Phobos

Github Page

The D Programming Language

Intro

Regular expressions are a commonly used method of pattern matching on strings, with  regex being a catchy word for a pattern in this domain specific language. Typical problems usually solved by regular expressions include validation of user input and the ubiquitous find & replace in text processing utilities.

Synopsis

import std.regex;
import std.stdio;
void main()
{
    // Print out all possible dd/mm/yy(yy) dates found in user input.

    // g - global: find all matches.

    auto r = regex(r"\b[0-9][0-9]?/[0-9][0-9]?/[0-9][0-9](?:[0-9][0-9])?\b", "g");
    foreach(line; stdin.byLine)
    {
      // Match returns a range that can be iterated

      // to get all subsequent matches.

      foreach(c; match(line, r))
          writeln(c.hit);
    }
}
...

// Create a static regex at compile-time, which contains fast native code.

enum ctr = ctRegex!(`^.*/([^/]+)/?$`);

// It works just like a normal regex:

auto m2 = match("foo/bar", ctr);   // First match found here, if any

assert(m2);   // Be sure to check if there is a match before examining contents!

assert(m2.captures[1] == "bar");   // Captures is a range of submatches: 0 = full match.


...

// The result of the match is directly testable with if/assert/while.

// e.g. test if a string consists of letters:

assert(match("Letter", `^\p{L}+$`));

Syntax and general information

The general usage guideline is to keep  regex complexity on the side of simplicity, as its capabilities reside in purely character-level manipulation. As such it's ill-suited for tasks involving higher level invariants like matching an integer number bounded in an [a,b] interval. Checks of this sort of are better addressed by additional post-processing.

The basic syntax shouldn't surprise experienced users of regular expressions. For an introduction to std. regex see a short tour of the module API and its abilities.

There are other web resources on regular expressions to help newcomers, and a good reference with tutorial can easily be found.

This library uses a remarkably common ECMAScript syntax flavor with the following extensions:
  • Named subexpressions, with Python syntax.
  • Unicode properties such as Scripts, Blocks and common binary properties e.g Alphabetic, White_Space, Hex_Digit etc.
  • Arbitrary length and complexity lookbehind, including lookahead in lookbehind and vise-versa.


Pattern syntax

std. regex operates on codepoint level, 'character' in this table denotes a single Unicode codepoint.
Pattern element Semantics
Atoms Match single characters
any character except [{|*+?()^$ Matches the character itself.
. In single line mode matches any character. Otherwise it matches any character except '\n' and '\r'.
[class] Matches a single character that belongs to this character class.
[^class] Matches a single character that does not belong to this character class.
\cC Matches the control character corresponding to letter C
\xXX Matches a character with hexadecimal value of XX.
\uXXXX Matches a character with hexadecimal value of XXXX.
\U00YYYYYY Matches a character with hexadecimal value of YYYYYY.
\f Matches a formfeed character.
\n Matches a linefeed character.
\r Matches a carriage return character.
\t Matches a tab character.
\v Matches a vertical tab character.
\d Matches any Unicode digit.
\D Matches any character except Unicode digits.
\w Matches any word character (note: this includes numbers).
\W Matches any non-word character.
\s Matches whitespace, same as \p{White_Space}.
\S Matches any character except those recognized as \s .
\\ Matches \ character.
\c where c is one of [|*+?() Matches the character c itself.
\p{PropertyName} Matches a character that belongs to the Unicode PropertyName set. Single letter abbreviations can be used without surrounding {,}.
\P{PropertyName} Matches a character that does not belong to the Unicode PropertyName set. Single letter abbreviations can be used without surrounding {,}.
\p{InBasicLatin} Matches any character that is part of the BasicLatin Unicode block.
\P{InBasicLatin} Matches any character except ones in the BasicLatin Unicode block.
\p{Cyrilic} Matches any character that is part of Cyrilic script.
\P{Cyrilic} Matches any character except ones in Cyrilic script.
Quantifiers Specify repetition of other elements
* Matches previous character/subexpression 0 or more times. Greedy version - tries as many times as possible.
*? Matches previous character/subexpression 0 or more times. Lazy version - stops as early as possible.
+ Matches previous character/subexpression 1 or more times. Greedy version - tries as many times as possible.
+? Matches previous character/subexpression 1 or more times. Lazy version - stops as early as possible.
{n} Matches previous character/subexpression exactly n times.
{n,} Matches previous character/subexpression n times or more. Greedy version - tries as many times as possible.
{n,}? Matches previous character/subexpression n times or more. Lazy version - stops as early as possible.
{n,m} Matches previous character/subexpression n to m times. Greedy version - tries as many times as possible, but no more than m times.
{n,m}? Matches previous character/subexpression n to m times. Lazy version - stops as early as possible, but no less then n times.
Other Subexpressions & alternations
( regex) Matches subexpression  regex, saving matched portion of text for later retrieval.
(?: regex) Matches subexpression  regex, not saving matched portion of text. Useful to speed up matching.
A|B Matches subexpression A, or failing that, matches B.
(?P<name> regex) Matches named subexpression  regex labeling it with name 'name'. When referring to a matched portion of text, names work like aliases in addition to direct numbers.
Assertions Match position rather than character
^ Matches at the begining of input or line (in multiline mode).
$ Matches at the end of input or line (in multiline mode).
\b Matches at word boundary.
\B Matches when not at word boundary.
(?= regex) Zero-width lookahead assertion. Matches at a point where the subexpression  regex could be matched starting from the current position.
(?! regex) Zero-width negative lookahead assertion. Matches at a point where the subexpression  regex could not be matched starting from the current position.
(?<= regex) Zero-width lookbehind assertion. Matches at a point where the subexpression  regex could be matched ending at the current position (matching goes backwards).
(?<! regex) Zero-width negative lookbehind assertion. Matches at a point where the subexpression  regex could not be matched ending at the current position (matching goes backwards).


Character classes

Pattern element Semantics
Any atom Has the same meaning as outside of a character class.
a-z Includes characters a, b, c, ..., z.
[a||b], [a--b], [a~~b], [a&&b] Where a, b are arbitrary classes, means union, set difference, symmetric set difference, and intersection respectively. Any sequence of character class elements implicitly forms a union.


Regex flags

Flag Semantics
g Global  regex, repeat over the whole input.
i Case insensitive matching.
m Multi-line mode, match ^, $ on start and end line separators as well as start and end of input.
s Single-line mode, makes . match '\n' and '\r' as well.
x Free-form syntax, ignores whitespace in pattern, useful for formatting complex regular expressions.


Unicode support



This library provides full Level 1 support* according to UTS 18. Specifically:
  • 1.1 Hex notation via any of \uxxxx, \U00YYYYYY, \xZZ.
  • 1.2 Unicode properties.
  • 1.3 Character classes with set operations.
  • 1.4 Word boundaries use the full set of "word" characters.
  • 1.5 Using simple casefolding to match case insensitively across the full range of codepoints.
  • 1.6 Respecting line breaks as any of \u000A | \u000B | \u000C | \u000D | \u0085 | \u2028 | \u2029 | \u000D\u000A.
  • 1.7 Operating on codepoint level.
*With exception of point 1.1.1, as of yet, normalization of input is expected to be enforced by user.

Replace format string



A set of functions in this module that do the substitution rely on a simple format to guide the process. In particular the table below applies to the format argument of replaceFirst and replaceAll .

The format string can reference parts of match using the following notation.
Format specifier Replaced by
$& the whole match.
$` part of input preceding the match.
$' part of input following the match.
$$ '$' character.
\c , where c is any character the character c itself.
\\ '\' character.
$1 .. $99 submatch number 1 to 99 respectively.


Slicing and zero memory allocations orientation



All matches returned by pattern matching functionality in this library are slices of the original input. The notable exception is the replace family of functions that generate a new string from the input.

In cases where producing the replacement is the ultimate goal replaceFirstInto and replaceAllInto could come in handy as functions that avoid allocations even for replacement.

License, Boost License 1.0.

Authors, Dmitry Olshansky,

API and utility constructs are modeled after the original std. regex by Walter Bright and Andrei Alexandrescu.

Source:
std/regex.d

struct  Regex(Char);

 Regex object holds regular expression pattern in compiled form. Instances of this object are constructed via calls to regex. This is an intended form for caching and storage of frequently used regular expressions.

const nothrow bool  empty();

Test if this object doesn't contain any compiled pattern.

Example:
Regex!char r;
assert(r.empty);
r = regex(""); // Note: "" is a valid regex pattern.

assert(!r.empty);

@property @safe auto  namedCaptures();

A range of all the named captures in the regex.

Example:
import std.range;
import std.algorithm;

auto re = regex(`(?P<name>\w+) = (?P<var>\d+)`);
auto nc = re.namedCaptures;
static assert(isRandomAccessRange!(typeof(nc)));
assert(!nc.empty);
assert(nc.length == 2);
assert(nc.equal(["name", "var"]));
assert(nc[0] == "name");
assert(nc[1..$].equal(["var"]));

NamedGroup[]  dict;



uint  ngroup;



uint  maxCounterDepth;



uint  hotspotTableSize;



uint  threadCount;



uint  flags;



const(Trie)[]  tries;



uint[]  backrefed;



Kickstart!Char  kickstart;



uint  isBackref(uint n);



void  checkIfOneShot();



void  lightPostprocess();



void  validate();



const void  print();



this(S)(Parser!S p);



struct  StaticRegex(Char);

A  StaticRegex is Regex object that contains specially generated machine code to speed up matching. Implicitly convertible to normal Regex, however doing so will result in losing this additional capability.

struct  Captures(R, DIndex = size_t) if (isSomeString!R);

 Captures object contains submatches captured during a call to match or iteration over RegexMatch range.

First element of range is the whole match.

Example, showing basic operations on  Captures: 

import std.regex;
import std.range;

void main()
{
    auto m = match("@abc#", regex(`(\w)(\w)(\w)`));
    auto c = m.captures;
    assert(c.pre == "@"); // Part of input preceeding match

    assert(c.post == "#"); // Immediately after match

    assert(c.hit == c[0] && c.hit == "abc"); // The whole match

    assert(c[2] =="b");
    assert(c.front == "abc");
    c.popFront();
    assert(c.front == "a");
    assert(c.back == "c");
    c.popBack();
    assert(c.back == "b");
    popFrontN(c, 2);
    assert(c.empty);
}

R  pre();

Slice of input prior to the match.

R  post();

Slice of input immediately after the match.

R  hit();

Slice of matched portion of input.

R  front();
R  back();
void  popFront();
void  popBack();
const bool  empty();
R  opIndex()(size_t i);

Range interface.

R  opIndex(String)(String i) if (isSomeString!String);

Lookup named submatch. 

import std.regex;
import std.range;

auto m = match("a = 42;", regex(`(?P<var>\w+)\s*=\s*(?P<value>\d+);`));
auto c = m.captures;
assert(c["var"] == "a");
assert(c["value"] == "42");
popFrontN(c, 2);
//named groups are unaffected by range primitives

assert(c["var"] =="a");
assert(c.front == "42");

const size_t  length();

Number of matches in this object.

@property ref auto  captures();

A hook for compatibility with original std.regex.

struct  RegexMatch(R, alias Engine = ThompsonMatcher) if (isSomeString!R);

A regex engine state, as returned by match family of functions.

Effectively it's a forward range of Captures!R, produced by lazily searching for matches in a given input.

alias Engine specifies an engine type to use during matching, and is automatically deduced in a call to match/bmatch.

R  pre();
R  post();
R  hit();

Shorthands for front. pre, front.post, front.hit.

@property auto  front();
void  popFront();
auto  save();

Functionality for processing subsequent matches of global regexes via range interface: 

import std.regex;
auto m = match("Hello, world!", regex(`\w+`, "g"));
assert(m.front.hit == "Hello");
m.popFront();
assert(m.front.hit == "world");
m.popFront();
assert(m.empty);

bool  empty();

Test if this match object is  empty.

T  opCast(T : bool)();

Same as !(x.empty), provided for its convenience in conditional statements.

@property auto  captures();

Same as .front, provided for compatibility with original std.regex.

auto  regex(S)(S pattern, const(char)[] flags = "") if (isSomeString!S);

Compile regular expression pattern for the later execution.

Returns, Regex object that works on inputs having the same character width as pattern.

Parameters,
S pattern Regular expression
const(char)[] flags The attributes (g, i, m and x accepted)

Throws, RegexException if there were any errors during compilation.

template  ctRegex(alias pattern, alias flags = [])

Experimental feature.

Compile regular expression using CTFE and generate optimized native machine code for matching it.

Returns, StaticRegex object for faster matching.

Parameters,
pattern Regular expression
flags The attributes (g, i, m and x accepted)

auto  match(R, RegEx)(R input, RegEx re) if (isSomeString!R && is(RegEx == Regex!(BasicElementOf!R)));
auto  match(R, String)(R input, String re) if (isSomeString!R && isSomeString!String);

Start matching input to regex pattern re, using Thompson NFA matching scheme.

The use of this function is discouraged - use either of matchAll or matchFirst .

Delegating the kind of operation to "g" flag is soon to be phased out along with the ability to choose the exact matching scheme. The choice of matching scheme to use depends highly on the pattern kind and can done automatically on case by case basis.

Returns, a RegexMatch object holding engine state after first  match.

auto  matchFirst(R, RegEx)(R input, RegEx re) if (isSomeString!R && is(RegEx == Regex!(BasicElementOf!R)));
auto  matchFirst(R, String)(R input, String re) if (isSomeString!R && isSomeString!String);

Find the first (leftmost) slice of the input that matches the pattern re. This function picks the most suitable regular expression engine depending on the pattern properties.

re parameter can be one of three types:

  • Plain string, in which case it's compiled to bytecode before matching.
  • Regex!char (wchar/dchar) that contains a pattern in the form of compiled bytecode.
  • StaticRegex!char (wchar/dchar) that contains a pattern in the form of compiled native machine code.

Returns, Captures containing the extent of a match together with all submatches if there was a match, otherwise an empty Captures object.

auto  matchAll(R, RegEx)(R input, RegEx re) if (isSomeString!R && is(RegEx == Regex!(BasicElementOf!R)));
auto  matchAll(R, String)(R input, String re) if (isSomeString!R && isSomeString!String);

Initiate a search for all non-overlapping matches to the pattern re in the given input. The result is a lazy range of matches generated as they are encountered in the input going left to right.

This function picks the most suitable regular expression engine depending on the pattern properties.

re parameter can be one of three types:

  • Plain string, in which case it's compiled to bytecode before matching.
  • Regex!char (wchar/dchar) that contains a pattern in the form of compiled bytecode.
  • StaticRegex!char (wchar/dchar) that contains a pattern in the form of compiled native machine code.

Returns, RegexMatch object that represents matcher state after the first match was found or an empty one if not present.

auto  bmatch(R, RegEx)(R input, RegEx re) if (isSomeString!R && is(RegEx == Regex!(BasicElementOf!R)));
auto  bmatch(R, String)(R input, String re) if (isSomeString!R && isSomeString!String);

Start matching of input to regex pattern re, using traditional backtracking matching scheme.

The use of this function is discouraged - use either of matchAll or matchFirst .

Delegating the kind of operation to "g" flag is soon to be phased out along with the ability to choose the exact matching scheme. The choice of matching scheme to use depends highly on the pattern kind and can done automatically on case by case basis.

Returns, a RegexMatch object holding engine state after first match.

R  replaceFirst(R, C, RegEx)(R input, RegEx re, const(C)[] format) if (isSomeString!R && is(C : dchar) && isRegexFor!(RegEx, R));

Construct a new string from input by replacing the first match with a string generated from it according to the format specifier.

To replace all matches use replaceAll .

Parameters,
R input string to search
RegEx re compiled regular expression to use
const(C)[] format format string to generate replacements from, see .

Returns, A string of the same type with the first match (if any) replaced. If no match is found returns the input string itself.

Example:
assert(replaceFirst("noon", regex("n"), "[$&]") == "[n]oon");

R  replaceFirst(alias fun, R, RegEx)(R input, RegEx re) if (isSomeString!R && isRegexFor!(RegEx, R));

This is a general replacement tool that construct a new string by replacing matches of pattern re in the input. Unlike the other overload there is no format string instead captures are passed to to a user-defined functor fun that returns a new string to use as replacement.

This version replaces the first match in input, see replaceAll to replace the all of the matches.

Returns, A new string of the same type as input with all matches replaced by return values of fun. If no matches found returns the input itself.

Example:
string list = "#21 out of 46";
string newList = replaceFirst!(cap => to!string(to!int(cap.hit)+1))
    (list, regex(`[0-9]+`));
assert(newList == "#22 out of 46");

void  replaceFirstInto(Sink, R, C, RegEx)(ref Sink sink, R input, RegEx re, const(C)[] format) if (isOutputRange!(Sink, dchar) && isSomeString!R && is(C : dchar) && isRegexFor!(RegEx, R));
void  replaceFirstInto(alias fun, Sink, R, RegEx)(Sink sink, R input, RegEx re) if (isOutputRange!(Sink, dchar) && isSomeString!R && isRegexFor!(RegEx, R));

A variation on replaceFirst that instead of allocating a new string on each call outputs the result piece-wise to the sink. In particular this enables efficient construction of a final output incrementally.

Like in replaceFirst family of functions there is an overload for the substitution guided by the format string and the one with the user defined callback.

Example:
import std.array;
string m1 = "first message\n";
string m2 = "second message\n";
auto result = appender!string();
replaceFirstInto(result, m1, regex(`([a-z]+) message`), "$1");
//equivalent of the above with user-defined callback

replaceFirstInto!(cap=>cap[1])(result, m2, regex(`([a-z]+) message`));
assert(result.data == "first\nsecond\n");

R  replaceAll(R, C, RegEx)(R input, RegEx re, const(C)[] format) if (isSomeString!R && is(C : dchar) && isRegexFor!(RegEx, R));

Construct a new string from input by replacing all of the fragments that match a pattern re with a string generated from the match according to the format specifier.

To replace only the first match use replaceFirst .

Parameters,
R input string to search
RegEx re compiled regular expression to use
const(C)[] format format string to generate replacements from, see .

Returns, A string of the same type as input with the all of the matches (if any) replaced. If no match is found returns the input string itself.

Example:
// Comify a number

auto com = regex(r"(?<=\d)(?=(\d\d\d)+\b)","g");
assert(replaceAll("12000 + 42100 = 54100", com, ",") == "12,000 + 42,100 = 54,100");

R  replaceAll(alias fun, R, RegEx)(R input, RegEx re) if (isSomeString!R && isRegexFor!(RegEx, R));

This is a general replacement tool that construct a new string by replacing matches of pattern re in the input. Unlike the other overload there is no format string instead captures are passed to to a user-defined functor fun that returns a new string to use as replacement.

This version replaces all of the matches found in input, see replaceFirst to replace the first match only.

Returns, A new string of the same type as input with all matches replaced by return values of fun. If no matches found returns the input itself.

Parameters,
R input string to search
RegEx re compiled regular expression
fun delegate to use

Example:
Capitalize the letters 'a' and 'r': 
string baz(Captures!(string) m)
{
    return std.string.toUpper(m.hit);
}
auto s = replaceAll!(baz)("Strap a rocket engine on a chicken.",
        regex("[ar]"));
assert(s == "StRAp A Rocket engine on A chicken.");

void  replaceAllInto(Sink, R, C, RegEx)(Sink sink, R input, RegEx re, const(C)[] format) if (isOutputRange!(Sink, dchar) && isSomeString!R && is(C : dchar) && isRegexFor!(RegEx, R));
void  replaceAllInto(alias fun, Sink, R, RegEx)(Sink sink, R input, RegEx re) if (isOutputRange!(Sink, dchar) && isSomeString!R && isRegexFor!(RegEx, R));

A variation on replaceAll that instead of allocating a new string on each call outputs the result piece-wise to the sink. In particular this enables efficient construction of a final output incrementally.

As with replaceAll there are 2 overloads - one with a format string, the other one with a user defined functor.

Example:
//swap all 3 letter words and bring it back

string text = "How are you doing?";
auto sink = appender!(char[])();
replaceAllInto!(cap => retro(cap[0]))(sink, text, regex(`\b\w{3}\b`));
auto swapped = sink.data.dup; // make a copy explicitly

assert(swapped == "woH era uoy doing?");
sink.clear();
replaceAllInto!(cap => retro(cap[0]))(sink, swapped, regex(`\b\w{3}\b`));
assert(sink.data == text);

R  replace(alias scheme = match, R, C, RegEx)(R input, RegEx re, const(C)[] format) if (isSomeString!R && isRegexFor!(RegEx, R));
R  replace(alias fun, R, RegEx)(R input, RegEx re) if (isSomeString!R && isRegexFor!(RegEx, R));

Old API for replacement, operation depends on flags of pattern re. With "g" flag it performs the equivalent of replaceAll otherwise it works the same as replaceFirst .

The use of this function is discouraged, please use replaceAll or replaceFirst explicitly.

struct  Splitter(Range, alias RegEx = Regex) if (isSomeString!Range && isRegexFor!(RegEx, Range));

Range that splits a string using a regular expression as a separator.

Example:
auto s1 = ", abc, de,  fg, hi, ";
assert(equal(splitter(s1, regex(", *")),
    ["", "abc", "de", "fg", "hi", ""]));

Range  front();
bool  empty();
void  popFront();
@property auto  save();

Forward range primitives.

Splitter!(Range, RegEx)  splitter(Range, RegEx)(Range r, RegEx pat) if (is(BasicElementOf!Range : dchar) && isRegexFor!(RegEx, Range));

A helper function, creates a Splitter on range r separated by regex pat. Captured subexpressions have no effect on the resulting range.

String[]  split(String, RegEx)(String input, RegEx rx) if (isSomeString!String && isRegexFor!(RegEx, String));

An eager version of splitter that creates an array with splitted slices of input.

class  RegexException: object.Exception;

Exception object thrown in case of errors during regex compilation.

this(string msg, string file = __FILE__, size_t line = __LINE__);