Transforming RFC7749-formatted XML through XSLT

rfc2629.xslt supports two kind of extension elements, using different XML namespaces.¶

The first set contains (hopefully) generally useful extensions, see Section 11.¶

The second set is used for change and issue tracking and currently is not documented here. Please email the author in case you're interested in using these extensions.¶

The following XSLT engines are believed to work well: ¶

• Windows: MSXML3 and MSXML4 (<http://msdn.microsoft.com/xml>; command line processor "msxsl" is available from Microsoft Download Center)
• Java: Saxon (<http://saxon.sourceforge.net/>)
• Java: Xalan (<http://xml.apache.org/xalan-j/>)
• C/C++: xsltproc (libxslt) (<http://xmlsoft.org/XSLT/>, make sure that you have a current version, also: does not support inclusion of "https" resources -- <https://bugzilla.gnome.org/show_bug.cgi?id=791491>)

The following browsers seem to work fine: ¶

• Internet Explorer 5.5 (Windows version, if MSXML3 is installed)

• Internet Explorer 6 and newer

• Firefox 3.0 and newer

  • Be aware that XSLT execution can be suppressed using NoScript
  • There seems to be a new problem in Firefox 4 where it occasionally does the initial rendering with the wrong width (people who can reproduce this problem please comment on <https://bugzilla.mozilla.org/show_bug.cgi?id=640390>.
  • Date computation is available in Firefox starting with Firefox 6 (see <https://bugzilla.mozilla.org/show_bug.cgi?id=603159>)

• Safari 3 (starting with version 3.0.4)

  • Date computation not available (see <https://bugs.webkit.org/show_bug.cgi?id=4079>)

• Google Chrome

  • Date computation not available (see <https://bugs.webkit.org/show_bug.cgi?id=4079>)

• Opera (starting with version 10)

  • Date computation not available

Note that browsers in general do not load external DTDs nor external entities (see, for instance, Mozilla Bug 22942) thus entities like &nbsp; need to be declared in the internal subset (Appendix C.1).¶

The following browsers are known not to work properly: ¶

• Firefox 1.*/2.*: (missing extension function - see change request at Mozilla BugZilla 193678)

• Opera 9.21: execution fails, potentially to a somewhat complex XPath expression (reported to Opera as bug 245725).

• Opera 9.5 and 9.6: transformation appears to work, but CSS isn't getting applied (reported to Opera as bug 337388 on 2008-06-12).

• Safari 2.* supports client-side XSLT as of MacOS X 10.4, but misses required extension functions. A problem with stylesheets producing non-ASCII output (such as NBSP characters) has been fixed as of OSX 10.4.4. Both problems have been reported through Apple's bug tracking system, see <http://drakken.dbc.mtview.ca.us/pipermail/xml2rfc/2005-May/002073.html> and <http://bugs.webkit.org/show_bug.cgi?id=4079>.

The transformation result is supposed to conform to HTML5 [HTML5]. This can be checked using the W3C's online validator at <http://validator.w3.org>.¶

XSLT 1.0 is not capable to directly emit the HTML doctype declaration, thus uses the SYSTEM ID "about:legacy-compat" instead (see Section 8.1.1 of [HTML5]).¶

with¶

LINK elements exist since HTML 2.0. They can be used to embed content-independant links inside the document. Unfortunately, only few user agents support this element. Firefox users may want to check the Link Widgets extension.¶

The figure below shows how Mozilla Firefox 1.0 displays the Site Navigation Bar for rfc2396.xml.

Unless turned off using the "rfc-ext support-rfc2731" processing instruction, the transformation will generate metadata according to [RFC2731] and [DC-HTML].¶

RFCs are immutable; once published, they do not change anymore. What does change though is their status, their relation to subsequent RFCs (such as when they are updated), and errata.¶

rfc2629toXHTML.xslt by default inserts code that will pull the relevant information from <https://tools.ietf.org>. This can be disabled by specifying the parameter "xml2rfc-ext-insert-metadata=no" (or by inserting the equivalent processing instruction into the source code).¶

An example for the generated information can be seen at <https://greenbytes.de/tech/webdav/rfc2616.html#rfc.meta>.¶

Unfortunately, the RFC Editor does not provide errata information in a well-defined machine readable format. What's available is "regular" HTML (and that could be the worst currently in use in standards bodies...).¶

parse-errata.xslt attempts to parse useful information out of these pages.¶

The code tries to make sense of the HTML, in particular it tries to detect what RFC sections each erratum applies to. The resulting XML format is work-in-progress and just contains the information that will be useful in subsequent formatting of the RFC.¶

When formatting the RFC for HTML output, the errata file can be passed as stylesheet parameter ("xml2rfc-ext-errata"). The output will include errata links at the beginnings of the section they apply to, or at the beginning of Section 1 when the location is unknown.¶

• "held for update": reviewed and decided to be "held for document update" (these are mostly editorial); "scissors" symbol.
• "submitted": submitted, but not reviewed; "letter" symbol.
• "verified": reviewed and found to be correct; "checkmark" symbol.

To recap: the errata information is passed into the transformation as additional parameter. The errata information will not be automatically retrieved from the RFC Editor web site.¶

Finally, here's an example for inserted errata links: <https://greenbytes.de/tech/webdav/rfc7230.html#transfer.codings>.¶

Transformation to XSL-FO [XSL-FO] format is available through rfc2629toFO.xslt (which includes rfc2629.xslt, so keep both in the same folder).¶

Compared to HTML user agents, XSL-FO engines unfortunately either come as open source (for instance, Apache FOP) or feature-complete (for instance, AntennaHouse XSL Formatter), but not both at the same time.¶

As Apache FOP needs special workarounds (index generation), and some popular extensions aren't standardized yet, the translation produces a generic output (hopefully) conforming to [XSL-FO]. Specific backends (xsl11toFop.xslt, xsl11toXep.xslt, xsl11toAn.xslt) then provide post-processing for the individual processors.¶

PDF output can also be produced directly from (X)HTML. One simple approach is to rely on the browser's printing function, and to use a printer driver that produces PDF. Depending on the brower's CSS capabilities, the output will behave properly with respect to table breaks etc.¶

An alternative is PrinceXML (see <http://www.princexml.com/>), which can produce PDF directly from (X)HTML input, based on the CSS printing information.¶

For instance, PDF output with text justification turned on can be produced with:

saxon input.xml rfc2629.xslt > output.html
prince output.html output.pdf

Converts the contained quoted string into a hex-encoded character sequence, for use in case-sensitive ABNF productions.¶

For instance, "<x:abnf-char-sequence>"HTTP"</x:abnf-char-sequence>" gets converted to "%x48.54.54.50".¶

Using its "value" attribute, this element allows the definition of an internal link target alias for the enclosing element. This alias can then be used with the <ref> element for intra-document references.¶

Note that the anchor alias is not subject to the naming constraints that apply to anchor elements (which are XML names).¶

This element can be used to include a contributor's contact information in place where a paragraph (<t>) would be allowed otherwise.¶

This element marks the content as being one of the normative keywords defined in [RFC2119].¶

The DOCTYPE definition below allows using these keywords using XML entity expansion: such as in "...server &MUST; accept...".

<!DOCTYPE rfc [
 <!ENTITY MAY "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'
   >MAY</bcp14>">
 <!ENTITY MUST "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'
   >MUST</bcp14>">
 <!ENTITY MUST-NOT "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'
   >MUST NOT</bcp14>">
 <!ENTITY OPTIONAL "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'
   >OPTIONAL</bcp14>">
 <!ENTITY RECOMMENDED "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'
   >RECOMMENDED</bcp14>">
 <!ENTITY REQUIRED "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'
   >REQUIRED</bcp14>">
 <!ENTITY SHALL "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'
   >SHALL</bcp14>">
 <!ENTITY SHALL-NOT "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'
   >SHALL NOT</bcp14>">
 <!ENTITY SHOULD "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'
   >SHOULD</bcp14>">
 <!ENTITY SHOULD-NOT "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'
   >SHOULD NOT</bcp14>">]>

Marking up a string as <bb> indicates that it represents the bottom line of a box drawing, replacing the "+" and "-" characters accordingly.¶

Marking up a string as <bc> indicates that it represents a center line of a box drawing, replacing the "|" character accordingly.¶

This element is like the <blockquote> element defined in Section 4.4.4 of [HTML5] (note this is a block-level element!). It should contain one or more <t> child elements.¶

Marking up a string as <bt> indicates that it represents the top line of a box drawing, replacing the "+" and "-" characters accordingly.¶

Contains mapping information for a single Unicode code points. Attributes are "c" (the actual character), "n" (the code point), and "d" (the name/description).¶

(The format is deliberately terse so that the size of a mapping file containing the whole Unicode character database is minimized).¶

This element is like the <dfn> element defined in Section 4.5.8 of [HTML5].¶

where "docname" is substituted by the document name, "section" is substituted by section title (number and name), and "ref" is substituted by a URI pointing to the section being referenced.¶

This element is like the "h" element in [XHTML2].¶

Used to highlight text passages, currently only allowed in <artwork>.¶

Note: this is stripped when generating input for xml2rfc, so please use with care.¶

This element can be used to insert the length of another formatted section (in decimal).¶

Example: computing the Content-Length header value

<artwork>
...
Content-Length: <x:length-of target="req"/>
  
<x:span anchor="req">123456789
<x:span><artwork/>

The lenght computation counts line ends as two characters (CRLF).¶

Note that indentation characters in artwork will be counted. The "indented" attribute allows to specify the amount of indentation to be substracted from the computed length.¶

This element can be added as a top-level child element below <rfc> to indicate additional link information. It's currently used only when generating HTML output, in which case an HTML <link> element with identical attributes gets generated.¶

Example: generating HTML link element

<x:link xmlns:x="http://purl.org/net/xml2rfc/ext"
    rel="Bookmark"
    title="IETF WEBDAV Working Group"
    href="http://ftp.ics.uci.edu/pub/ietf/webdav/"/>

If the attribute "basename" is present, it is used to compute the target href based on the output format being generated (this is handy for "next"/"prev" links in a series of documents. In this case, the href attribute is not required.¶

For instance:

<x:link xmlns:x="http://purl.org/net/xml2rfc/ext"
    rel="next"
    title="Part2"
    basename="draft-foobar-protocol-p2-latest"/>

Used for grouping multiple <t> elements into a single list item.¶

Can be used to add a note, usually indented by a few characters. It should contain one or more <t> child elements.¶

This element instructs the processor to parse the contents as XML and to warn when there's a problem (requires either MSXML or Saxon8 or newer).¶

This element can be used inside <reference> to add plain text (before the date, when present).¶

See also <refcontent> (Section 12.16). ¶

This element is like the <q> element defined in Section 4.5.7 of [HTML5].¶

This element is a simplified variant of the <xref> element, in that no "target" attribute needs to be specified, instead the text contents acts as identifier. That in itself wouldn't be terribly useful, but together with the <anchor-alias>, it allows referring to other parts of the document with minimal additional markup.¶

For instance, given an alias definition such as¶

the following simple references¶

Can be used to enhance a <reference> with information about the location for the XML source. This can be used by the <xref> processing code to automatically extract the target section number.¶

For example:

...
<xref target="RFC2616" x:fmt="of" x:rel="#PUT" />
...

<reference target="RFC2616"/>
  ...
  <x:source href="rfc2616.xml"/>
  ...
    

This element is like the <sup> element in Section 4.5.16 of [HTML5].¶

Note: the down conversion to RFC7749 format replaces "x^y" by "x^y".¶

See also <sup> (Section 12.24). ¶

Needed for the lookup of Unicode character database information; this element can either appear in-band in the source document, or off-band as specified using the xml2rfc-ext-ucd-file parameter/PI.¶

Contains multiple <x:c> elements (see Section 11.9).¶

The utility XSLT convert-ucd.xslt can be used to create a mapping file based on the Unicode XML database format, as available from <https://unicode.org/ucd/#UCDinXML>.¶

Sometimes, artwork occurs inside lists. To get it indent properly in xml2rfc's text output, it needs to be indented in the source. This is sub-optimal, as this whitespace will also appear in the HTML output, where it's already indented due to HTML's semantics. As a workaround, a "x:indent-with" attribute can be specified, containing a string that will be prepended to each line when clean-for-DTD.xslt is run (see Section 13.4).¶

Furthermore, documents can contain code that might need to be marked as "code component" (<http://www.ietf.org/iesg/statement/copyright.html>). This can be done using "x:is-code-component".¶

Finally, when allowing pretty-printing of code (see "html-pretty-print" in Section 3.3, the "x:lang" attribute can used to explicitly opt into pretty-printing. If the pretty printer can figure out the code type without assistance, an empty value will be sufficient. Otherwise, the language can be specified in the attribute (such as "html"), which will in turn be coded into the CSS class as "lang-" concatenated with the language name.¶

The attribute "x:include-day" ("true", "false") can be used to control whether the day-of-month should be included in the output. This can be used to adjust the rendering of dates for April-1st-RFCs to the desired special format, including the "1".¶

In particular: on the front page, the attribute defaults to "true", unless when generating RFCs. The output format also varies depending on RFC-ness: it's "day month year" for RFCs, but "month day, year" otherwise. In references, the attribute defaults to "false", and when set to true, the day is just inserted in front of the month.¶

The extension attribute below is allowed on the standard <iref> element: ¶

• x:for-anchor specifies that the <iref> will also be automatically inserted whenever the specified anchor is cross-referenced -- this may save entering lots of <iref> instances. As a special case, a value of "" (empty string) refers to the anchor attribute of the closest ancestor.

The extension attribute below is allowed on the standard <list> element: ¶

• x:indent specifies the amount of indentation for list items in hanging lists. This can be useful when the output format, such as XSL-FO, does not support automatical formatting. The value takes an XSL-FO width, such as "5em". The default is length of longest label in characters times 0.8em.

Also, the <list> element can take <x:lt> child elements instead of <t>, allowing to insert multiple paragraphs into a single list item.¶

The extension attributes below are allowed on the standard <rfc> element: ¶

• grddl:transformation can be used to reference a GRDDL transform.
• x:maturity-level can be used to specify the IETF Standards Track Maturity Level of "proposed", "draft" or "internet" (see Section 4.1 of [RFC2026]).

The extension attribute below is allowed on the standard <list> element: ¶

• x:fixed-section-number can be used to specify a fixed section number. This can be useful when formatting historic documents that used a different numbering style.

Three extension attributes are allowed on the standard <xref> element: ¶

1. x:sec can be specified to point to a specific section of the referenced document,
2. x:rel may specify a relative reference to use when linking into the referenced document (if linking by section number is not available),
3. x:fmt defines the text format to be used.

, (Comma)

[reference], Section sec

()

[reference] (Section sec)

of

Section sec of [reference]

number

sec

none

No output (can be used to have xrefs to references without having them rendered as such)

sec

Section sec

These extensions are currently only supported for <xref> elements without child nodes.¶

If the processor knows how to reference the target section, it will generate a link directly to the target section, such as in [RFC2119], Section 5.¶

See Section 3.1.1 of [V3IMPNOTES].¶

See Section 2.6 of [RFC7991bis].¶

See Section 2.9 of [RFC7991bis].¶

See Section 2.10 of [RFC7991bis].¶

See Section 2.10 of [RFC7991bis].¶

See Section 2.18 of [RFC7991bis].¶

See Section 2.17 of [RFC7991bis].¶

See Section 2.19 of [RFC7991bis].¶

See Section 2.20 of [RFC7991bis].¶

See Section 2.21 of [RFC7991bis].¶

See Section 2.28 of [RFC7991bis].¶

See Section 2.29 of [RFC7991bis].¶

See Section 2.31 of [RFC7991bis]. Currently only supported inside <references> and <section>.¶

See Section 2.33 of [RFC7991bis].¶

See Section 2.37 of [RFC7991bis].¶

See Section 2.38 of [RFC7991bis].¶

See Section 2.40 of [RFC7991bis].¶

See Section 2.43 of [RFC7991bis].¶

EXPERIMENTAL, might be removed again - see <https://github.com/rfc-format/draft-iab-xml2rfc-v3-bis/issues/26>. ¶

See Section 2.47 of [RFC7991bis].¶

See Section 2.39 of [RFC7991bis].¶

See Section 2.45 of [RFC7991bis].¶

See Section 2.49 of [RFC7991bis].¶

See Section 2.50 of [RFC7991bis].¶

See Section 2.51 of [RFC7991bis].¶

See Section 4 of [RFC7991bis].¶

See Section 2.53 of [RFC7991bis].¶

See Section 2.54 of [RFC7991bis].¶

See Section 2.55 of [RFC7991bis].¶

See Section 2.56 of [RFC7991bis].¶

See Section 2.57 of [RFC7991bis].¶

See Section 2.58 of [RFC7991bis].¶

See Section 2.60 of [RFC7991bis].¶

See Section 2.61 of [RFC7991bis].¶

See Appendix A.1 of [V3IMPNOTES].¶

1. When in XSLT 1.0: to compute the Unicode code point of a given character (when non-ASCII),
2. In any case: to lookup the Unicode name of a character.

The mapping information can be supplied inline using the <x:u-map> element, or in an external file. See Section 11.24 for details.¶

See Section 2.62 of [RFC7991bis].¶

See Appendix B.1 of [RFC7991bis].¶

EXPERIMENTAL and INCOMPLETE - only supported as a child element of <references>, and only supporting parse type "XML". ¶

check-references.xslt can be used to check all references to RFC- and ID-series IETF publications and to W3C publications (note this script requires local copies of <ftp://ftp.isi.edu/in-notes/rfc-index.xml> and <http://www.w3.org/2002/01/tr-automation/tr.rdf> and will use the XML status information provided at <http://tools.ietf.org/>).¶

If the document is supposed to be published on the IETF standards track, the desired level can be specified using the parameter intended-level as 'proposed', 'draft' or 'internet'. Alternatively, it can be specified inside the document using the attribute x:maturity-level on the <rfc> element (see Section 11.29).¶

Note: Downward references should be annotated using the <annotate> element, containing an <xref> to [BCP97].¶

When an XSLT 2.0 processor is used, links in the document can be checked as well using the link-check parameter ('yes' or 'no'). Note that this only works for http links to documents of type text/*.¶

For instance, as of 2008-07-12, the script produces for <https://greenbytes.de/tech/webdav/rfc2518.xml>:

> saxon rfc2518.xml check-references.xslt intended-status=PROPOSED \
  link-check=yes

Normative References:
ISO-11578: not checked
ISO-639: not checked
ISO-8601: not checked
REC-xml-19980210: [FirstEdition] obsoleted by REC-xml-20001006
REC-xml-names-19990114: [FirstEdition] obsoleted by
 REC-xml-names-20060816
RFC1766: [PROPOSED STANDARD] obsoleted by RFC3066 RFC3282
RFC2068: [PROPOSED STANDARD] obsoleted by RFC2616
RFC2069: [PROPOSED STANDARD] obsoleted by RFC2617
RFC2119: [BEST CURRENT PRACTICE] (-> BCP0014) ok
RFC2141: [PROPOSED STANDARD] ok
RFC2277: [BEST CURRENT PRACTICE] (-> BCP0018) ok
RFC2396: [DRAFT STANDARD] obsoleted by RFC3986
RFC2279: [DRAFT STANDARD] obsoleted by RFC3629

Informational References:
REC-PICS-labels-961031: [REC] ok
RFC1807: [INFORMATIONAL] ok
RFC2026: [BEST CURRENT PRACTICE] (-> BCP0009) ok
RFC2291: [INFORMATIONAL] ok
RFC2376: [INFORMATIONAL] obsoleted by RFC3023
RFC2413: [INFORMATIONAL] obsoleted by RFC5013
USMARC: not checked
WF: not checked

Link Targets
<http://www.w3.org/TR/1998/REC-xml-19980210>: ok
<http://www.w3.org/TR/1999/REC-xml-names-19990114>: ok
<http://www.dlib.org/dlib/july96/lagoze/07lagoze.html>: ok
<http://www.w3.org/pub/WWW/TR/REC-PICS-labels-961031.html>: ok

Recognized formats in the <seriesInfo> element are: ¶

• for RFCs, the name attribute must be "RFC", and the value attribute must be the number of the RFC,
• for Internet Drafs, the name attribute must be "ID" or "Internet-Draft", and the value attribute must be the file name of the draft (including the two-digit running number, but excluding a file extension),
• for W3C documents, the name attribute must be "W3C", must start with "W3C ", or must start with "World Wide Web Consortium ", and the value attribute must be the "shorthand" name of the specification, such as "REC-xml-19980210".

gen-reference-graph.xslt generates a graph of RFC dependencies, using the same base data as in check-references.xslt (see Section 13.1). Its output is a "dot" file, to be processed by GraphViz (see <http://www.graphviz.org/>).¶

The picture below shows the RFC dependencies in RFC2629.

amazon-asin.xslt uses the Amazon web services to generate a <reference> element for a given ASIN (ISBN).¶

For instance:

<?xml version="1.0" encoding="utf-8"?>
<references>
 <reference target="urn:isbn:0134516591">
   <front>
     <title>Simple Book, The: An Introduction to Internet Management,
               Revised Second Edition</title>
     <author surname="Rose"
                fullname="Marshall T. Rose" initials="M. T. ">
       <organization/>
     </author>
     <author surname="Marshall"
                fullname="Rose T. Marshall" initials="R. T.">
       <organization/>
     </author>
     <date year="1996" month="March"/>
   </front>
   <seriesInfo name="Prentice Hall" value=""/>
 </reference>
</references>

Note that the resulting XML usually requires checking, in this case Amazon's database is playing tricks with Marshall's name...¶

clean-for-DTD.xslt can be used to down-convert some extensions to a format that is supported by the existing xml2rfc formatters, mainly for the purpose of generating plain-text output. Note that these extensions are experimental (feedback appreciated).¶

The following mappings are done: ¶

• <iref> elements inside <artwork> elements are moved in front of the enclosing <figure> element.
• <xref> elements inside <artwork> are expanded just like in regular text (that is, the markup is stripped, but the element is replaced by the applicable replacement text).
• <x:anchor-alias> elements get stripped.
• <x:bcp14> elements get stripped.
• <x:bb>, <x:bc> and <x:bt> elements get stripped.
• <x:blockquote> elements get converted to indented text (through a <list> element).
• <x:dfn> elements get stripped.
• <x:h> elements get stripped.
• <x:link> elements get stripped.
• <x:lt> elements get collapsed into a single <lt> element with added <vspace> added to simulate paragraph breaks.
• <x:note> elements get converted to indented text (through a <list> element).
• <x:q> elements get stripped, with apostrophes added around the text.
• <x:prose> elements are transformed into <seriesInfo> elements (which is an abuse of the element and only a workaround until xml2rfc gets a matching extension).
• <x:ref> elements get replaced by <xref> elements, targetting either the anchor or another anchor with matching <x:anchor-alias> child element.

As the output formatters evolve to support the V3 format (proposed in [RFC7991bis]), clean-for-DTD.xslt will start taking advantage of these changes. Right now, it supports three modes, one of which being used for the historic TCL processor, and the other ones being used with xml2rfc 2.5.2 and xml2rfc 2.6.0 (see https://pypi.python.org/pypi/xml2rfc/).¶

201510

For xml2rfc.tcl and versions of the Python formatter before version 2.5.2.

201610

Maps the V3 extension for unnumbered section numbers to something the 2.5.2 version of the Python formatter understands (but note https://trac.tools.ietf.org/tools/xml2rfc/trac/ticket/313).

201706

Copies the V3 extension for unnumbered section numbers to the output.

With extract-artwork.xslt, artwork elements named through the "name" attribute can be extracted. This can be used to automatically check their syntax (for instance, when ABNFs appear within a figure element).¶

For instance:

saxon rfc3986.xml extract-artwork.xslt name=uri.abnf

rfc2629grddl.xslt extracts RDF information. This is experimental work-in-progress. See <http://www.w3.org/TR/grddl/> for more information.¶

Experimental ¶

The "HTML Live Refresh" mode allows to run a text editor and a browser side-by-side, with the browser auto-updating every few seconds, displaying the updated HTML, and automatically navigating to the part of the page that changed last.¶

1. A browser that supports the DOMParser and XSLTProcessor APIs.
2. The ability to reload the source code and the XSLT code from within Javascript; in some browsers this is forbidden for "file:" URIs due to perceived security problems.

1. Mozilla Firefox (where, to be able to reload from the local filesystem, Firefox needs to be configured with security.fileuri.strict_origin_policy set to false in about:config, see Mozilla documentation), and
2. Google Chome (where, to be able to reload from the local filesystem, Chrome needs to be started with the command line option --allow-file-access-from-files).

The optional parameters refresh-interval and refresh-xslt support changing the refresh interval (in seconds, defaulting to 10), and the name of the XSLT file to use (defaulting to "rfc2629.xslt").¶

To test this feature, start with a minimal source file like this:

<?xml-stylesheet type='text/xsl' href='rfc2629.xslt'?>
<?rfc-ext refresh-from="draft-foo-bar-00.xml"?> 
<rfc docName="draft-foo-bar-00" ipr="trust200902">
  <front>
    <title>Title Goes Here</title>
    <abstract>
      <t>Abstract</t>
    </abstract>
  </front>
</rfc>

...open it in both text editor and browser, start editing and of course ocasionally save. And don't forget to put rfc2629.xslt into the same folder.

There are many methods for automatic inclusion of material in the XML source, such as the "include" processing instruction (see Section 3.1), external entities (Appendix C.1), or XInclude. In general, those share a common problem: the XML source file isn't self-contained, which makes it harder to submit it as Internet Draft.¶

The tool refresh-inclusions.sh does in-place replacement: it scans the source file for inclusion directives (expressed as XML processing instructions), and refreshes the included text with data from an external file. It will not modify the source file unless included material did actually change. When it does modify the source file, it will copy the original source to a backup file.¶

<?BEGINESCAPEDINC refresh-inclusions.sh ?>
...
<?ENDSCAPEDINC refresh-inclusions.sh ?>

(note that the SP character at the end of the directive is significant)

refresh-inclusions.sh:

#!/bin/sh
# Refresh file inclusions based on XML processing instructions
# 
# Copyright (c) 2006-2016, Julian Reschke (julian.reschke@greenbytes.de)
# All rights reserved.
# 
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
# 
# * Redistributions of source code must retain the above copyright notice,
#   this list of conditions and the following disclaimer.
# * Redistributions in binary form must reproduce the above copyright notice,
#   this list of conditions and the following disclaimer in the documentation
#   and/or other materials provided with the distribution.
# * Neither the name of Julian Reschke nor the names of its contributors
#   may be used to endorse or promote products derived from this software
#   without specific prior written permission.
# 
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
# ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
# LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
# CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
# SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
# INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
# CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
# ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
# POSSIBILITY OF SUCH DAMAGE.

expand() {

  # remember whether we started with CRLF (assumes that we have dos2unix)
  CRLF=$(dos2unix -ic "$1" 2>/dev/null | tr -d ' ')
  
  cat "$1" | awk ' 

  function filecontents(filename) {
    while (getline < filename > 0) {
      fc[filename] = fc[filename] $0 "\n"
    }
    return fc[filename]
  }

  BEGIN {
    includefile = "";
    includeescapedfile = "";
  }
  
  # start include (verbatim mode)
  /<\?BEGININC .* \?>$/ {
    print
    keyword = "<?BEGININC " 
    extract = match($0, /<\?BEGININC .* \?>$/)
    includefile = substr($0, RSTART + length(keyword),
                    RLENGTH - 3 - length(keyword))
    output = filecontents(includefile)
    printf("%s", output)
  }
  
  # start include (escape-for-XML mode)
  /<\?BEGINESCAPEDINC .* \?>$/ {
    print
    keyword = "<?BEGINESCAPEDINC " 
    extract = match($0, /<\?BEGINESCAPEDINC .* \?>$/)
    includeescapedfile = substr($0, RSTART + length(keyword),
                           RLENGTH - 3 - length(keyword))
    output = filecontents(includeescapedfile)
    # escape ampersand, less-than, and greater-than
    # when part of a CDATA end marker
    gsub(/&/, "\\&amp;", output)
    gsub(/</, "\\&lt;", output)
    gsub(/]]>/, "]]\\&gt;", output)
    printf("%s", output)
  }

  # end include (verbatim mode)
  /^<\?ENDINC .* \?>/ {
    if ($2 != includefile) {
      printf ("unexpected ENDINC, got %s but expected %s\n", $2,
        includefile) >> "/dev/stderr"
    }
    includefile = "";
  }
  
  # end include (escape-for-XML mode)
  /^<\?ENDESCAPEDINC .* \?>/ {
    if ($2 != includeescapedfile) {
      printf ("unexpected ENDESCAPEDINC, got %s but expected %s\n", $2,
        includeescapedfile) >> "/dev/stderr"
    }
    includeescapedfile = "";
  }

  #default
  {
    if (includefile == "" && includeescapedfile == "") {
      print
    }
  }
  
  END {
    if (includefile != "") {
      printf ("missing ENDINC for %s\n",
        includefile) >> "/dev/stderr"
    }
    if (includeescapedfile != "") {
      printf ("missing ENDESCAPEDINC for %s\n",
        includeescapedfile) >> "/dev/stderr"
    }
  }
  
  ' > $$
  
  # restore CRLF if needed
  if [ -n "$CRLF" ]; then
    FNN=$(echo "$1" | tr -d ' ')
    [ "$FNN" = "$CRLF" ] && unix2dos -q $$
  fi
  
  # check for changes
  cmp -s "$1" $$ || (
    cp -v "$1" "$1".ri.bak
    cp $$ "$1"
    echo "$1" updated )
  
  rm -f $$
}

[ $# -ne 0 ] || ( echo "refresh-inclusions.sh file..." >&2 ; exit 2 )

for i in $*
do
  expand $i
done

<?xml version="1.0"?>
<?xml-stylesheet type='text/xsl' href='rfc2629.xslt' ?>

<!DOCTYPE rfc SYSTEM "rfc2629.dtd" [

  <!-- use "&MAY;" for a BCP 14 "MAY", see Section 11.4 -->
  <!ENTITY MAY
  "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'>MAY</bcp14>">

  <!-- re-declare "&nbsp;" as code point 160 (non-breaking space) -->
  <!-- you may need this for UAs that do not read external DTDs -->
  <!ENTITY nbsp
  "&#160;">

  <!-- allow later RFC2616 reference using "&rfc2616;" -->
  <!-- the data will be fetched from xml.resource.org -->
  <!ENTITY rfc2616 SYSTEM
  "http://xml.resource.org/public/rfc/bibxml/reference.RFC.2616.xml">

  <!-- allow a custom reference using "&mydraft;" -->
  <!-- the data will be fetched from the same location as the 
       source file -->
  <!ENTITY mydraft SYSTEM "reference.mydraft.xml">
]>

Note: including entities from a remote site will not work in browsers due to the Same-Origin policy.

The XSLT code can be customized by creating a custom XSLT file that uses <xsl:import> to include the original code, and just overrides particular rules.¶

For instance, the code below overrides several attributes in rfc2629toFO.xslt, changing the color, spacing and font family for headers.

<xsl:transform xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
               version="1.0">

  <xsl:import href="rfc2629toFO.xslt"/>
  
  <xsl:attribute-set name="h1">
    <xsl:attribute name="color">darkblue</xsl:attribute>
    <xsl:attribute name="font-family">sans-serif</xsl:attribute>
    <xsl:attribute name="space-before">24pt</xsl:attribute>
  </xsl:attribute-set>
  
  <xsl:attribute-set name="h2">
    <xsl:attribute name="color">darkblue</xsl:attribute>
    <xsl:attribute name="font-family">sans-serif</xsl:attribute>
    <xsl:attribute name="space-before">18pt</xsl:attribute>
    <xsl:attribute name="space-after">3pt</xsl:attribute>
  </xsl:attribute-set>
  
  <xsl:attribute-set name="h3">
    <xsl:attribute name="color">darkblue</xsl:attribute>
    <xsl:attribute name="font-family">sans-serif</xsl:attribute>
    <xsl:attribute name="space-before">16pt</xsl:attribute>
    <xsl:attribute name="space-after">2pt</xsl:attribute>
  </xsl:attribute-set>

</xsl:transform>