LEDA document and graphics conversions

What to expect

LEDA uses a battery of programs to convert your RTF file into HTML and PDF versions. Most of the time these will produce converted files that are reasonably faithful to the original document. However, all conversion programs have limitations, and it's possible that conversions will be less than perfect -- not surprising, when you consider just how many things you can do with a word processor.

Here are some things you can do that will help your RTF documents look as good as possible once converted:

• In general, RTF generated using Microsoft Word is superior to RTF generated by WordPerfect; having written the standard, Microsoft is a little better at conforming to it. Under some circumstances it may benefit you to run your WordPerfect document into Word, then save as RTF. A list of known conversion bugs is below.
• In order to be converted, your graphics have to be incorporated in the RTF file (as opposed to being linked as external objects). This is done differently by different word processors; consult online help for the word processor you use.
• Sizing and scaling of graphics is done by applying a one-size-fits-all scaling algorithm that sometimes won't fit any; it tends to err on the large side. If you like, you can adjust this; see the sections on conversions below.
• The PDF conversion (in particular) is fairly "broad-brush" -- fonts are available in only five sizes (extra small, small, medium, large, and extra-large) and a few families (eg. proportional-Times-Roman-like, sans-serif-Helvetica-like, and so on). Fine gradations of type size and style probably won't come through. As with many other problems it is well to bear in mind that the reader will have access to the actual document as it originally appeared in the RTF version, if fine points of appearance are critical.
• Some special characters are not available in HTML, and hence will not translate.
• Other character-translation problems arise because the various conversion programs we use don't understand some characters in some character sets. In PDF versions, this usually results in an empty rectangle (looking something like a check box) in your document where the character in question should be. We can improve this in future versions if you report the problem to us .

It's worth remembering that readers have access to the original RTF version if an absolutely accurate version is needed, and that submitters have the option of replacing our converted files with their own should they require something different or better than what LEDA provides.

We are especially eager to hear about conversion problems -- so let us know of any that you encounter.

Conversion into HTML -- the technical process.

LEDA uses the popular rtf2html program for conversion to HTML. Its behavior is controlled by configuration files found in the /usr/local/leda/etc directory, and in /usr/local/leda/bin/rtf2html . You also have control over the translation of special characters and the like, and of image scaling. We have found that rtf2html is sufficiently feature-rich that we don't need to supplement it with pre- or post-filtering scripts, though we may encounter a need to do so in the future.

Full documentation for rtf2html is in HTML format on your LEDA server at /leda/manual/r2hdocs/guide.htm. The documentation on the Logictran site is for a newer, commercial successor to rtf2html but most of the same features are present. You may alter the configuration files to suit your needs -- and specify alternate configurations for different series you set up in LEDA. This is particularly helpful if you want different logos or stylesheets for different journals, etc.

To set up a different HTML "look" for a series:

• Create a new html-trn file to control the conversion. It will probably be easiest to copy the "vanilla" html-trn file (/usr/local/leda/etc/html-trn) under a new name, then edit the new file in accordance with the documents at /leda/manual/r2hdocs/guide.htm .
• Associate the new html-trn file with the series, by editing via /leda/review/view_all_serials.php3

Conversion into PDF -- the technical process

Unfortunately, Adobe does not sell a version of Acrobat Distiller that will run under Linux (at least not at a reasonable price; there is an enterprise-scale document conversion server product that creates PDF that goes for around $5K at the time of this writing). As a result, we use a multistep process that first transforms the RTF document into LaTeX, and from there to PDF:

• The RTF document is pre-filtered. This step is basically used to suppress features (like document comments) that we don't want to see in the PDF document.
• The RTF is converted to laTeX using the rtf2latex2e package.
• The intermediate laTeX file is filtered; this adjusts footnote numbering, takes care of graphics scaling, and so on.
• The laTeX file is converted to pdf.

If you're a serious conversion hacker, you can work magic by customizing the various scripts involved. Depending on what you want to do, you may find it easier to alter one or the other of our filtering scripts. We've found it easier to use the RTF pre-filter to remove information and the laTeX filter to add or alter information.

Where to find more

/usr/local/leda/bin/ledacvt.sh is the shell script that controls the conversion of LEDA documents; it in turn calls other programs found (for the most part) in the /usr/local/leda/bin directory tree. Looking at the source is probably the best way to understand exactly what's being done.

What to do if our conversions don't satisfy you

Known conversion bugs (last modified 3 January 2002):

• Hanging indents don't indent if the document originated in WordPerfect.