<?php
// Additional preg_* options
define ( 'PREG_WIPE_MATCHES' , 0x00010000 ) ; // Wipes unnamed captures
/**
* A class that implements extra regex functions.
*
* This class provides static methods for :
* - Converting an Msdos- or Unix-style wildcard string to a regular expression
* - Matching a string against a wildcard expression
* - Developing wildcard expressions
* - Using regular expressions that can have several times the same named captures
* - Meta-matching groups of lines based on a set of supplied regular expressions
*
* @Package String
* @Category Regex
* @version 1.0
* @author Christian Vigh (christian.vigh@orange.fr)
*
*/
class Regex
{
/**
* Returns a set of combinations from an array.
*
* Takes an array containing values and nested arrays (no more than one level of nesting is authorized) and
* generates all the possible combinations, each nested array providing alternatives for the generation.
*
* For example, the following input array :
*
* [ [ 'a', 'b' ], 1, 2, [ 'x', 'y', 'z' ] ]
*
* will generate the following results :
*
* [
* [0] => [ 'a', 1, 2, 'x' ]
* [1] => [ 'b', 1, 2, 'x' ]
* [2] => [ 'a', 1, 2, 'y' ]
* [3] => [ 'b', 1, 2, 'y' ]
* [4] => [ 'a', 1, 2, 'z' ]
* [5] => [ 'b', 1, 2, 'z' ]
* ]
*
* Note that the combination generation is computed from left to right.
*
* @param array $array
* Input array;
*
* @param int $max_results
* This parameter is provided to limit exponential combinations generation.
*
* @return array
* Returns an array of arrays containing all the combinations of the supplied input array.
*/
// __develop_non_recursive -
// A non-recursive version for the CombinationsOf() method, that allows only one nesting level of arrays.
private static function __develop_non_recursive ( &$results, $array, $max_results = 10000 )
{
$array_count = count ( $array ) ;
// Loop through input array, which is processed from right to left to generate the resulting array
for ( $i = $array_count - 1 ; $i >= 0 ; $i -- )
{
$item = $array [$i] ;
$result_count = count ( $results ) ;
// Current item is a nested array : we will have to demultiplicate each result in the $results array
// to hold each $array items
if ( is_array ( $item ) )
{
$item_count = count ( $item ) ;
// Check for possible quota overflow
if ( $result_count * $item_count > $max_results )
throw ( new Exception ( "Array development limit of $max_results results exceeded." ) ) ;
// Then check that each subitem of this array is not an array in itself
foreach ( $item as $subitem )
{
if ( is_array ( $subitem ) )
throw ( new Exception ( "No more than one nesting level is allowed in arrays to be developed." ) ) ;
}
// Demultiplicate existing results
if ( $result_count )
{
$new_results = [] ;
for ( $j = 0 ; $j < $result_count ; $j ++ )
{
for ( $k = 0 ; $k < $item_count ; $k ++ )
{
$subitem = $item [$k] ;
$new_results [] = array_merge ( [ $subitem ], $results [$j] ) ;
}
}
$results = $new_results ;
}
// ... but for the first iteration, only add the elements of the current array item
else
{
foreach ( $item as $subitem )
$results [] = [ $subitem ] ;
}
}
// Current item is not an array
else
{
// Prepend it to the existing items
if ( $result_count )
{
foreach ( $results as &$result )
array_unshift ( $result, $item ) ;
}
// ... or simply add it, if the results array is empty
else
{
$results [] = [ $item ] ;
}
}
}
}
public static function CombinationsOf ( $array, $max_results = 10000 )
{
$array_count = count ( $array ) ;
if ( ! $array_count )
return ( [] ) ;
$results = [] ;
self::__develop_non_recursive ( $results, $array, $max_results ) ;
return ( $results ) ;
}
/**
*
* Expands a factorized expression.
*
* Expands (develops) a factorized string expression.
* Sometimes, it is necessary to represent a set of values with a factorized expression,
* such as the shell allows us to match a set of files using a pattern.
* The DevelopExpression() method allows for input strings that can contain "character
* classes" such as :
*
* [a-c].somestring
*
* which will expand to the following array of values :
*
* a.somestring
* b.somestring
* c.somestring
*
* Character classes can be either alphabetic or alphanumeric, such as in the following
* example :
*
* [a-b][0-1].somestring
*
* which will expand to :
*
* a0.somestring
* a1.somestring
* b0.somestring
* b1.somestring
*
* Numeric values can be zero-padded, using an optional width construct, like in the
* following example :
*
* [a-b][0-1]/4.somestring
*
* which will expand to :
*
* a0000.somestring
* a0001.somestring
* b0000.somestring
* b0001.somestring
*
* For alphabetic character classes, the case of the first character determines the case
* of the expanded result ; for example :
*
* [A-c].somestring
*
* will give :
*
* A.somestring
* B.somestring
* C.somestring
*
* Finally, brackets can be escaped using the backslash character.
*
*
* @param string $expression
* Expression to be developed.
*
* @param int $limit
* Maximum number of developed expressions to be returned. This limit is arbitrarily fixed to 10000.
* When null or negative, no limit is applied.
*
* @return array
* Returns an array containing the developed expression results. The returned value always contain
* at least one element (the supplied input value) when no factorization expression are found.
*
* @todo
* . Allow for real character classes, such as :
* [A] (same as [A-A])
* [A-] (same as [A-Z])
* [-Z] (same as [A-Z])
* [abc012] (low/high will cycle through the characters "abc012")
*
*/
public static function DevelopExpression ( $expression, $limit = 10000 )
{
// Regular expression that matches :
// . Either an escaped bracket
// . Or a factorization expression
static $re = '#
(?P<escape> \\\\ [\[\]] )
|
(?P<range>
\[
(?P<rangelo> ( [a-z] | [0-9]+ ) )
-
(?P<rangehi> ( [a-z] | [0-9]+ ) )
\]
(
\/
(?P<width> \d+)
)?
)
#imsx' ;
// First simplication case : the input string contain no factorization expression, return
// it as is
if ( ! self::PregMatchAll ( $re, $expression, $matches, PREG_OFFSET_CAPTURE ) )
return ( [ $expression ] ) ;
// Second elimination case : the input string does not contain any factorization expression,
// but escaped brackets are present (thus generating a "match")
$found_range = false ;
foreach ( $matches [ 'range' ] as $match )
{
if ( $match )
{
$found_range = true ;
break ;
}
}
if ( ! $found_range )
return ( [ $expression ] ) ;
// Initializations
$results = [] ; // The results that will be returned
$result_count = 0 ; // Number of generated results - checked against the $limit parameter
$ranges = [] ; // Associative arrays that describe each input string portion, either plain constant or factorization expression
$range_indexes = [] ; // Indexes into $ranges of the entries containing a factorization expression
$expression_length = strlen ( $expression ) ; // Compute it once and for all
/***
Loop through the matches found ; upon exit, the $ranges array will contain associative arrays with
the following keys :
- 'expression' (boolean) :
True if the entry is a factorization expression, false otherwise.
- 'offset' (integer) :
Start character.
- 'length' (integer) :
Length of the expression or plain substring.
- 'capture' (string) :
Original string (either the plain substring or the factorization expression).
For factorization expressions, the following keys are also present :
- 'low' (integer) :
Low value of the factorization expression.
- 'high' (integer) :
High value of the factorization expression.
- 'numeric' (boolean) :
True if the factorization expression relates to numeric values, false for alphabetic values.
- 'width' (boolean) :
False for alphabetic values.
For numeric values, specifies the zero-pad width.
- 'current' (integer or string) :
Current index in the generation loop. Initialized to the value of the 'low' entry.
***/
// If first match does not start at string offset zero, create a plain string entry in the $ranges array
if ( $matches [ 'range' ] [0] [1] != 0 )
$ranges [] =
[
'expression' => false,
'offset' => 0,
'length' => $matches [ 'range' ] [0] [1],
'capture' => substr ( $expression, 0, $matches [ 'range' ] [0] [1] )
] ;
// Loop through the matches
for ( $i = 0 ; $i < count ( $matches [ 'range' ] ) ; $i ++ )
{
// $matches [ 'range' ] [$i] will never be an array if an escaped bracket is met
if ( ! $matches [ 'range' ] [$i] )
continue ;
// Create the range entry
$range =
[
'expression' => true,
'low' => $matches [ 'rangelo' ] [$i] [0],
'high' => $matches [ 'rangehi' ] [$i] [0],
'offset' => $matches [ 'range' ] [$i] [1],
'current' => $matches [ 'rangelo' ] [$i] [0],
'length' => strlen ( $matches [ 'range' ] [$i] [0] ),
'capture' => $matches [ 'range' ] [$i] [0],
'width' => ( $matches [ 'width' ] [$i] ) ? $matches [ 'width' ] [$i] [0] : false
] ;
// If low value is alphabetic...
if ( ctype_alpha ( $range [ 'low' ] ) )
{
// ... and if high value is also alphabetic
if ( ctype_alpha ( $range [ 'high' ] ) )
{
// ... the case of the factorized expressions will be the case of the 'low' value
if ( ctype_lower ( $range [ 'low' ] ) )
$range [ 'high' ] = strtolower ( $range [ 'high' ] ) ;
else
$range [ 'high' ] = strtoupper ( $range [ 'high' ] ) ;
}
// ... but complain if out of alphabetic range
else
throw ( new Exception ( "Invalid combination of letters and numbers in factorized expression : " .
$matches [ 'range' ] [$i] [0] ) ) ;
$range [ 'numeric' ] = false ; // This entry is alphabetic
}
// Low value is not alphabetic : due to the regex construct, this must be an integer, so check that the high
// value is also an integer
else if ( ! is_numeric ( $range [ 'high' ] ) )
throw ( new Exception ( "Invalid combination of letters and numbers in factorized expression : " .
$matches [ 'range' ] [$i] [0] ) ) ;
// Both low and high values are numeric
else
$range [ 'numeric' ] = true ;
// Check that low range is not greater than upper range
if ( $range [ 'numeric' ] )
{
if ( ( integer ) $range [ 'low' ] > ( integer ) $range [ 'high' ] )
throw ( new Exception ( "Lower numeric value cannot be greater than upper value in factorized expression : " .
$matches [ 'range' ] [$i] [0] ) ) ;
}
else if ( $range [ 'low' ] > $range [ 'high' ] )
throw ( new Exception ( "Lower alphabetic value cannot be greater than upper value in factorized expression : " .
$matches [ 'range' ] [$i] [0] ) ) ;
// If there is a plain string between the previous range and this one, then we need to add
// a plain string entry before it
$count = count ( $ranges ) ;
if ( $count ) // ... but only if we already have found ranges
{
$previous_range = $ranges [ $count - 1 ] ;
$last_found = $previous_range [ 'offset' ] + $previous_range [ 'length' ] ;
if ( $last_found < $range [ 'offset' ] )
{
$ranges [] =
[
'expression' => false,
'offset' => $last_found,
'length' => $range [ 'offset' ] - $last_found,
'capture' => substr ( $expression, $last_found, $range [ 'offset' ] - $last_found )
] ;
}
}
// Since the $ranges array contains both plain strings and factorization expressions, we need an array
// to store factorization expression indexes, in order to simplify the generation process
$range_indexes [] = count ( $ranges ) ;
// Add this factorization expression to the list
$ranges [] = $range ;
}
// If the string does not end with a factorization expression, then we have to add this plain string to
// the $ranges array
$count = count ( $ranges ) ;
if ( $count )
{
$previous_range = $ranges [ $count - 1 ] ;
$last_found = $previous_range [ 'offset' ] + $previous_range [ 'length' ] ;
if ( $last_found < $expression_length )
{
$ranges [] =
[
'expression' => false,
'offset' => $last_found,
'length' => $expression_length - $last_found,
'capture' => substr ( $expression, $last_found, $expression_length - $last_found )
] ;
}
}
// After all this preparatory stuff, we are finally ready to generate the developed values !
$range_index_count = count ( $range_indexes ) ;
while ( true )
{
$current = "" ;
// Generate the current developed value
foreach ( $ranges as $range )
{
if ( $range [ 'expression' ] )
{
if ( $range [ 'width' ] )
$current .= sprintf ( "%0{$range [ 'width' ]}d", $range [ 'current' ] ) ;
else
$current .= $range [ 'current' ] ;
}
else
$current .= $range [ 'capture' ] ;
}
$results [] = $current ;
$result_count ++ ;
// Stop if we reached the limit (if any...)
if ( $limit > 0 && $result_count == $limit )
break ;
// This loop finds the next increment value, when multiple factorization expressions are specified
for ( $i = $range_index_count - 1 ; $i >= 0 ; $i -- )
{
// Look only the entries in the $range array that are factorization expressions, starting
// with the very last one
$range_index = $range_indexes [$i] ;
$current_range = &$ranges [ $range_index ] ;
// Increment the value
if ( $current_range [ 'numeric' ] )
$current_range [ 'current' ] = $current_range [ 'current' ] + 1 ;
else
$current_range [ 'current' ] = chr ( ord ( $current_range [ 'current' ] ) + 1 ) ;
// When the current (last) value reaches the high-end of the range...
if ( $current_range [ 'current' ] > $current_range [ 'high' ] )
{
// Just start again with the low-end value
if ( $i )
$current_range [ 'current' ] = $current_range [ 'low' ] ;
// ... but only if there are still values to be incremented
else
break 2 ;
}
// Current value is still within the [low..high] range
else
break ;
}
}
// All done, return
return ( $results ) ;
}
/**
* Allows a regex match result to have captures with the same name.
*
* After the result of calling a preg_* function, takes the resulting matches and groups
* them using the specified replacement associative array.
* The $replacement array comes from a possible call to Regex::RenumberNamedCaptures()
* method ; each key defines the renumbered capture names, whereas each value is the
* original name.
* There is a 1 -> n relationship between the new and old names, allowing a single
* regular expression to contain more than one capture having the same name.
*
* @param mixed $match
* A match result, as can be returned by the self::PregMatch*() functions.
*
* @param array $replacements
* An associative array returned by the Regex::RenumberNamedCaptures() method,
* having the following shape :
* - key : New capture name
* - value : old capture name
*
* @return array
*/
public static function GroupNamedCaptures ( $match, $replacements )
{
$new_match = [] ;
foreach ( $match as $key => $value )
{
// If the key corresponds to one of the matched replacements, group it into an array
if ( isset ( $replacements [ $key ] ) )
{
if ( isset ( $new_match [ $replacements [ $key ] ] ) )
$new_match [ $replacements [ $key ] ] [] = $value ;
else
$new_match [ $replacements [ $key ] ] = [ $value ] ;
}
else
{
if ( isset ( $new_match [ $key ] ) )
$new_match [ $key ] [] = $value ;
else
$new_match [ $key ] = [ $value ] ;
}
}
return ( $new_match ) ;
}
/**
*
* Checks if the specified string is a valid regular expression.
*
* @param string $str
* String to be checked.
*
* @param string $delimiter
* Regular expression delimiter. If not specified, the delimiter is taken from
* the first character of the specified input string.
*
* @return bool
* Returns true if the specified string represents a regular expression, false otherwise.
*/
public static function IsRegex ( $str, $delimiter = false )
{
static $pcre_options = 'imsxe' ;
$length = strlen ( $str ) ;
// Basic re's must have at least 3 characters : 2 delimiters and a character inside
if ( $length < 3 )
return ( false ) ;
// Check that the starting character is not an alphanumeric, backslash or whitespace character
if ( $str [0] == '\\' || ctype_alnum ( $str [0] ) || ctype_space ( $str [0] ) )
return ( false ) ;
// If a delimiter has been specified, check that the supplied string starts with it
if ( $delimiter !== false )
{
if ( $str [0] != $delimiter )
return ( false ) ;
}
// Otherwise, the delimiter will be the first character of the string
else
$delimiter = $str [0] ;
// Handle asymetric delimiters
switch ( $delimiter )
{
case '<' : $end_delimiter = '>' ; break ;
case '{' : $end_delimiter = '}' ; break ;
case '[' : $end_delimiter = ']' ; break ;
case '(' : $end_delimiter = ')' ; break ;
default : $end_delimiter = $delimiter ;
}
// Find the trailing delimiter from the end of the string
for ( $i = $length - 1 ; $i > 0 ; $i -- )
{
$ch = $str [$i] ;
// Delimiter found : we can safely say this is a regex
if ( $ch == $end_delimiter )
{
// ... but only if the previous character is not an escape character
if ( $i > 0 && $str [ $i - 1 ] == '\\' )
return ( false ) ;
return ( true ) ;
}
// But if we find a non-pcre option after the trailing delimiter, then this definitely is not a regex
else if ( stripos ( $pcre_options, $ch ) === false )
return ( false ) ;
}
// No trailing delimiter found
return ( false ) ;
}
/**
*
* Checks if a filename corresponds to a filemask.
*
* @param string $file
* Filename to be checked.
*
* @param string $pattern
* Wildcard pattern. See the Regex::WildcardToRegex() method for an explanation on wildcard expressions syntax.
*
* @param bool $case_sensitive
* When true, comparisons are case-sensitive.
*
* @return bool|int
* Returns true if $pattern matches $file, false otherwise.
*
*/
public static function Matches ( $file, $pattern, $case_sensitive = false )
{
$length = strlen ( $pattern ) ;
$newpattern = "" ;
for ( $i = 0 ; $i < $length ; $i ++ )
{
$char = $pattern [ $i ] ;
$depth = 0 ;
switch ( $char )
{
case '/' : case '\\' :
$newpattern .= '[\\/\\\\]' ;
break ;
case '.' : case '+' : case '^' : case '$' : case '(' : case ')' : case '|' :
case '{' : case '}' : case '=' : case '!' : case '<' : case '>' : case '/' :
$newpattern .= '\\' . $char ;
break ;
case '?' : case '*' :
$newpattern .= '[^\\/\\\\]' . $char ;
break ;
case '[' :
$newpattern .= '[' ;
$depth ++ ;
break ;
case ']' :
if ( ! $depth )
return ( false ) ;
$newpattern .= ']' ;
break ;
default :
$newpattern .= $char ;
}
}
if ( $case_sensitive )
$extra = '' ;
else
$extra = 'i' ;
$status = preg_match ( '/^' . $newpattern . '$/' . $extra, $file ) ;
return ( ( $status ) ? true : false ) ;
}
/**
* A meta-matching artefact for regular expressions.
*
* Suppose you have to scan a sequence of lines, such as in a log file. You want to
* recognize which sequence follows which pattern.
*
* A sequence in an example log file could be, for example :
* - A line containing "message start"
* - Any number of lines starting with "log:" and followed by any sequence of characters
* - A line containing "message end"
*
* The following example gives a layout of such a log file :
*
* message start
* log: message 1
* log: message 2
* ...
* log: message n
* message end
*
* The purpose is to check whether a sequence of lines would match this scheme ; a set
* of regular expressions would be first needeed to match every particular line in a sequence :
*
* $regex_list =
* [
* '1' => '/message start/',
* '2' => 'log: \s* (?P<logmessage> .*),
* '3' => '/message end/'
* ] ;
*
* Then, to match a set of lines containing 'message start', having an unlimited number
* of lines starting with 'log:', then ending with a line containing 'message end', you
* would have to provide a regular expression using a backreference-style syntax
* referencing the keys of our $regex_list array, which would give :
*
* $sequence = '\1 \2* \3' ;
*
* meaning :
* - The first line must be the one identified by '\1', ie 'message start'
* - There can be any number of lines identified by '\2', ie starting with 'log:'
* - The last line must be 'message end'
*
* Note that each $regex_list item is a regular expression which can contain group
* captures, either named or not.
* If it does not contain re delimiters, then '/ /imsx' is assumed, so do not forget that
* spaces will not be significant.
*
* Thus, checking if a set of lines (in an array) matches the regular expressions
* specified in $sequence and defined in $regex_list, a simple call will be enough :
*
* $status = Regex::MetaPregMatchEx ( $sequence, $regex_list, $lines ) ;
*
* @param string $sequence
* A regular expression containing preg backreference-style constructs that
* refer to array keys in the $regex_list array.
*
* The following preg-style backreferences are supported ('x' stands for a
* sequence of digits, 'name' for a group capture name) :
* - \x
* - \gx
* - \g{x}
* - (?P=name)
* - \k<name>
* - \k'name'
* - \k{name}
* - \g{name}
*
* @param array $regex_list
* An associative array whose keys are backreference ids (either the 'x' or the
* 'name' string described in the $sequence parameter help) and whose values
* are regular expressions.
* Each entry is meant to match one or more lines of a sequence of lines.
*
* If no delimiter encloses the regex, then a default delimiter '/' will be used,
* and the 'imsx' preg options will be automatically added before performing the
* match.
*
*
* @param array $subject_array
* Array of input lines to be matched against the specified sequence.
*
* @param array $matches
* Reference to an array which will receive the individual matches.
* Each entry is an associative array having the following keys :
* - 'reference' :
* The original string reference.
* - 'regex' :
* The regex that matched the line.
* - 'matches' :
* Array of matches. Note that since the method uses PregMatchEx(),
* an additional level of indirection is added with regards to self::PregMatch,
* since several captures can have the same name.
*
* @param int $flags
* PREG_* Flags for the self::PregMatch*() function.
*
* @param array $missing_matches
* When specified, the indexes of non-matching lines will be stored in this array.
*
* @return bool
* Returns true if the sequence of lines matches the specified sequence, false otherwise.
*/
public static function MetaPregMatchEx ( $sequence, $regex_list, $subject_array, &$matches = null, $flags = 0, $match_all = false, $missing_matches = [] )
{
$match_list = [] ; // List of matched results in $subject_array
$match_references = [] ; // List of corresponding match references (keys in regex_list array)
// Normalize the sequence
$new_sequence = self::NormalizeMetaSequence ( $sequence ) ;
$subject_index = 0 ;
// Loop through each log line
foreach ( $subject_array as $subject )
{
$found = false ;
// Compare with each item in regex list
$regex_index = 0 ;
foreach ( $regex_list as $regex_key => $regex_value )
{
// Normalize regular expression
if ( ! self::IsRegex ( $regex_value ) )
$regex_value = '/^ \s* ' . str_replace ( '/', '\\/', $regex_value ) . '/imsx' ;
// Call either the PregMatchEx or PregMatchAllEx method
if ( $match_all )
$status = self::PregMatchAllEx ( $regex_value, $subject, $match_result, $flags ) ;
else
$status = self::PregMatchEx ( $regex_value, $subject, $match_result, $flags ) ;
// Match found : add this line to the match list
if ( $status )
{
$reference = '\\' . $regex_key ;
$new_match =
[
'reference' => $regex_key,
'subject' => $subject,
'subject-index' => $subject_index,
'regex' => $regex_value,
'regex-index' => $regex_index,
'matches' => $match_result
] ;
// Handle the case where several regex match the same string
// For example, if both regex indexed by "\4" and "\5" match the same string, we will have
// to construct a matching regex having the following contents :
// ( (\4) | (\5) )
if ( isset ( $match_list [ $subject_index ] ) )
{
$match_list [ $regex_key ] [] = $new_match ;
$match_references [ $subject_index ] [] = $reference ;
}
else
{
$match_list [ $regex_key ] = [ $new_match ] ;
$match_references [ $subject_index ] = [ $reference ] ;
}
$found = true ;
}
$regex_index ++ ;
}
// If no matching regex has been found, record the index of the current line
if ( ! $found )
$missing_matches [] = $subject_index ;
$subject_index ++ ;
}
// Not all subject lines have a match in the $regex_list array : consider this match has failed
$status = false ;
if ( ! count ( $missing_matches ) )
{
// Since an input string may correspond to several matches, get all the possible combination of matches
$key_subjects = self::CombinationsOf ( $match_references ) ;
// Make $sequence a regex if not already specified : add delimiters and options
if ( ! self::IsRegex ( $new_sequence ) )
$new_sequence = '/^ ' . str_replace ( '/', '\\/', $new_sequence ) . ' $/imsx' ;
// Find a match
$status = false ;
foreach ( $key_subjects as $key_subject )
{
$subject_string = implode ( '', $key_subject ) ;
if ( self::PregMatchAll ( $new_sequence, $subject_string, $seq_match ) )
{
$matches = $match_list ;
$status = true ;
}
}
}
// Return match status
return ( $status ) ;
}
/**
* Performs multiple inline substrings replacements.
*
* @param string $subject
* String where replacements are to be performed.
*
* @param array $replacements
* Array of arrays, each of them containing 3 elements :
* - Element 1 :
* The string to be replaced in $subject.
* - Element 2 :
* The replacement string.
* - Element 3 :
* The offset, in $subject, of the string to be replaced.
*
* @return string
* The input string, with all replacements having taken place.
*
*/
public static function MultiSubstrReplace ( $subject, $replacements )
{
// First, sort entries by ascending offset
if ( count ( $replacements ) > 1 )
usort ( $replacements, function ( $a, $b ) { return ( $a [2] - $b [2] ) ; } ) ;
// Parts that are extracted from the subject
$list = [] ;
// Last seen subject string offset so far
$last_seen_offset = 0 ;
$subject_length = strlen ( $subject ) ;
// Loop through replacement strings
foreach ( $replacements as $replacement )
{
$length = strlen ( $replacement [0] ) ;
// Delta between current offset and last seen offset : add this substring to the list of items to be joined at the end
if ( $replacement [2] > $last_seen_offset )
{
$list [] = substr ( $subject, $last_seen_offset, $replacement [2] - $last_seen_offset ) ;
$last_seen_offset = $replacement [2] + $length ;
}
// Include the replacement string
$list [] = $replacement [1] ;
// Update currently last seen offset (= current offset + current string length)
$last_seen_offset = $replacement [2] + $length ;
}
// Don't forget the trailing characters
if ( $last_seen_offset < $subject_length )
$list [] = substr ( $subject, $last_seen_offset ) ;
// All done, return
return ( implode ( '', $list ) ) ;
}
/**
*
* Normalizes a sequence for the MetaPregMatchEx method.
*
* Normalizes a meta-sequence, which uses preg-like backreference syntax to reference
* regular expressions indexed by the backreference value in the $match_definitions array.
* The method accepts all the backreference syntaxes that are recognized by the preg_replace
* function ('x' stands for a sequence of digits, 'name' for a group capture name) :
* - \x
* - \gx
* - \g{x}
* - (?P=name)
* - \k<name>
* - \k'name'
* - \k{name}
* - \g{name}
* All those forms are normalized in the input sequence as :
* (\x)
* or :
* (\name)
* Note the enclosing parentheses to prevent side effects when performing the match.
*
* @param string $sequence
* Sequence to be normalized.
*
* @param mixed $subsequences
* Associative array whose keys are sequence references and whose values are
* regular expressions to be matched.
* When specified, all references specified in the $sequence parameter are checked
* for existence in this array.
*
* @return string
* Returns the normalized sequence.
*
*/
// A regular expression matching all the possible backreferences allowed in PHP
// ('x' stands for a sequence of digits, 'name' for a group capture name) :
// - \x
// - \gx
// - \g{x}
// - (?P=name)
// - \k<name>
// - \k'name'
// - \k{name}
// - \g{name}
// Because of the (?P=name) form, we cannot unify all the regular expressions matching the
// above syntaxes, so we need to split the matching in two groups
const META_SEQUENCE_MATCH = '# [^\\\\]?
(?P<match>
(?P<group1>
\\\\
(
(
g (?P<number_reference_2> \d+)
)
|
(
g \{ (?P<number_reference_3> \d+) \}
)
|
(
k \< (?P<name_reference_1> [\w.\-]+) \>
)
|
(
k \' (?P<name_reference_2> [\w.\-]+) \'
)
|
(
[kg] \{ (?P<name_reference_3> [\w.\-]+) \}
)
|
(
(?P<name_reference_4> [\w.\-]+)
)
|
(
(?P<number_reference_1> \d+)
)
)
)
|
(?P<group2>
\( \? P = (?P<named_reference> [\w_.\-]+) \)
)
)
#imsx' ;
public static function NormalizeMetaSequence ( $sequence, &$subsequences = null )
{
// Are there any references to match items in the sequence string ?
if ( preg_match_all ( self::META_SEQUENCE_MATCH, $sequence, $matches, PREG_OFFSET_CAPTURE ) )
{
$substrings = [] ;
// Loop through found matches
foreach ( $matches as $key => $match )
{
// Numeric keys simply indicate unnamed group captures - ignore them
if ( is_numeric ( $key ) )
continue ;
// In order to obtain the real capture length, we need to determine whether the match comes
// from <group1> or <group2>
$group = false ;
if ( ! strncmp ( $key, 'number', 5 ) || ! strncmp ( $key, 'name', 4 ) )
$group = '1' ;
else if ( ! strncmp ( $key, 'named', 5 ) )
$group = '2' ;
else
continue ;
$groupname = "group$group" ;
$index = 0 ;
// Loop through each match found
foreach ( $matches [ $key ] as $value )
{
// ... but only if the value is an array and its offset item is not negative
if ( is_array ( $value ) && $value [1] != -1 )
{
// Add it to the substrings array which will be passed to the Regex::MultiSubstrReplace() method
// Note the extra 4th element, not used by Regex::MultiSubstrReplace, which serves to determine
// if the reference exists as a key in the $match_definitions array, when specified
$substrings [] =
[
$matches [ 'match' ] [ $index ] [0],
'(\\\\' . $value [0] . ')',
$matches [ $groupname ] [ $index ] [1],
$matches [ $key ] [ $index ] [0]
] ;
}
$index ++ ;
}
}
// Replace all reference syntaxes allowed with a much simpler "(\reference)" form
$new_sequence = Regex::MultiSubstrReplace ( $sequence, $substrings ) ;
// If matches have been specified, check that all the references found in $sequence have a corresponding entry
// in the $match_definitions array
if ( $subsequences !== null )
$subsequences = $substrings ;
return ( $new_sequence ) ;
}
// Useless sequence, since it contains no reference to a match string
else
throw new Exception ( "The following sequence does not contain any reference to match strings :\n$sequence" ) ;
}
/**
*
* Removes unnamed captures from the result of a call to a preg_* function.
*
* @param array $matches
* Matches returned as the 3rd parameter of a preg_* call.
*
* @param int $flags
* PREG_* flags specified when calling a preg_* function. When the PREG_OFFSET_CAPTURE flag is
* specified, it happens that some results are an empty string instead of a 2-elements array.
* This helps removing unnecessary entries.
*
*/
public static function PregWipeMatches ( &$matches, $flags )
{
$new_matches = [] ;
foreach ( $matches as $key => $value )
{
if ( is_numeric ( $key ) )
continue ;
if ( $flags & PREG_OFFSET_CAPTURE )
{
if ( ! is_array ( $value ) )
continue ;
if ( $value [1] == -1 )
continue ;
}
$new_matches [ $key ] = $value ;
}
$matches = $new_matches ;
}
/**
*
* Encapsulates the preg_match() function and optionnally wipes unnamed captures if the PREG_WIPE_MATCHES flag
* is specified.
*
*/
public static function PregMatch ( $pattern, $subject, &$matches = null, $flags = 0, $offset = 0 )
{
// Perform the match and handle potential error
$status = @preg_match ( $pattern, $subject, $matches, $flags, $offset ) ;
if ( $status === false )
throw ( new Exception ( "Invalid regex : $pattern" ) ) ;
// If needed, wipe any unnamed captures
if ( $flags & PREG_WIPE_MATCHES )
self::PregWipeMatches ( $matches, $flags ) ;
// All done, return
return ( $status ) ;
}
/**
*
* Encapsulates the preg_match_all() function and optionnally wipes unnamed captures if the PREG_WIPE_MATCHES flag
* is specified.
*
*/
public static function PregMatchAll ( $pattern, $subject, &$matches = null, $flags = PREG_PATTERN_ORDER, $offset = 0 )
{
// Perform the match and handle potential error
$status = @preg_match_all ( $pattern, $subject, $matches, $flags, $offset ) ;
if ( $status === false )
throw ( new Exception ( "Invalid regex : $pattern" ) ) ;
// If needed, wipe any unnamed captures
if ( $flags & PREG_WIPE_MATCHES )
self::PregWipeMatches ( $matches, $flags ) ;
// All done, return
return ( $status ) ;
}
/**
*
* Encapsulates the preg_replace() function.
*
*/
public static function PregReplace ( $pattern, $replacement, $subject, $limit = -1, $count = null )
{
$status = @preg_replace ( $pattern, $replacement, $subject, $limit, $count ) ;
if ( $status === false )
throw ( new Exception ( "Invalid regex : $pattern" ) ) ;
return ( $status ) ;
}
/**
*
* An extended version of the preg_match() function, that allows for specifying multiple named captures
* with the same name.
*
* @param array $matches
* Strings matched by the captures. Since named captures can be specified more than once, each array item
* will contain an additional level of indirection, an array for each matched item.
* Thus, the elements of a capture named <pat> will be accessible through the following expressions :
* - $match [ 'pat' ] [0] will yield to the first capture of a group named 'pat'
* - count ( $match [ 'pat' ] ) will give the number of expressions matched by the named capture 'pat'
* The same applies to unnamed captures.
*
*/
public static function PregMatchEx ( $pattern, $subject, &$matches = null, $flags = 0, $offset = 0 )
{
$newpattern = self::RenumberNamedCaptures ( $pattern, $replacements ) ;
if ( $status = self::PregMatch ( $newpattern, $subject, $matches, $flags, $offset ) )
{
$matches = self::GroupNamedCaptures ( $matches, $replacements ) ;
}
return ( $status ) ;
}
/**
*
* An extended version of the preg_match_all() function, that allows for specifying multiple named captures
* with the same name.
*
*/
public static function PregMatchAllEx ( $pattern, $subject, &$matches = null, $flags = 0, $offset = 0 )
{
$replacements = [] ;
$newpattern = self::RenumberNamedCaptures ( $pattern, $replacements ) ;
if ( $status = self::PregMatchAll ( $newpattern, $subject, $matches, $flags, $offset ) )
{
$matches = self::GroupNamedCaptures ( $matches, $replacements ) ;
// Cancel one level of indirection in the results ; the resulting match array will have
// the same shape as the one returned by the PregMatchEx() method
foreach ( $matches as $key => $value )
{
// Non-numeric keys are named pattern matches : get rid of one indirection level and collect only
// the successful matches ; this is a deviation from the normal preg_match_all() function, where the
// results could include as well an empty string or a sub-array with an empty string and an offset of -1
if ( ! is_numeric ( $key ) )
{
$new_array = [] ;
foreach ( $value as $subvalue )
{
foreach ( $subvalue as $subsubkey => $subsubvalue )
{
if ( isset ( $subsubvalue ) && is_array ( $subsubvalue ) && $subsubvalue [1] != -1 )
$new_array [ $subsubkey ] = $subsubvalue ;
}
}
$matches [ $key ] = $new_array ;
}
// For numeric keys, when the sub-array contains only one element, then this is a normal preg_match result,
// so get rid of the level of indirection that has been added by the GroupNamedCaptures() method
else if ( count ( $value ) == 1 )
$matches [ $key ] = $value [0] ;
}
}
return ( $status ) ;
}
/*--------------------------------------------------------------------------------------------------------------
NAME
PregStrReplace - Replace string(s) using regular expression(s)
PROTOTYPE
$result = PdfToText::PregStrReplace ( $pattern, $replacement, $subject, $limit = -1,
&$match_count = null )
DESCRIPTION
This function behaves like a mix of str_replace() and preg_replace() ; it allows to search for strings
using regular expressions, but the replacements are plain-text strings and no reference to a capture
specified in the regular expression will be interpreted.
This is useful when processing replacement strings that contain constructs such as "\00" or "$", which
are interpreted by preg_replace() as references to captures.
This function has the same parameters as preg_replace().
RETURN VALUE
Returns the substituted text.
*-------------------------------------------------------------------------------------------------------------*/
public static function PregStrReplace ( $pattern, $replacement, $subject, $limit = -1, &$match_count = null )
{
// Make sure that $pattern and $replacement become arrays of the same size
if ( is_array ( $pattern ) )
{
if ( is_array ( $replacement ) )
{
if ( count ( $pattern ) !== count ( $replacement ) )
{
trigger_error ( "The \$replacement parameter should have the same number of element as \$pattern." ) ;
return ( $subject ) ;
}
}
else
$replacement = array_fill ( $replacement, count ( $pattern ), $replacement ) ;
}
else
{
if ( is_array ( $replacement ) )
{
trigger_error ( "Expected string for the \$replacement parameter." ) ;
return ( $subject ) ;
}
$pattern = array ( $pattern ) ;
$replacement = array ( $replacement ) ;
}
// Upper limit
if ( $limit < 1 )
$limit = PHP_INT_MAX ;
// Loop through each supplied pattern
$current_subject = $subject ;
$count = 0 ;
for ( $i = 0, $pattern_count = count ( $pattern ) ; $i < $pattern_count ; $i ++ )
{
$regex = $pattern [$i] ;
// Get all matches for this pattern
if ( preg_match_all ( $regex, $current_subject, $matches, PREG_OFFSET_CAPTURE ) )
{
$result = '' ; // Current output result
$last_offset = 0 ;
// Process each match
foreach ( $matches [0] as $match )
{
$offset = ( integer ) $match [1] ;
// Append data from the last seen offset up to the current one
if ( $last_offset < $offset )
$result .= substr ( $current_subject, $last_offset, $offset - $last_offset ) ;
// Append the replacement string for this match
$result .= $replacement [$i] ;
// Compute next offset in $current_subject
$last_offset = $offset + strlen ( $match [0] ) ;
// Limit checking
$count ++ ;
if ( $count > $limit )
break 2 ;
}
// Append the last part of the subject that has not been matched by anything
$result .= substr ( $current_subject, $last_offset ) ;
// The current subject becomes the string that has been built in the steps above
$current_subject = $result ;
}
}
/// All done, return
return ( $current_subject ) ;
}
/**
*
* Gives a unique id to each named capture within a regex.
*
* Reassigns unique identifiers to named captures within a regular expression. The new
* identifiers will have the form "prefix_x", where "prefix" is given by the $prefix
* parameter, and "x" a unique identifier starting from 0.
*
* @param string $pattern
* Regex pattern containing potential named captures to be renamed.
*
* @param array $correspondances
* On output, will hold an associative array whose keys are the new capture group names, and values the old ones.
*
* @param string $prefix
* Prefix string for capture name replacements.
*
* @return string
* Returns the input pattern with all named captures replaced by unique identifiers.
*
* @notes
* This method, along with the GroupNamedCaptures() one, is used by the Preg*Ex methods
* to allow processing of regular expressions having duplicate named captures.
*
*/
public static function RenumberNamedCaptures ( $pattern, &$correspondances = [], $prefix = 'match_' )
{
static $re = '/
\( \? P <
(?P<pattern> [^>]+ )
>
/imsx' ;
// Get named captures
if ( self::PregMatchAll ( $re, $pattern, $matches, PREG_OFFSET_CAPTURE ) )
{
$index = 0 ;
$pattern_matches = [] ;
// Loop through pattern matches
foreach ( $matches [ 'pattern' ] as $match )
{
$pname = $match [0] ;
$poffset = $match [1] ;
$newpattern = "$prefix$index" ;
// Build the correspondance array
$correspondances [ $newpattern ] = $pname ;
// Add this entry (old name, new name, offset) into an array for the Regex::MultiSubsrReplace() method
$pattern_matches [] = [ $pname, $newpattern, $poffset ] ;
$index ++ ;
}
// Perform the multiple-string replace
$new_pattern = Regex::MultiSubstrReplace ( $pattern, $pattern_matches ) ;
}
else
$new_pattern = $pattern ;
// All done, return
return ( $new_pattern ) ;
}
/**
*
* Replaces patterns in a regular expression string.
*
* Replace named patterns in a string. This function uses the result of self::PregMatchAll()
* to match named patterns with the supplied input array $replacements.
*
* @param string $pattern
* A pattern matching subpart(s) of the specified subject string.
*
* @param string $subject
* String to be matched against.
*
* @param array $replacements
* Associative array whose keys are the pattern name (as specified in the
* (?P<name> re) parts of a regular expression) and whose values are also
* an associative array. Each entry in the array have the following meaning :
*
* - key : A regular expression specifying the value of the named pattern
* name. Do not put anchors nor delimiters in this pattern since
* they are automatically added.
* - value : The replacement value for the named pattern specified by the key
* value.
*
* @param int $options
* PREG_* options.
*
* @return string
* Returns the substituted string.
*
* @example
* The following example will replace :
* - every matched pattern named "toto" by one of the string matching this pattern,
* either 'the new replacement for the pattern string' or 'the replacement of thepattern',
* for captures matching either 'pattern.*' or 'thepattern', respectively.
* The (?P<pattern1> ...) or (?P<pattern2> ...) match the first item.
* - Every matched pattern name "aaa" by the string 'replacement of aaa'.
*
* $subject = "aaa (?P<pattern1> coucou) zzzz (?P<pattern2> coucou2) aaa (?P<thepattern> zzzz)";
* $pattern = "# ( \( \? P < (?P<toto> [^>]+) > .*? \) ) | (?P<aaa> aaa*) #imsx" ;
* $replacement =
* [
* "toto" =>
* [
* 'pattern.*' => 'the new replacement for the pattern string',
* 'thepattern' => 'the replacement of thepattern'
* ],
* "aaa" =>
* [
* 'aaa' => 'replacement of aaa'
* ]
* ] ;
* $result = String::RegReplaceNamedPatterns ( $pattern, $subject, $replacement ) ) ;
*
* The resulting string will be :
* replacement of aaa (?P<the new replacement for the pattern string> coucou) zzzz (?P<the
* new replacement for the pattern string> coucou2) replacement of aaa (?P<the replacement of thepattern> zzzz)
*
*/
public static function ReplaceNamedPatterns ( $pattern, $subject, $replacements, $options = null )
{
// Process the substitutions only if there is a match...
if ( self::PregMatchAll ( $pattern, $subject, $matches, PREG_OFFSET_CAPTURE | $options ) )
{
$substitutions = [] ;
// Loop through matches
foreach ( $matches as $match_key => $match_values )
{
// ... then through each matched value
foreach ( $match_values as $match_value )
{
// Ignore empty matches
if ( ! $match_value || $match_value [1] == -1 )
continue ;
// Loop through replacement patterns
foreach ( $replacements as $replacement_key => $replacement_values )
{
// Skip non-matching ones
if ( strcmp ( $match_key, $replacement_key ) )
continue ;
// Find the replacement values for each matched pattern
foreach ( $replacement_values as $replacement_re => $replacement_value )
{
if ( self::PregMatch ( "/^$replacement_re\$/imsx", $match_value [0] ) )
{
$substitutions [] =
[
$match_value [0], // Original value to be replaced
$replacement_value, // Replacement value
$match_value [1] // Original value offset
] ;
}
}
}
}
}
// Replace catched captures
$new_subject = self::MultiSubstrReplace ( $subject, $substitutions ) ;
return ( $new_subject ) ;
}
// No match found : return the subject as is
else
return ( $subject ) ;
}
/**
* Converts an Msdos or Unix-style wildcard string to a regular expression.
*
* Converts an Msdos or Unix-style wildcard string to a regular expression. Allowed wildcard forms are :
* '?' -
* Matches 0 or 1 character Path separator cannot be matched with this expression.
* '*' -
* Matches 0 or more characters. Path separator cannot be matched with this expression.
* '[cclass]' -
* A character class that matches one character, for example "[a-z]" to match any lowercase alphabetic character, or
* "[^a-z]" to match anything but a lowercase alphabetic character.
*
* Note that the path separator can either be '/' or '\' : this method does not take care of the host operating system conventions.
*
* @param string $pattern
* Wildcard pattern to be converted.
*
* @param string $escaped_chars
* Optional characters to be escaped in the regular expression (regex special characters are systematically escaped).
*
* @return bool|string
* Returns false if the wilcard string cannot be converted to a regular expression (typically because there is an angle
* bracket mismatch) or the regular expression corresponding to the supplied wildcard pattern.
* Note that the returned value does not include the leading and trailing delimiters, so that it can be included into
* an existing regex.
*
*/
public static function WildcardToRegex ( $pattern, $escaped_chars = "" )
{
$length = strlen ( $pattern ) ;
$newpattern = "" ;
for ( $i = 0 ; $i < $length ; $i ++ )
{
$char = $pattern [ $i ] ;
$depth = 0 ;
switch ( $char )
{
case '.' : case '+' : case '^' : case '$' : case '(' : case ')' : case '|' :
case '{' : case '}' : case '=' : case '!' : case '<' : case '>' : case '/' :
$newpattern .= '\\' . $char ;
break ;
case '?' : case '*' :
$newpattern .= '[^\\/]' . $char ;
break ;
case '[' :
$newpattern .= '[' ;
$depth ++ ;
break ;
case ']' :
if ( ! $depth )
return ( false ) ;
$newpattern .= ']' ;
break ;
default :
if ( strpos ( $escaped_chars, $char ) !== false )
$newpattern .= '\\' ;
$newpattern .= $char ;
}
}
return ( $newpattern ) ;
}
}
|
Download
