|
|
|
|
|
|
|
|
|
|
How much does GutenMark cost? It's free, licensed to you under the GPL.What if I want to use GutenMark in a commercial application, such as print-on-demand, or to set up a website for automatically prettifying PG texts? Fine, do it! I'll be happy to encourage this in any way I can, if you let me know what you're trying to do. Again, consult the GPL if you are thinking of distributing the GutenMark program itself. Also, you will want to carefully read the Project Gutenberg " fine print" to determine if what you want to do is acceptable to Project Gutenberg.
Why use GutenMark rather than another text-to-html markup utility? Although GutenMark is a text-to-html markup tool, it is not a general-purpose utility. It is designed to correct the deficiencies of books that are in plain-vanilla ASCII format, and (specifically) Project Gutenberg etexts. The goal is 100% automatic publishable-quality markup. In other words, to produce books that look as if they had been published. General-purpose tools are not really suited for this, but it doesn't hurt to try them. Let me know if any of them do a really good job, and I'll take a look at them. Here are some samples from the most promising general-purpose text-to-HTML and text-to-PDF tools I've found. To get an apples-to-apples comparison, all samples have been converted to PDF, using the Times font. (In viewing them, ignore things like goofed-up page headings, because these are my fault from not wanting to spend a lot of time figuring the tools out.) Remember: No manual markup or editing has taken place.
Can GutenMark be used to make PDF for my non-PG etexts? Yes and no:What's the status of GutenMark? GutenMark has reached the stage of being pretty suitable for personal use. It should also be very useful for anyone intending to manually mark up a PG etext in HTML, since it does most of the work for you. For a commercial printing operation -- e.g., a print-on-demand service -- GutenMark can use some improvement. For a list of things that GutenMark can't do (or perhaps, can't do well), look at the buglist.
- Yes. GutenMark is tailored to Project Gutenberg etexts, but for versions 20011227 and later, it does work with plain-vanilla ASCII etexts from sources other than PG. Try it and see, I guess.
- No. If the etexts you have are already in HTML, then GutenMark can do nothing for you since its only real functionality is to produce HTML. However, you can use the same freely-available utilities that GutenMark uses to convert HTML to PDF (namely, html2ps and ghostscript), and you can use GutenMark 's configuration files for html2ps if you like the sample pages you've seen here. Refer to the download and usage webpages.
GutenMark seems to work much more slowly than the early versions. That's true. GutenMark is now able to use wordlists and namelists to help it work more intelligently, but this intelligence comes at the cost of speed. Processing the wordlists isn't really dependent on the etext file's size -- in other words, it adds roughly the same amount of time for big etexts as for small etexts -- and so the speed difference seems more obvious, and is more objectionable, if you process a small test etext. Here's a rough speed comparison made on my 450 MHz iMac, processing the 400 Kbyte etext file TMOTB10.TXT.
GutenMark
"profile"Wordlists Processing
time--profile=english
(the default)American names, English, French, German, Italian, Latin, Spanish 25 seconds --profile=none none 4 seconds --profile=english_all American names, Danish, English, Finnish, French, Gaelic, German (2), Italian, Latin, Norwegian (monstrously big), Spanish, Swedish 66 seconds I find this acceptable, but if you don't, here are a few things you can do about it, short of getting a faster computer. :-)
(NOTE: Since the Q/A above was written, the available wordlists have increased somewhat: as of 12/23/01, they contain about 4 million words, 12 Mbytes compressed, 45 Mbytes uncompressed. I assume that the benchmarks using the complete set of wordlists would slow down proportionally.)
- The most drastic thing you can do, though I don't recommend it, is to delete the wordlists and namelists ("*.names.gz" and "*.words.gz") by simply erasing them, or to run GutenMark with the command-line switch "--profile=none". You will restore most of GutenMark's speed, but will eliminate the many advantages that the wordlists provide.
- Edit GutenMark.cfg file to create a profile having just the minimum set of languages compatible with your interests. How much speed you regain depends on the languages you choose. Frankly, you only need wordlists for languages with words likely to appear in the texts you're reading. So, if you're reading Horatio Alger and L. Frank Baum exclusively, feel free to eliminate everything but English. If you're reading Thomas Carlyle, you're going to need proper names, English, German, French, Latin, Italian, and probably some of the Scandinavian languages as well!
- Use uncompressed wordlists. GutenMark can use the wordlists and namelists in either compressed or uncompressed form. As distributed, they are compressed. If you gunzip the wordlists and namelists, and modify GutenMark.cfg to show the uncompressed files (english.words rather than english.words.gz, etc.), you may find that the program operates more quickly. For example, the '--profile=english_all' example in the table above was reduced to 42 seconds. Unfortunately, the wordlists are normally compressed by about a 5-to-1 ratio, so the wordlists in this example will be increased from about 6 Mbytes to over 27 Mbytes. Ultimately, it's a question of sheer immensity of data; these wordlists contain over 2.2 million words, and obviously this takes some time to process. And, I intend to add more and more wordlists as time goes on.
Why does GutenMark discard the Project Gutenberg file header by default? Is this even legal? If you refer to the Project Gutenberg standard file header (an example of which may be seen here), under the section titled DISTRIBUTION UNDER "PROJECT GUTENBERG-tm", you'll note that Project Gutenberg specifically requires the header (and all other references to PG) to be removed if the etext has been changed. It's unclear whether GutenMark changes the etext sufficiently to activate this clause, but in any case removal of the header is always allowed. Therefore, the default is to remove the header. You can restore the PG header with GutenMark 's "--yes-header" command-line option. If you do so, please keep in mind that complying with PG's requirements is entirely your responsibility.
Why does the HTML produced by GutenMark look funny in my browser? Each browser tends to have its own individual quirks that limit the accuracy with which it can display HTML correctly. In other words, browsers (even very popular ones that I won't name) don't necessarily follow the HTML standards as closely as you might like. You can check some of your own browser's capabilities by looking at the following table.
Description What your browser displays long dash (em-dash) - short dash (en-dash) - soft hyphen should be invisible curly left double-quote " curly right double-quote " curly left single-quote ` curly right single-quote ' And, there's an additional problem. HTML allows two separate ways of representing special symbols (like those in the table above), the numeric way and the symbolic way, and your browser's quirks may be different in the numeric mode than in the symbolic mode! By default, GutenMark uses the numeric mode, because browsers tend to support it better. To use the symbolic mode instead, the '--force-symbolic' command-line switch is available. This may or may not work differently with your browser, but will definitely produce more readable raw HTML if additional manual markups are going to be made.
Remember, though, that the goal of the GutenMark project is to produce good-looking printouts, or good-looking PDF-based online displays, and only secondarily to produce good-looking browser-based online displays.
Why does GutenMark waste disk space by using custom "wordlists" instead of the spelling dictionary already installed on the computer? GutenMark is designed to be very portable, and the types of spelling dictionaries available differ greatly from one computer platform to the next. Most GutenMark wordlists are derived from the spelling dictionaries of the ispell program, which are installed on many Linux computers. However, even on a Linux platform, some of these ispell dictionaries have technical deficiencies from GutenMark 's standpoint. For some languages (such as Latin), no comprehensive ispell dictionary has been previously available. Certainly there is no ispell dictionary of personal and geographical proper names. Therefore, the choice has been made to produce a set of custom wordlists used only by GutenMark, even if it has the unfortunate side effect of increased download-times and disk-usage.
What if I want to mirror the GutenMark website? Swell! Just make sure you use the material as-is, without change. Let me know about it, and I'll provide a link.
©2001 Ronald S. Burkey
Last updated 12/28/01 by RSB. Contact me.