sindict

(Re-)Building the dictionary

These procedures describe how to build an HTML or PDF version of the dictionary from the core source XML lexicon.

Prerequisites

You will need the following command-line tools on your system:

• xsltproc - XSLT processor.

  The provided XSLT style-sheets are in XSLT 1.0 format and rely on a few EXSLT extensions. You may try another XSLT processor, but there is no guarantee it will work.

• node - Node.js JavaScript runtime.

  Any recent version should work. It is currently only used to clean-up the HTML output from the XSLT style-sheets, using regular expressions to replace some patterns. You can do without it and skip that step, but the output will not be perfect.

• Optional: bash - Shell command-line scripting.

  It is needed in order to use the provided shell scripts described below. You can do without it, but you will then need to adapt the commands to your operating system.

• sile - The SILE typesetter

  Version 0.12 (at the time of writing) was used to generate a PDF version of the lexicon, along with a set of specialized classes and packages from the Omikhleia SILE packages repository.

Principles

The core source lexicon in XML is src/dict-sd-fr-en.xml

Quick conversion to HTML

To just convert the core source lexicon to HTML, the steps are the following:

1. Apply to the lexicon the scripts/tei2html/tei-lite.xsl XSLT style-sheet, to generate a raw HTML file.
2. Apply the Node script scripts/tei2html/post-process.js on that generated HTML to clean it. This step fixes a few issues (e.g. spacing, punctuations, blank lines, etc.) that would have been hard to manage at the XSLT level.

The provided convert.sh script applies these two steps in a row, and puts the converted HTML in docs/dict-sd.html

Complete (post-processed) conversion to HTML

To produce a much better version of the lexicon, several additional tasks are performed.

1. Apply the (post-)processing XSLT style-sheets in order (see below), each using as input the output of the previous task.
2. Apply the scripts/tei2html/tei-lite.xsl XSLT style-sheet to the final XML, to generate a raw HTML file.
3. Apply the Node script scripts/tei2html/post-process.js on that generated HTML to clean it.

The provided process.sh script applies all the necessary steps in a row, and puts the converted HTML in docs/dict-sd.html

The post-processing tasks may be updated and the list below may become outdated, please check the script in case of doubts.

But basically, for the record, the main following steps should be performed:

1. Expand cross-references from alternate and inflected forms (scripts/process/expand-xref-pass1.xsl): this task creates dictionary entries for all entry forms which are not main headwords, i.e. appear in an entry as alternative variants or inflected forms.
2. Expand cross-references from related forms (scripts/process/expand-xref-pass2.xsl): this task creates dictionary entries for secondary entries (a.k.a. related sub-entries) which are not marked as already having an existing main entry.
3. Sort the resulting lexicon in alphabetical order (scripts/process/sort.xsl): obviously, the previous tasks bluntly added new entries that won’t be ordered. (And the original core source XML is not guaranteed, anyway, to be ordered correctly.)
4. Re-number homographs (scripts/process/expand-renum.xsl)
5. Add sections (scripts/process/add-sections.xsl): that is, insert an alphabetical section milestone when the first letter of entries changes.

As an illustration, assuming an original entry:

form1 S. XX/xxx (variant1 S. YY/yyy, variant2 N. ZZ/zzz), pl. plural1 S. TT/ttt (plvariant N. UU/uuu) n. Some definitions - subform S. MM/mmm n. abst.

The expanded post-processed lexicon would yield to something such as:

F

form1 S. XX/xxx (variant1 S. YY/yyy, variant2 N. ZZ/zzz), pl. plural1 S. TT/ttt (plvariant N. UU/uuu) n. Some definitions - subform S. MM/mmm n. abst.

P

plural1 S. TT/ttt pl. → form1

plvariant N. UU/uuu pl. → form1

S

subform S. MM/mmm n. abst. → form1

V

variant1 S. YY/yyy → form1

variant2 N. ZZ/zzz → form1

Complete (post-processed) conversion to PDF

To start with, we need the post-processed version of the lexicon in XML. The initial steps are therefore the same as for the HTML output, without the final (HTML-only) steps.

The provided process-to-xml-only.sh script applies all the necessary steps in a row, and puts the converted (expanded) XML in docs/dict-sd.xml

As noted above, PDF generation requires:

• The SILE Typesetter
• Specialized classes and packages from Omikhleia SILE packages repository.

Installing SILE and our custom classes and packages is an exercise left to the reader. When properly set up:

• For the English PDF:

  sile -I preambles/dict-sd-en-preamble.sil docs/dict-sd.xml
  

• For the French PDF

  sile -I preambles/dict-sd-fr-preamble.sil docs/dict-sd.xml
  

This site is open source. Improve this page.