Syntax::Highlight::Perl::Improved(3) Highlighting of Perl Syntactical Structures


This module provides syntax highlighting for Perl code. The design bias is roughly line-oriented and streamed (ie, processing a file line-by-line in a single pass). Provisions may be made in the future for tasks related to ``back-tracking'' (ie, re-doing a single line in the middle of a stream) such as speeding up state copying.


The only constructor provided is "new()". When called on an existing object, "new()" will create a new copy of that object. Otherwise, "new()" creates a new copy of the (internal) Default Object. Note that the use of the procedural syntax modifies the Default Object and that those changes will be reflected in any subsequent "new()" calls.


Formatting is done using the "format_string()" method. Call "format_string()" with one or more strings to format, or it will default to using $_.

Setting and Getting Formats

You can set the text used for formatting a syntax element using "set_format()" (or set the start and end format individually using "set_start_format()" and "set_end_format()", respectively).

You can also retrieve the text used for formatting for an element via "get_start_format()" or "get_end_format". Bulk retrieval of the names or values of defined formats is possible via "get_format_names_list()" (names), "get_start_format_values_list()" and "get_end_format_values_list()".

See ``FORMAT TYPES'' later in this document for information on what format elements can be used.

Checking and Setting the State

You can check certain aspects of the state of the formatter via the methods: "in_heredoc()", "in_string()", "in_pod()", "was_pod()", "in_data()", and "line_count()".

You can reset all of the above states (and a few other internal ones) using "reset()".

Stable and Unstable Formatting Modes

You can set or check the stability of formatting via "unstable()".

In unstable (TRUE) mode, formatting is not considered to be persistent with nested formats. Or, put another way, when unstable, the formatter can only ``remember'' one format at a time and must reinstate formatting for each token. An example of unstable formatting is using ANSI color escape sequences in a terminal.

In stable (FALSE) mode (the default), formatting is considered persistent within arbitrarily nested formats. Even in stable mode, however, formatting is never allowed to span multiple lines; it is always fully closed at the end of the line and reinstated at the beginning of a new line, if necessary. This is to ensure properly balanced tags when only formatting a partial code snippet. An example of stable formatting is HTML.


Using "define_substitution()", you can have the formatter substitute certain strings with others, after the original string has been parsed (but before formatting is applied). This is useful for escaping characters special to the output mode (eg, > and < in HTML) without them affecting the way the code is parsed.

You can retrieve the current substitutions (as a hash-ref) via "substitutions()".


The Syntax::Highlight::Perl::Improved formatter recognizes and differentiates between many Perl syntactical elements. Each type of syntactical element has a Format Type associated with it. There is also a 'DEFAULT' type that is applied to any element who's Format Type does not have a value.

Several of the Format Types have underscores in their name. This underscore is special, and indicates that the Format Type can be ``generalized.'' This means that you can assign a value to just the first part of the Format Type name (the part before the underscore) and that value will be applied to all Format Types with the same first part. For example, the Format Types for all types of variables begin with ``Variable_''. Thus, if you assign a value to the Format Type ``Variable'', it will be applied to any type of variable. Generalized Format Types take precedence over non-generalized Format Types. So the value assigned to ``Variable'' would be applied to ``Variable_Scalar'', even if ``Variable_Scalar'' had a value explicitly assigned to it.

You can also define a ``short-cut'' name for each Format Type that can be generalized. The short-cut name would be the part of the Format Type name after the underscore. For example, the short-cut for ``Variable_Scalar'' would be ``Scalar''. Short-cut names have the least precedence and are only assigned if neither the generalized Type name, nor the full Type name have values.

Following is a list of all the syntactical elements that Syntax::Highlight::Perl::Improved currently recognizes, along with a short description of what each would be applied to.

A normal Perl comment. Starts with '#' and goes until the end of the line.
Inline documentation. Starts with a line beginning with an equal sign ('=') followed by a word (eg: '=pod') and continuing until a line beginning with '=cut'.
Either the ``she-bang'' line at the beginning of the file, or a line directive altering what the compiler thinks the current line and file is.
A loop or statement label (to be the target of a goto, next, last or redo).
Any string or character that begins or ends a String. Including, but not necessarily limited to: quote-like regular expression operators ("m//", "s///", "tr///", etc), a Here-Document terminating line, the lone period terminating a format, and, of course, normal quotes ("'", """, "`", "q{}", "qq{}", "qr{}", "qx{}").
Any text within quotes, "format"s, Here-Documents, Regular Expressions, and the like.
The identifier used to define, identify, or call a subroutine (or method). Note that Syntax::Highlight::Perl::Improved cannot recognize a subroutine if it is called without using parentheses or an ampersand, or methods called using the indirect object syntax. It formats those as barewords.
A scalar variable.

Note that (theoretically) this format is not applied to non-scalar variables that are being used as scalars (ie: array or hash lookups, nor references to anything other than scalars). Syntax::Highlight::Perl::Improved figures out (or at least tries to) the actual type of the variable being used (by looking at how you're subscripting it) and formats it accordingly. The first character of the variable (ie, the "$", "@", "%", or "*") tells you the type of value being used, and the color (hopefully) tells you the type of variable being used to get that value.

(See ``KNOWN ISSUES'' for information about when this doesn't work quite right.)

An array variable (but not usually a slice; see above).
A hash variable.
A typeglob. Note that typeglobs not beginning with an asterisk (*) (eg: filehandles) are formatted as barewords. This is because, well, they are.
Whitespace. Not usually formatted but it can be.
A special, or backslash-escaped, character. For example: "\n" (newline), or "\d" (digits).

Only occurs within strings or regular expressions.

A Perl keyword. Some examples include: my, local, sub, next.

Note that Perl does not make any distinction between keywords and built-in functions (at least not in the documentation). Thus I had to make a subjective call as to what would be considered keywords and what would be built-in functions.

The list of keywords can be found (and overloaded) in the variable $Syntax::Highlight::Perl::Improved::keyword_list_re as a pre-compiled regular expression.

A Perl built-in function, called as a function (ie, using parentheses).

The list of built-in functions can be found (and overloaded) in the variable $Syntax::Highlight::Perl::Improved::builtin_list_re as a pre-compiled regular expression.

A Perl built-in function, called as a list or unary operator (ie, without using parentheses).

The list of built-in functions can be found (and overloaded) in the variable $Syntax::Highlight::Perl::Improved::builtin_list_re as a pre-compiled regular expression.

A Perl operator.

The list of operators can be found (and overloaded) in the variable $Syntax::Highlight::Perl::Improved::operator_list_re as a pre-compiled regular expression.

A bareword. This can be user-defined subroutine called without parentheses, a typeglob used without an asterisk (*), or just a plain old bareword.
The name of a package or pragmatic module.

Note that this does not apply to the package portion of a fully qualified variable name.

A numeric literal.
A symbol (ie, non-operator punctuation).
The special tokens that signal the end of executable code and the begining of the DATA section. Specifically, '"__END__"' and '"__DATA__"'.
Anything in the DATA section (see "CodeTerm").


Creates a new object. If called on an existing object, creates a new copy of that object (which is thenceforth totally separate from the original).
Resets the object's internal state. This breaks out of strings and here-docs, ends PODs, resets the line-count, and otherwise gets the object back into a ``normal'' state to begin processing a new stream.

Note that this does not reset any user options (including formats and format stability).

unstable EXPR
Returns true if the formatter is in unstable mode.

If called with a non-zero number, puts the formatter into unstable formatting mode.

In unstable mode, it is assumed that formatting is not persistent one token to the next and that each token must be explicitly formatted.

Returns true if the next string to be formatted will be inside a Here-Document.
Returns true if the next string to be formatted will be inside a multi-line string.
Returns true if the formatter would consider the next string passed to it as begin within a POD structure. This is false immediately before any POD instigators ("=pod", "=head1", "=item", etc), true immediately after an instigator, throughout the POD and immediately before the POD terminator ("=cut"), and false immediately after the POD terminator.
Returns true if the last line of the string just formatted was part of a POD structure. This includes the "/^=\w+/" POD instigators and terminators.
Returns true if the next string to be formatted will be inside the DATA section (ie, follows a "__DATA__" or "__END__" tag).
Returns the number of lines processed by the formatter.
Returns a reference to the substitution table used. The substitution table is a hash whose keys are the strings to be replaced, and whose values are what to replace them with.
define_substitution HASH_REF
define_substitution LIST
Allows user to define certain characters that will be substituted before formatting is done (but after they have been processed for meaning).

If the first parameter is a reference to a hash, the formatter will replace it's own hash with the given one, and subsequent changes to the hash outside the formatter will be reflected.

Otherwise, it will copy the arguments passed into it's own hash, and any substitutions already defined (but not in the parameter list) will be preserved. (ie, the new substitutions will be added, without destroying what was there already.)

set_start_format HASH_REF
set_start_format LIST
Given either a list of keys/values, or a reference to a hash of keys/values, copy them into the object's Formats list.
set_end_format HASH_REF
set_end_format LIST
Given either a list of keys/values, or a reference to a hash of keys/values, copy them into the object's Formats list.
set_format LIST
Sets the formatting string for one or more formats.

You should pass a list of keys/values where the keys are the format names and the values are references to arrays containing the starting and ending formatting strings (in that order) for that format.

get_start_format LIST
Retrieve the string that is inserted to begin a given format type (starting format string).

The names are looked for in the following order:

First: Prefer the names joined by underscore, from most general to least. For example, given (``Variable'', ``Scalar''): ``Variable'' then ``Variable_Scalar''.

Second: Then try each name singly, in reverse order. For example, ``Scalar'' then ``Variable''.

See ``FORMAT TYPES'' for more information.

get_end_format LIST
Retrieve the string that is inserted to end a given format type (ending format string).
Returns a list of the names of all the Formats defined.
Returns a list of the values of all the start Formats defined (in the same order as the names returned by "get_format_names_list()").
Returns a list of the values of all the end Formats defined (in the same order as the names returned by "get_format_names_list()").
format_string LIST
Formats one or more strings of Perl code. If no strings are specified, defaults to $_. Returns the list of formatted strings (or the first string formatted if called in scalar context).

Note: The end of the string is considered to be the end of a line, regardless of whether or not there is a trailing line-break (but trailing line-breaks will not cause an extra, empty line).

Another Note: The function actually uses $/ to determine line-breaks, unless $/ is set to "\n" (newline). If $/ is "\n", then it looks for the first match of "m/\r?\n|\n?\r/" in the string and uses that to determine line-breaks. This is to make it easy to handle non-unix text. Whatever characters it ends up using as line-breaks are preserved.

format_token TOKEN, LIST
Returns TOKEN wrapped in the start and end Formats corresponding to LIST (as would be returned by "get_start_format( LIST )" and "get_end_format( LIST )", respectively).

No syntax checking is done on TOKEN but substitutions defined with "define_substitution()" are performed.


  • Barewords used as keys to a hash are formatted as strings. This is Good. They should not be, however, if they are not the only thing within the curly braces. That can be fixed.
  • This version does not handle formats (see perlform(1)) very well. It treats them as Here-Documents and ignores the rules for comment lines, as well as the fact that picture lines are not supposed to be interpolated. Thus, your picture lines will look strange with the '@'s being formatted as array variables (albeit, invalid ones). Ideally, it would also treat value lines as normal Perl code and format accordingly. I think I'll get to the comment lines and non-interpolating picture lines first. If/When I do get this fixed, I will most likely add a format type of 'Format' or something, so that they can be formatted differently, if so desired.
  • This version does not handle Regular Expression significant characters. It simply treats Regular Expressions as interpolated strings.
  • User-defined subroutines, called without parentheses, are formatted as barewords. This is because there is no way to tell them apart from barewords without parsing the code, and would require us to go as far as perl does when doing the "-c" check (ie, executing BEGIN and END blocks and the like). That's not going to happen.
  • If you are indexing (subscripting) an array or hash, the formatter tries to figure out the ``real'' variable class by looking at how you index the variable. However, if you do something funky (but legal in Perl) and put line-breaks or comments between the variable class character ($) and your identifier, the formatter will get confused and treat your variable as a scalar. Until it finds the index character. Then it will format the scalar class character ($) as a scalar and your identifier as the ``correct'' class.
  • If you put a line-break between your variable identifier and it's indexing character (see above), which is also legal in Perl, the formatter will never find it and treat your variable as a scalar.
  • If you put a line-break between a bareword hash-subscript and the hash variable, or between a bareword and its associated "=>" operator, the bareword will not be formatted correctly (as a string). (Noticing a pattern here?)