedPerl5PatternException
- If there is an error in
the expression. You are not forced to catch this exception
because it is derived from RuntimeException.
- 1.0
substitute
substitute
public int substitute(StringBuffer result,
String expression,
String input)
throws MalformedPerl5PatternException
Substitutes a pattern in a given input with a replacement string.
The substitution expression is specified in Perl5 native format:
s/pattern/replacement/[g][i][m][o][s][x]
The
s
prefix is mandatory and the meaning of the optional
trailing options are:
- g
- Substitute all occurrences of pattern with replacement.
The default is to replace only the first occurrence.
- i
- perform a case insensitive match
- m
- treat the input as consisting of multiple lines
- o
- If variable interopolation is used, only evaluate the
interpolation once (the first time). This is equivalent
to using a numInterpolations argument of 1 in
Util.substitute()
.
The default is to compute each interpolation independently.
See
Util.substitute()
and Perl5Substitution
for more details on variable interpolation in
substitutions.
- s
- treat the input as consisting of a single line
- x
- enable extended expression syntax incorporating whitespace
and comments
As with Perl, any non-alphanumeric character can be used in lieu of
the slashes. This is helpful to avoid backslashing. For example,
using slashes you would have to do:
numSubs = util.substitute(result, "s/foo\\/bar/goo\\/\\/baz/", input);
when you could more easily write:
numSubs = util.substitute(result, "s#foo/bar#goo//baz#", input);
where the hashmarks are used instead of slashes.
There is a special case of backslashing that you need to pay attention
to. As demonstrated above, to denote a delimiter in the substituted
string it must be backslashed. However, this can be a problem
when you want to denote a backslash at the end of the substituted
string. As of PerlTools 1.3, a new means of handling this
situation has been implemented.
In previous versions, the behavior was that
"... a double backslash (quadrupled in the Java String) always
represents two backslashes unless the second backslash is followed
by the delimiter, in which case it represents a single backslash."
The new behavior is that a backslash is always a backslash
in the substitution portion of the expression unless it is used to
escape a delimiter. A backslash is considered to escape a delimiter
if an even number of contiguous backslashes preceed the backslash
and the delimiter following the backslash is not the FINAL delimiter
in the expression. Therefore, backslashes preceding final delimiters
are never considered to escape the delimiter. The following, which
used to be an invalid expression and require a special-case extra
backslash, will now replace all instances of / with \:
numSubs = util.substitute(result, "s#/#\\#g", input);
result
- The StringBuffer in which to store the result of the
substitutions. The buffer is only appended to.expression
- The Perl5 substitution regular expression.input
- The input on which to perform substitutions.
- The number of substitutions made.
MalformedPerl5PatternException
- If there is an error in
the expression. You are not forced to catch this exception
because it is derived from RuntimeException.
- 2.0.6
toString
public String toString()
Returns the same as group(0).
- toString in interface MatchResult
- A string containing the entire match.