NAME

 HTML::PrettyPrinter - generate nice HTML files from HTML syntax trees

SYNOPSIS

  use HTML::TreeBuilder;
  # generate a HTML syntax tree
  my $tree = new HTML::TreeBuilder;
  $tree->parse_file($file_name);
  # modify the tree if you want

  use HTML::PrettyPrinter;
  my $hpp = new HTML::PrettyPrinter ('linelength' => 130,
                                     'quote_attr' => 1);
  # configure
  $tree->address("0.1.0")->attr(_hpp_indent,0);    # for an individual element
  $hpp->set_force_nl(1,qw(body head));             # for tags
  $hpp->set_force_nl(1,qw(@SECTIONS));             # as above
  $hpp->set_nl_inside(0,'default!');               # for all tags

  # format the source
  my $linearray_ref = $hpp->format($tree);
  print @$linearray_ref;

  # alternative: print directly to filehandle
  use FileHandle;
  my $fh = new FileHandel ">$filenaem2";
  if (defined $fh) {
    $hpp->select($fh);
    $hpp->format();
    undef $fh;
    $hpp->select(undef),  
  }

DESCRIPTION

HTML::PrettyPrinter produces nicely formatted HTML code from a HTML syntax tree. It is especially usefull if the produced HTML file shall be read or edited manually afterwards. Various parameters let you adapt the output to different styles and requirements.

If you don't care how the HTML source looks like as long as it is valid and readable by browsers, you should use the as_HTML() method of HTML::Element instead of the pretty printer. It is about five times faster.

The pretty printer will handle line wrapping, indention and structuring by the way the whitespace in the tree is represented in the output. Furthermore upper/lowercase markup and markup minimization, quoting of attribute values, the encoding of entities and the presence of optional end tags are configurable.

There are two types of parameters to influence the output, individual parameters that are set on a per element and per tag basis and common parameters that are set only once for each instance of a pretty printer.

In order to faciliate the configuration a mechanism to handle tag groups is provided. Thus, it is possible to modify a parameter for a group of tags (e.g. all known block elements) without writing each tag name explicitly. Perhaps the code for tag groups will move to an other Perl module in the future.

For HTML::Elements that require a special treatment like <PRE>, <XMP>, <SCRIPT>, comments and declarations, pretty printer will fall back to the method as_HTML() of the HTML elements.

INDIVIDUAL PARAMETERS

Following individual paramters exist

indent n

The indent of new lines inside the element is increased by n coloumns. Default is 2 for all tags.

skip bool

If true, the element and its content is skipped from output. Default is false.

nl_before n

Number of newlines before the start tag. Default is 0 for inline elements and 1 for other elements.

nl_inside n

Number of newlines between the tags and the contents of an element. Default is 0.

nl_after n

Number of newlines after an element. Default is 0 for inline elements and 1 for other elements.

force_nl bool

Force linebreaks before and after an element even if the HTML tree does not contain whitespace at this place. Default is false for inline elements and true for all other elements. This parameter is superseded if the common parameter allow_forced_nl is set to false.

endtag bool

Print an optional endtag. Default is true.

Access Methods

Following access methods exist for each individual paramenter. Replace parameter by the respective name.

$hpp->parameter($element)

Takes a reference to an HTML element as argument. Returns the value of the parameter for that element. The priority to retrieve the value is:

  1. The value of the element's internal attribute _hpp_parameter.

  2. The value specified inside the pretty printer for the tag of the element.

  3. The value specified inside the pretty printer for 'default!'.

$hpp->parameter('tag')

Like parameter($element), except that only priorities 2 and 3 are evaluated.

$hpp->set_parameter($value,'tag1','tag2',...)

Sets the parameter for each tag in the list to $value.

If $value is undefined, the entries for the tags are deleted.

Beside individual tags the list may include tag groups like '@BLOCK' (see below) and 'default!'. Individual tag names are written in lower case, the names of tag groups start with an '@' and are written in upper case letters. Tag groups are expanded during the call of set_parameter(). 'default!' sets the default value, which is retrived if no value is defined for the individual element or tag.

$hpp->set_parameter($value,'all!')

Deletes all existing settings for parameter inside the pretty printer and sets the default to $value..

COMMON PARAMETERS

tabify n

If non zero, each n spaces at the beginnig of a line are converted into one TAB. Default is 8.

linelength n

The maximum number of character a line should have. Default is 80.

The linelength may be exceeded if there is no proper way to break a line without modifying the content, e.g. inside <PRE> and other special elements or if there is no whitespace.

min_bool_attr bool

Minimize boolean attributes, e.g. print <UL COMPACT> instead of <UL COMPACT=COMPACT>. Default is true.

quote_attr bool

Always quote attribute values. If false, attribute values consisting entirely of letters, digits, periods and hyphens only are not put into quotes. Default is false.

entities string

The string contains all characters that are escaped to their entity names. Default is the bare minimum of "&<>" plus the non breaking space 'nbsp' (because otherwise it is difficult for the human eye to distiguish it from a normal space in most editors).

wrap_at_tagend NEVER|AFTER_ATTR|ALWAYS

May pretty printer wrap lines before the closing ankle of a start tag? Supported values are the predifined constants NEVER (allow line wraps at white space only ), AFTER_ATTR (allow line wraps at the end of tags that contain attributes only) and ALWAYS (allow line wraps at the end of every start tag). Default is AFTER_ATTR.

allow_forced_nl bool

Allow the addition of white space, that is not in the HTML tree. If set to false (the default) the force_nl parameter is ignored. It is recomended to set this parameter to true if the HTML tree was generated with ignore_ignorable_whitespace set to true.

uppercase bool

Use uppercase letters for markup. Default is the value of $HTML::Element::html_uc at the time the constructor is called.

Access Method

$hpp->paramter([value])

Retrieves and optionaly sets the parameter.

OTHER METHODS

$hpp = HTML::PrettyPrinter->new(%common_paremeters)

This class method creates a new HTML::PrettyPrinter and returns it. Key/value pair arguments may be provided to overwrite the default settings of common parameters. There is currently no mechanism to overwrite the default values for individual parameters at construction. Use the $hpp-set_parameter()> methods instead.

$hpp->select($fh)

Select a FileHandle object for output.

If a FileHandle is selected the generated HTML is printed directly to that file. With $hpp->select(undef) you can switch back to the default behaviour.

$line_array_ref = $hpp->format($tree,[$indent],[$line_array_ref])

Format the HTML syntax (sub-) tree.

$tree is not restricted to the root of the HTML syntax tree. A reference to any HTML::Element will do.

The optional $indent indents the first element by n characters

Return value is the reference to an array with the generated lines. If such a reference is provided as third argument, the lines will be appended to that array. Otherwise a new array will be created.

If a FileHandle is selected by a previous call of the $hpp-select($fh)> method, the lines are printed to the FileHandle object directly. The array of lines is not changed in this case.

TAG GROUPS

Tag groups are lists that contain the names of tags and other tag groups which are considered as subsets. This reflects the way allowed content is specified in HTML DTDs, where e.g. %flow consists of all %block and %inline elements and %inline covers several subsets like %phrase.

If you add a tag name to a group A, it will be seen in any group that contains group A. Thus, it is easy to maintain groups of tags with similar properties. (and configure HTML pretty printer for these tags).

The names of tag groups are written in upper case letters with a leading '@' (e.g. '@BLOCK'). The names of simple tags are written all lower case.

Functions

All the functions to handle and modify tag groups are included in the @EXPORT_OK list of HTML::PrettyPrinter.

@tag_groups = list_groups()

Returns a list with the names of all defined tag groups

@tags = group_expand('tag_or_tag_group0',['tag_or_tag_group1',...])

Returns a list of every tag in the tag groups and their subgroups Each tag is listed once only. The order of the list is not specified.

@tag_groups = sub_group('tag_group0',['tag_group1',...])

Returns a list of every tag group and sub group in the list. Each group is listed once only. The order of the list is not specified.

group_get('@NAME')

Return the (unexpanded) contents of a tag group.

group_set('@NAME',['tag_or_tag_group0',...])

Set a tag group.

group_add('@NAME','tag_or_tag_group0',['tag_or_tag_group1',...])

Add tags and tag groups to a group.

group_remove('@NAME','tag_or_tag_group0',['tag_or_tag_group1',...])

Remove tags or tag groups from a group. Subgroups are not expanded. Thus, group_remove('@A','@B') will remove '@B' from '@A' if it is included directly. Tags included in '@B' will not be removed from '@A'. Nor will '@A' be changed if '@B' is included in a aubgroup of '@A' but not in '@A' directly.

Predefined Tag Groups

There are a couple of predefined tag groups. Use foreach my $tg (list_groups()) { print "'$tg' => qw(".join(',',group_get($tg)).")\n"; } to get a list.

Examples for tag groups

1. create some groups

group_set('@A',qw(a1 a2 a3)); group_set('@B',qw(b1 b2)); group_set('@C',qw(@A @B c1 @D)); # @D needs to be defined when @C is expannded group_set('@D',qw(d1 @B)); group_set('@E',qw(e1 @D)); group_set('@F',qw(f1 @A));

2. add tags

group_add('@A',qw(a4 a5)); # @A contains (a1 a2 a3 a4 a5) group_add('@D',qw(d1)); # @D contains (d1 @B d1) group_add('@F',group_exapand('@B'),'@F'); # @F contains (f1 @A b1 b2 f1 @F)

3. evaluate

group_exapand('@E'); # returns e1, d1, b1, b2 sub_groups('@E'); # returns @B, @D sub_groups(qw(@E @F)); # returns @A, @B, @D group_get('@F')); # returns f1, @A, b1, b2, f1, @F

4. remove tags

group_remove('@E','@C'); # @E not changed, because it doesn't contain @C group_remove('@E','@D'); # @D removed from @E group_remove('@D','d1'); # all d1's are removed. Now @D contains @B only group_remove('@C','@B'); # @C now contains (@a c1 @D), Thus sub_groups('@C'); # still returns @A, @B, @D, # because @B is included in @D, too

5. application

# set the indent for tags b1, b2, e1, g1 to 0 $hpp->set_indent(0,qw(@D @E g1));

If the groups @D or @E are modified afterwards, the configuration of the pretty printer is not affected, because set_indent() will expand the tag groups.

EXAMPLE

Consider the following HTML tree

    <html> @0
      <head> @0.0
        <title> @0.0.0
          "Demonstrate HTML::PrettyPrinter"
      <body> @0.1
        <h1> @0.1.0
          "Headline"
        <p align="JUSTIFY"> @0.1.1
          "Some text in "
          <b> @0.1.1.1
            "bold"
          " and "
          <i> @0.1.1.3
            "italics"
          " and with 'ä' & 'ü'."
        <table align="LEFT" border=0> @0.1.2
          <tr> @0.1.2.0
            <td align="RIGHT"> @0.1.2.0.0
              "top right"
          <tr> @0.1.2.1
            <td align="LEFT"> @0.1.2.1.0
              "bottom left"
        <hr noshade="NOSHADE" size=5> @0.1.3
        <address> @0.1.4
          <a href="mailto:schotten@gmx.de"> @0.1.4.0
            "Claus Schotten"

and $hpp = HTML::PrettyPrinter-new('uppercase' => 1); print @{$hpp->format($tree)}; >

will print

  <HTML><HEAD><TITLE>Demonstrate
        HTML::PrettyPrinter</TITLE></HEAD><BODY><H1>Headline</H1><P
        ALIGN=JUSTIFY>Some text in <B>bold</B> and
        <I>italics</I> and with 'ä' &amp; 'ü'.</P><TABLE
        ALIGN=LEFT BORDER=0><TR><TD ALIGN=RIGHT>top
            right</TD></TR><TR><TD ALIGN=LEFT>bottom
            left</TD></TR></TABLE><HR NOSHADE SIZE=5
        ><ADDRESS><A HREF="mailto:schotten@gmx.de"
          >Claus&nbsp;Schotten</A></ADDRESS></BODY></HTML>

That doesn't look very nice. What went wrong? By default HTML::PrettyPrinter takes a conservative approach on whitespace. It will enlarge existing whitespace, but it will not introduce new whitespace outside of tags, because that might change the way a browser renders the HTML document. However the HTML tree was constructed with ignore_ignorable_whitespace> turned on. Thus, there is no whitespace between block elements that the pretty printer could format. So pretty printer does line wrapping and indention only. E.g. the title is in the third level of the tree. Thus, the second line is indented six characters. The table cells in the fifth level are indented by ten characters. Furthermore, you see that there is a whitespace inserted after the last attribute of the <A> tag.

Let's set $hpp->allow_forced_nl(1);. Now the forced_nl parameters are enabled. By default, they are set for all non-inline tags. That creates

 <HTML>
   <HEAD>
     <TITLE>Demonstrate HTML::PrettyPrinter</TITLE>
   </HEAD>
   <BODY>
     <H1>Headline</H1>
     <P ALIGN=JUSTIFY>Some text in <B>bold</B> and
       <I>italics</I> and with 'ä' &amp; 'ü'.</P>
     <TABLE ALIGN=LEFT BORDER=0>
       <TR>
         <TD ALIGN=RIGHT>top right</TD>
       </TR>
       <TR>
         <TD ALIGN=LEFT>bottom left</TD>
       </TR>
     </TABLE>
     <HR NOSHADE SIZE=5>
     <ADDRESS><A HREF="mailto:schotten@gmx.de"
         >Claus&nbsp;Schotten</A></ADDRESS>
   </BODY>
 </HTML>
  

Much better, isn't it? Now let's improve the structuring. $hpp->set_nl_before(2,qw(body table)); $hpp->set_nl_after(2,qw(table)); will require two new lines in front of <body> and <table> tags and after <table> tags.

 <HTML>
   <HEAD>
     <TITLE>Demonstrate HTML::PrettyPrinter</TITLE>
   </HEAD>
  
   <BODY>
     <H1>Headline</H1>
     <P ALIGN=JUSTIFY>Some text in <B>bold</B> and
       <I>italics</I> and with 'ä' &amp; 'ü'.</P>
  
     <TABLE ALIGN=LEFT BORDER=0>
       <TR>
         <TD ALIGN=RIGHT>top right</TD>
       </TR>
       <TR>
         <TD ALIGN=LEFT>bottom left</TD>
       </TR>
     </TABLE>
  
     <HR NOSHADE SIZE=5>
     <ADDRESS><A HREF="mailto:schotten@gmx.de"
         >Claus&nbsp;Schotten</A></ADDRESS>
   </BODY>
 </HTML>

Currently the mail address is the only attribute value which is quoted. Here the quotes are required by the '@' character. For all other attribute values quotes are optional and thus ommited by default. $hpp->quote_attr(1); will turn the quotes on.

$hpp->set_endtag(0,'all!') turns all optional endtags off. This affects the </p> (and should affect </tr> and </td>, see below). Alternatively, we could use $hpp->set_endtag(0,'default!'). That would turn the default off, too. But it wouldn't delete settings for individual tags that supersede the default.

$hpp->set_nl_after(3,'head') requires three new lines after the <head> element. Because there are already two new lines required by the start of <body> only one additional line is added.

$hpp->set_force_nl(0,'td') will inhibit the introduction of whitespace alround <td>. Thus, the table cells are now on the same line as the table rows.

  <HTML>
    <HEAD>
      <TITLE>Demonstrate HTML::PrettyPrinter</TITLE>
    </HEAD> 
    
    
    <BODY>
      <H1>Headline</H1>
      <P ALIGN="JUSTIFY">Some text in <B>bold</B> and
        <I>italics</I> and with 'ä' &amp; 'ü'.
   
      <TABLE ALIGN="LEFT" BORDER="0">
        <TR><TD ALIGN="RIGHT">top right</TD></TR>
        <TR><TD ALIGN="LEFT">bottom left</TD></TR>
      </TABLE>
   
      <HR NOSHADE SIZE="5">
      <ADDRESS><A HREF="mailto:schotten@gmx.de"
          >Claus&nbsp;Schotten</A></ADDRESS>
    </BODY>
  </HTML>

The end tags </td> and </tr> are printed because HTML:Tagset says they are mandatory. map {$HTML::Tagset::optionalEndTag{$_}=1} qw(td tr th); will fix that.

The additional new line after </head> doesn't look nice. With $hpp->set_nl_after(undef,'head') we will reset the parameter for the <head> tag.

$hpp->entities($hpp->entities().'ä'); will enforce the entity encoding of 'ä'.

$hpp->min_bool_attr(0); will inhibt the minimizyation of the NOSHADE attribute to <hr>.

Let's fiddle with the indention: $hpp->set_indent(8,'@TEXTBLOCK'); $hpp->set_indent(0,'html');

New lines inside text blocks (here inside <h1>, <p> and <address>) will be indented by 8 characters instead of two, whereas the code directly under <html> will not be indented.

 <HTML>
 <HEAD>
   <TITLE>Demonstrate HTML::PrettyPrinter</TITLE>
 </HEAD>
  
 <BODY>
   <H1>Headline</H1>
   <P ALIGN="JUSTIFY">Some text in <B>bold</B> and
           <I>italics</I> and with '&auml;' &amp; 'ü'.
  
   <TABLE ALIGN="LEFT" BORDER="0">
     <TR><TD ALIGN="RIGHT">top right
     <TR><TD ALIGN="LEFT">bottom left
   </TABLE>
  
   <HR NOSHADE="NOSHADE" SIZE="5">
   <ADDRESS><A HREF="mailto:schotten@gmx.de"
             >Claus&nbsp;Schotten</A></ADDRESS>
 </BODY>
 </HTML>

$hpp->wrap_at_tagend(HTML::PrettyPrinter::NEVER); will disable the line wrap between the attribute and the '>' of the <a> tag. The resulting line excedes the target line length by far, but the is no point left, where the pretty printer could legaly break this line.

$hpp->set_endtag(1,'tr') will overwrite the default. Thus, the </tr> appears in the code whereas the other optional endtags are still omitted.

Finally, we customize some individual elements:

$tree-address('0.1.1')->attr('_hpp_skip',1)>

will skip the <p> and its content from the output

$tree-address('0.1.2.1.0')->attr('_hpp_force_nl',1)>

will force new lines arround the second <td>, but will not affect the first. <td>.

 <HTML>
 <HEAD>
   <TITLE>Demonstrate HTML::PrettyPrinter</TITLE>
 </HEAD>
 
 <BODY>
   <H1>Headline</H1>
  
   <TABLE ALIGN="LEFT" BORDER="0">
     <TR><TD ALIGN="RIGHT">top right</TR>
     <TR>
       <TD ALIGN="LEFT">bottom left
     </TR>
   </TABLE>
  
   <HR NOSHADE="NOSHADE" SIZE="5">
   <ADDRESS><A
             HREF="mailto:schotten@gmx.de">Claus&nbsp;Schotten</A></ADDRESS>
 </BODY>
 </HTML>

KNOWN BUGS

  • This is early alpha code. The interfaces are subject to changes.

  • The module is tested with perl 5.005_03 only. It should work with perl 5.004 though.

  • The predefined tag groups are incomplete. Several tags need to be added.

  • Attribute values from a fixed set given in the DTD (e.g. ALIGN=LEFT|RIGHT etc.) should be converted to upper or lower case depending on the value of the uppercase parameter. Currently, they are printed as given in the HTML tree.

  • No optimization for performance was done.

SEE ALSO

HTML::TreeBuilder, HTML::Element, HTML::Tagset

COPYRIGHT

Copyright 2000 Claus Schotten schotten@gmx.de

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself.

AUTHOR

Claus Schotten <schotten@gmx.de>

1 POD Error

The following errors were encountered while parsing the POD:

Around line 954:

Non-ASCII character seen before =encoding in 'print "'$tg''. Assuming CP1252