hxindex - insert an index into an HTML document



hxindex − insert an index into an HTML document


hxindex [−t] [−x] [−n|−N] [−f] [−r] [−c class[,class...]] [−b base] [−i indexdb] [−s template] [−u phrase] [−O element[,element...]] [−X element[,element...]] [--] [file−or−URL]


The hxindex looks for terms to be indexed in a document, collects them, turns them into target anchors and creates a sorted index as an HTML list, which is inserted at the place of a placeholder in the document. The resulting document is written to standard output.

The index is inserted at the place of a comment of the form


or between two comments of the form


In the latter case, all existing content between the two comments is removed first.

Index terms are either elements of type <dfn> or elements with a class attribute of "index". (For backward compatibility, also class attributes "index-inst" and "index-def" are recognized.) <dfn> elements (and class "index-def") are considered more important than elements with class "index" and will appear in bold in the generated index.

The option −c adds additional classes, that are aliases for "index".

By default, the contents of the element are taken as the index term. Here are two examples of occurrences of the index term "shoe":

A <dfn>shoe</dfn> is a piece of clothing that...
completed by a leather <span class="index">shoe</span>...

If the term to be indexed is not equal to the contents of the element, the title attribute can be used to give the correct term:

... <dfn title="shoe">Shoes</dfn> are pieces of clothing that...
... with two leather <span class="index" title="shoe">shoes</span>...

The title attribute must also be used when the index term is a subterm of another. Subterms appear indented in the index, under their head term. To define a subterm, use a title attribute with two exclamation marks ("!!") between the term and the subterm, like this:

<dfn title="shoe!!leather">...</dfn>
<dfn title="shoe!!invention of">...</dfn>
<em class="index" title="shoe!!protective!!steel nosed">...</em>

As the last example above shows, there can be multiple levels of sub-subterms.

The title attribute also allows multiple index terms to be associated with a single occurrence. The multiple terms are separated with a vertical bar ("|"). Compare the following examples with the ones above:

<dfn title="shoe|boot">...</dfn>
<dfn title="shoe!!invention of|inventions!!shoe">...</dfn>

These two elements both insert two terms into the index. Note that the second example above combines subterms and multiple terms.

It is possible to run index on a file that already has an index. The old target anchors and the old index will be removed before being re-generated.


The following options are supported:


By default, hxindex adds an ID attribute to the element that contains the occurrence of a term and also inserts an <a> element inside it with a name attribute equal to the ID. This is to allow old browsers that ignore ID attributes, such as Netscape 4, to find the target as well. The −t option suppresses the <a> element.


This option turns on XML syntax conventions: empty elements will end in /> instead of > as in HTML. −x implies −t.

−i indexdb

hxindex can read an initial index from a file and write the merged collection of index terms back to that file. This allows an index to span several documents. The −i option is used to give the name of the file that contains the index.

−b base

This option is useful in combination with −i to give the base URL reference of the document. By default, hxindex will store links to occurrences in the indexdb file in the form #anchor, but when −b is given, the links will look like base#anchor instead.

When used in combination with −n, the title attributes of the links will contain the title of the document that contains the term. The title is inserted before the template (see option −s) and separated from it with a comma and a space. E.g., if hxindex is called with

hxindex -i termdb -n -base myfile.html myfile.html

and the termdb already contains an entry for "foo" in in section "3.1" of a document called "file2.html" with title "The foos", then the generated index will contain an entry such as this:

foo, <a href="file2.html#foo"
title="The foos, section 3.1">3.1</a>

−c class,class,...

Normal index terms are recognized because they have a class of "index". The −c option adds additional, comma-separated class names that will be considered aliases for "index". E.g., −c instance will make sure that <span class="instance">term</span> is recognized as a term for the index.


By default, the index consists of links with "#" as the anchor text. Option −n causes the link text to consist of the section numbers of the sections in which the terms occur, falling back to "without number" (see option −u below) if no section number could be found. Section numbers are found by looking for the nearest preceding start tag with a class of "secno" or "no-num". In the case of "secno", the contents of that element are taken as the section number. In the case of "no-num" the section is assumed to have no number and hxindex will print "without number" instead. These classes are also used by hxnum(1), so it is useful to run hxindex after hxnum, e.g.,

hxnum myfile.html | hxindex −n >mynewfile.html


With this option, the anchor text of the links in the index is the full title of the section in which the term occurs. The title of the section is the nearest preceding H1, H2, H3, H4, H5 or H6 element, or the document’s title if there is no preceding H* element. This option cannot be used together with −n. If both are used, the last one specified wins.

−s template

When option −n is used, the link will have a title attribute and the template determines what it contains. The default is "section %s", where the %s is a placeholder for the section number. In other words, the index will contain entries like this:

term, <a href="#term" title="section 7.8">7.8</a>

Some examples:

hxindex -n -s ’chapter %s’
hxindex -n -s ’part %s’
hxindex -n -s ’hoofdstuk %s’ -u ’zonder nummer’

This option is only useful in combination with −n

−u phrase

When option −n is used to display section numbers, references for which no section number can be found are shown as phrase instead. The default is "??".

This option is only useful in combination with −n


Remove title attributes that were used for the index as well as the comments that delimit the inserted index. This avoids that browsers display these attributes. Note that hxindex cannot be run again on its own output if this option is used. (Mnemonic: "freeze" or "final".)


Do not ignore trailing punctuation when sorting index terms. E.g., if two terms are written as

<dfn>foo,</dfn>... <span class=index>foo</span>

hxindex will normally ignore the comma and treat them as the same term, but with −r, they are treated as different. This affects trailing commas (,), semicolons (;), colons (:), exclamations mark (!), question marks (?) and full stops (.). A final full stop is never ignored if there are two or more in the term, to protect abbreviations ("B.C.") and ellipsis ("more..."). This does not affect how the index term is printed (it is always printed as it appears in the text), only how it is compared to similar terms. (Mnemonic: "raw".)

−O element,element,...

If −O is present, only elements with the given names will be indexed. E.g.,

hxindex -O span,i,em

means that hxindex will only look for class="index" (and other classes, according to −c) on the elements span, i and em. The argument of −O must be a comma-separated list of element names. Note that this does not affect the element dfn. It will always be indexed as a defining instance.

−X element,element,...

The option −X excludes the given elements from being indexed. E.g.,

hxindex -X ul,ol

makes sure that ul and ol elements are not indexed, even if they have a class="index" attribute. This does not exclude their children from being indexed. E.g.,

<ul class=index>
<li class=index>foo
<li class=index>bar

will add foo and bar to the index, but not the whole content of the ul element (foo bar baz). If both −O and −X are given and an element occurs in both options, it will be excluded. E.g.,

hxindex -X p,h1,ul -O em,span,h1,h2

will cause hxindex to only look for class attributes on em, span and h2, because h1 is excluded.


The following operand is supported:

The name of an HTML or XML file or the URL of one. If absent, or if the file is "-", standard input is read instead.


The following exit values are returned:


Successful completion.


An error occurred in parsing the HTML file.


The input is assumed to be in UTF-8, but the current locale is used to determine the sorting order of the index terms. I.e., hxindex looks at the LANG, LC_ALL and/or LC_COLLATE environment variables. See locale(1).

To use a proxy to retrieve remote files, set the environment variables http_proxy or ftp_proxy. E.g., http_proxy="http://localhost:8080/"


Assumes UTF-8 as input. Doesn’t expand character entities (apart from the standard ones: "&amp;", "&lt;", "&gt" and "&quot"). Instead, pipe the input through hxunent(1) and, if needed, asc2xml(1) to convert it to UTF-8.

Remote files (specified with a URL) are currently only supported for HTTP. Password-protected files or files that depend on HTTP "cookies" are not handled. (You can use tools such as curl(1) or wget(1) to retrieve such files.)

The accessibility of an index, even when generated with option −n, is poor.


asc2xml(1), hxnormalize(1), hxnum(1), hxprune(1), hxtoc(1), hxunent(1), xml2asc(1), locale(1), UTF-8 (RFC 2279)

Updated 2024-01-29 - jenkler.se | uex.se