| HTML::TextToHTML - convert plain text file to HTML |
HTML::TextToHTML - convert plain text file to HTML
From the command line:
perl -MHTML::TextToHTML -e run_txt2html -- I<arguments>;
(calls the txt2html method with the given arguments)
From Scripts:
use HTML::TextToHTML;
# create a new object
my $conv = new HTML::TextToHTML();
# convert a file
$conv->txt2html(infile=>[$text_file],
outfile=>$html_file,
title=>"Wonderful Things",
mail=>1,
]);
# reset arguments
$conv->args(infile=>[], mail=>0);
# convert a string
$newstring = $conv->process_chunk($mystring)
HTML::TextToHTML converts plain text files to HTML.
It supports headings, tables, lists, simple character markup, and hyperlinking, and is highly customizable. It recognizes some of the apparent structure of the source document (mostly whitespace and typographic layout), and attempts to mark that structure explicitly using HTML. The purpose for this tool is to provide an easier way of converting existing text documents to HTML format.
There are two ways to use this module: (1) called from a perl script (2) call run_txt2html from the command line
The first usage requires one to create a HTML::TextToHTML object, and then call the txt2html or process_para method with suitable arguments. One can also pass arguments in when creating the object, or call the args method to pass arguments in.
The second usage allows one to pass arguments in from the command line, by calling perl and executing the module, and calling run_txt2html which creates an object for you and parses the command line.
Either way, the arguments are the same. See OPTIONS for the arguments; see METHODS for the methods of the HTML::TextToHTML object.
The following are the exported functions of this module:
run_txt2html()
This is what is used to run this module from the command-line. It creates a HTML::TextToHTML object and parses the command-line arguments, and passes them to the object, and runs the txt2html method. It takes no arguments.
All arguments can be set when the object is created, and further options can be set when calling the actual txt2html method. Arguments to methods can take either a hash of arguments, or a reference to an array (which will then be processed as if it were a command-line, which makes this easy to use from scripts even if you don't wish to use the commonly used Getopt::Long module in your script).
Note that all option-names must match exactly -- no abbreviations are allowed.
The arguments get treated differently depending on whether they are given in a hash or a reference to an array. When the arguments are in a hash, the argument-keys are expected to have values matching those required for that argument -- whether that be a boolean, a string, a reference to an array or a reference to a hash. These will replace any value for that argument that might have been there before.
When the arguments are in a reference to an array, it is treated somewhat as if it were a command-line: option names are expected to start with '--' or '-', boolean options are set to true as soon as the option is given (no value is expected to follow), boolean options with the word ``no'' prepended set the option to false, string options are expected to have a string value following, and those options which are internally arrays or hashes are treated as cumulative; that is, the value following the --option is added to the current set for that option, to add more, one just repeats the --option with the next value, and in order to reset that option to empty, the special value of ``CLEAR'' must be added to the list.
append_file=>I<filename>
If you want something appended by default, put the filename here. The appended text will not be processed at all, so make sure it's plain text or decent HTML. i.e. do not have things like: Mary Andersen <kitty@example.com> but instead, have: Mary Andersen <kitty@example.com>
(default: nothing)
append_head=>I<filename>
If you want something appended to the head by default, put the filename here. The appended text will not be processed at all, so make sure it's plain text or decent HTML. i.e. do not have things like: Mary Andersen <kitty@example.com> but instead, have: Mary Andersen <kitty@example.com>
(default: nothing)
body_deco=>I<string>
Body decoration string: a string to be added to the BODY tag so that one can set attributes to the BODY (such as class, style, bgcolor etc) For example, ``class='withimage'''.
bullets=>I<string>
This defines what single characters are taken to be ``bullet'' characters for unordered lists. Note that because this is used as a character class, if you use '-' it must come first. (default:-=o*\267)
bullets_ordered=>I<string>
This defines what single characters are taken to be ``bullet'' placeholder characters for ordered lists. Ordered lists are normally marked by a number or letter followed by '.' or ')' or ']' or ':'. If an ordered bullet is used, then it simply indicates that this is an ordered list, without giving explicit numbers.
Note that because this is used as a character class, if you use '-' it must come first. (default:nothing)
caps_tag=>I<tag>
Tag to put around all-caps lines (default: STRONG) If an empty tag is given, then no tag will be put around all-caps lines.
custom_heading_regexp=>I<regexp>
Add a regexp for headings. Header levels are assigned by regexp in order seen When a line matches a custom header regexp, it is tagged as a header. If it's the first time that particular regexp has matched, the next available header level is associated with it and applied to the line. Any later matches of that regexp will use the same header level. Therefore, if you want to match numbered header lines, you could use something like this: -H '^ *\d+\. \w+' -H '^ *\d+\.\d+\. \w+' -H '^ *\d+\.\d+\.\d+\. \w+'
Then lines like
" 1. Examples "
" 1.1. Things"
and " 4.2.5. Cold Fusion"
Would be marked as H1, H2, and H3 (assuming they were found in that order, and that no other header styles were encountered). If you prefer that the first one specified always be H1, the second always be H2, the third H3, etc, then use the -EH/--explicit-headings option.
This is a multi-valued option.
(default: none)
debug=>1
Enable copious script debugging output (don't bother, this is for the developer) (default: false)
default_link_dict=>I<filename>
The name of the default ``user'' link dictionary. (default: ``$ENV{'HOME'}/.txt2html.dict'' -- this is the same as for the txt2html script)
dict_debug=>I<n>
Debug mode for link dictionaries Bitwise-Or what you want to see: 1: The parsing of the dictionary 2: The code that will make the links 4: When each rule matches something 8: When each tag is created
(default: 0)
doctype=>I<doctype>
This gets put in the DOCTYPE field at the top of the document, unless it's empty. (default : ``-//W3C//DTD HTML 3.2 Final//EN'') If --xhtml is true, the contents of this is ignored, unless it's empty, in which case no DOCTYPE declaration is output.
eight_bit_clean=>1
disable Latin-1 character entity naming (default: false)
escape_HTML_chars=>1
turn & < > into & > < (default: true)
explicit_headings=>1
Don't try to find any headings except the ones specified in the --custom_heading_regexp option. Also, the custom headings will not be assigned levels in the order they are encountered in the document, but in the order they are specified on the command line. (default: false)
extract=>1
Extract Mode; don't put HTML headers or footers on the result, just the plain HTML (thus making the result suitable for inserting into another document (or as part of the output of a CGI script). (default: false)
hrule_min=>I<n>
Min number of ---s for an HRule. (default: 4)
indent_width=>I<n>
Indents this many spaces for each level of a list. (default: 2)
indent_par_break=>1
Treat paragraphs marked solely by indents as breaks with indents. That is, instead of taking a three-space indent as a new paragraph, put in a <BR> and three non-breaking spaces instead. (see also --preserve_indent) (default: false)
infile=>\@my_files
infile=>['chapter1.txt', 'chapter2.txt']
"--infile", "chapter1.txt", "--infile", "chapter2.txt"
The name of the input file(s). When the arguments are given as a hash, this expects a reference to an array of filenames. When the arguments are given as a reference to an array, then the ``--infile'' option must be repeated for each new file added to the list. If you want to reset the list to be empty, give the special value of ``CLEAR''.
(default:-)
links_dictionaries=>\@my_link_dicts
links_dictionaries=>['url_links.dict', 'format_links.dict']
"--links_dictionaries", "url_links.dict", "--links_dictionaries", "format_links.dict"
File(s) to use as a link-dictionary. There can be more than one of
these. These are in addition to the System Link Dictionary and the User
Link Dictionary. When the arguments are given as a hash, this expects a
reference to an array of filenames. When the arguments are given as a
reference to an array, then the ``--links_dictionaries'' option must be
repeated for each new file added to the list. If you want to reset the
list to be empty, give the special value of ``CLEAR''.
link_only=>1
Do no escaping or marking up at all, except for processing the links dictionary file and applying it. This is useful if you want to use the linking feature on an HTML document. If the HTML is a complete document (includes HTML,HEAD,BODY tags, etc) then you'll probably want to use the --extract option also. (default: false)
mailmode=>1
Deal with mail headers & quoted text (default: false)
make_anchors=>0
Should we try to make anchors in headings? (default: true)
make_links=>0
Should we try to build links? If this is false, then the links dictionaries are not consulted and only structural text-to-HTML conversion is done. (default: true)
make_tables=>1
Should we try to build tables? If true, spots tables and marks them up appropriately. See Input File Format for information on how tables should be formatted.
This overrides the detection of lists; if something looks like a table, it is taken as a table, and list-checking is not done for that paragraph.
(default: false)
min_caps_length=>I<n>
min sequential CAPS for an all-caps line (default: 3)
outfile=>I<filename>
The name of the output file. If it is ``-'' then the output goes to Standard Output. (default: - )
par_indent=>I<n>
Minumum number of spaces indented in first lines of paragraphs. Only used when there's no blank line preceding the new paragraph. (default: 2)
preformat_trigger_lines=>I<n>
How many lines of preformatted-looking text are needed to switch to <PRE> <= 0 : Preformat entire document 1 : one line triggers >= 2 : two lines trigger
(default: 2)
endpreformat_trigger_lines=>I<n>
How many lines of unpreformatted-looking text are needed to switch from <PRE> <= 0 : Never preformat within document 1 : one line triggers >= 2 : two lines trigger (default: 2)
NOTE for preformat_trigger_lines and endpreformat_trigger_lines: A zero takes precedence. If one is zero, the other is ignored. If both are zero, entire document is preformatted.
preformat_start_marker=>I<regexp>
What flags the start of a preformatted section if --use_preformat_marker is true.
(default: ``^(:?(:?<)|<)PRE(:?(:?>)|>)\$'')
preformat_end_marker=>I<regexp>
What flags the end of a preformatted section if --use_preformat_marker is true.
(default: ``^(:?(:?<)|<)/PRE(:?(:?>)|>)\$'')
preformat_whitespace_min=>I<n>
Minimum number of consecutive whitespace characters to trigger normal preformatting. NOTE: Tabs are expanded to spaces before this check is made. That means if tab_width is 8 and this is 5, then one tab may be expanded to 8 spaces, which is enough to trigger preformatting. (default: 5)
prepend_file=>I<filename>
If you want something prepended to the processed body text, put the filename here. The prepended text will not be processed at all, so make sure it's plain text or decent HTML.
(default: nothing)
preserve_indent=>1
Preserve the first-line indentation of paragraphs marked with indents by replacing the spaces of the first line with non-breaking spaces. (default: false)
short_line_length=>I<n>
Lines this short (or shorter) must be intentionally broken and are kept that short. (default: 40)
style_url=>I<url>
This gives the URL of a stylesheet; a LINK tag will be added to the output.
system_link_dict=>I<filename>
The name of the default ``system'' link dictionary. (default: ``/usr/share/txt2html/txt2html.dict'')
tab_width=>I<n>
How many spaces equal a tab? (default: 8)
table_type=>{ ALIGN=>0, PGSQL=>0, BORDER=>1, DELIM=>0 }
This determines which types of tables will be recognised when ``make_tables'' is true. The possible types are ALIGN, PGSQL, BORDER and DELIM. (default: all types are true)
title=>I<title>
You can specify a title. Otherwise it will use a blank one. (default: nothing)
titlefirst=>1
Use the first non-blank line as the title.
underline_length_tolerance=>I<n>
How much longer or shorter can underlines be and still be underlines? (default: 1)
underline_offset_tolerance=>I<n>
How far offset can underlines be and still be underlines? (default: 1)
unhyphenation=>0
Enables unhyphenation of text. (default: true)
use_mosaic_header=>1
Use this option if you want to force the heading styles to match what Mosaic outputs. (Underlined with ``***''s is H1, with ``===''s is H2, with ``+++'' is H3, with ``---'' is H4, with ``~~~'' is H5 and with ``...'' is H6) This was the behavior of txt2html up to version 1.10. (default: false)
use_preformat_marker=>1
Turn on preformatting when encountering ``<PRE>'' on a line by itself, and turn it off when there's a line containing only ``</PRE>''. (default: off)
xhtml=>1
Try to make the output conform to the XHTML standard, including closing all open tags and marking empty tags correctly. This turns on --lower_case_tags and overrides the --doctype option. Note that if you add a header or a footer file, it is up to you to make it conform; the header/footer isn't touched by this. Likewise, if you make link-dictionary entries that break XHTML, then this won't fix them, except to the degree of putting all tags into lower-case.
$conv = new HTML::TextToHTML()
$conv = new HTML::TextToHTML(\@args)
$conv = new HTML::TextToHTML(titlefirst=>1,
...
);
Create a new object with new. If one argument is given, it is assumed to be a reference to an array of arguments. If more than one argument is given, it is assumed to be a hash of arguments. These arguments will be used in invocations of other methods.
See OPTIONS for the possible values of the arguments.
$conv->args(\@args)
$conv->args(short_line_length=>60,
titlefirst=>1,
....
);
Updates the current arguments/options of the HTML::TextToHTML object. Takes either a hash, or a reference to an array of arguments, which will be used in invocations of other methods. See OPTIONS for the possible values of the arguments.
$newstring = $conv->process_chunk($mystring);
Convert a string to a HTML fragment. This assumes that this string is at the least, a single paragraph, but it can contain more than that. This returns the processed string. If you want to pass arguments to alter the behaviour of this conversion, you need to do that earlier, either when you create the object, or with the the args manpage method.
$newstring = $conv->process_chunk($mystring,
close_tags=>0);
If there are open tags (such as lists) in the input string, process_chunk will now automatically close them, unless you specify not to, with the close_tags option.
$newstring = $conv->process_chunk($mystring,
is_fragment=>1);
If you want this string to be treated as a fragment, and not assumed to be a paragraph, set is_fragment to true. If there is more than one paragraph in the string (ie it contains blank lines) then this option will be ignored.
$newstring = $conv->process_para($mystring);
Convert a string to a HTML fragment. This assumes that this string is at the most a single paragraph, with no blank lines in it. If you don't know whether your string will contain blank lines or not, use the the process_chunk manpage method instead.
This returns the processed string. If you want to pass arguments to alter the behaviour of this conversion, you need to do that earlier, either when you create the object, or with the the args manpage method.
$newstring = $conv->process_para($mystring,
close_tags=>0);
If there are open tags (such as lists) in the input string, process_para will now automatically close them, unless you specify not to, with the close_tags option.
$newstring = $conv->process_para($mystring,
is_fragment=>1);
If you want this string to be treated as a fragment, and not assumed to be a paragraph, set is_fragment to true.
$conv->txt2html(\@args);
$conv->txt2html(%args);
Convert a text file to HTML. Takes a hash of arguments, or a reference to an array of arguments to customize the conversion; (this includes saying what file to convert!) See OPTIONS for the possible values of the arguments. Arguments which have already been set with new or args will remain as they are, unless they are overridden.
There are two files which are used which can affect the outcome of the conversion. One is the link dictionary, which contains patterns (of how to recognise http links and other things) and how to convert them. The other is, naturally, the format of the input file itself.
A link dictionary file contains patterns to match, and what to convert them to. It is called a ``link'' dictionary because it was intended to be something which defined what a href link was, but it can be used for more than that. However, if you wish to define your own links, it is strongly advised to read up on regular expressions (regexes) because this relies heavily on them.
The file consists of comments (which are lines starting with #) and blank lines, and link entries. Each entry consists of a regular expression, a -> separator (with optional flags), and a link ``result''.
In the simplest case, with no flags, the regular expression defines the pattern to look for, and the result says what part of the regular expression is the actual link, and the link which is generated has the href as the link, and the whole matched pattern as the visible part of the link. The first character of the regular expression is taken to be the separator for the regex, so one could either use the traditional / separator, or something else such as | (which can be helpful with URLs which are full of / characters).
So, for example, an ftp URL might be defined as:
|ftp:[\w/\.:+\-]+| -> $&
This takes the whole pattern as the href, and the resultant link has the same thing in the href as in the contents of the anchor.
But sometimes the href isn't the whole pattern.
/<URL:\s*(\S+?)\s*>/ --> $1
With the above regex, a () grouping marks the first subexpression, which is represented as $1 (rather than $& the whole expression). This entry matches a URL which was marked explicity as a URL with the pattern <URL:foo> (note the < is shown as the entity, not the actual character. This is because by the time the links dictionary is checked, all such things have already been converted to their HTML entity forms) This would give us a link in the form <A HREF=``foo''><URL:foo></A>
The h flag
However, if we want more control over the way the link is constructed, we can construct it ourself. If one gives the h flag, then the ``result'' part of the entry is taken not to contain the href part of the link, but the whole link.
For example, the entry:
/<URL:\s*(\S+?)\s*>/ -h-> <A HREF="$1">$1</A>
will take <URL:foo> and give us <A HREF=``foo''>foo</A>
However, this is a very powerful mechanism, because it can be used to construct custom tags which aren't links at all. For example, to flag *italicised words* the following entry will surround the words with EM tags.
/\B\*([a-z][a-z -]*[a-z])\*\B/ -hi-> <EM>$1</EM>
The i flag
This turns on ignore case in the pattern matching.
The e flag
This turns on execute in the pattern substitution. This really only makes sense if h is turned on too. In that case, the ``result'' part of the entry is taken as perl code to be executed, and the result of that code is what replaces the pattern.
The o flag
This marks the entry as a once-only link. This will convert the first instance of a matching pattern, and ignore any others further on.
For example, the following pattern will take the first mention of HTML::TextToHTML and convert it to a link to the module's home page.
"HTML::TextToHTML" -io-> http://www.katspace.com/tools/text_to_html/
For the most part, this module tries to use intuitive conventions for determining the structure of the text input. Unordered lists are marked by bullets; ordered lists are marked by numbers or letters; in either case, an increase in indentation marks a sub-list contained in the outer list.
Headers (apart from custom headers) are distinguished by ``underlines'' underneath them; headers in all-capitals are distinguished from those in mixed case. All headers, both normal and custom headers, are expected to start at the first line in a ``paragraph''.
In other words, the following is a header:
I am Head Man
-------------
But the following does not have a header:
I am not a head Man, man
I am Head Man
-------------
Tables require a more rigid convention. A table must be marked as a separate paragraph, that is, it must be surrounded by blank lines. Tables come in different types. For a table to be parsed, its --table_type option must be on, and the --make_tables option must be true.
ALIGN Table Type
Columns must be separated by two or more spaces (this prevents accidental incorrect recognition of a paragraph where interword spaces happen to line up). If there are two or more rows in a paragraph and all rows share the same set of (two or more) columns, the paragraph is assumed to be a table. For example
-e File exists.
-z File has zero size.
-s File has nonzero size (returns size).
becomes
<TABLE>
<TR><TD>-e</TD><TD>File exists.</TD></TR>
<TR><TD>-z</TD><TD>File has zero size.</TD></TR>
<TR><TD>-s</TD><TD>File has nonzero size (returns size).</TD></TR>
</TABLE>
This guesses for each column whether it is intended to be left, centre or right aligned.
BORDER Table Type
This table type has nice borders around it, and will be rendered with a border, like so:
+---------+---------+
| Column1 | Column2 |
+---------+---------+
| val1 | val2 |
| val3 | val3 |
+---------+---------+
The above becomes
<TABLE border="1">
<THEAD><TR><TH>Column1</TH><TH>Column2</TH></TR></THEAD>
<TBODY>
<TR><TD>val1</TD><TD>val2</TD></TR>
<TR><TD>val3</TD><TD>val3</TD></TR>
</TBODY>
</TABLE>
It can also have an optional caption at the start.
My Caption
+---------+---------+
| Column1 | Column2 |
+---------+---------+
| val1 | val2 |
| val3 | val3 |
+---------+---------+
PGSQL Table Type
This format of table is what one gets from the output of a Postgresql query.
Column1 | Column2
---------+---------
val1 | val2
val3 | val3
(2 rows)
This can also have an optional caption at the start. This table is also rendered with a border and table-headers like the BORDER type.
DELIM Table Type
This table type is delimited by non-alphanumeric characters, and has to have at least two rows and two columns before it's recognised as a table.
This one is delimited by the '| character:
| val1 | val2 |
| val3 | val3 |
But one can use almost any suitable character such as : # $ % + and so on. This is clever enough to figure out what you are using as the delimiter if you have your data set up like a table. Note that the line has to both begin and end with the delimiter, as well as using it to separate values.
This can also have an optional caption at the start.
use HTML::TextToHTML;
=head2 Create a new object
my $conv = new HTML::TextToHTML();
my $conv = new HTML::TextToHTML(title=>"Wonderful Things",
system_link_dict=>$my_link_file,
);
my $conv = new HTML::TextToHTML(\@ARGV);
$conv->args(short_line_length=>60,
preformat_trigger_lines=>4,
caps_tag=>"strong",
);
$conv->txt2html(infile=>[$text_file],
outfile=>$html_file,
title=>"Wonderful Things",
mail=>1
);
$conv->txt2html(["--file", $text_file,
"--outfile", $html_file,
"--title", "Wonderful Things",
"--mail"
]);
Tell me about them.
HTML::TextToHTML requires Perl 5.005_03 or later.
It also requires Data::Dumper (only for debugging purposes)
run_txt2html
Kathryn Andersen, <http//www.katspace.com> 2002,2003 Original txt2html script copyright (C) 2000 Seth Golub <seth AT aigeek.com>
the perl manpage. the txt2html manpage. Data::Dumper
| HTML::TextToHTML - convert plain text file to HTML |