% dolines.doc  --- version 1.0 (15 April 1991)
% This is the documentation for dolines.tex.
% To read this file, process it with (plain) TeX: tex dolines.doc
% (To do that, you need manmac.tex, which has
% the macros for typesetting the TeXbook, as explained in appendix E.
% It comes free with the standard distribution of TeX.)
%
% NOTE: Processing this file with TeX will create empty files
%       index.tex and answers.tex in your directory. And if you
%	have a file pages.tex in your directory, TeX will probably
%	get very confused.
%	The manmac format will try to load a font called cmman; if
%	TeX complains that it cannot find that font you can savely
%	ignore that, since the font is not actually used.
%
% This file: copyright (c) 1991  Marcel van der Goot.
%	You may redistribute this file and printed versions of it.
%
% original: csvax.cs.caltech.edu [131.215.131.131] in pub/tex
%	    (use anonymous ftp). Also in various archives.
%
% Marcel van der Goot
% marcel@cs.caltech.edu
%   Caltech 256--80
%   Pasadena, CA 91125
%   (818) 356--4603

% ------ definitions for manual pages:
% The documentation is typeset using Knuth's ``manmac'' macros, with a
% few additional things defined here. You are free to use or adapt these
% extra definitions.

\input manmac % the macros used for typesetting The TeXbook (appendix E)
\catcode`^=7 % manmac makes ^ active for creating an index
	% we don't make an index, and we need ^ to be non-active

{\catcode`\^^M=12 \endlinechar=-1 % 12 = other
\long\gdef\manpagehead#1^^M#2^^M#3^^M^^M% tex-file, description, version
   {\egroup % matches \bgroup in \manpage
    \par\vfill\eject
    \edef\rhead{\manualname: #1\unskip}%
    \setbox1=\hbox{\titlefont #1\quad}%
    \setbox3=\vtop{\advance\hsize by-\wd1 \parindent=0pt \raggedright
		   #2\par
		   #3\par
		   \cnotice
		  }%
    \line{\box1\box3}%
    \bigskip\bigskip\bigskip\noindent
   }%
} % restore \catcode`\^^M

\outer\def\manpage
   {\bgroup\catcode`\^^M=12
    \manpagehead
   }

\hsize=6.5in\vsize=8.9in % default plain TeX values
\pagewidth=\hsize \pageheight=\vsize

\def\cnotice{Copyright \copyright\ 1991 Marcel R. van der Goot}
\def\manualname{Midnight Macros}

% LaTeX logo, from latex.tex (except for the capital `A': manmac
% doesn't have a real smallcaps font)
\def\LaTeX{{\rm L\kern-.36em\raise.3ex\hbox{\sc A}\kern-.15em
    T\kern-.1667em\lower.7ex\hbox{E}\kern-.125emX}}

\tenpoint

% (end of extra definitions)

% -------------------- dolines.doc:

\pageno=9 % follows loop.doc

\manpage dolines.tex
Meta-macros to separate arguments by newlines and by empty lines.
Version: 1.0 (15 April 1991) (doc: 15 Apr 1991)

The macros in dolines.tex are intended to help you write your own
macros, rather than to typeset text directly. Consequently, it is
assumed that you have some experience writing macros; in other words,
assume that this whole section is preceded by a ``dangerous bend'' symbol.

Dolines.tex\footnote*{To use dolines.tex you also need Midnight/loop.tex,
which will be |\input| when you |\input dolines|.}
helps to write macros for which line breaks inserted
by the user are important. When the definitions from dolines.tex
are in effect, text consists of paragraphs separated by one
or more empty lines (as usual). Each paragraph consist of one
or more lines, where a line is terminated by a normal newline character,
`|^^M|' (normally, \TeX\ treats newlines as if they are spaces). Note that,
as far as dolines.tex is concerned, a paragraph can only be
terminated by an empty line, not by, e.g., |\par| or |\vskip|.
To use dolines.tex, you have to define three macros, |\beforelines|,
|\everyline#1|, and |\afterlines|. When dolines.tex detects the beginning
of a paragraph, |\beforelines| is called. Next, for every line,
|\everyline| is called with as argument a single macro. That macro
has as its expansion the line that was read. (Usually, this is almost
the same as saying that |\everyline| is called with the line as argument.)
After the last line of the paragraph, |\afterlines| is called.
No output is generated, other than what is produced by your own macros.

{\def\fnm{\<\rm file name>}
There are two ways of using dolines.tex. One is to read lines from
a file, the other is to take the lines from the current input file.
If you say
\begintt
\filedolines{|fnm}
\endtt
dolines.tex will attempt to read the file, calling |\beforelines|,
|\everyline|, and |\afterlines| as described above. The file is read
inside a group, so you may need to use |\global| if you do assignments
within, e.g., |\everyline|.
}% \fnm

As an example, we define a ``poetry environment'' which reads a poem
from a file, preserving the poet's line breaks. We draw horizontal
lines around the poem, and put every stanza in a vbox to prevent
page breaks. (It is certainly possible to write better poetry environments,
we are just illustrating dolines.tex.)
\beginlines
|\input dolines|
\nobreak\smallskip\nobreak
|\def\readpoem#1{\medskip \hrule \filedolines{#1}\hrule \medskip}|
\smallbreak
|\def\beforelines{\medskip \vbox\bgroup}|
|\def\everyline#1{\hbox{\kern\parindent \it #1}}|
|\def\afterlines{\egroup \medskip} % end of the vbox|
\endlines
If file ``epic.tex'' contains a poem, we can typeset it with
|\readpoem{epic}|.

However, if you are typesetting several short rhymes, it is somewhat
inconvenient to have each of them in a separate file. This is where
the second method of using dolines.tex is useful. You have to define
a macro |\enddolines|, the expansion of which should be
the text you want to use to terminate
dolines.tex's effect. You also have to define a macro |\finishdolines|.
When you say |\begindolines|, dolines.tex starts reading lines,
calling |\everyline| etc. This continues until dolines.tex reads a
line equal to the (top level) expansion of |\enddolines|. Then your
macro |\finishdolines| is called, and \TeX\ resumes normal processing.
|\begindolines| and the line equal to |\enddolines| form a group
(as with |\filedolines|); |\finishdolines| is called {\sl outside}
this group.

Here is how we can extend the poetry environment with two control
sequences |\beginpoem| and |\endpoem|. We keep the above definitions
of |\beforelines| etc.
\beginlines
|\def\beginpoem{\medskip \hrule \begindolines}|
|\def\enddolines{\endpoem}|
|\def\finishdolines{\hrule \medskip}|
\endlines
Now we can typeset a poem as follows (the English stanza is from The
\TeX book):
\beginlines
|\beginpoem|
\nobreak\medskip\nobreak
|Roses are red,|
|\quad Violets are blue;|
|Rhymes can be typeset|
|\quad With boxes and glue.|
\nobreak\medskip\nobreak
|Blauw de viooltjes,|
|\quad Rood zijn de rozen;|
|Een rijm kan gezet|
|\quad Met plaksel en dozen.|
\nobreak
|\endpoem|
\endlines
Note that we never defined |\endpoem|, as it is never expanded. In fact,
there is no reason to terminate the poem with a control sequence. If we
define |\def\enddolines{***}|, three stars on a single line will play
the r\^ole of |\endpoem| above.

\danger There are a few things to be careful about. First, a line
must have properly nested |{...}| groups. (Because a line is read as
an argument delimited by a \<return> character.) You should not use
this to fold long lines, because that will insert an explicit \<return>
character with catcode 12 in the line. Second, spaces at the beginning
of a line are skipped, unless you change the catcode of `\]'. This means
that, for instance, there can be spaces in front of |\endpoem|.
(Spaces are skipped at the beginning
of a line because at that point \TeX\ is in ``state N'' --- see Chapter
8 of The \TeX book.) Third, when |\everyline| is called, the line
that is its argument is already tokenized. This means that a catcode
change within |\everyline| cannot effect this line. Furthermore,
when |\beforelines| is called, the first line is already read and
tokenized.

\danger |\filedolines| and |\begindolines| treat lines in pretty much
the same way. There is however a significant difference in the way
comments are treated. If a line has a comment (i.e., it ends with
`|%...|'), |\begindolines| will miss the \<return> at the end of the
line. This is normal for \TeX, it can be used to fold long lines.
However, if you are using |\filedolines|, comments will be skipped,
but lines are not concatenated: |\filedolines| will not miss the
\<return>! (This is because |\filedolines| uses \TeX's |\read| primitive.)
A rather obscure difference
is that a line consisting of just `|\par|' is read as an empty line
by |\filedolines|, whereas it is read as a non-empty line by |\begindolines|.
If you want to use |\begindolines| to read from a file, you can say
|\expandafter\begindolines\input ...|.

It is possible for |\finishdolines| to take arguments, but they have to appear
on the line following the line with the expansion of |\enddolines|. Suppose
we change the poem environment by defining |\finishdolines| as
\begintt
\def\finishdolines#1%
   {\line{\strut \hss --- #1\hss\hss}\hrule \medskip}
\endtt
Hence, |\finishdolines| will print the author's name under the poem.
The argument should be written on the line following |\endpoem|:
\begintt
\beginpoem
   ...
\endpoem
{A.U. Thor}
\endtt                     

Sometimes we want to use dolines.tex to read just one paragraph by making
|\enddolines| equal to an empty line. For instance, a macro |\address| which
reads an address terminated by an empty line. The problem with defining
|\def\enddolines{}| is that if there is an empty line before the address
(or even if the rest of the line after |\address| is empty), no lines
will be read at all. To ensure that at least one paragraph is read, we use
|\beforelines| to define |\enddolines|:
\begintt
\def\beforelines{\normalbeforelines\def\enddolines{}}
\let\enddolines=\relax
\endtt
Here, |\normalbeforelines| denotes whatever must be done at the beginning
of the address. Initially |\enddolines| is equal to |\relax|, but as soon
as the first non-empty line is found, |\enddolines|
is redefined as an empty line. Following that, an empty line causes calls
of |\afterlines| and |\finishdolines|. (A way to include an empty line in
the address is to write a tie (`|~|') on a line of its own.)

\danger As a more complicated example, we design a simple verbatim
environment. The verbatim environment treats horizontal tabs as spaces,
and considers multiple empty lines equal to a single empty line.
We first write macros |\setupverbatim| and |\finishverbatim|. The first
sets up the character codes for verbatim printing, the second resets
those character codes.

\beginlines
|\def\setupverbatim|
\nobreak
|   {\medskip \hrule \bgroup|
|    \def\do##1{\catcode`##1=12 }\dospecials|
|    \obeyspaces \tt|
|   }|
\medskip
|\def\finishverbatim{\egroup \hrule \medskip}|
\medskip
|{\obeyspaces\global\let|\]|=\|\]|}|
\endlines
Grouping is used to keep the catcode changes local. Plain \TeX's |\dospecials|
macro applies the macro |\do| to all special characters (`|\|', `|$|', etc.),
which, in this case, gives them all catcode 12 (``other''). But we don't want
catcode 12 for spaces, because the |\tt| font has a visible space character
`\]'. Therefore, spaces are made active with |\obeyspaces|, and in the
last line we make an active space equal to \hbox{|\|\]}.

To start setting verbatim text, we use |\beginverbatim|; |\readverbatim|
is used to typeset a whole file in verbatim mode.
The macros for typesetting lines and paragraphs are very straightforward.
\beginlines
|\def\beginverbatim{\setupverbatim\begindolines}|
|\let\finishdolines=\finishverbatim|
\medbreak
|\def\readverbatim#1{\setupverbatim\filedolines{#1}\finishverbatim}|
\medbreak
|\def\beforelines{\smallskip}|
|\def\everyline#1{\hbox{\strut #1}}|
\nobreak
|\def\afterlines{\smallskip}|
\endlines

Finally, we have to define |\enddolines|. We want to match |\beginverbatim|
with |\endverbatim|, so we would like to do
\begintt
\def\enddolines{\endverbatim}
\endtt
The problem is that this makes |\enddolines| expand to the {\sl control
sequence} |\endverbatim|. But while we are in verbatim mode, no control
sequences are recognized, because `|\|' is not an escape character.
What we really want is to make the expansion of |\enddolines| equal
to the {\sl string} ``$\backslash$endverbatim''. This is done by
temporarily making `|\|' a normal character and `|/|' an escape
character:
\beginlines
|{\catcode`\/=0 /catcode`/\=12|
| /gdef/enddolines{\endverbatim}|
\nobreak
|}|
\endlines
This completes the verbatim environment. Note that |\endverbatim| is
only recognized if it appears on a line by its own and is not preceded
by spaces. If you are not afraid of confusion, you can change the
name |\finishverbatim| to |\endverbatim|.

\bye