| 1 | CREATING THE EXIM DOCUMENTATION |
| 2 | |
| 3 | "You are lost in a maze of twisty little scripts." |
| 4 | |
| 5 | |
| 6 | This document describes how the various versions of the Exim documentation, in |
| 7 | different output formats, are created from DocBook XML, and also how the |
| 8 | DocBook XML is itself created. |
| 9 | |
| 10 | |
| 11 | BACKGROUND: THE OLD WAY |
| 12 | |
| 13 | From the start of Exim, in 1995, the specification was written in a local text |
| 14 | formatting system known as SGCAL. This is capable of producing PostScript and |
| 15 | plain text output from the same source file. Later, when the "ps2pdf" command |
| 16 | became available with GhostScript, that was used to create a PDF version from |
| 17 | the PostScript. (A few earlier versions were created by a helpful user who had |
| 18 | bought the Adobe distiller software.) |
| 19 | |
| 20 | A demand for a version in "info" format led me to write a Perl script that |
| 21 | converted the SGCAL input into a Texinfo file. Because of the somewhat |
| 22 | restrictive requirements of Texinfo, this script always needed a lot of |
| 23 | maintenance, and was never totally satisfactory. |
| 24 | |
| 25 | The HTML version of the documentation was originally produced from the Texinfo |
| 26 | version, but later I wrote another Perl script that produced it directly from |
| 27 | the SGCAL input, which made it possible to produce better HTML. |
| 28 | |
| 29 | There were a small number of diagrams in the documentation. For the PostScript |
| 30 | and PDF versions, these were created using Aspic, a local text-driven drawing |
| 31 | program that interfaces directly to SGCAL. For the text and texinfo versions, |
| 32 | alternative ascii-art diagrams were used. For the HTML version, screen shots of |
| 33 | the PostScript output were turned into gifs. |
| 34 | |
| 35 | |
| 36 | A MORE STANDARD APPROACH |
| 37 | |
| 38 | Although in principle SGCAL and Aspic could be generally released, they would |
| 39 | be unlikely to receive much (if any) maintenance, especially after I retire. |
| 40 | Furthermore, the old production method was only semi-automatic; I still did a |
| 41 | certain amount of hand tweaking of spec.txt, for example. As the maintenance of |
| 42 | Exim itself was being opened up to a larger group of people, it seemed sensible |
| 43 | to move to a more standard way of producing the documentation, preferable fully |
| 44 | automated. However, we wanted to use only non-commercial software to do this. |
| 45 | |
| 46 | At the time I was thinking about converting (early 2005), the "obvious" |
| 47 | standard format in which to keep the documentation was DocBook XML. The use of |
| 48 | XML in general, in many different applications, was increasing rapidly, and it |
| 49 | seemed likely to remain a standard for some time to come. DocBook offered a |
| 50 | particular form of XML suited to documents that were effectively "books". |
| 51 | |
| 52 | Maintaining an XML document by hand editing is a tedious, verbose, and |
| 53 | error-prone process. A number of specialized XML text editors were available, |
| 54 | but all the free ones were at a very primitive stage. I therefore decided to |
| 55 | keep the master source in AsciiDoc format, from which a secondary XML master |
| 56 | could be automatically generated. |
| 57 | |
| 58 | The first "new" versions of the documents, for the 4.60 release, were generated |
| 59 | this way. However, there were a number of problems with using AsciiDoc for a |
| 60 | document as large and as complex as the Exim manual. As a result, I wrote a new |
| 61 | application called xfpt ("XML From Plain Text") which creates XML from a |
| 62 | relatively simple and consistent markup language. This application has been |
| 63 | released for general use, and the master sources for the Exim documentation are |
| 64 | now in xfpt format. |
| 65 | |
| 66 | All the output formats are generated from the XML file. If, in the future, a |
| 67 | better way of maintaining the XML source becomes available, this can be adopted |
| 68 | without changing any of the processing that produces the output documents. |
| 69 | Equally, if better ways of processing the XML become available, they can be |
| 70 | adopted without affecting the source maintenance. |
| 71 | |
| 72 | A number of issues arose while setting this all up, which are best summed up by |
| 73 | the statement that a lot of the technology was (in 2006) still very immature. |
| 74 | Trying to do this conversion any earlier would probably not have been anywhere |
| 75 | near as successful. The main issues that bother me in the XML-generated |
| 76 | documentation are described in the penultimate section of this document. |
| 77 | |
| 78 | Initially, the major problems were in producing PostScript and PDF outputs. The |
| 79 | available free software for doing this was and still is (we are now in 2007) |
| 80 | cumbersome and slow, and does not support certain output features that I would |
| 81 | like. My response to this was, over a period of two years, to write an XML |
| 82 | processor called SDoP (Simple DocBook Processor). This program reads DocBook |
| 83 | XML and writes PostScript, without using any of the heavyweight apparatus that |
| 84 | is required for xmlto and fop (the previously used software). |
| 85 | |
| 86 | An experimental first version of SDoP was used for the Exim 4.67 |
| 87 | documentation. Subsequently SDoP was released for general use. SDoP's output |
| 88 | includes features that are missing when xmlto/fop is used, and it also runs |
| 89 | about 60 times faster. The main manual can be formatted in 2.5 seconds instead |
| 90 | of 2.5 minutes, which makes checking and fixing mistakes much easier. |
| 91 | |
| 92 | The Makefile that is used to build the various forms of output will, for the |
| 93 | moment, support both ways of producing PostScript and PDF output, though the |
| 94 | default is now to use SDoP. |
| 95 | |
| 96 | The following sections describe the processes by which the xfpt files are |
| 97 | transformed into the final output documents. In practice, the details are coded |
| 98 | into a Makefile that specifies the chain of commands for each output format. |
| 99 | |
| 100 | |
| 101 | REQUIRED SOFTWARE |
| 102 | |
| 103 | Installing software to process XML puts lots and lots of stuff on your box. I |
| 104 | run Gentoo Linux, and a lot of things have been installed as dependencies that |
| 105 | I am not fully aware of. This is what I know about (version numbers are current |
| 106 | at the time of writing): |
| 107 | |
| 108 | . xfpt 0.03 |
| 109 | |
| 110 | This converts the master source file into a DocBook XML file. |
| 111 | |
| 112 | . sdop 0.03 |
| 113 | |
| 114 | This is my new DocBook-to-PostScript processor. |
| 115 | |
| 116 | . ps2pdf |
| 117 | |
| 118 | This is a wrapper script that is part of the GhostScript distribution. It |
| 119 | converts a PostScript file into a PDF file. It is used to process the output |
| 120 | from SDoP. It is not required when xmlto/fop is being used to generate PDF |
| 121 | output. |
| 122 | |
| 123 | . xmlto 0.0.18 |
| 124 | |
| 125 | This is a shell script that drives various XML processors. It is used to |
| 126 | produce "formatted objects" when PostScript and PDF output is being generated |
| 127 | using fop (the old way) rather than SDoP. It is always used to produce HTML |
| 128 | output. It uses xsltproc, libxml, libxslt, libexslt, and possibly other |
| 129 | things that I have not figured out, to apply the DocBook XSLT stylesheets. |
| 130 | |
| 131 | . libxml 1.8.17 |
| 132 | libxml2 2.6.28 |
| 133 | libxslt 1.1.20 |
| 134 | |
| 135 | These are all installed on my box; I do not know which of libxml or libxml2 |
| 136 | the various scripts are actually using. |
| 137 | |
| 138 | . xsl-stylesheets-<version> |
| 139 | |
| 140 | These are the standard DocBook XSL stylesheets. |
| 141 | |
| 142 | The documents use http://docbook.sourceforge.net/release/xsl/current/ which |
| 143 | should be mapped to an appropriate local path via the system catalogs. |
| 144 | |
| 145 | . fop 0.93 |
| 146 | |
| 147 | FOP is a processor for "formatted objects". It is written in Java. The fop |
| 148 | command is a shell script that drives it. It required only if you do not |
| 149 | want to use SDoP and ps2pdf to generate PostScript and PDF output. |
| 150 | |
| 151 | . w3m 0.5.2 |
| 152 | |
| 153 | This is a text-oriented web brower. It is used to produce the ASCII form of |
| 154 | the Exim documentation (spec.txt) from a specially-created HTML format. It |
| 155 | seems to do a better job than lynx. |
| 156 | |
| 157 | . docbook2texi (part of docbook2X 0.8.5) |
| 158 | |
| 159 | This is a wrapper script for a two-stage conversion process from DocBook to a |
| 160 | Texinfo file. It uses db2x_xsltproc and db2x_texixml. Unfortunately, there |
| 161 | are two versions of this command; the old one is based on an earlier fork of |
| 162 | docbook2X and does not work. |
| 163 | |
| 164 | . db2x_xsltproc and db2x_texixml (part of docbook2X 0.8.5) |
| 165 | |
| 166 | More wrapping scripts (see previous item). |
| 167 | |
| 168 | . makeinfo 4.8 |
| 169 | |
| 170 | This is used to make an "info" file from a Texinfo file. |
| 171 | |
| 172 | In addition, there are a number of locally written Perl scripts. These are |
| 173 | described below. |
| 174 | |
| 175 | |
| 176 | THE MAKEFILE |
| 177 | |
| 178 | The makefile supports a number of targets of the form x.y, where x is one of |
| 179 | "filter", "spec", or "test", and y is one of "xml", "fo", "ps", "pdf", "html", |
| 180 | "txt", or "info". The intermediate targets "x.xml" and "x.fo" are provided for |
| 181 | testing purposes. The other five targets are production targets. For example: |
| 182 | |
| 183 | make spec.pdf |
| 184 | |
| 185 | This runs the necessary tools in order to create the file spec.pdf from the |
| 186 | original source spec.xfpt. A number of intermediate files are created during |
| 187 | this process, including the master DocBook source, called spec.xml. Of course, |
| 188 | the usual features of "make" ensure that if this already exists and is |
| 189 | up-to-date, it is not needlessly rebuilt. |
| 190 | |
| 191 | Because there are now two ways of creating the PostScript and PDF outputs, |
| 192 | there are two targets for each one. For example fop-spec.ps makes PostScript |
| 193 | using fop, and sdop-spec.ps makes it using SDoP. The generic targets spec.ps |
| 194 | and spec.pdf now point to the SDoP versions. |
| 195 | |
| 196 | The "test" series of targets were created so that small tests could easily be |
| 197 | run fairly quickly, because processing even the shortish XML document takes |
| 198 | a bit of time, and processing the main specification takes ages -- except when |
| 199 | using SDoP for PostScript and PDF. |
| 200 | |
| 201 | Another target is "exim.8". This runs a locally written Perl script called |
| 202 | x2man, which extracts the list of command line options from the spec.xml file, |
| 203 | and creates a man page. There are some XML comments in the spec.xml file to |
| 204 | enable the script to find the start and end of the options list. |
| 205 | |
| 206 | There is also a "clean" target that deletes all the generated files. |
| 207 | |
| 208 | |
| 209 | CREATING DOCBOOK XML FROM XFPT INPUT |
| 210 | |
| 211 | The small amount of local configuration for xfpt is included at the start of |
| 212 | the two .xfpt files; there are no separate local xfpt configuration files. |
| 213 | Running the xfpt command creates a .xml file from a .xfpt file. When this |
| 214 | succeeds, there is no output. |
| 215 | |
| 216 | |
| 217 | DOCBOOK PROCESSING |
| 218 | |
| 219 | Processing a .xml file into the five different output formats is not entirely |
| 220 | straightforward. For a start, the same XML is not suitable for all the |
| 221 | different output styles. When the final output is in a text format (.txt, |
| 222 | .texinfo) for instance, all non-ASCII characters in the input must be converted |
| 223 | to ASCII transliterations because the current processing tools do not do this |
| 224 | correctly automatically. |
| 225 | |
| 226 | In order to cope with these issues in a flexible way, a Perl script called |
| 227 | Pre-xml was written. This is used to preprocess the .xml files before they are |
| 228 | handed to the main processors. Adding one more tool onto the front of the |
| 229 | processing chain does at least seem to be in the spirit of XML processing. |
| 230 | |
| 231 | The XML processors other than SDoP make use of style files, which can be |
| 232 | overridden by local versions. There is one that applies to all styles, called |
| 233 | MyStyle.xsl, and others for the different output formats. I have included |
| 234 | comments in these style files to explain what changes I have made. Some of the |
| 235 | changes are quite significant. |
| 236 | |
| 237 | |
| 238 | XSL INCLUDES |
| 239 | |
| 240 | References to XSL paths should use the public URLs, such as: |
| 241 | http://docbook.sourceforge.net/release/xsl/current/xhtml/docbook.xsl |
| 242 | If this fails to work for you, then there is a problem with your system |
| 243 | catalogs. As a work-around, you can adjust the OS-Fixups script and then: |
| 244 | $ make os-fixup |
| 245 | |
| 246 | As an example of how this should normally work, on a FreeBSD system the |
| 247 | resolution goes to /usr/local/share/xml/catalog which contains a directive: |
| 248 | <nextCatalog catalog="/usr/local/share/xml/catalog.ports" /> |
| 249 | to pull in the file automatically maintained by the Ports system. That file |
| 250 | will contain: |
| 251 | <delegateSystem |
| 252 | systemIdStartString="http://docbook.sourceforge.net/release/xsl/" |
| 253 | catalog="file:///usr/local/share/xsl/docbook/catalog" /> |
| 254 | <delegateURI |
| 255 | uriStartString="http://docbook.sourceforge.net/release/xsl/" |
| 256 | catalog="file:///usr/local/share/xsl/docbook/catalog" /> |
| 257 | and that catalog file contains: |
| 258 | <rewriteSystem |
| 259 | systemIdStartString="http://docbook.sourceforge.net/release/xsl/current" |
| 260 | rewritePrefix="file:///usr/local/share/xsl/docbook" /> |
| 261 | <rewriteURI |
| 262 | uriStartString="http://docbook.sourceforge.net/release/xsl/current" |
| 263 | rewritePrefix="file:///usr/local/share/xsl/docbook" /> |
| 264 | and the full path is thus eventually arrived at. |
| 265 | |
| 266 | See also the tools: |
| 267 | xmlcatalog(1) from libxml2 |
| 268 | xmlcatmgr(1) for a lightweight tool written for the NetBSD Packages system. |
| 269 | |
| 270 | |
| 271 | THE PRE-XML SCRIPT |
| 272 | |
| 273 | The Pre-xml script copies a .xml file, making certain changes according to the |
| 274 | options it is given. The currently available options are as follows: |
| 275 | |
| 276 | -ascii |
| 277 | |
| 278 | This option is used for ASCII output formats. It makes the following |
| 279 | character replacements: |
| 280 | |
| 281 | ’ => ' apostrophe |
| 282 | © => (c) copyright |
| 283 | † => * dagger |
| 284 | ‡ => ** double dagger |
| 285 | => a space hard space |
| 286 | – => - en dash |
| 287 | |
| 288 | The apostrophe is specified numerically because that is what xfpt generates |
| 289 | from an ASCII single quote character. Non-ASCII characters that are not in |
| 290 | this list should not be used without thinking about how they might be |
| 291 | converted for the ASCII formats. |
| 292 | |
| 293 | In addition to the character replacements, this option causes quotes to be |
| 294 | put round <literal> text items, and <quote> and </quote> to be replaced by |
| 295 | ASCII quote marks. You would think the stylesheet would cope with the latter, |
| 296 | but it seems to generate non-ASCII characters that w3m then turns into |
| 297 | question marks. |
| 298 | |
| 299 | -bookinfo |
| 300 | |
| 301 | This option causes the <bookinfo> element to be removed from the XML. It is |
| 302 | used for the PostScript/PDF forms of the filter document, in order to avoid |
| 303 | the generation of a full title page. |
| 304 | |
| 305 | -fi |
| 306 | |
| 307 | Replace any occurrence of "fi" by the ligature fi except when it is |
| 308 | inside an XML element, or inside a <literal> part of the text. |
| 309 | |
| 310 | The use of ligatures would be nice for the PostScript and PDF formats. Sadly, |
| 311 | it turns out that fop cannot at present handle the FB01 character correctly. |
| 312 | Happily this problem is now avoided when SDoP is used to generate PostScript |
| 313 | (and thence PDF) because SDoP automatically uses an "fi" ligature for |
| 314 | non-fixed-width fonts. |
| 315 | |
| 316 | The only xmlto format that handles FB01 is the HTML format, but when I used |
| 317 | this in the test version, people complained that it made searching for words |
| 318 | difficult. So this option is in practice not used at all. |
| 319 | |
| 320 | -noindex |
| 321 | |
| 322 | Remove the XML to generate a Concept Index and an Options index. The source |
| 323 | document has three types of index entry, for variables, options, and concept |
| 324 | indexes. However, no index is required for the .txt and .texinfo outputs. |
| 325 | |
| 326 | -oneindex |
| 327 | |
| 328 | Remove the XML to generate separate variables, options, and concept indexes, |
| 329 | and add XML to generate a single index. The only output processors that |
| 330 | support multiple indexes are SDoP and the processor that produces "formatted |
| 331 | objects" for PostScript and PDF output for fop. The HTML processor ignores |
| 332 | the XML settings for multiple indexes and just makes one unified index. |
| 333 | Specifying three indexes gets you three copies of the same index, so this has |
| 334 | to be changed. |
| 335 | |
| 336 | -optbreak |
| 337 | |
| 338 | Look for items of the form <option>...</option> and <varname>...</varname> in |
| 339 | ordinary paragraphs, and insert ​ after each underscore in the |
| 340 | enclosed text. The same is done for any word containing four or more upper |
| 341 | case letters (compile-time options in the Exim specification). The character |
| 342 | ​ is a zero-width space. This means that the line may be split after |
| 343 | one of these underscores, but no hyphen is inserted. |
| 344 | |
| 345 | |
| 346 | CREATING POSTSCRIPT AND PDF |
| 347 | |
| 348 | These two output formats are created either by using my new SDoP program to |
| 349 | produce PostScript which can then be run through ps2pdf to make a PDF, or by |
| 350 | using xmlto and fop in the old way. |
| 351 | |
| 352 | |
| 353 | USING SDOP TO CREATE POSTSCRIPT AND PDF |
| 354 | |
| 355 | PostScript output is created in two stages. First, the XML is pre-processed by |
| 356 | the Pre-xml script. For the filter document, the <bookinfo> element is removed |
| 357 | so that no title page is generated. For the main specification, the only change |
| 358 | is to insert line breakpoints via -optbreak. |
| 359 | |
| 360 | The SDoP program is then used to create PostScript output directly from the XML |
| 361 | input. Then the ps2pdf command is used to generated a PDF from the PostScript. |
| 362 | There are no external stylesheets that are used by SDoP. Any variations to the |
| 363 | default format are specified inline using "processing instructions". |
| 364 | |
| 365 | |
| 366 | USING XMLTO AND FOP TO CREATE POSTSCRIPT AND PDF |
| 367 | |
| 368 | This is the original way of creating PostScript and PDF output. The processing |
| 369 | happens in three stages, with an additional fourth stage for PDF. First, the |
| 370 | XML is pre-processed by the Pre-xml script. For the filter document, the |
| 371 | <bookinfo> element is removed so that no title page is generated. For the main |
| 372 | specification, the only change is to insert line breakpoints via -optbreak. |
| 373 | |
| 374 | Second, the xmlto command is used to produce a "formatted objects" (.fo) file. |
| 375 | This process uses the following stylesheets: |
| 376 | |
| 377 | (1) Either MyStyle-filter-fo.xsl or MyStyle-spec-fo.xsl |
| 378 | (2) MyStyle-fo.xsl |
| 379 | (3) MyStyle.xsl |
| 380 | (4) MyTitleStyle.xsl |
| 381 | |
| 382 | The last of these is not used for the filter document, which does not have a |
| 383 | title page. The first three stylesheets were created manually, either by typing |
| 384 | directly, or by coping from the standard style sheet and editing. |
| 385 | |
| 386 | The final stylesheet has to be created from a template document, which is |
| 387 | called MyTitlepage.templates.xml. This was copied from the standard styles and |
| 388 | modified. The template is processed with xsltproc to produce the stylesheet. |
| 389 | All this apparatus is appallingly heavyweight. The processing is also very slow |
| 390 | in the case of the specification document. However, there should be no errors. |
| 391 | |
| 392 | The reference book that saved my life while I was trying to get all this to |
| 393 | work is "DocBook XSL, The Complete Guide", third edition (2005), by Bob |
| 394 | Stayton, published by Sagehill Enterprises. |
| 395 | |
| 396 | In the third part of the processing, the .fo file that is produced by the xmlto |
| 397 | command is processed by the fop command to generate either PostScript or PDF. |
| 398 | This is also very slow, and you get a whole slew of errors, of which these are |
| 399 | a sample: |
| 400 | |
| 401 | [ERROR] property - "background-position-horizontal" is not implemented yet. |
| 402 | |
| 403 | [ERROR] property - "background-position-vertical" is not implemented yet. |
| 404 | |
| 405 | [INFO] JAI support was not installed (read: not present at build time). |
| 406 | Trying to use Jimi instead |
| 407 | Error creating background image: Error creating FopImage object (Error |
| 408 | creating FopImage object |
| 409 | (http://docbook.sourceforge.net/release/images/draft.png) : |
| 410 | org.apache.fop.image.JimiImage |
| 411 | |
| 412 | [WARNING] table-layout=auto is not supported, using fixed! |
| 413 | |
| 414 | [ERROR] Unknown enumerated value for property 'span': inherit |
| 415 | |
| 416 | [ERROR] Error in span property value 'inherit': |
| 417 | org.apache.fop.fo.expr.PropertyException: No conversion defined |
| 418 | |
| 419 | [ERROR] Areas pending, text probably lost in lineinclude parts matched in the |
| 420 | response by response_pattern by means of numeric variables such as |
| 421 | |
| 422 | The last one is particularly meaningless gobbledegook. Some of the errors and |
| 423 | warnings are repeated many times. Nevertheless, it does eventually produce |
| 424 | usable output, though I have a number of issues with it (see a later section of |
| 425 | this document). Maybe one day there will be a new release of fop that does |
| 426 | better. In the meantime, I have written my own program for making PostScript |
| 427 | output -- see the previous section -- because the problems with xmlto/fop were |
| 428 | sufficiently annoying. |
| 429 | |
| 430 | The PDF file that is produced by this process has one problem: the pages, as |
| 431 | shown by acroread in its thumbnail display, are numbered sequentially from one |
| 432 | to the end. Those numbers do not correspond with the page numbers of the body |
| 433 | of the document, which makes finding a page from the index awkward. There is a |
| 434 | facility in the PDF format to give pages appropriate "labels", but I cannot |
| 435 | find a way of persuading fop to generate these. Fortunately, it is possibly to |
| 436 | fix up the PDF to add page labels. I wrote a script called PageLabelPDF which |
| 437 | does this. They are shown correctly by acroread and xpdf, but not by |
| 438 | GhostScript (gv). |
| 439 | |
| 440 | |
| 441 | THE PAGELABELPDF SCRIPT |
| 442 | |
| 443 | This script reads the standard input and writes the standard output. It is used |
| 444 | to "tidy up" the PDF output that is produced by fop. It is not needed when |
| 445 | PDF output is generated from SDoP's output using ps2pdf. |
| 446 | |
| 447 | The PageLabelPDF script searches for the PDF object that sets data in its |
| 448 | "Catalog", and adds appropriate information about page labels. The number of |
| 449 | front-matter pages (those before chapter 1) is hard-wired into this script as |
| 450 | 12 because I could not find a way of determining it automatically. As the |
| 451 | current table of contents finishes near the top of the 11th page, there is |
| 452 | plenty of room for expansion, so it is unlikely to be a problem. |
| 453 | |
| 454 | Having added data to the PDF file, the script then finds the xref table at the |
| 455 | end of the file, and adjusts its entries to allow for the added text. This |
| 456 | simple processing seems to be enough to generate a new, valid, PDF file. |
| 457 | |
| 458 | |
| 459 | CREATING HTML |
| 460 | |
| 461 | Only two stages are needed to produce HTML, but the main specification is |
| 462 | subsequently postprocessed. The Pre-xml script is called with the -optbreak and |
| 463 | -oneindex options to preprocess the XML. Then the xmlto command creates the |
| 464 | HTML output directly. For the specification document, a directory of files is |
| 465 | created, whereas the filter document is output as a single HTML page. The |
| 466 | following stylesheets are used: |
| 467 | |
| 468 | (1) Either MyStyle-chunk-html.xsl or MyStyle-nochunk-html.xsl |
| 469 | (2) MyStyle-html.xsl |
| 470 | (3) MyStyle.xsl |
| 471 | |
| 472 | The first stylesheet references the chunking or non-chunking standard DocBook |
| 473 | stylesheet, as appropriate. |
| 474 | |
| 475 | You may see a number of these errors when creating HTML: "Revisionflag on |
| 476 | unexpected element: literallayout (Assuming block)". They seem to be harmless; |
| 477 | the output appears to be what is intended. |
| 478 | |
| 479 | The original HTML that I produced from the SGCAL input had hyperlinks back from |
| 480 | chapter and section titles to the table of contents. These links are not |
| 481 | generated by xmlto. One of the testers pointed out that the lack of these |
| 482 | links, or simple self-referencing links for titles, makes it harder to copy a |
| 483 | link name into, for example, a mailing list response. |
| 484 | |
| 485 | I could not find where to fiddle with the stylesheets to make such a change, if |
| 486 | indeed the stylesheets are capable of it. Instead, I wrote a Perl script called |
| 487 | TidyHTML-spec to do the job for the specification document. It updates the |
| 488 | index.html file (which contains the the table of contents) setting up anchors, |
| 489 | and then updates all the chapter files to insert appropriate links. |
| 490 | |
| 491 | The index.html file as built by xmlto contains the whole table of contents in a |
| 492 | single line, which makes is hard to debug by hand. Since I was postprocessing |
| 493 | it anyway, I arranged to insert newlines after every '>' character. |
| 494 | |
| 495 | The TidyHTML-spec script also processes every HTML file, to tidy up some of the |
| 496 | untidy features therein. It turns <div class="literallayout"><p> into <div |
| 497 | class="literallayout"> and a matching </p></div> into </div> to get rid of |
| 498 | unwanted vertical white space in literallayout blocks. Before each occurrence |
| 499 | of </td> it inserts so that the table's cell is a little bit wider than |
| 500 | the text itself. |
| 501 | |
| 502 | The TidyHTML-spec script also takes the opportunity to postprocess the |
| 503 | spec_html/ix01.html file, which contains the document index. Again, the index |
| 504 | is generated as one single line, so it splits it up. Then it creates a list of |
| 505 | letters at the top of the index and hyperlinks them both ways from the |
| 506 | different letter portions of the index. |
| 507 | |
| 508 | People wanted similar postprocessing for the filter.html file, so that is now |
| 509 | done using a similar script called TidyHTML-filter. It was easier to use a |
| 510 | separate script because filter.html is a single file rather than a directory, |
| 511 | so the logic is somewhat different. |
| 512 | |
| 513 | |
| 514 | CREATING TEXT FILES |
| 515 | |
| 516 | This happens in four stages. The Pre-xml script is called with the -ascii, |
| 517 | -optbreak, and -noindex options to convert the input to ASCII characters, |
| 518 | insert line break points, and disable the production of an index. Then the |
| 519 | xmlto command converts the XML to a single HTML document, using these |
| 520 | stylesheets: |
| 521 | |
| 522 | (1) MyStyle-txt-html.xsl |
| 523 | (2) MyStyle-html.xsl |
| 524 | (3) MyStyle.xsl |
| 525 | |
| 526 | The MyStyle-txt-html.xsl stylesheet is the same as MyStyle-nochunk-html.xsl, |
| 527 | except that it contains an addition item to ensure that a generated "copyright" |
| 528 | symbol is output as "(c)" rather than the Unicode character. This is necessary |
| 529 | because the stylesheet itself generates a copyright symbol as part of the |
| 530 | document title; the character is not in the original input. |
| 531 | |
| 532 | The w3m command is used with the -dump option to turn the HTML file into ASCII |
| 533 | text, but this contains multiple sequences of blank lines that make it look |
| 534 | awkward. Furthermore, chapter and section titles do not stand out very well. A |
| 535 | local Perl script called Tidytxt is used to post-process the output. First, it |
| 536 | converts sequences of blank lines into a single blank lines. Then it searches |
| 537 | for chapter and section headings. Each chapter heading is uppercased, and |
| 538 | preceded by an extra two blank lines and a line of equals characters. An extra |
| 539 | newline is inserted before each section heading, and they are underlined with |
| 540 | hyphens. |
| 541 | |
| 542 | The output of xmlto also contains non-ASCII Unicode characters that w3m passes |
| 543 | through. Fortunately, they are few, and Tidytxt cleans them up as well. Some |
| 544 | headings use "box drawing" characters in the range U+2500 to U+253F which are |
| 545 | translated into -+| as appropriate, and U+00A0 (hard space) and U+25CF (bullet) |
| 546 | are translated into plain spaces and asterisks. (It might be possible to do all |
| 547 | this in the same way as I dealt with copyright - see above - but adding a few |
| 548 | lines of Perl to an existing script was a lot easier.) |
| 549 | |
| 550 | |
| 551 | CREATING INFO FILES |
| 552 | |
| 553 | This process starts with the same Pre-xml call as for text files. Non-ascii |
| 554 | characters in the source are transliterated, and the <index> elements are |
| 555 | removed. The docbook2texi script is then called to convert the XML file into a |
| 556 | Texinfo file. However, this is not quite enough. The converted file ends up |
| 557 | with "conceptindex" and "optionindex" items, which are not recognized by the |
| 558 | makeinfo command. These have to be changed to "cindex" and "findex" |
| 559 | respectively in the final .texinfo file. Furthermore, the main menu lacks a |
| 560 | pointer to the index, and indeed the index node itself is missing. These |
| 561 | problems are fixed by running the file through a script called TidyInfo. |
| 562 | Finally, a call of makeinfo creates a .info file. |
| 563 | |
| 564 | There is one apparently unconfigurable feature of docbook2texi: it does not |
| 565 | seem possible to give it a file name for its output. It chooses a name based on |
| 566 | the title of the document. Thus, the main specification ends up in a file |
| 567 | called the_exim_mta.texi and the filter document in exim_filtering.texi. These |
| 568 | files are removed after their contents have been copied and modified by the |
| 569 | TidyInfo script, which writes to a .texinfo file. |
| 570 | |
| 571 | |
| 572 | CREATING THE MAN PAGE |
| 573 | |
| 574 | I wrote a Perl script called x2man to create the exim.8 man page from the |
| 575 | DocBook XML source. I deliberately did NOT start from the xfpt source, |
| 576 | because it is the DocBook source that is the "standard". This comment line in |
| 577 | the DocBook source marks the start of the command line options: |
| 578 | |
| 579 | <!-- === Start of command line options === --> |
| 580 | |
| 581 | A similar line marks the end. If at some time in the future another way other |
| 582 | than xfpt is used to maintain the DocBook source, it needs to be capable of |
| 583 | maintaining these comments. |
| 584 | |
| 585 | |
| 586 | UNRESOLVED PROBLEMS |
| 587 | |
| 588 | There are a number of unresolved problems with producing the Exim documentation |
| 589 | in the manner described above. I will describe them here in the hope that in |
| 590 | future some way round them can be found. Some of the problems are solved by |
| 591 | using SDoP instead of xmlto/fop to produce PostScript and PDF output. |
| 592 | |
| 593 | (1) When a whole chain of tools is processing a file, an error somewhere |
| 594 | in the middle is often very hard to debug. For instance, an error in the |
| 595 | xfpt file might not show up until an XML processor throws a wobbly because |
| 596 | the generated XML is bad. You have to be able to read XML and figure out |
| 597 | what generated what. One of the reasons for creating the "test" series of |
| 598 | targets was to help in checking out these kinds of problem. |
| 599 | |
| 600 | (2) There is a mechanism in XML for marking parts of the document as |
| 601 | "revised", and I have arranged for xfpt markup to use it. However, the |
| 602 | only xmlto output format that pays attention to this is the HTML output, |
| 603 | which sets a green background. If xmlto/fop is used to generate PostScript |
| 604 | and PDF, there are no revision marks (change bars). This problem |
| 605 | is not present when SDoP is used. However, the text and Texinfo output |
| 606 | format lack revision indications. |
| 607 | |
| 608 | (3) The index entries in the HTML format take you to the top of the section |
| 609 | that is referenced, instead of to the point in the section where the index |
| 610 | marker was set. |
| 611 | |
| 612 | (4) The HTML output supports only a single index, so the variable, options, |
| 613 | and concept index entries have to be merged. |
| 614 | |
| 615 | (5) The index for the PostScript/PDF output created by xmlto/fop does not |
| 616 | merge identical page numbers, which makes some entries look ugly. This is |
| 617 | not a problem when SDoP is used. |
| 618 | |
| 619 | (6) The HTML index and the PostScript/PDF indexes, when made with xmlto/fop, |
| 620 | make no use of textual markup; the text is all roman, without any italic |
| 621 | or boldface. For PostScript/PDF, this is not a problem when SDoP is used. |
| 622 | |
| 623 | (7) I turned off hyphenation in the PostScript/PDF output produced by |
| 624 | xmlto/fop, because it was being done so badly. Needless to say, I made |
| 625 | SDoP do a better job. These comments apply to xmlto/fop: |
| 626 | |
| 627 | (a) It seems to force hyphenation if it is at all possible, without |
| 628 | regard to the "tightness" or "looseness" of the line. Decent |
| 629 | formatting software should attempt hyphenation only if the line is |
| 630 | over some "looseness" threshold; otherwise you get far too many |
| 631 | hyphenations, often for several lines in succession. |
| 632 | |
| 633 | (b) It uses an algorithmic form of hyphenation that doesn't always produce |
| 634 | acceptable word breaks. (I prefer to use a hyphenation dictionary, |
| 635 | which is what SDoP does.) |
| 636 | |
| 637 | (8) The PostScript/PDF output produced by xmlto/fop is badly paginated: |
| 638 | |
| 639 | (a) There seems to be no attempt to avoid "widow" and "orphan" lines on |
| 640 | pages. A "widow" is the last line of a paragraph at the top of a page, |
| 641 | and an "orphan" is the first line of a paragraph at the bottom of a |
| 642 | page. |
| 643 | |
| 644 | (b) There seems to be no attempt to prevent section headings being placed |
| 645 | last on a page, with no following text on the page. |
| 646 | |
| 647 | Neither of these problems occurs when SDoP is used to produce the |
| 648 | PostScript/PDF output. |
| 649 | |
| 650 | (9) The fop processor does not support "fi" ligatures, not even if you put the |
| 651 | appropriate Unicode character into the source by hand. Again, this is not |
| 652 | a problem if SDoP is used. |
| 653 | |
| 654 | (10) There are no diagrams in the new documentation. This is something I hope |
| 655 | to work on. The previously used Aspic command for creating line art from a |
| 656 | textual description can output Encapsulated PostScript or Scalar Vector |
| 657 | Graphics, which are two standard diagram representations. Aspic could be |
| 658 | formally released and used to generate output that could be included in at |
| 659 | least some of the output formats. |
| 660 | |
| 661 | (11) The use of a "zero-width space" works well as a way of specifying that |
| 662 | Exim option names can be split, without hyphens, over line breaks. |
| 663 | |
| 664 | However, when xmlto/fop is being used and an option is not split, if the |
| 665 | line is very "loose", the zero-width space is expanded, along with other |
| 666 | spaces. This is a totally crazy thing to, but unfortunately it is |
| 667 | suggested by the Unicode definition of the zero-width space, which says |
| 668 | "its presence between two characters does not prevent increased letter |
| 669 | spacing in justification". It seems that the implementors of fop have |
| 670 | understood "letter spacing" also to include "word spacing". Sigh. |
| 671 | |
| 672 | This problem does not arise when SDoP is used. |
| 673 | |
| 674 | The consequence of (7), (8), and (9) is that the PostScript/PDF output as |
| 675 | produced by xmlto/fop looks as if it comes from some of the very early attempts |
| 676 | at text formatting of around 20 years ago. We can only hope that 20 years' |
| 677 | progress is not going to get lost, and that things will improve in this area. |
| 678 | My small contribution to this has been to write SDoP, which, though simple and |
| 679 | "non-standard", does get some of these formatting issues right. |
| 680 | |
| 681 | |
| 682 | LIST OF FILES |
| 683 | |
| 684 | Markup.txt Describes the xfpt markup that is used |
| 685 | HowItWorks.txt This document |
| 686 | Makefile The makefile |
| 687 | MyStyle-chunk-html.xsl Stylesheet for chunked HTML output |
| 688 | MyStyle-filter-fo.xsl Stylesheet for filter fo output |
| 689 | MyStyle-fo.xsl Stylesheet for any fo output |
| 690 | MyStyle-html.xsl Stylesheet for any HTML output |
| 691 | MyStyle-nochunk-html.xsl Stylesheet for non-chunked HTML output |
| 692 | MyStyle-spec-fo.xsl Stylesheet for spec fo output |
| 693 | MyStyle-txt-html.xsl Stylesheet for HTML=>text output |
| 694 | MyStyle.xsl Stylesheet for all output |
| 695 | MyTitleStyle.xsl Stylesheet for spec title page |
| 696 | MyTitlepage.templates.xml Template for creating MyTitleStyle.xsl |
| 697 | Myhtml.css Experimental css stylesheet for HTML output |
| 698 | PageLabelPDF Script to postprocess xmlto/fop PDF output |
| 699 | Pre-xml Script to preprocess XML |
| 700 | TidyHTML-filter Script to tidy up the filter HTML output |
| 701 | TidyHTML-spec Script to tidy up the spec HTML output |
| 702 | TidyInfo Script to sort index problems in Texinfo output |
| 703 | Tidytxt Script to compact multiple blank lines |
| 704 | filter.xfpt xfpt source of the filter document |
| 705 | spec.xfpt xfpt source of the specification document |
| 706 | x2man Script to make the Exim man page from the XML |
| 707 | |
| 708 | |
| 709 | (Originally, and for the most part: Philip Hazel) |
| 710 | The Exim Maintainers |
| 711 | Last updated: 5 July 2010 |