First Blood

Below is the sample paragraph by 3B2.

It has many disadvantages:

• quite loose lines
• no ligatures
• no optical margins
• use of tracking

Disproportionate Response

My first try was with the state of the art pdftex with activated font expansion and optical margins through the LaTeX package microtype without any tweaking. The result is smashing: pdftex produces much better, thight lines, ligatures, kerning, and optical margins. Hyphenation is also correct.

Round 2: Classical K.O.

In my second try I used the “classical” TeX without the pdftex-extensions. I adjusted the default LaTeX typesetting parameter to discourage the hyphenation of consequent lines. Even so, the result ist still much better than the paragraph from 3B2 (as matter of fact, it is quite identical with the one above):

Conclusion

As expected, TeX typesetted the paragraph better than 3B2. I don’t know, if there are better options for line breaking in 3B2, but since this is an example from a real book, which is reviewed by the publisher and the typesetter, I suppose 3B2 should have already been set for best possible line breaking.

The LaTeX code follows:


\documentclass[a4paper]{article}
\usepackage[T1]{fontenc}
\usepackage[latin1]{inputenc}
\usepackage[ngerman]{babel}
\renewcommand{\rmdefault}{padj}
\usepackage{soul}
\pagestyle{empty}
%
%% parameter for pdftex
% \usepackage[activate]{microtype}
%% parameter for pdftex
%
%% parameter for plain TeX
\tolerance=1000
\doublehyphendemerits=100000
\hyphenpenalty=500
%% parameter for plain TeX
%
\newcommand*{\cdu}{{\fontsize{8.97bp}{13pt}\selectfont C\kern1ptD\kern0.6ptU}}
\begin{document}
\setlength{\textwidth}{250.7bp}%
\noindent\fontsize{10pt}{13pt}\selectfont
Die \cdu{} vereinte nach \caps{1945} eine Reihe von Konfliktgegnern aus der
Zeit vor \caps{1933}: Süddeutsche und Norddeutsche, Bürger und Arbeiter,
Katholiken und Protestanten, Deutschnationale und auch Liberale. Eine
sichere Identität hatte die Partei infolgedessen nicht. Die Mitglieder und
Anhänger teilten nicht kollektiv die Herkunft, nicht die Weltanschauung, nicht
die sozialen Interessen, nicht die Tradition. Daher konnte die \cdu{} auch nur
Union sein und durfte lange nicht Partei werden. Sie musste die heterogenen
bürgerlichen und konfessionellen Milieus gewissermaßen friedlich und autonom
koexistieren lassen, durfte sie nicht in eine offene Diskussionsschlacht über
Programme und Richtungen schicken. Eine freie und lebendige Diskussionskultur,
eine moderne, partizipatorische Parteistruktur hätte die \cdu{} zerstört,
hätte die neuerliche parteipolitische Zersplitterung der bürgerlichen und
kirchlichen Lebenswelten ziemlich gewiss zur Folge gehabt.
\end{document}


1. same entries
2. same sorting

Further editing of the online references includes for example

• abreviation of the author/editor names
• inclundig and formatting the needed fields

The publisher’s demands for the online references are often included in the
DTD describing the XML tags for the online documents. In the worst case the
online formatting differs from the print formatting. Nevertheless, if there is
already a LaTeX/BibTeX workflow, and especially a BibTeX style file, it is
better to use directliy BibTeX (the program) to produce the XML output instead
of external converters. Using BibTeX guaranties automatically that the both
items above are true. Further, if one is familiar with the BibTeX programming
syntax and with the used style file, it is easy to modify the functions to
produce the XML output, while keeping the framework unchanged.

Below is an example of a BibTeX function which formats the author names
acoording to the Springer A++ DTD. It is part of an BibTeX style file, which I
wrote years ago for the print and online references in the Springer series
Topics in Applied Physics. The first names are abbreviated and put in
<Initials> tag, the whole name is in <BibAthorName> and so on.

FUNCTION {format.authornames.online}
{ 'bibinfo :=
  duplicate$ empty$ 'skip$ {
  's :=
  "" 't :=
  #1 'nameptr :=
  s num.names$ 'numnames :=
  numnames 'namesleft :=
    { namesleft #0 > }
    { " <BibAuthorName> "
      s nameptr "{f.}" format.name$
      duplicate$ empty$
        { " <NoInitials/>" * * }
        {  "<Initials>" swap$ * * "</Initials>" * }
        if$
      s  nameptr
      " <FamilyName>{ll}</FamilyName> " format.name$ *
      s nameptr "{vv{}}" format.name$
      duplicate$ empty$
        { * }
        { "<Particle>" swap$ * * "</Particle>" * }
        if$
      s nameptr "{jj{}}" format.name$
      duplicate$ empty$
        { * }
        { "<Suffix>" swap$ * * "</Suffix>" * }
        if$
      "</BibAuthorName> " *
      bibinfo bibinfo.check
      't :=
      nameptr #1 >
        {
          namesleft #1 >
            {  " " * t * }
            {
              "" *
              s nameptr "{ll}" format.name$ duplicate$ "others" =
                { 't := }
                { pop$ }
              if$
              t "others" =
                {
                  " " * bbl.etal.online *
                }
                { " " * t * }
              if$
            }
          if$
        }
        't
      if$
      nameptr #1 + 'nameptr :=
      namesleft #1 - 'namesleft :=
    }
  while$
  } if$
}

So, processing for example the bib entry

yields to the XML code

The XML code can be then validated and integrataded in the whole online file.

Der Haken an der Sache mit dem Häkchen liegt aber in der deutschen Rechtschreibung. Denn sie bestimmt, kurz gesagt, dass keine Ligaturen an den Wortfugen gesetzt werden dürfen. Während der TeX-Quelltext Listen- (rmligs) oder Silben-basiert (ligatex) bereinigt werden kann, scheint die Ligaturentrennung für InDesign & Co. (hier und hier) eine ziemlich aufwändige Gelegenheit zu sein. Und so breitet sich die Cool-Feature-Plage mit den Wortfugenligaturen in gigantischen Ausmaßen aus. Keinen Verlagshersteller und keinen Lektor scheint diese Art des Analphabetentums zu stören, nein schon “Auflage” im Impressum wird mit der falschen Ligatur gesetzt.

Fazit: InDesign & Co. sind für den deutschsprachige Werk- und Mengensatz ungeeignet!

PS Ich lese zur Zeit ein Taschenbuch, das in Minion mit Ligaturen an den Wortfugen gesetzt ist. Immerwieder stolpere ich an die häufig vorkommenden “Dorfleute” und “auffinden”, und wäre es nicht wie immer die grandiöse Erzählung Appelfelds, hätte ich das Buch längst beiseite gelegt.

PPS Da im Deutschen beliebig zusamengesetzte Wörter erschaffen werden können, ist es ratsam, die Ligaturen vor dem Endsatz zu überprüfen. Die Funktion “occur” von Emacs ist sehr hilfreich für den schnellen Überblick.

\include uses \@input@ in its definition to load a file and \@input@ is defined as

The simplest way to change \@input@ ist to \let it to \@iinput, the macro, which issues an error if file is not found.

Since \@input@ is not only used in the \include macro, this redefinition affects the input of .fd and .bbl files. You can see this as some kind of strict (or final, because of BibTeX) mode LaTeX, producing error on missing files.

Of course, we can redefine only the \include macro, just by copying its definition and replacing \@input@

#001 Springer, Anne; Münch, Karsten; Munz, Dietrich (Hg.)
#002 Sexualitäten
#003 Eine Publikation der DGPT
#004
#005 Bestellnr.: 854 #006 ISBN 978-3-89806-854-3
#007 2008 #006 455 Seiten #006 Gebunden
#008 ? (D) 39,90
#009 Die Autorinnen und Autoren dieses Bandes reflektieren aus heutiger Sicht
mit unterschiedlichen Perspektiven ein zentrales Thema der Psychoanalyse in
Überlegungen zur Theorie und Behandlungstechnik: die Sexualität. Im Zentrum
steht dabei die Aktualität des Triebkonzepts.
#010
#011 9783898068543

After automatic cleaning and processing the file, we produced a LaTeX formated data:

The pictures for the book covers were delivered as jpeg files. We used ImageMagic to add shadows to them. The \cover macro also examines the year of publication and adds via overpic a “NEW” label to the recent publications. Author names and titles are linked using hyperref to the particular publisher’s website. Clicking on an author’s name for example leeds to the website with all publications from the same person listed. Since every person was marked, it was easy to produce an index as well.

At the end, some few corrections in the content and the layout had to be made. Also the hyphenation was corrected and some pagebreaks were shifted. As a result we produced a fully linkend, (almost) automatically produced PDF file. It is online now and available here.

And the best part was, no XML was ever envolved in the production

We finally received a heavy marked text as commented PDF with the index number in each comment. Almost every page looked like this:

The index had to be generated very quickly and obviously there was no way to do that by marking each entry in the source TeX file. Fortunately it was confirmed, that the book preview is OK and no page should change.

We used the Acrobat feature of summarizing the comments in one PDF file to merge all the comments with the page numbers:

We converted the PDF file with the comments to plain text and used sed &co. to obtain a LaTeX-like markup in the end:

The first argument of the \sachindex macro is the page number, in the second argument all the index numbers are placed as a comma separated list.

The list with the numbered index entries was also converted to a number dependent macro:

The last step was to program both macros to write the apropriate \indexentry in an .idx file.

Special attention was required to the beginings end endings of an index range. For this purpose we generate 3 macros from an index number: normal, begin range and end range.

Finally we obtained an idx file with 3087 index markings. After processing with MakeIndex they resulted in 749 index entries.

This way a job could be completed within a few hours when normally it would take days.

1. create new phonetic glyphs from the base font
2. create albanian hyphenation patterns based on the turkish ones with some additions like not to break on digraphs
3. vertical bars and underrules used in the main entry words for showing morphology and intonation disappear (automagically) in the running head, so there is only the whole word typesetted (example below)

The main work actually was to clear and normalize the Word sources. Emacs did a good job as usually.