lisp/muse-docbook.el

   1 ;;; muse-docbook.el --- publish DocBook files
   2
   3 ;; Copyright (C) 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
   4
   5 ;; This file is part of Emacs Muse.  It is not part of GNU Emacs.
   6
   7 ;; Emacs Muse is free software; you can redistribute it and/or modify
   8 ;; it under the terms of the GNU General Public License as published
   9 ;; by the Free Software Foundation; either version 3, or (at your
  10 ;; option) any later version.
  11
  12 ;; Emacs Muse is distributed in the hope that it will be useful, but
  13 ;; WITHOUT ANY WARRANTY; without even the implied warranty of
  14 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15 ;; General Public License for more details.
  16
  17 ;; You should have received a copy of the GNU General Public License
  18 ;; along with Emacs Muse; see the file COPYING.  If not, write to the
  19 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  20 ;; Boston, MA 02110-1301, USA.
  21
  22 ;;; Commentary:
  23
  24 ;;; Contributors:
  25
  26 ;; Dale P. Smith (dpsm AT en DOT com) improved the markup
  27 ;; significantly and made many valuable suggestions.
  28
  29 ;;; Code:
  30
  31 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
  32 ;;
  33 ;; Muse DocBook XML Publishing
  34 ;;
  35 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
  36
  37 (require 'muse-publish)
  38 (require 'muse-regexps)
  39 (require 'muse-xml-common)
  40
  41 (defgroup muse-docbook nil
  42   "Options controlling the behavior of Muse DocBook XML publishing.
  43 See `muse-docbook' for more information."
  44   :group 'muse-publish)
  45
  46 (defcustom muse-docbook-extension ".xml"
  47   "Default file extension for publishing DocBook XML files."
  48   :type 'string
  49   :group 'muse-docbook)
  50
  51 (defcustom muse-docbook-header
  52   "<?xml version=\"1.0\" encoding=\"<lisp>
  53   (muse-docbook-encoding)</lisp>\"?>
  54 <!DOCTYPE article PUBLIC \"-//OASIS//DTD DocBook V4.2//EN\"
  55                   \"http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd\"<lisp>(muse-docbook-entities)</lisp>>
  56 <article>
  57   <articleinfo>
  58     <title><lisp>(muse-publishing-directive \"title\")</lisp></title>
  59     <author><lisp>(muse-docbook-get-author
  60                     (muse-publishing-directive \"author\"))</lisp></author>
  61     <pubdate><lisp>(muse-publishing-directive \"date\")</lisp></pubdate>
  62   </articleinfo>
  63   <!-- Page published by Emacs Muse begins here -->\n"
  64   "Header used for publishing DocBook XML files.
  65 This may be text or a filename."
  66   :type 'string
  67   :group 'muse-docbook)
  68
  69 (defcustom muse-docbook-footer "
  70   <!-- Page published by Emacs Muse ends here -->
  71 <lisp>(muse-docbook-bibliography)</lisp></article>\n"
  72   "Footer used for publishing DocBook XML files.
  73 This may be text or a filename."
  74   :type 'string
  75   :group 'muse-docbook)
  76
  77 (defcustom muse-docbook-markup-regexps
  78   `(;; Beginning of doc, end of doc, or plain paragraph separator
  79     (10000 ,(concat "\\(\\(\n\\(?:[" muse-regexp-blank "]*\n\\)*"
  80                     "\\([" muse-regexp-blank "]*\n\\)\\)"
  81                     "\\|\\`\\s-*\\|\\s-*\\'\\)")
  82            3 muse-docbook-markup-paragraph))
  83   "List of markup rules for publishing a Muse page to DocBook XML.
  84 For more on the structure of this list, see `muse-publish-markup-regexps'."
  85   :type '(repeat (choice
  86                   (list :tag "Markup rule"
  87                         integer
  88                         (choice regexp symbol)
  89                         integer
  90                         (choice string function symbol))
  91                   function))
  92   :group 'muse-docbook)
  93
  94 (defcustom muse-docbook-markup-functions
  95   '((anchor . muse-xml-markup-anchor)
  96     (table . muse-xml-markup-table))
  97   "An alist of style types to custom functions for that kind of text.
  98 For more on the structure of this list, see
  99 `muse-publish-markup-functions'."
 100   :type '(alist :key-type symbol :value-type function)
 101   :group 'muse-docbook)
 102
 103 (defcustom muse-docbook-markup-strings
 104   '((image-with-desc . "<mediaobject>
 105 <imageobject>
 106 <imagedata fileref=\"%1%.%2%\" format=\"%2%\" />
 107 </imageobject>
 108 <caption><para>%3%</para></caption>
 109 </mediaobject>")
 110     (image           . "<inlinemediaobject><imageobject>
 111 <imagedata fileref=\"%1%.%2%\" format=\"%2%\" />
 112 </imageobject></inlinemediaobject>")
 113     (image-link      . "<ulink url=\"%1%\"><inlinemediaobject><imageobject>
 114 <imagedata fileref=\"%2%.%3%\" format=\"%3%\" />
 115 </imageobject></inlinemediaobject></ulink>")
 116     (anchor-ref      . "<link linkend=\"%s\">%s</link>")
 117     (url             . "<ulink url=\"%s\">%s</ulink>")
 118     (link            . "<ulink url=\"%s\">%s</ulink>")
 119     (link-and-anchor . "<ulink url=\"%s#%s\">%s</ulink>")
 120     (email-addr      . "<email>%s</email>")
 121     (anchor          . "<anchor id=\"%s\" />\n")
 122     (emdash          . "%s&mdash;%s")
 123     (comment-begin   . "<!-- ")
 124     (comment-end     . " -->")
 125     (rule            . "")
 126     (no-break-space  . "&nbsp;")
 127     (enddots         . "....")
 128     (dots            . "...")
 129     (section         . "<section><title>")
 130     (section-end     . "</title>")
 131     (subsection      . "<section><title>")
 132     (subsection-end  . "</title>")
 133     (subsubsection   . "<section><title>")
 134     (subsubsection-end . "</title>")
 135     (section-other   . "<section><title>")
 136     (section-other-end . "</title>")
 137     (section-close   . "</section>")
 138     (footnote        . "<footnote><para>")
 139     (footnote-end    . "</para></footnote>")
 140     (begin-underline . "")
 141     (end-underline   . "")
 142     (begin-literal   . "<systemitem>")
 143     (end-literal     . "</systemitem>")
 144     (begin-emph      . "<emphasis>")
 145     (end-emph        . "</emphasis>")
 146     (begin-more-emph . "<emphasis role=\"strong\">")
 147     (end-more-emph   . "</emphasis>")
 148     (begin-most-emph . "<emphasis role=\"strong\"><emphasis>")
 149     (end-most-emph   . "</emphasis></emphasis>")
 150     (begin-verse     . "<literallayout>\n")
 151     (verse-space     . "  ")
 152     (end-verse       . "</literallayout>")
 153     (begin-example   . "<programlisting>")
 154     (end-example     . "</programlisting>")
 155     (begin-center    . "<para role=\"centered\">\n")
 156     (end-center      . "\n</para>")
 157     (begin-quote     . "<blockquote>\n")
 158     (end-quote       . "\n</blockquote>")
 159     (begin-cite      . "<citation role=\"%s\">")
 160     (begin-cite-author . "<citation role=\"%s\">A:")
 161     (begin-cite-year . "<citation role=\"%s\">Y:")
 162     (end-cite        . "</citation>")
 163     (begin-quote-item . "<para>")
 164     (end-quote-item  . "</para>")
 165     (begin-uli       . "<itemizedlist mark=\"bullet\">\n")
 166     (end-uli         . "\n</itemizedlist>")
 167     (begin-uli-item  . "<listitem><para>")
 168     (end-uli-item    . "</para></listitem>")
 169     (begin-oli       . "<orderedlist>\n")
 170     (end-oli         . "\n</orderedlist>")
 171     (begin-oli-item  . "<listitem><para>")
 172     (end-oli-item    . "</para></listitem>")
 173     (begin-dl        . "<variablelist>\n")
 174     (end-dl          . "\n</variablelist>")
 175     (begin-dl-item   . "<varlistentry>\n")
 176     (end-dl-item     . "\n</varlistentry>")
 177     (begin-ddt       . "<term>")
 178     (end-ddt         . "</term>")
 179     (begin-dde       . "<listitem><para>")
 180     (end-dde         . "</para></listitem>")
 181     (begin-table     . "<informaltable>\n")
 182     (end-table       . "</informaltable>")
 183     (begin-table-group . "  <tgroup cols='%s'>\n")
 184     (end-table-group . "  </tgroup>\n")
 185     (begin-table-row . "    <row>\n")
 186     (end-table-row   . "    </row>\n")
 187     (begin-table-entry . "      <entry>")
 188     (end-table-entry . "</entry>\n"))
 189   "Strings used for marking up text.
 190 These cover the most basic kinds of markup, the handling of which
 191 differs little between the various styles."
 192   :type '(alist :key-type symbol :value-type string)
 193   :group 'muse-docbook)
 194
 195 (defcustom muse-docbook-encoding-default 'utf-8
 196   "The default Emacs buffer encoding to use in published files.
 197 This will be used if no special characters are found."
 198   :type 'symbol
 199   :group 'muse-docbook)
 200
 201 (defcustom muse-docbook-charset-default "utf-8"
 202   "The default DocBook XML charset to use if no translation is
 203 found in `muse-docbook-encoding-map'."
 204   :type 'string
 205   :group 'muse-docbook)
 206
 207 (defun muse-docbook-encoding ()
 208   (muse-xml-transform-content-type
 209    (or (and (boundp 'buffer-file-coding-system)
 210             buffer-file-coding-system)
 211        muse-docbook-encoding-default)
 212    muse-docbook-charset-default))
 213
 214 (defun muse-docbook-markup-paragraph ()
 215   (catch 'bail-out
 216     (let ((end (copy-marker (match-end 0) t)))
 217       (goto-char (match-beginning 0))
 218       (when (save-excursion
 219               (save-match-data
 220                 (and (re-search-backward
 221                       "<\\(/?\\)\\(para\\|footnote\\|literallayout\\)[ >]"
 222                       nil t)
 223                      (cond ((string= (match-string 2) "literallayout")
 224                             (and (not (string= (match-string 1) "/"))
 225                                  (throw 'bail-out t)))
 226                            ((string= (match-string 2) "para")
 227                             (and
 228                              (not (string= (match-string 1) "/"))
 229                              ;; don't mess up nested lists
 230                              (not (and (muse-looking-back "<listitem>")
 231                                        (throw 'bail-out t)))))
 232                            ((string= (match-string 2) "footnote")
 233                             (string= (match-string 1) "/"))
 234                            (t nil)))))
 235         (when (get-text-property (1- (point)) 'end-list)
 236           (goto-char (previous-single-property-change (1- (point)) 'end-list)))
 237         (muse-insert-markup "</para>"))
 238       (goto-char end))
 239     (cond
 240      ((eobp)
 241       (unless (bolp)
 242         (insert "\n")))
 243      ((eq (char-after) ?\<)
 244       (when (looking-at (concat "<\\(emphasis\\|systemitem\\|inlinemediaobject"
 245                                 "\\|u?link\\|anchor\\|email\\)[ >]"))
 246         (muse-insert-markup "<para>")))
 247      (t
 248       (muse-insert-markup "<para>")))))
 249
 250 (defun muse-docbook-get-author (&optional author)
 251   "Split the AUTHOR directive into separate fields.
 252 AUTHOR should be of the form: \"Firstname Other Names Lastname\",
 253 and anything after `Firstname' is optional."
 254   (setq author (save-match-data (split-string author)))
 255   (let ((num-el (length author)))
 256     (cond ((eq num-el 1)
 257            (concat "<firstname>" (car author) "</firstname>"))
 258           ((eq num-el 2)
 259            (concat "<firstname>" (nth 0 author) "</firstname>"
 260                    "<surname>" (nth 1 author) "</surname>"))
 261           ((eq num-el 3)
 262            (concat "<firstname>" (nth 0 author) "</firstname>"
 263                    "<othername>" (nth 1 author) "</othername>"
 264                    "<surname>" (nth 2 author) "</surname>"))
 265           (t
 266            (let (first last)
 267              (setq first (car author))
 268              (setq author (nreverse (cdr author)))
 269              (setq last (car author))
 270              (setq author (nreverse (cdr author)))
 271              (concat "<firstname>" first "</firstname>"
 272                      "<othername>"
 273                      (mapconcat 'identity author " ")
 274                      "</othername>"
 275                      "<surname>" last "</surname>"))))))
 276
 277 (defun muse-docbook-fixup-images ()
 278   (goto-char (point-min))
 279   (while (re-search-forward (concat "<imagedata fileref=\"[^\"]+\""
 280                                     " format=\"\\([^\"]+\\)\" />$")
 281                             nil t)
 282     (replace-match (upcase (match-string 1)) t t nil 1)))
 283
 284 (defun muse-docbook-fixup-citations ()
 285   ;; remove the role attribute if there is no role
 286   (goto-char (point-min))
 287   (while (re-search-forward "<\\(citation role=\"nil\"\\)>" nil t)
 288     (replace-match "citation" t t nil 1))
 289   ;; replace colons in multi-head citations with semicolons
 290   (goto-char (point-min))
 291   (while (re-search-forward "<citation.*>" nil t)
 292     (let ((start (point))
 293           (end (re-search-forward "</citation>")))
 294       (save-restriction
 295         (narrow-to-region start end)
 296         (goto-char (point-min))
 297         (while (re-search-forward "," nil t)
 298           (replace-match ";"))))))
 299
 300 (defun muse-docbook-munge-buffer ()
 301   (muse-docbook-fixup-images)
 302   (muse-docbook-fixup-citations))
 303
 304 (defun muse-docbook-entities ()
 305   (save-excursion
 306     (goto-char (point-min))
 307     (if (re-search-forward "<citation" nil t)
 308         (concat
 309          " [\n<!ENTITY bibliography SYSTEM \""
 310          (if (string-match ".short$" (muse-page-name))
 311              (substring (muse-page-name) 0 -6)
 312            (muse-page-name))
 313          ".bib.xml\">\n]")
 314       "")))
 315
 316 (defun muse-docbook-bibliography ()
 317   (save-excursion
 318     (goto-char (point-min))
 319     (if (re-search-forward "<citation" nil t)
 320         "&bibliography;\n"
 321       "")))
 322
 323 (defun muse-docbook-finalize-buffer ()
 324   (when (boundp 'buffer-file-coding-system)
 325     (when (memq buffer-file-coding-system '(no-conversion undecided-unix))
 326       ;; make it agree with the default charset
 327       (setq buffer-file-coding-system muse-docbook-encoding-default))))
 328
 329 ;;; Register the Muse DocBook XML Publisher
 330
 331 (muse-define-style "docbook"
 332                    :suffix     'muse-docbook-extension
 333                    :regexps    'muse-docbook-markup-regexps
 334                    :functions  'muse-docbook-markup-functions
 335                    :strings    'muse-docbook-markup-strings
 336                    :specials   'muse-xml-decide-specials
 337                    :before-end 'muse-docbook-munge-buffer
 338                    :after      'muse-docbook-finalize-buffer
 339                    :header     'muse-docbook-header
 340                    :footer     'muse-docbook-footer
 341                    :browser    'find-file)
 342
 343 (provide 'muse-docbook)
 344
 345 ;;; muse-docbook.el ends here