Perl 5 version 32.0 documentation

Unicode::UCD

NAME

Unicode::UCD - Unicode character database

SYNOPSIS

  1. use Unicode::UCD 'charinfo';
  2. my $charinfo = charinfo($codepoint);
  3. use Unicode::UCD 'charprop';
  4. my $value = charprop($codepoint, $property);
  5. use Unicode::UCD 'charprops_all';
  6. my $all_values_hash_ref = charprops_all($codepoint);
  7. use Unicode::UCD 'casefold';
  8. my $casefold = casefold($codepoint);
  9. use Unicode::UCD 'all_casefolds';
  10. my $all_casefolds_ref = all_casefolds();
  11. use Unicode::UCD 'casespec';
  12. my $casespec = casespec($codepoint);
  13. use Unicode::UCD 'charblock';
  14. my $charblock = charblock($codepoint);
  15. use Unicode::UCD 'charscript';
  16. my $charscript = charscript($codepoint);
  17. use Unicode::UCD 'charblocks';
  18. my $charblocks = charblocks();
  19. use Unicode::UCD 'charscripts';
  20. my $charscripts = charscripts();
  21. use Unicode::UCD qw(charscript charinrange);
  22. my $range = charscript($script);
  23. print "looks like $script\n" if charinrange($range, $codepoint);
  24. use Unicode::UCD qw(general_categories bidi_types);
  25. my $categories = general_categories();
  26. my $types = bidi_types();
  27. use Unicode::UCD 'prop_aliases';
  28. my @space_names = prop_aliases("space");
  29. use Unicode::UCD 'prop_value_aliases';
  30. my @gc_punct_names = prop_value_aliases("Gc", "Punct");
  31. use Unicode::UCD 'prop_values';
  32. my @all_EA_short_names = prop_values("East_Asian_Width");
  33. use Unicode::UCD 'prop_invlist';
  34. my @puncts = prop_invlist("gc=punctuation");
  35. use Unicode::UCD 'prop_invmap';
  36. my ($list_ref, $map_ref, $format, $missing)
  37. = prop_invmap("General Category");
  38. use Unicode::UCD 'search_invlist';
  39. my $index = search_invlist(\@invlist, $code_point);
  40. # The following function should be used only internally in
  41. # implementations of the Unicode Normalization Algorithm, and there
  42. # are better choices than it.
  43. use Unicode::UCD 'compexcl';
  44. my $compexcl = compexcl($codepoint);
  45. use Unicode::UCD 'namedseq';
  46. my $namedseq = namedseq($named_sequence_name);
  47. my $unicode_version = Unicode::UCD::UnicodeVersion();
  48. my $convert_to_numeric =
  49. Unicode::UCD::num("\N{RUMI DIGIT ONE}\N{RUMI DIGIT TWO}");

DESCRIPTION

The Unicode::UCD module offers a series of functions that provide a simple interface to the Unicode Character Database.

code point argument

Some of the functions are called with a code point argument, which is either a decimal or a hexadecimal scalar designating a code point in the platform's native character set (extended to Unicode), or a string containing U+ followed by hexadecimals designating a Unicode code point. A leading 0 will force a hexadecimal interpretation, as will a hexadecimal digit that isn't a decimal digit.

Examples:

  1. 223 # Decimal 223 in native character set
  2. 0223 # Hexadecimal 223, native (= 547 decimal)
  3. 0xDF # Hexadecimal DF, native (= 223 decimal)
  4. '0xDF' # String form of hexadecimal (= 223 decimal)
  5. 'U+DF' # Hexadecimal DF, in Unicode's character set
  6. (= LATIN SMALL LETTER SHARP S)

Note that the largest code point in Unicode is U+10FFFF.

charinfo()

  1. use Unicode::UCD 'charinfo';
  2. my $charinfo = charinfo(0x41);

This returns information about the input code point argument as a reference to a hash of fields as defined by the Unicode standard. If the code point argument is not assigned in the standard (i.e., has the general category Cn meaning Unassigned ) or is a non-character (meaning it is guaranteed to never be assigned in the standard), undef is returned.

Fields that aren't applicable to the particular code point argument exist in the returned hash, and are empty.

For results that are less "raw" than this function returns, or to get the values for any property, not just the few covered by this function, use the charprop() function.

The keys in the hash with the meanings of their values are:

  • code

    the input native code point argument expressed in hexadecimal, with leading zeros added if necessary to make it contain at least four hexdigits

  • name

    name of code, all IN UPPER CASE. Some control-type code points do not have names. This field will be empty for Surrogate and Private Use code points, and for the others without a name, it will contain a description enclosed in angle brackets, like <control> .

  • category

    The short name of the general category of code. This will match one of the keys in the hash returned by general_categories().

    The prop_value_aliases() function can be used to get all the synonyms of the category name.

  • combining

    the combining class number for code used in the Canonical Ordering Algorithm. For Unicode 5.1, this is described in Section 3.11 Canonical Ordering Behavior available at http://www.unicode.org/versions/Unicode5.1.0/

    The prop_value_aliases() function can be used to get all the synonyms of the combining class number.

  • bidi

    bidirectional type of code. This will match one of the keys in the hash returned by bidi_types().

    The prop_value_aliases() function can be used to get all the synonyms of the bidi type name.

  • decomposition

    is empty if code has no decomposition; or is one or more codes (separated by spaces) that, taken in order, represent a decomposition for code. Each has at least four hexdigits. The codes may be preceded by a word enclosed in angle brackets, then a space, like <compat> , giving the type of decomposition

    This decomposition may be an intermediate one whose components are also decomposable. Use Unicode::Normalize to get the final decomposition in one step.

  • decimal

    if code represents a decimal digit this is its integer numeric value

  • digit

    if code represents some other digit-like number, this is its integer numeric value

  • numeric

    if code represents a whole or rational number, this is its numeric value. Rational values are expressed as a string like 1/4 .

  • mirrored

    Y or N designating if code is mirrored in bidirectional text

  • unicode10

    name of code in the Unicode 1.0 standard if one existed for this code point and is different from the current name

  • comment

    As of Unicode 6.0, this is always empty.

  • upper

    is, if non-empty, the uppercase mapping for code expressed as at least four hexdigits. This indicates that the full uppercase mapping is a single character, and is identical to the simple (single-character only) mapping. When this field is empty, it means that the simple uppercase mapping is code itself; you'll need some other means, (like charprop() or casespec() to get the full mapping.

  • lower

    is, if non-empty, the lowercase mapping for code expressed as at least four hexdigits. This indicates that the full lowercase mapping is a single character, and is identical to the simple (single-character only) mapping. When this field is empty, it means that the simple lowercase mapping is code itself; you'll need some other means, (like charprop() or casespec() to get the full mapping.

  • title

    is, if non-empty, the titlecase mapping for code expressed as at least four hexdigits. This indicates that the full titlecase mapping is a single character, and is identical to the simple (single-character only) mapping. When this field is empty, it means that the simple titlecase mapping is code itself; you'll need some other means, (like charprop() or casespec() to get the full mapping.

  • block

    the block code belongs to (used in \p{Blk=...} ). The prop_value_aliases() function can be used to get all the synonyms of the block name.

    See Blocks versus Scripts.

  • script

    the script code belongs to. The prop_value_aliases() function can be used to get all the synonyms of the script name. Note that this is the older "Script" property value, and not the improved "Script_Extensions" value.

    See Blocks versus Scripts.

Note that you cannot do (de)composition and casing based solely on the decomposition, combining, lower, upper, and title fields; you will need also the casespec() function and the Composition_Exclusion property. (Or you could just use the lc, uc, and ucfirst functions, and the Unicode::Normalize module.)

charprop()

  1. use Unicode::UCD 'charprop';
  2. print charprop(0x41, "Gc"), "\n";
  3. print charprop(0x61, "General_Category"), "\n";
  4. prints
  5. Lu
  6. Ll

This returns the value of the Unicode property given by the second parameter for the code point argument given by the first.

The passed-in property may be specified as any of the synonyms returned by prop_aliases().

The return value is always a scalar, either a string or a number. For properties where there are synonyms for the values, the synonym returned by this function is the longest, most descriptive form, the one returned by prop_value_aliases() when called in a scalar context. Of course, you can call prop_value_aliases() on the result to get other synonyms.

The return values are more "cooked" than the charinfo() ones. For example, the "uc" property value is the actual string containing the full uppercase mapping of the input code point. You have to go to extra trouble with charinfo to get this value from its upper hash element when the full mapping differs from the simple one.

Special note should be made of the return values for a few properties:

  • Block

    The value returned is the new-style (see Old-style versus new-style block names).

  • Decomposition_Mapping

    Like charinfo(), the result may be an intermediate decomposition whose components are also decomposable. Use Unicode::Normalize to get the final decomposition in one step.

    Unlike charinfo(), this does not include the decomposition type. Use the Decomposition_Type property to get that.

  • Name_Alias

    If the input code point's name has more than one synonym, they are returned joined into a single comma-separated string.

  • Numeric_Value

    If the result is a fraction, it is converted into a floating point number to the accuracy of your platform.

  • Script_Extensions

    If the result is multiple script names, they are returned joined into a single comma-separated string.

When called with a property that is a Perl extension that isn't expressible in a compound form, this function currently returns undef, as the only two possible values are true or false (1 or 0 I suppose). This behavior may change in the future, so don't write code that relies on it. Present_In is a Perl extension that is expressible in a bipartite or compound form (for example, \p{Present_In=4.0} ), so charprop accepts it. But Any is a Perl extension that isn't expressible that way, so charprop returns undef for it. Also charprop returns undef for all Perl extensions that are internal-only.

charprops_all()

  1. use Unicode::UCD 'charprops_all';
  2. my $%properties_of_A_hash_ref = charprops_all("U+41");

This returns a reference to a hash whose keys are all the distinct Unicode (no Perl extension) properties, and whose values are the respective values for those properties for the input code point argument.

Each key is the property name in its longest, most descriptive form. The values are what charprop() would return.

This function is expensive in time and memory.

charblock()

  1. use Unicode::UCD 'charblock';
  2. my $charblock = charblock(0x41);
  3. my $charblock = charblock(1234);
  4. my $charblock = charblock(0x263a);
  5. my $charblock = charblock("U+263a");
  6. my $range = charblock('Armenian');

With a code point argument charblock() returns the block the code point belongs to, e.g. Basic Latin . The old-style block name is returned (see Old-style versus new-style block names). The prop_value_aliases() function can be used to get all the synonyms of the block name.

If the code point is unassigned, this returns the block it would belong to if it were assigned. (If the Unicode version being used is so early as to not have blocks, all code points are considered to be in No_Block .)

See also Blocks versus Scripts.

If supplied with an argument that can't be a code point, charblock() tries to do the opposite and interpret the argument as an old-style block name. On an ASCII platform, the return value is a range set with one range: an anonymous array with a single element that consists of another anonymous array whose first element is the first code point in the block, and whose second element is the final code point in the block. On an EBCDIC platform, the first two Unicode blocks are not contiguous. Their range sets are lists containing start-of-range, end-of-range code point pairs. You can test whether a code point is in a range set using the charinrange() function. (To be precise, each range set contains a third array element, after the range boundary ones: the old_style block name.)

If the argument to charblock() is not a known block, undef is returned.

charscript()

  1. use Unicode::UCD 'charscript';
  2. my $charscript = charscript(0x41);
  3. my $charscript = charscript(1234);
  4. my $charscript = charscript("U+263a");
  5. my $range = charscript('Thai');

With a code point argument, charscript() returns the script the code point belongs to, e.g., Latin , Greek , Han . If the code point is unassigned or the Unicode version being used is so early that it doesn't have scripts, this function returns "Unknown" . The prop_value_aliases() function can be used to get all the synonyms of the script name.

Note that the Script_Extensions property is an improved version of the Script property, and you should probably be using that instead, with the charprop() function.

If supplied with an argument that can't be a code point, charscript() tries to do the opposite and interpret the argument as a script name. The return value is a range set: an anonymous array of arrays that contain start-of-range, end-of-range code point pairs. You can test whether a code point is in a range set using the charinrange() function. (To be precise, each range set contains a third array element, after the range boundary ones: the script name.)

If the charscript() argument is not a known script, undef is returned.

See also Blocks versus Scripts.

charblocks()

  1. use Unicode::UCD 'charblocks';
  2. my $charblocks = charblocks();

charblocks() returns a reference to a hash with the known block names as the keys, and the code point ranges (see charblock()) as the values.

The names are in the old-style (see Old-style versus new-style block names).

prop_invmap(block) can be used to get this same data in a different type of data structure.

prop_values(Block) can be used to get all the known new-style block names as a list, without the code point ranges.

See also Blocks versus Scripts.

charscripts()

  1. use Unicode::UCD 'charscripts';
  2. my $charscripts = charscripts();

charscripts() returns a reference to a hash with the known script names as the keys, and the code point ranges (see charscript()) as the values.

prop_invmap(script) can be used to get this same data in a different type of data structure. Since the Script_Extensions property is an improved version of the Script property, you should instead use prop_invmap(scx).

prop_values(Script) can be used to get all the known script names as a list, without the code point ranges.

See also Blocks versus Scripts.

charinrange()

In addition to using the \p{Blk=...} and \P{Blk=...} constructs, you can also test whether a code point is in the range as returned by charblock() and charscript() or as the values of the hash returned by charblocks() and charscripts() by using charinrange() :

  1. use Unicode::UCD qw(charscript charinrange);
  2. $range = charscript('Hiragana');
  3. print "looks like hiragana\n" if charinrange($range, $codepoint);

general_categories()

  1. use Unicode::UCD 'general_categories';
  2. my $categories = general_categories();

This returns a reference to a hash which has short general category names (such as Lu , Nd , Zs , S ) as keys and long names (such as UppercaseLetter , DecimalNumber , SpaceSeparator , Symbol ) as values. The hash is reversible in case you need to go from the long names to the short names. The general category is the one returned from charinfo() under the category key.

The prop_values() and prop_value_aliases() functions can be used as an alternative to this function; the first returning a simple list of the short category names; and the second gets all the synonyms of a given category name.

bidi_types()

  1. use Unicode::UCD 'bidi_types';
  2. my $categories = bidi_types();

This returns a reference to a hash which has the short bidi (bidirectional) type names (such as L , R ) as keys and long names (such as Left-to-Right , Right-to-Left ) as values. The hash is reversible in case you need to go from the long names to the short names. The bidi type is the one returned from charinfo() under the bidi key. For the exact meaning of the various bidi classes the Unicode TR9 is recommended reading: http://www.unicode.org/reports/tr9/ (as of Unicode 5.0.0)

The prop_values() and prop_value_aliases() functions can be used as an alternative to this function; the first returning a simple list of the short bidi type names; and the second gets all the synonyms of a given bidi type name.

compexcl()

WARNING: Unicode discourages the use of this function or any of the alternative mechanisms listed in this section (the documentation of compexcl() ), except internally in implementations of the Unicode Normalization Algorithm. You should be using Unicode::Normalize directly instead of these. Using these will likely lead to half-baked results.

  1. use Unicode::UCD 'compexcl';
  2. my $compexcl = compexcl(0x09dc);

This routine returns undef if the Unicode version being used is so early that it doesn't have this property.

compexcl() is included for backwards compatibility, but as of Perl 5.12 and more modern Unicode versions, for most purposes it is probably more convenient to use one of the following instead:

  1. my $compexcl = chr(0x09dc) =~ /\p{Comp_Ex};
  2. my $compexcl = chr(0x09dc) =~ /\p{Full_Composition_Exclusion};

or even

  1. my $compexcl = chr(0x09dc) =~ /\p{CE};
  2. my $compexcl = chr(0x09dc) =~ /\p{Composition_Exclusion};

The first two forms return true if the code point argument should not be produced by composition normalization. For the final two forms to return true, it is additionally required that this fact not otherwise be determinable from the Unicode data base.

This routine behaves identically to the final two forms. That is, it does not return true if the code point has a decomposition consisting of another single code point, nor if its decomposition starts with a code point whose combining class is non-zero. Code points that meet either of these conditions should also not be produced by composition normalization, which is probably why you should use the Full_Composition_Exclusion property instead, as shown above.

The routine returns false otherwise.

casefold()

  1. use Unicode::UCD 'casefold';
  2. my $casefold = casefold(0xDF);
  3. if (defined $casefold) {
  4. my @full_fold_hex = split / /, $casefold->{'full'};
  5. my $full_fold_string =
  6. join "", map {chr(hex($_))} @full_fold_hex;
  7. my @turkic_fold_hex =
  8. split / /, ($casefold->{'turkic'} ne "")
  9. ? $casefold->{'turkic'}
  10. : $casefold->{'full'};
  11. my $turkic_fold_string =
  12. join "", map {chr(hex($_))} @turkic_fold_hex;
  13. }
  14. if (defined $casefold && $casefold->{'simple'} ne "") {
  15. my $simple_fold_hex = $casefold->{'simple'};
  16. my $simple_fold_string = chr(hex($simple_fold_hex));
  17. }

This returns the (almost) locale-independent case folding of the character specified by the code point argument. (Starting in Perl v5.16, the core function fc() returns the full mapping (described below) faster than this does, and for entire strings.)

If there is no case folding for the input code point, undef is returned.

If there is a case folding for that code point, a reference to a hash with the following fields is returned:

  • code

    the input native code point argument expressed in hexadecimal, with leading zeros added if necessary to make it contain at least four hexdigits

  • full

    one or more codes (separated by spaces) that, taken in order, give the code points for the case folding for code. Each has at least four hexdigits.

  • simple

    is empty, or is exactly one code with at least four hexdigits which can be used as an alternative case folding when the calling program cannot cope with the fold being a sequence of multiple code points. If full is just one code point, then simple equals full. If there is no single code point folding defined for code, then simple is the empty string. Otherwise, it is an inferior, but still better-than-nothing alternative folding to full.

  • mapping

    is the same as simple if simple is not empty, and it is the same as full otherwise. It can be considered to be the simplest possible folding for code. It is defined primarily for backwards compatibility.

  • status

    is C (for common ) if the best possible fold is a single code point (simple equals full equals mapping). It is S if there are distinct folds, simple and full (mapping equals simple). And it is F if there is only a full fold (mapping equals full; simple is empty). Note that this describes the contents of mapping. It is defined primarily for backwards compatibility.

    For Unicode versions between 3.1 and 3.1.1 inclusive, status can also be I which is the same as C but is a special case for dotted uppercase I and dotless lowercase i:

    • * If you use this I mapping

      the result is case-insensitive, but dotless and dotted I's are not distinguished

    • * If you exclude this I mapping

      the result is not fully case-insensitive, but dotless and dotted I's are distinguished

  • turkic

    contains any special folding for Turkic languages. For versions of Unicode starting with 3.2, this field is empty unless code has a different folding in Turkic languages, in which case it is one or more codes (separated by spaces) that, taken in order, give the code points for the case folding for code in those languages. Each code has at least four hexdigits. Note that this folding does not maintain canonical equivalence without additional processing.

    For Unicode versions between 3.1 and 3.1.1 inclusive, this field is empty unless there is a special folding for Turkic languages, in which case status is I , and mapping, full, simple, and turkic are all equal.

Programs that want complete generality and the best folding results should use the folding contained in the full field. But note that the fold for some code points will be a sequence of multiple code points.

Programs that can't cope with the fold mapping being multiple code points can use the folding contained in the simple field, with the loss of some generality. In Unicode 5.1, about 7% of the defined foldings have no single code point folding.

The mapping and status fields are provided for backwards compatibility for existing programs. They contain the same values as in previous versions of this function.

Locale is not completely independent. The turkic field contains results to use when the locale is a Turkic language.

For more information about case mappings see http://www.unicode.org/unicode/reports/tr21

all_casefolds()

  1. use Unicode::UCD 'all_casefolds';
  2. my $all_folds_ref = all_casefolds();
  3. foreach my $char_with_casefold (sort { $a <=> $b }
  4. keys %$all_folds_ref)
  5. {
  6. printf "%04X:", $char_with_casefold;
  7. my $casefold = $all_folds_ref->{$char_with_casefold};
  8. # Get folds for $char_with_casefold
  9. my @full_fold_hex = split / /, $casefold->{'full'};
  10. my $full_fold_string =
  11. join "", map {chr(hex($_))} @full_fold_hex;
  12. print " full=", join " ", @full_fold_hex;
  13. my @turkic_fold_hex =
  14. split / /, ($casefold->{'turkic'} ne "")
  15. ? $casefold->{'turkic'}
  16. : $casefold->{'full'};
  17. my $turkic_fold_string =
  18. join "", map {chr(hex($_))} @turkic_fold_hex;
  19. print "; turkic=", join " ", @turkic_fold_hex;
  20. if (defined $casefold && $casefold->{'simple'} ne "") {
  21. my $simple_fold_hex = $casefold->{'simple'};
  22. my $simple_fold_string = chr(hex($simple_fold_hex));
  23. print "; simple=$simple_fold_hex";
  24. }
  25. print "\n";
  26. }

This returns all the case foldings in the current version of Unicode in the form of a reference to a hash. Each key to the hash is the decimal representation of a Unicode character that has a casefold to other than itself. The casefold of a semi-colon is itself, so it isn't in the hash; likewise for a lowercase "a", but there is an entry for a capital "A". The hash value for each key is another hash, identical to what is returned by casefold() if called with that code point as its argument. So the value all_casefolds()->{ord("A")}' is equivalent to casefold(ord("A")) ;

casespec()

  1. use Unicode::UCD 'casespec';
  2. my $casespec = casespec(0xFB00);

This returns the potentially locale-dependent case mappings of the code point argument. The mappings may be longer than a single code point (which the basic Unicode case mappings as returned by charinfo() never are).

If there are no case mappings for the code point argument, or if all three possible mappings (lower, title and upper) result in single code points and are locale independent and unconditional, undef is returned (which means that the case mappings, if any, for the code point are those returned by charinfo()).

Otherwise, a reference to a hash giving the mappings (or a reference to a hash of such hashes, explained below) is returned with the following keys and their meanings:

The keys in the bottom layer hash with the meanings of their values are:

  • code

    the input native code point argument expressed in hexadecimal, with leading zeros added if necessary to make it contain at least four hexdigits

  • lower

    one or more codes (separated by spaces) that, taken in order, give the code points for the lower case of code. Each has at least four hexdigits.

  • title

    one or more codes (separated by spaces) that, taken in order, give the code points for the title case of code. Each has at least four hexdigits.

  • upper

    one or more codes (separated by spaces) that, taken in order, give the code points for the upper case of code. Each has at least four hexdigits.

  • condition

    the conditions for the mappings to be valid. If undef, the mappings are always valid. When defined, this field is a list of conditions, all of which must be true for the mappings to be valid. The list consists of one or more locales (see below) and/or contexts (explained in the next paragraph), separated by spaces. (Other than as used to separate elements, spaces are to be ignored.) Case distinctions in the condition list are not significant. Conditions preceded by "NON_" represent the negation of the condition.

    A context is one of those defined in the Unicode standard. For Unicode 5.1, they are defined in Section 3.13 Default Case Operations available at http://www.unicode.org/versions/Unicode5.1.0/. These are for context-sensitive casing.

The hash described above is returned for locale-independent casing, where at least one of the mappings has length longer than one. If undef is returned, the code point may have mappings, but if so, all are length one, and are returned by charinfo(). Note that when this function does return a value, it will be for the complete set of mappings for a code point, even those whose length is one.

If there are additional casing rules that apply only in certain locales, an additional key for each will be defined in the returned hash. Each such key will be its locale name, defined as a 2-letter ISO 3166 country code, possibly followed by a "_" and a 2-letter ISO language code (possibly followed by a "_" and a variant code). You can find the lists of all possible locales, see Locale::Country and Locale::Language. (In Unicode 6.0, the only locales returned by this function are lt, tr, and az .)

Each locale key is a reference to a hash that has the form above, and gives the casing rules for that particular locale, which take precedence over the locale-independent ones when in that locale.

If the only casing for a code point is locale-dependent, then the returned hash will not have any of the base keys, like code , upper , etc., but will contain only locale keys.

For more information about case mappings see http://www.unicode.org/unicode/reports/tr21/

namedseq()

  1. use Unicode::UCD 'namedseq';
  2. my $namedseq = namedseq("KATAKANA LETTER AINU P");
  3. my @namedseq = namedseq("KATAKANA LETTER AINU P");
  4. my %namedseq = namedseq();

If used with a single argument in a scalar context, returns the string consisting of the code points of the named sequence, or undef if no named sequence by that name exists. If used with a single argument in a list context, it returns the list of the ordinals of the code points.

If used with no arguments in a list context, it returns a hash with the names of all the named sequences as the keys and their sequences as strings as the values. Otherwise, it returns undef or an empty list depending on the context.

This function only operates on officially approved (not provisional) named sequences.

Note that as of Perl 5.14, \N{KATAKANA LETTER AINU P} will insert the named sequence into double-quoted strings, and charnames::string_vianame("KATAKANA LETTER AINU P") will return the same string this function does, but will also operate on character names that aren't named sequences, without you having to know which are which. See charnames.

num()

  1. use Unicode::UCD 'num';
  2. my $val = num("123");
  3. my $one_quarter = num("\N{VULGAR FRACTION 1/4}");
  4. my $val = num("12a", \$valid_length); # $valid_length contains 2

num() returns the numeric value of the input Unicode string; or undef if it doesn't think the entire string has a completely valid, safe numeric value. If called with an optional second parameter, a reference to a scalar, num() will set the scalar to the length of any valid initial substring; or to 0 if none.

If the string is just one character in length, the Unicode numeric value is returned if it has one, or undef otherwise. If the optional scalar ref is passed, it would be set to 1 if the return is valid; or 0 if the return is undef. Note that the numeric value returned need not be a whole number. num("\N{TIBETAN DIGIT HALF ZERO}") , for example returns -0.5.

If the string is more than one character, undef is returned unless all its characters are decimal digits (that is, they would match \d+ ), from the same script. For example if you have an ASCII '0' and a Bengali '3', mixed together, they aren't considered a valid number, and undef is returned. A further restriction is that the digits all have to be of the same form. A half-width digit mixed with a full-width one will return undef. The Arabic script has two sets of digits; num will return undef unless all the digits in the string come from the same set. In all cases, the optional scalar ref parameter is set to how long any valid initial substring of digits is; hence it will be set to the entire string length if the main return value is not undef.

num errs on the side of safety, and there may be valid strings of decimal digits that it doesn't recognize. Note that Unicode defines a number of "digit" characters that aren't "decimal digit" characters. "Decimal digits" have the property that they have a positional value, i.e., there is a units position, a 10's position, a 100's, etc, AND they are arranged in Unicode in blocks of 10 contiguous code points. The Chinese digits, for example, are not in such a contiguous block, and so Unicode doesn't view them as decimal digits, but merely digits, and so \d will not match them. A single-character string containing one of these digits will have its decimal value returned by num , but any longer string containing only these digits will return undef.

Strings of multiple sub- and superscripts are not recognized as numbers. You can use either of the compatibility decompositions in Unicode::Normalize to change these into digits, and then call num on the result.

prop_aliases()

  1. use Unicode::UCD 'prop_aliases';
  2. my ($short_name, $full_name, @other_names) = prop_aliases("space");
  3. my $same_full_name = prop_aliases("Space"); # Scalar context
  4. my ($same_short_name) = prop_aliases("Space"); # gets 0th element
  5. print "The full name is $full_name\n";
  6. print "The short name is $short_name\n";
  7. print "The other aliases are: ", join(", ", @other_names), "\n";
  8. prints:
  9. The full name is White_Space
  10. The short name is WSpace
  11. The other aliases are: Space

Most Unicode properties have several synonymous names. Typically, there is at least a short name, convenient to type, and a long name that more fully describes the property, and hence is more easily understood.

If you know one name for a Unicode property, you can use prop_aliases to find either the long name (when called in scalar context), or a list of all of the names, somewhat ordered so that the short name is in the 0th element, the long name in the next element, and any other synonyms are in the remaining elements, in no particular order.

The long name is returned in a form nicely capitalized, suitable for printing.

The input parameter name is loosely matched, which means that white space, hyphens, and underscores are ignored (except for the trailing underscore in the old_form grandfathered-in "L_" , which is better written as "LC" , and both of which mean General_Category=Cased Letter ).

If the name is unknown, undef is returned (or an empty list in list context). Note that Perl typically recognizes property names in regular expressions with an optional "Is_" (with or without the underscore) prefixed to them, such as \p{isgc=punct} . This function does not recognize those in the input, returning undef. Nor are they included in the output as possible synonyms.

prop_aliases does know about the Perl extensions to Unicode properties, such as Any and XPosixAlpha , and the single form equivalents to Unicode properties such as XDigit , Greek , In_Greek , and Is_Greek . The final example demonstrates that the "Is_" prefix is recognized for these extensions; it is needed to resolve ambiguities. For example, prop_aliases('lc') returns the list (lc, Lowercase_Mapping) , but prop_aliases('islc') returns (Is_LC, Cased_Letter) . This is because islc is a Perl extension which is short for General_Category=Cased Letter . The lists returned for the Perl extensions will not include the "Is_" prefix (whether or not the input had it) unless needed to resolve ambiguities, as shown in the "islc" example, where the returned list had one element containing "Is_" , and the other without.

It is also possible for the reverse to happen: prop_aliases('isc') returns the list (isc, ISO_Comment) ; whereas prop_aliases('c') returns (C, Other) (the latter being a Perl extension meaning General_Category=Other . Properties accessible through Unicode::UCD in perluniprops lists the available forms, including which ones are discouraged from use.

Those discouraged forms are accepted as input to prop_aliases , but are not returned in the lists. prop_aliases('isL&') and prop_aliases('isL_') , which are old synonyms for "Is_LC" and should not be used in new code, are examples of this. These both return (Is_LC, Cased_Letter) . Thus this function allows you to take a discouraged form, and find its acceptable alternatives. The same goes with single-form Block property equivalences. Only the forms that begin with "In_" are not discouraged; if you pass prop_aliases a discouraged form, you will get back the equivalent ones that begin with "In_" . It will otherwise look like a new-style block name (see. Old-style versus new-style block names).

prop_aliases does not know about any user-defined properties, and will return undef if called with one of those. Likewise for Perl internal properties, with the exception of "Perl_Decimal_Digit" which it does know about (and which is documented below in prop_invmap()).

prop_values()

  1. use Unicode::UCD 'prop_values';
  2. print "AHex values are: ", join(", ", prop_values("AHex")),
  3. "\n";
  4. prints:
  5. AHex values are: N, Y

Some Unicode properties have a restricted set of legal values. For example, all binary properties are restricted to just true or false ; and there are only a few dozen possible General Categories. Use prop_values to find out if a given property is one such, and if so, to get a list of the values:

  1. print join ", ", prop_values("NFC_Quick_Check");
  2. prints:
  3. M, N, Y

If the property doesn't have such a restricted set, undef is returned.

There are usually several synonyms for each possible value. Use prop_value_aliases() to access those.

Case, white space, hyphens, and underscores are ignored in the input property name (except for the trailing underscore in the old-form grandfathered-in general category property value "L_" , which is better written as "LC" ).

If the property name is unknown, undef is returned. Note that Perl typically recognizes property names in regular expressions with an optional "Is_" (with or without the underscore) prefixed to them, such as \p{isgc=punct} . This function does not recognize those in the property parameter, returning undef.

For the block property, new-style block names are returned (see Old-style versus new-style block names).

prop_values does not know about any user-defined properties, and will return undef if called with one of those.

prop_value_aliases()

  1. use Unicode::UCD 'prop_value_aliases';
  2. my ($short_name, $full_name, @other_names)
  3. = prop_value_aliases("Gc", "Punct");
  4. my $same_full_name = prop_value_aliases("Gc", "P"); # Scalar cntxt
  5. my ($same_short_name) = prop_value_aliases("Gc", "P"); # gets 0th
  6. # element
  7. print "The full name is $full_name\n";
  8. print "The short name is $short_name\n";
  9. print "The other aliases are: ", join(", ", @other_names), "\n";
  10. prints:
  11. The full name is Punctuation
  12. The short name is P
  13. The other aliases are: Punct

Some Unicode properties have a restricted set of legal values. For example, all binary properties are restricted to just true or false ; and there are only a few dozen possible General Categories.

You can use prop_values() to find out if a given property is one which has a restricted set of values, and if so, what those values are. But usually each value actually has several synonyms. For example, in Unicode binary properties, truth can be represented by any of the strings "Y", "Yes", "T", or "True"; and the General Category "Punctuation" by that string, or "Punct", or simply "P".

Like property names, there is typically at least a short name for each such property-value, and a long name. If you know any name of the property-value (which you can get by prop_values(), you can use prop_value_aliases () to get the long name (when called in scalar context), or a list of all the names, with the short name in the 0th element, the long name in the next element, and any other synonyms in the remaining elements, in no particular order, except that any all-numeric synonyms will be last.

The long name is returned in a form nicely capitalized, suitable for printing.

Case, white space, hyphens, and underscores are ignored in the input parameters (except for the trailing underscore in the old-form grandfathered-in general category property value "L_" , which is better written as "LC" ).

If either name is unknown, undef is returned. Note that Perl typically recognizes property names in regular expressions with an optional "Is_" (with or without the underscore) prefixed to them, such as \p{isgc=punct} . This function does not recognize those in the property parameter, returning undef.

If called with a property that doesn't have synonyms for its values, it returns the input value, possibly normalized with capitalization and underscores, but not necessarily checking that the input value is valid.

For the block property, new-style block names are returned (see Old-style versus new-style block names).

To find the synonyms for single-forms, such as \p{Any} , use prop_aliases() instead.

prop_value_aliases does not know about any user-defined properties, and will return undef if called with one of those.

prop_invlist()

prop_invlist returns an inversion list (described below) that defines all the code points for the binary Unicode property (or "property=value" pair) given by the input parameter string:

  1. use feature 'say';
  2. use Unicode::UCD 'prop_invlist';
  3. say join ", ", prop_invlist("Any");
  4. prints:
  5. 0, 1114112

If the input is unknown undef is returned in scalar context; an empty-list in list context. If the input is known, the number of elements in the list is returned if called in scalar context.

perluniprops gives the list of properties that this function accepts, as well as all the possible forms for them (including with the optional "Is_" prefixes). (Except this function doesn't accept any Perl-internal properties, some of which are listed there.) This function uses the same loose or tighter matching rules for resolving the input property's name as is done for regular expressions. These are also specified in perluniprops. Examples of using the "property=value" form are:

  1. say join ", ", prop_invlist("Script_Extensions=Shavian");
  2. prints:
  3. 66640, 66688
  4. say join ", ", prop_invlist("ASCII_Hex_Digit=No");
  5. prints:
  6. 0, 48, 58, 65, 71, 97, 103
  7. say join ", ", prop_invlist("ASCII_Hex_Digit=Yes");
  8. prints:
  9. 48, 58, 65, 71, 97, 103

Inversion lists are a compact way of specifying Unicode property-value definitions. The 0th item in the list is the lowest code point that has the property-value. The next item (item [1]) is the lowest code point beyond that one that does NOT have the property-value. And the next item beyond that ([2]) is the lowest code point beyond that one that does have the property-value, and so on. Put another way, each element in the list gives the beginning of a range that has the property-value (for even numbered elements), or doesn't have the property-value (for odd numbered elements). The name for this data structure stems from the fact that each element in the list toggles (or inverts) whether the corresponding range is or isn't on the list.

In the final example above, the first ASCII Hex digit is code point 48, the character "0", and all code points from it through 57 (a "9") are ASCII hex digits. Code points 58 through 64 aren't, but 65 (an "A") through 70 (an "F") are, as are 97 ("a") through 102 ("f"). 103 starts a range of code points that aren't ASCII hex digits. That range extends to infinity, which on your computer can be found in the variable $Unicode::UCD::MAX_CP . (This variable is as close to infinity as Perl can get on your platform, and may be too high for some operations to work; you may wish to use a smaller number for your purposes.)

Note that the inversion lists returned by this function can possibly include non-Unicode code points, that is anything above 0x10FFFF. Unicode properties are not defined on such code points. You might wish to change the output to not include these. Simply add 0x110000 at the end of the non-empty returned list if it isn't already that value; and pop that value if it is; like:

  1. my @list = prop_invlist("foo");
  2. if (@list) {
  3. if ($list[-1] == 0x110000) {
  4. pop @list; # Defeat the turning on for above Unicode
  5. }
  6. else {
  7. push @list, 0x110000; # Turn off for above Unicode
  8. }
  9. }

It is a simple matter to expand out an inversion list to a full list of all code points that have the property-value:

  1. my @invlist = prop_invlist($property_name);
  2. die "empty" unless @invlist;
  3. my @full_list;
  4. for (my $i = 0; $i < @invlist; $i += 2) {
  5. my $upper = ($i + 1) < @invlist
  6. ? $invlist[$i+1] - 1 # In range
  7. : $Unicode::UCD::MAX_CP; # To infinity.
  8. for my $j ($invlist[$i] .. $upper) {
  9. push @full_list, $j;
  10. }
  11. }

prop_invlist does not know about any user-defined nor Perl internal-only properties, and will return undef if called with one of those.

The search_invlist() function is provided for finding a code point within an inversion list.

prop_invmap()

  1. use Unicode::UCD 'prop_invmap';
  2. my ($list_ref, $map_ref, $format, $default)
  3. = prop_invmap("General Category");

prop_invmap is used to get the complete mapping definition for a property, in the form of an inversion map. An inversion map consists of two parallel arrays. One is an ordered list of code points that mark range beginnings, and the other gives the value (or mapping) that all code points in the corresponding range have.

prop_invmap is called with the name of the desired property. The name is loosely matched, meaning that differences in case, white-space, hyphens, and underscores are not meaningful (except for the trailing underscore in the old-form grandfathered-in property "L_" , which is better written as "LC" , or even better, "Gc=LC" ).

Many Unicode properties have more than one name (or alias). prop_invmap understands all of these, including Perl extensions to them. Ambiguities are resolved as described above for prop_aliases() (except if a property has both a complete mapping, and a binary Y /N mapping, then specifying the property name prefixed by "is" causes the binary one to be returned). The Perl internal property "Perl_Decimal_Digit, described below, is also accepted. An empty list is returned if the property name is unknown. See Properties accessible through Unicode::UCD in perluniprops for the properties acceptable as inputs to this function.

It is a fatal error to call this function except in list context.

In addition to the two arrays that form the inversion map, prop_invmap returns two other values; one is a scalar that gives some details as to the format of the entries of the map array; the other is a default value, useful in maps whose format name begins with the letter "a" , as described below in its subsection; and for specialized purposes, such as converting to another data structure, described at the end of this main section.

This means that prop_invmap returns a 4 element list. For example,

  1. my ($blocks_ranges_ref, $blocks_maps_ref, $format, $default)
  2. = prop_invmap("Block");

In this call, the two arrays will be populated as shown below (for Unicode 6.0):

  1. Index @blocks_ranges @blocks_maps
  2. 0 0x0000 Basic Latin
  3. 1 0x0080 Latin-1 Supplement
  4. 2 0x0100 Latin Extended-A
  5. 3 0x0180 Latin Extended-B
  6. 4 0x0250 IPA Extensions
  7. 5 0x02B0 Spacing Modifier Letters
  8. 6 0x0300 Combining Diacritical Marks
  9. 7 0x0370 Greek and Coptic
  10. 8 0x0400 Cyrillic
  11. ...
  12. 233 0x2B820 No_Block
  13. 234 0x2F800 CJK Compatibility Ideographs Supplement
  14. 235 0x2FA20 No_Block
  15. 236 0xE0000 Tags
  16. 237 0xE0080 No_Block
  17. 238 0xE0100 Variation Selectors Supplement
  18. 239 0xE01F0 No_Block
  19. 240 0xF0000 Supplementary Private Use Area-A
  20. 241 0x100000 Supplementary Private Use Area-B
  21. 242 0x110000 No_Block

The first line (with Index [0]) means that the value for code point 0 is "Basic Latin". The entry "0x0080" in the @blocks_ranges column in the second line means that the value from the first line, "Basic Latin", extends to all code points in the range from 0 up to but not including 0x0080, that is, through 127. In other words, the code points from 0 to 127 are all in the "Basic Latin" block. Similarly, all code points in the range from 0x0080 up to (but not including) 0x0100 are in the block named "Latin-1 Supplement", etc. (Notice that the return is the old-style block names; see Old-style versus new-style block names).

The final line (with Index [242]) means that the value for all code points above the legal Unicode maximum code point have the value "No_Block", which is the term Unicode uses for a non-existing block.

The arrays completely specify the mappings for all possible code points. The final element in an inversion map returned by this function will always be for the range that consists of all the code points that aren't legal Unicode, but that are expressible on the platform. (That is, it starts with code point 0x110000, the first code point above the legal Unicode maximum, and extends to infinity.) The value for that range will be the same that any typical unassigned code point has for the specified property. (Certain unassigned code points are not "typical"; for example the non-character code points, or those in blocks that are to be written right-to-left. The above-Unicode range's value is not based on these atypical code points.) It could be argued that, instead of treating these as unassigned Unicode code points, the value for this range should be undef. If you wish, you can change the returned arrays accordingly.

The maps for almost all properties are simple scalars that should be interpreted as-is. These values are those given in the Unicode-supplied data files, which may be inconsistent as to capitalization and as to which synonym for a property-value is given. The results may be normalized by using the prop_value_aliases() function.

There are exceptions to the simple scalar maps. Some properties have some elements in their map list that are themselves lists of scalars; and some special strings are returned that are not to be interpreted as-is. Element [2] (placed into $format in the example above) of the returned four element list tells you if the map has any of these special elements or not, as follows:

  • s

    means all the elements of the map array are simple scalars, with no special elements. Almost all properties are like this, like the block example above.

  • sl

    means that some of the map array elements have the form given by "s" , and the rest are lists of scalars. For example, here is a portion of the output of calling prop_invmap () with the "Script Extensions" property:

    1. @scripts_ranges @scripts_maps
    2. ...
    3. 0x0953 Devanagari
    4. 0x0964 [ Bengali, Devanagari, Gurumukhi, Oriya ]
    5. 0x0966 Devanagari
    6. 0x0970 Common

    Here, the code points 0x964 and 0x965 are both used in Bengali, Devanagari, Gurmukhi, and Oriya, but no other scripts.

    The Name_Alias property is also of this form. But each scalar consists of two components: 1) the name, and 2) the type of alias this is. They are separated by a colon and a space. In Unicode 6.1, there are several alias types:

    • correction

      indicates that the name is a corrected form for the original name (which remains valid) for the same code point.

    • control

      adds a new name for a control character.

    • alternate

      is an alternate name for a character

    • figment

      is a name for a character that has been documented but was never in any actual standard.

    • abbreviation

      is a common abbreviation for a character

    The lists are ordered (roughly) so the most preferred names come before less preferred ones.

    For example,

    1. @aliases_ranges @alias_maps
    2. ...
    3. 0x009E [ 'PRIVACY MESSAGE: control', 'PM: abbreviation' ]
    4. 0x009F [ 'APPLICATION PROGRAM COMMAND: control',
    5. 'APC: abbreviation'
    6. ]
    7. 0x00A0 'NBSP: abbreviation'
    8. 0x00A1 ""
    9. 0x00AD 'SHY: abbreviation'
    10. 0x00AE ""
    11. 0x01A2 'LATIN CAPITAL LETTER GHA: correction'
    12. 0x01A3 'LATIN SMALL LETTER GHA: correction'
    13. 0x01A4 ""
    14. ...

    A map to the empty string means that there is no alias defined for the code point.

  • a

    is like "s" in that all the map array elements are scalars, but here they are restricted to all being integers, and some have to be adjusted (hence the name "a" ) to get the correct result. For example, in:

    1. my ($uppers_ranges_ref, $uppers_maps_ref, $format, $default)
    2. = prop_invmap("Simple_Uppercase_Mapping");

    the returned arrays look like this:

    1. @$uppers_ranges_ref @$uppers_maps_ref Note
    2. 0 0
    3. 97 65 'a' maps to 'A', b => B ...
    4. 123 0
    5. 181 924 MICRO SIGN => Greek Cap MU
    6. 182 0
    7. ...

    and $default is 0.

    Let's start with the second line. It says that the uppercase of code point 97 is 65; or uc("a") == "A". But the line is for the entire range of code points 97 through 122. To get the mapping for any code point in this range, you take the offset it has from the beginning code point of the range, and add that to the mapping for that first code point. So, the mapping for 122 ("z") is derived by taking the offset of 122 from 97 (=25) and adding that to 65, yielding 90 ("z"). Likewise for everything in between.

    Requiring this simple adjustment allows the returned arrays to be significantly smaller than otherwise, up to a factor of 10, speeding up searching through them.

    Ranges that map to $default , "0" , behave somewhat differently. For these, each code point maps to itself. So, in the first line in the example, ord(uc(chr(0))) is 0, ord(uc(chr(1))) is 1, .. ord(uc(chr(96))) is 96.

  • al

    means that some of the map array elements have the form given by "a" , and the rest are ordered lists of code points. For example, in:

    1. my ($uppers_ranges_ref, $uppers_maps_ref, $format, $default)
    2. = prop_invmap("Uppercase_Mapping");

    the returned arrays look like this:

    1. @$uppers_ranges_ref @$uppers_maps_ref
    2. 0 0
    3. 97 65
    4. 123 0
    5. 181 924
    6. 182 0
    7. ...
    8. 0x0149 [ 0x02BC 0x004E ]
    9. 0x014A 0
    10. 0x014B 330
    11. ...

    This is the full Uppercase_Mapping property (as opposed to the Simple_Uppercase_Mapping given in the example for format "a" ). The only difference between the two in the ranges shown is that the code point at 0x0149 (LATIN SMALL LETTER N PRECEDED BY APOSTROPHE) maps to a string of two characters, 0x02BC (MODIFIER LETTER APOSTROPHE) followed by 0x004E (LATIN CAPITAL LETTER N).

    No adjustments are needed to entries that are references to arrays; each such entry will have exactly one element in its range, so the offset is always 0.

    The fourth (index [3]) element ($default ) in the list returned for this format is 0.

  • ae

    This is like "a" , but some elements are the empty string, and should not be adjusted. The one internal Perl property accessible by prop_invmap is of this type: "Perl_Decimal_Digit" returns an inversion map which gives the numeric values that are represented by the Unicode decimal digit characters. Characters that don't represent decimal digits map to the empty string, like so:

    1. @digits @values
    2. 0x0000 ""
    3. 0x0030 0
    4. 0x003A: ""
    5. 0x0660: 0
    6. 0x066A: ""
    7. 0x06F0: 0
    8. 0x06FA: ""
    9. 0x07C0: 0
    10. 0x07CA: ""
    11. 0x0966: 0
    12. ...

    This means that the code points from 0 to 0x2F do not represent decimal digits; the code point 0x30 (DIGIT ZERO) represents 0; code point 0x31, (DIGIT ONE), represents 0+1-0 = 1; ... code point 0x39, (DIGIT NINE), represents 0+9-0 = 9; ... code points 0x3A through 0x65F do not represent decimal digits; 0x660 (ARABIC-INDIC DIGIT ZERO), represents 0; ... 0x07C1 (NKO DIGIT ONE), represents 0+1-0 = 1 ...

    The fourth (index [3]) element ($default ) in the list returned for this format is the empty string.

  • ale

    is a combination of the "al" type and the "ae" type. Some of the map array elements have the forms given by "al" , and the rest are the empty string. The property NFKC_Casefold has this form. An example slice is:

    1. @$ranges_ref @$maps_ref Note
    2. ...
    3. 0x00AA 97 FEMININE ORDINAL INDICATOR => 'a'
    4. 0x00AB 0
    5. 0x00AD SOFT HYPHEN => ""
    6. 0x00AE 0
    7. 0x00AF [ 0x0020, 0x0304 ] MACRON => SPACE . COMBINING MACRON
    8. 0x00B0 0
    9. ...

    The fourth (index [3]) element ($default ) in the list returned for this format is 0.

  • ar

    means that all the elements of the map array are either rational numbers or the string "NaN" , meaning "Not a Number". A rational number is either an integer, or two integers separated by a solidus ("/" ). The second integer represents the denominator of the division implied by the solidus, and is actually always positive, so it is guaranteed not to be 0 and to not be signed. When the element is a plain integer (without the solidus), it may need to be adjusted to get the correct value by adding the offset, just as other "a" properties. No adjustment is needed for fractions, as the range is guaranteed to have just a single element, and so the offset is always 0.

    If you want to convert the returned map to entirely scalar numbers, you can use something like this:

    1. my ($invlist_ref, $invmap_ref, $format) = prop_invmap($property);
    2. if ($format && $format eq "ar") {
    3. map { $_ = eval $_ if $_ ne 'NaN' } @$map_ref;
    4. }

    Here's some entries from the output of the property "Nv", which has format "ar" .

    1. @numerics_ranges @numerics_maps Note
    2. 0x00 "NaN"
    3. 0x30 0 DIGIT 0 .. DIGIT 9
    4. 0x3A "NaN"
    5. 0xB2 2 SUPERSCRIPTs 2 and 3
    6. 0xB4 "NaN"
    7. 0xB9 1 SUPERSCRIPT 1
    8. 0xBA "NaN"
    9. 0xBC 1/4 VULGAR FRACTION 1/4
    10. 0xBD 1/2 VULGAR FRACTION 1/2
    11. 0xBE 3/4 VULGAR FRACTION 3/4
    12. 0xBF "NaN"
    13. 0x660 0 ARABIC-INDIC DIGIT ZERO .. NINE
    14. 0x66A "NaN"

    The fourth (index [3]) element ($default ) in the list returned for this format is "NaN" .

  • n

    means the Name property. All the elements of the map array are simple scalars, but some of them contain special strings that require more work to get the actual name.

    Entries such as:

    1. CJK UNIFIED IDEOGRAPH-<code point>

    mean that the name for the code point is "CJK UNIFIED IDEOGRAPH-" with the code point (expressed in hexadecimal) appended to it, like "CJK UNIFIED IDEOGRAPH-3403" (similarly for CJK COMPATIBILITY IDEOGRAPH-<code point>).

    Also, entries like

    1. <hangul syllable>

    means that the name is algorithmically calculated. This is easily done by the function charnames::viacode(code) in charnames.

    Note that for control characters (Gc=cc ), Unicode's data files have the string "<control> ", but the real name of each of these characters is the empty string. This function returns that real name, the empty string. (There are names for these characters, but they are considered aliases, not the Name property name, and are contained in the Name_Alias property.)

  • ad

    means the Decomposition_Mapping property. This property is like "al" properties, except that one of the scalar elements is of the form:

    1. <hangul syllable>

    This signifies that this entry should be replaced by the decompositions for all the code points whose decomposition is algorithmically calculated. (All of them are currently in one range and no others outside the range are likely to ever be added to Unicode; the "n" format has this same entry.) These can be generated via the function Unicode::Normalize::NFD().

    Note that the mapping is the one that is specified in the Unicode data files, and to get the final decomposition, it may need to be applied recursively. Unicode in fact discourages use of this property except internally in implementations of the Unicode Normalization Algorithm.

    The fourth (index [3]) element ($default ) in the list returned for this format is 0.

Note that a format begins with the letter "a" if and only the property it is for requires adjustments by adding the offsets in multi-element ranges. For all these properties, an entry should be adjusted only if the map is a scalar which is an integer. That is, it must match the regular expression:

  1. / ^ -? \d+ $ /xa

Further, the first element in a range never needs adjustment, as the adjustment would be just adding 0.

A binary search such as that provided by search_invlist(), can be used to quickly find a code point in the inversion list, and hence its corresponding mapping.

The final, fourth element (index [3], assigned to $default in the "block" example) in the four element list returned by this function is used with the "a" format types; it may also be useful for applications that wish to convert the returned inversion map data structure into some other, such as a hash. It gives the mapping that most code points map to under the property. If you establish the convention that any code point not explicitly listed in your data structure maps to this value, you can potentially make your data structure much smaller. As you construct your data structure from the one returned by this function, simply ignore those ranges that map to this value. For example, to convert to the data structure searchable by charinrange(), you can follow this recipe for properties that don't require adjustments:

  1. my ($list_ref, $map_ref, $format, $default) = prop_invmap($property);
  2. my @range_list;
  3. # Look at each element in the list, but the -2 is needed because we
  4. # look at $i+1 in the loop, and the final element is guaranteed to map
  5. # to $default by prop_invmap(), so we would skip it anyway.
  6. for my $i (0 .. @$list_ref - 2) {
  7. next if $map_ref->[$i] eq $default;
  8. push @range_list, [ $list_ref->[$i],
  9. $list_ref->[$i+1],
  10. $map_ref->[$i]
  11. ];
  12. }
  13. print charinrange(\@range_list, $code_point), "\n";

With this, charinrange() will return undef if its input code point maps to $default . You can avoid this by omitting the next statement, and adding a line after the loop to handle the final element of the inversion map.

Similarly, this recipe can be used for properties that do require adjustments:

  1. for my $i (0 .. @$list_ref - 2) {
  2. next if $map_ref->[$i] eq $default;
  3. # prop_invmap() guarantees that if the mapping is to an array, the
  4. # range has just one element, so no need to worry about adjustments.
  5. if (ref $map_ref->[$i]) {
  6. push @range_list,
  7. [ $list_ref->[$i], $list_ref->[$i], $map_ref->[$i] ];
  8. }
  9. else { # Otherwise each element is actually mapped to a separate
  10. # value, so the range has to be split into single code point
  11. # ranges.
  12. my $adjustment = 0;
  13. # For each code point that gets mapped to something...
  14. for my $j ($list_ref->[$i] .. $list_ref->[$i+1] -1 ) {
  15. # ... add a range consisting of just it mapping to the
  16. # original plus the adjustment, which is incremented for the
  17. # next time through the loop, as the offset increases by 1
  18. # for each element in the range
  19. push @range_list,
  20. [ $j, $j, $map_ref->[$i] + $adjustment++ ];
  21. }
  22. }
  23. }

Note that the inversion maps returned for the Case_Folding and Simple_Case_Folding properties do not include the Turkic-locale mappings. Use casefold() for these.

prop_invmap does not know about any user-defined properties, and will return undef if called with one of those.

The returned values for the Perl extension properties, such as Any and Greek are somewhat misleading. The values are either "Y" or "N". All Unicode properties are bipartite, so you can actually use the "Y" or "N" in a Perl regular expression for these, like qr/\p{ID_Start=Y/} or qr/\p{Upper=N/}. But the Perl extensions aren't specified this way, only like /qr/\p{Any} , etc. You can't actually use the "Y" and "N" in them.

Getting every available name

Instead of reading the Unicode Database directly from files, as you were able to do for a long time, you are encouraged to use the supplied functions. So, instead of reading Name.pl directly, which changed formats in 5.32, and may do so again without notice in the future or even disappear, you ought to use prop_invmap() like this:

  1. my (%name, %cp, %cps, $n);
  2. # All codepoints
  3. foreach my $cat (qw( Name Name_Alias )) {
  4. my ($codepoints, $names, $format, $default) = prop_invmap($cat);
  5. # $format => "n", $default => ""
  6. foreach my $i (0 .. @$codepoints - 2) {
  7. my ($cp, $n) = ($codepoints->[$i], $names->[$i]);
  8. # If $n is a ref, the same codepoint has multiple names
  9. foreach my $name (ref $n ? @$n : $n) {
  10. $name{$cp} //= $name;
  11. $cp{$name} //= $cp;
  12. }
  13. }
  14. }
  15. # Named sequences
  16. { my %ns = namedseq();
  17. foreach my $name (sort { $ns{$a} cmp $ns{$b} } keys %ns) {
  18. $cp{$name} //= [ map { ord } split "" => $ns{$name} ];
  19. }
  20. }

search_invlist()

  1. use Unicode::UCD qw(prop_invmap prop_invlist);
  2. use Unicode::UCD 'search_invlist';
  3. my @invlist = prop_invlist($property_name);
  4. print $code_point, ((search_invlist(\@invlist, $code_point) // -1) % 2)
  5. ? " isn't"
  6. : " is",
  7. " in $property_name\n";
  8. my ($blocks_ranges_ref, $blocks_map_ref) = prop_invmap("Block");
  9. my $index = search_invlist($blocks_ranges_ref, $code_point);
  10. print "$code_point is in block ", $blocks_map_ref->[$index], "\n";

search_invlist is used to search an inversion list returned by prop_invlist or prop_invmap for a particular code point argument. undef is returned if the code point is not found in the inversion list (this happens only when it is not a legal code point argument, or is less than the list's first element). A warning is raised in the first instance.

Otherwise, it returns the index into the list of the range that contains the code point.; that is, find i such that

  1. list[i]<= code_point < list[i+1].

As explained in prop_invlist(), whether a code point is in the list or not depends on if the index is even (in) or odd (not in). And as explained in prop_invmap(), the index is used with the returned parallel array to find the mapping.

Unicode::UCD::UnicodeVersion

This returns the version of the Unicode Character Database, in other words, the version of the Unicode standard the database implements. The version is a string of numbers delimited by dots ('.' ).

Blocks versus Scripts

The difference between a block and a script is that scripts are closer to the linguistic notion of a set of code points required to represent languages, while block is more of an artifact of the Unicode code point numbering and separation into blocks of consecutive code points (so far the size of a block is some multiple of 16, like 128 or 256).

For example the Latin script is spread over several blocks, such as Basic Latin , Latin 1 Supplement, Latin Extended-A , and Latin Extended-B . On the other hand, the Latin script does not contain all the characters of the Basic Latin block (also known as ASCII): it includes only the letters, and not, for example, the digits nor the punctuation.

For blocks see http://www.unicode.org/Public/UNIDATA/Blocks.txt

For scripts see UTR #24: http://www.unicode.org/unicode/reports/tr24/

Matching Scripts and Blocks

Scripts are matched with the regular-expression construct \p{...} (e.g. \p{Tibetan} matches characters of the Tibetan script), while \p{Blk=...} is used for blocks (e.g. \p{Blk=Tibetan} matches any of the 256 code points in the Tibetan block).

Old-style versus new-style block names

Unicode publishes the names of blocks in two different styles, though the two are equivalent under Unicode's loose matching rules.

The original style uses blanks and hyphens in the block names (except for No_Block ), like so:

  1. Miscellaneous Mathematical Symbols-B

The newer style replaces these with underscores, like this:

  1. Miscellaneous_Mathematical_Symbols_B

This newer style is consistent with the values of other Unicode properties. To preserve backward compatibility, all the functions in Unicode::UCD that return block names (except as noted) return the old-style ones. prop_value_aliases() returns the new-style and can be used to convert from old-style to new-style:

  1. my $new_style = prop_values_aliases("block", $old_style);

Perl also has single-form extensions that refer to blocks, In_Cyrillic , meaning Block=Cyrillic . These have always been written in the new style.

To convert from new-style to old-style, follow this recipe:

  1. $old_style = charblock((prop_invlist("block=$new_style"))[0]);

(which finds the range of code points in the block using prop_invlist , gets the lower end of the range (0th element) and then looks up the old name for its block using charblock ).

Note that starting in Unicode 6.1, many of the block names have shorter synonyms. These are always given in the new style.

Use with older Unicode versions

The functions in this module work as well as can be expected when used on earlier Unicode versions. But, obviously, they use the available data from that Unicode version. For example, if the Unicode version predates the definition of the script property (Unicode 3.1), then any function that deals with scripts is going to return undef for the script portion of the return value.

AUTHOR

Jarkko Hietaniemi. Now maintained by perl5 porters.