TeX 1995 July/TeX CD-ROM July 1995 (Disc 1)(Walnut Creek)(1995).ISO/macros/text1/texix/texix.web

Text File | 1991-10-23 | 49.7 KB | 1,578 lines

% Version 1.3 makes it easier to create other changes files (6/89)
\font\twelvept=cmbx12
\font\tentex=cmr10
\def\topofcontents{\null\vfill\eject
    \def\titlepage{T}
    \centerline{{\twelvept The \TeX IX Index Program}}
    \centerline{{\twelvept For IBM VM/CMS Pascal/VS}}
    \vskip15pt
    \centerline{Version 1.3, June 1989}
    \hbox{\vbox{\hsize\the\hsize This work is
   protected as an unpublished work under U.S. copyright laws.
   Copyright $\copyright$ 1988 by WSUCSC.  All rights Reserved.}}
   \vskip18pt
    \hbox{\vbox{\hsize\the\hsize This software is furnished under a
    license for
    use only on a single computer system and may be copied only
  with the inclusion of the above copyright notice.
 This software, or any other copies
 thereof, may not be provided or otherwise made available to any
 other person except for use on such system and to one who agrees to
 these license terms.  Title to and ownership of the software shall
 at all times remain in WSUCSC.}}
    \vfill}
\let\tentex=\tt
\def\_{\leavevmode \kern.06em \vbox{\hrule width.3em}}%
%
\def\}{\ifmmode \rbrace
          \else $\rbrace$\fi}%
%
\def\{{\ifmmode \lbrace
          \else $\lbrace$\fi}%
%
\def\us#1{$\underline{\smash{\hbox{#1}}}$}%
\def\lin#1{\par
                  \leftskip 0pt
                  \advance \leftskip by #1
                  }%
\def\vs#1{\vskip #1\relax}
\catcode`*=11
% labeled definition macro
%
\newdimen\la*w   \la*w=1in         % Label width
\newtoks\la*g    \la*g={1em}       % Label gutter
\newtoks\la*s    \la*s={12pt}      % Skip before label
\newtoks\la*f    \la*f={\rm}       % Label font
\newbox\tempbox
%
\long\def\label#1{\par\vskip\the\la*s
             \setbox\tempbox=\vtop{\hsize=\la*w
                                  \leftskip=0pt
                                  \rightskip=0pt plus2em
                                  \tolerance=1600
                                  \noindent
                                  \the\la*f
                                  #1}%
             \hangindent=\la*w
             \advance \hangindent by \the\la*g
             \hangafter=1
             \noindent
             \setbox0=\hbox{\rlap{\box\tempbox}}\ht0=0pt\dp0=0pt\box0%
             \hskip\la*w
             \hskip\the\la*g
             \ignorespaces}
\let\la=\label
\catcode`*=12
\let\nin=\noindent
\def\lbr{\null\hfil\break}
\def\PASCAL{{\sc PASCAL/VS}}
\def\hs#1{\hskip#1}
\def\text#1{{\it \TeX T#1}}
\def\9{\it}
\def\bs{$\backslash$}
% Version 1.0; Dean Guenther; 4/1/86
%
@* Introduction.
This procedure was created to give \text1\  the ability to produce a sorted
index in the same one pass, instead of creating the index, sorting it, and
printing the index out in three separate steps.

To accomplish this feat, a new control sequence, \bs{}sortindex, was added to
the basic \TeX\ program which gives the \text1\  user up to three sorted
index files per run. The format of this new control sequence is
\par{\tt\indent\bs{}sortindex\it n}
\par
\nin where {\it n} is the index file number: 1, 2, or 3.

@ The @^TEXIX@>
program is written entirely in WEB, except for an external
procedure, {\it plsort}, which is written in @^PL/1@> PL/1.
This was necessary since
\PASCAL\  cannot call
@:plsort@>
Syncsort @^SYNCSORT@>
to do an internal sort, and PL/1 can.

@
Also, it should be pointed out early that I made use of \PASCAL'
nifty string handling capability, in particular, the |string|,
|index| and |substr| functions. If needed, these should not be too
difficult to translate into another \PASCAL\
compiler (famous last words).

@p
segment texix;
procedure texix(ix:char); external;
procedure texix;
const @<Global Constants@>
type @<Global Types@>
var @! m,k,i,j :integer;@/
@<Global Variables@>

@ The file {\it ix\_file}
is the file opened for output in the \text1\  session
itself. It will be closed in TEXIX, opened for input, closed again, and
finally opened for output where the final sorted index file will be
written to, complete with all \text1\  markup necessary for printing
the index.

@<Global Var...@>=
sysprint:text;
@!ix_file :text;

@ The {\it sort\_file} is used for the internal sort only. The \text1\  user
never has access to it.

@<Global Var...@>=
@!sort_file :file of sort_type;

@* Macros and definitions.
Here are some macros and definitions used throughout.

@d incr(#) == # := # + 1
@d decr(#) == # := # - 1

@d othercases == otherwise
@d endcases == end;

@d dosubstr == substr
@d doindex == index

@f othercases == else
@f endcases == end

@d getout == return
@d messages == sysprint

@d max_field = 300
@d max_levels = 3
@d max_lrecl = 2048

@d remove_characters(#) == if length(in_record) > # then
in_record := dosubstr(in_record,1 + #)
else in_record := ''

@d do_nothing == begin end

@d do_sort_as == begin remove_characters(6);
i := doindex(in_record,'{$}');
write_sort_chars(i-1);
remove_characters(3);
end

@d check_case == begin
if not respect_case then for j:= 1 to
sort_part[i].field_lngth do sort_part[i].field_level[j] :=
upper_case(sort_part[i].field_level[j]); end

@d string_type(#) == string(#)

@d ccat_temp ==  @t\hs{2em}@>@=||@>@t\hs{2em}@>


@d add_comma(#) == # := ccat(#,',')
@d add_cmma_blnk(#) == # := ccat(#,', ')

@d do_when_bold == begin
if length(bold_string) > 0 then add_comma(bold_string);
bold_string:=
ccat(ccat(ccat(bold_string,'{\bd '),curr_str_page),'}');
@.\bs{}bd@>
end

@d do_when_underscore == begin
if length(us_string) > 0 then add_comma(us_string);
us_string:=
ccat(ccat(ccat(us_string,'\us{'),curr_str_page),'}');
@.\bs{}us@>
end

@d do_when_italic == begin
if length(rm_string) > 0 then add_comma(rm_string);
rm_string:=
ccat(ccat(ccat(rm_string,'{\it '),curr_str_page),'}');
@.\bs{}it@>
end

@d do_when_roman == begin
if length(rm_string) > 0 then add_comma(rm_string);
rm_string:=ccat(rm_string,curr_str_page);
end

@d do_write(#)==write(#)

@d do_lnwrite(#)==writeln(#)

@ This is arbitrary, but
there should never be more than 8 indicies.

@d s_file_number == '9'

@ This is used to print the subentry1 header if there are no page
numbers under its subentry and there are under subentry2.

@d id2_missing_check==
curr_level:=2;
if sort_record.sort_part[3].field_lngth=0
then do_nothing else begin
id2_is_missing:=true;
write_header(sort_record.print[2].field_lngth,
sort_record.print[2].field_level);
id2_is_missing:=false;
curr_level:=3;end

@ This is used to print the primary header if there are no page
numbers under the primary header, and there are page numbers under
either subentry1 or subentry2.

@d id1_missing_check==if sort_record.sort_part[2].field_lngth = 0
then begin curr_level:=1;
if sort_record.sort_part[3].field_lngth = 0
then do_nothing else begin
id1_is_missing:=true;
write_header(sort_record.print[1].field_lngth,
sort_record.print[1].field_level);
id1_is_missing:=false;
curr_level:=3;end;end
else begin curr_level:=1;
id1_is_missing:=true;
write_header(sort_record.print[1].field_lngth,
sort_record.print[1].field_level);
id1_is_missing:=false;
id2_missing_check;
end

@ This is a little macro used to access {\it write\_header}.

@d write_prev_header==
begin case curr_level of
1: write_header(prev_ln1,prev_pn1);
2: write_header(prev_ln2,prev_pn2);
3: write_header(prev_ln3,prev_pn3);
othercases do_nothing
endcases
end;

@ The function ccat will concatenate two strings together.

@^System dependent code@>
@p function ccat(x,y:string_pass):string_pass;
begin
  ccat := x ccat_temp y;
end;

@* Opening and Closing the files.
This is all very dependent on \PASCAL.

@ The {\it reset\_file} procedure is used to open all files.
One time this is done when first entering TEXIX.
Note that the file name is ``{\tt IX\it n}'' where ``{\it n}'' is
1--3. This can easily be extended to 4--8 by chaning the \bs{}index
markup in TEXT1@@.
The other time the open is done is when opening the sort file, after it
has already been sorted.
This is system dependent. The following will work for \PASCAL.

@^System dependent code@>
@p procedure reset_file(file_number:char);
begin
if file_number = '9' then reset(sort_file,
'NAME=TEXT1$$.OUTSORT.A,LRECL=1857,RECFM=V')
else reset(ix_file,
ccat(ccat('NAME=TEXT1$$.IX',str(file_number)),'.A'));
end;

@
The {\it sort\_file} is opened for output to write the sort records to.
The {\it ix\_file} is opened for output after the sort records have
been sorted and processed again.
This is system dependent. The following will work for \PASCAL.

@^System dependent code@>
@p procedure file_rewrite(file_number:char);
begin
if file_number = '9'
then
rewrite(sort_file,'NAME=TEXT1$$.INSORT.A,LRECL=1857,RECFM=V')
else
rewrite(ix_file,
ccat(ccat('NAME=TEXT1$$.IX',str(file_number)),
'.A,LRECL=2048,RECFM=V'));
end;

@* The Sort Record Description. The record type called
{\it sort\_type} is the record written to the {\it sort\_file}.
If consists of the following:
\la{\bf Bytes}
\la{1}
A one byte {\it record\_type} (0={\bf bold page number};
2= roman page number (the default); 4=\us{underscored} \us{page}
\us{number};
6={\it italic page number}; 9=blind entry).
\la{2--5}
The integer page number. If the page
number is in the preface part of the document, then the page number will
be negative.
\la{6--9}
Used in sorting decending. That means that -1 will be at the top.
If the page number is plus to begin with, then this is set to zero
for sorting, so it will be at the bottom of the list.
\la{10-29} This is the same as page number, unless the page number is
negative, in which case this field is a roman numeral.
\la{30--33\lbr34--333\lbr334--337\lbr338--637\lbr638--641\lbr642--941}
The sort field. There are three arrays in this field. Each array
consists of a 4 byte length, followed by the {\it field\_array} which is
the length of the sort entry or subentry.
It is these fields which will get sorted by @^Syncsort@> Syncsort.
If there was a {\it sort\_as}
used for a sort entry or subentry, that value passed in the {\it sort\_as}
is placed here.
\vs{24pt}
\la{942--945\lbr946--1245\lbr1246--1249\lbr
1250--1549\lbr1550--1553\lbr1554--1853}
The print field.
The description is the same as the sort field.
This is the way the index entry will print, but not necessarily how
it will sort.
\vs{96pt}

@ @<Global Constants@>=
@!max_pn_alpha=20;
@!max_pnum=9;

@ @<Global Types@>=
@!string_pass=string_type(max_lrecl);
@!pass_pn_alpha=string_type(max_pn_alpha);
@!pn_type=packed array[1..max_pnum] of char;
@!pn_alpha_type=packed array[1..max_pn_alpha] of char;
@!field_array  = packed array[1..max_field] of char;
@!field_type = packed record@|
@!field_lngth :integer;
@!field_level  :field_array;
end;
@!sort_type = packed record@/
@!record_type :char;
@!page_number :integer;
@!abs_page_number :integer;
@!page_string :pn_alpha_type;
@!sort_part :packed array[1..3] of field_type;
@!print :packed array[1..3] of field_type;
end;

@* Subroutines. For many of the following subroutines, the following
global variable, {\it current\_level} is needed to indicate what
index level we are presently processing.
0 = the primary index; 1 = subentry 1, and
2 = subentry 2.
{\it on\_a\_roll} is `true' if we have a series of consecutive page
numbers going. `false' otherwise.
In other places, {\it print\_style}=1 for the paragraph style; it is
2 for the dash style; and it is 3 for the indent style.

@<Global Var...@>=
@!curr_level :integer;
@!on_a_roll:boolean;

@ The {\it write\_header} procedure will write out the primary and
subentry level titles from {\it sort\_record.print}.

@p procedure write_header(print_length:integer;print_field:field_array);
var i:integer;
begin
case curr_level of
1: begin
  do_write(ix_file,'\goodbreak\hp ');
@.\bs{}leavevmode@>
@.\bs{}goodbreak@>
@.\bs{}hp@>
  for i:=1 to print_length do do_write(ix_file,print_field[i]);
  case print_style of
   '1':begin
       do_write(ix_file,', ');
       end;
   '2','3':begin
       if id1_is_missing or dot_leadering
          then do_lnwrite(ix_file,' ')
          else do_lnwrite(ix_file,', ');
       end;
   othercases do_nothing
   endcases; end;
2: begin case print_style of
   '1':do_nothing;
   '2':begin
         do_write(ix_file,'\indentsubentry',ix,'1---');
@.\bs{}indentsubentry@>
         end;
   '3':begin
         do_write(ix_file,'\indentsubentry',ix,'1');
       end;
   othercases do_nothing
   endcases;
  for i:=1 to print_length do do_write(ix_file,print_field[i]);
  if (print_style <> '1') or id2_is_missing or dot_leadering
     then do_lnwrite(ix_file,' ')
     else do_lnwrite(ix_file,', ');
  end;
3: begin case print_style of
   '1':do_nothing;
   '2':begin
         do_write(ix_file,'\indentsubentry',ix,'2---');
         end;
   '3':begin
         do_write(ix_file,'\indentsubentry',ix,'2');
       end;
   othercases do_nothing
   endcases;
  for i:=1 to print_length do do_write(ix_file,print_field[i]);
  if dot_leadering
  then do_lnwrite(ix_file,' ')
  else do_lnwrite(ix_file,', ');
  end;
othercases do_nothing
endcases
end;

@ The {\it numeric} function will take a packed array of length
max\_pnum and convert
that array (which is really the page number) into an integer.

@p function numeric(simple_array:pn_type):integer;
var @!i,j_mult,pn:integer;
begin pn := 0;
i := max_pnum;j_mult:=1;
repeat
if i = 1 then if simple_array[1] = '-'
then pn := -1 * pn
else pn := pn + ((ord(simple_array[1])-ord('0')) * j_mult)
else pn := pn + ((ord(simple_array[i])-ord('0')) * j_mult);
j_mult := j_mult * 10;
decr(i);
until i < 1;
numeric:=pn;
end;

@ The {\it get\_numeric}
function will take a packed array of length max\_pn\_alpha and convert
that array (which is really the page number) into an integer.

@p function get_numeric(x_string:string_type(max_pn_alpha)):integer;
var @!i:integer;
begin
readstr(x_string,i);
get_numeric:=i;
end;

@ The {\it strvalue} function takes an integer and converts it into
a string.

@^System dependent code@>
@p procedure strvalue(x:integer; var results:pass_pn_alpha);
var
temp:string_type(max_pn_alpha);
begin
writestr(temp,x);
results:=ltrim(temp);
end;

@ The {\it strconv} function takes an array and converts it into
a string.

@p function strconv(x:pn_alpha_type):string_type(max_pn_alpha);
var i:integer;
temp:string_type(max_pn_alpha);
begin
temp:='';i := 1;
with sort_record do begin
repeat
if x[i] <> ' ' then
temp := ccat(temp,str(x[i]));
incr(i);
until (i > max_pn_alpha) or (x[i] = ' ');
end;
strconv:=temp;
end;

@ The procedure {\it write\_print\_chars} will write the number of
characters indicated
in the parm field to the appropriate print field. There are three levels of
print fields. Level 0 is the primary index, level 1 is the subentry 1 index,
and level 2 is the subentry 2 index. After writing the number of characters
to the print field, that number of characters is removed from the input record.
You might notice that this procedure is very similar to the
write\_sort\_chars
procedure. The only difference in the two is that this procedure deals
with how the index is
to be printed after sorting. The former deals with how the
index is to be sorted.
Also, this procedure is called by {\it write\_sort\_chars}.

@^System dependent code@>
@p procedure write_print_chars(number_of_characters:integer);
var m:integer;
begin with sort_record.print[curr_level+1] do
for m := 1 to number_of_characters do begin
incr(field_lngth);
field_level[field_lngth] := in_record[m];
end;
remove_characters(number_of_characters);
end;

@ The procedure {\it write\_sort\_chars}
will write the number of characters indicated
in the parm field to the appropriate sort field. There are three levels of
sort fields. Level 0 is the primary index, level 1 is the subentry 1 index,
and level 2 is the subentry 2 index. After writing the characters to
the sort fields, they will be removed from the input string.

@^System dependent code@>
@p procedure write_sort_chars(number_of_characters:integer);
var m:integer;
begin with sort_record.sort_part[curr_level+1] do
for m := 1 to number_of_characters do begin
incr(field_lngth);
field_level[field_lngth] := in_record[m];
end;
remove_characters(number_of_characters);
end;

@
This procedure processes subentries within the \bs{}index command. For example,
the phrase ``Mt. St. Helens'' would be processed here for the entry created
by the markup:
\vs{12pt}
{\lin{.5in}
{\tt\bs{}index\{volcanos\bs{}subentry1\{Mt. St. Helens\}\}}\par}
\vs{12pt}
\nin Note
that there are two levels of subentries, \bs{}subentry1 and \bs{}subentry2.
You can
use \bs{}sortas within a \bs{}subentry. But you cannot use \bs{}subentry2 within
\bs{}subentry1 and vise versa. If you think it should be permitted, don't.
Also, \bs{}blindentry is not permitted within \bs{}subentry.
(I can't be flexible {\bf everywhere}.)

@ First save the current level, then set the current level to the
subentry level. Process until the subentry is completely digested.

@^System dependent code@>
@p procedure process_subentry(@!entry_level:integer);
var temp_level :integer;
digest :boolean;
begin remove_characters(1);{Throw away the `1' or `2'}@/
temp_level := curr_level;
curr_level := entry_level;
digest := true;
repeat
@<Digest Subentry@>
until not digest;
curr_level := temp_level;
end;

@ If the next character is a dollar sign, then we {\bf might} have a \bs{}sortas
If the next character is a left curly brace, then check to see if we've
reached the end of this subentry. If the character is not `\$' or `\{',
then write the character to the print file.

@<Digest Subentry@>=
if in_record[1] = '$'
then @<Check for Sortas@>
else if in_record[1] = '{'
then @<Check for Subentry End@>
else write_print_chars(1);

@ If the next two characters are `\$\}' then we are are the end of
the subentry.

@<Check for Subentry End@>=
begin
if (in_record[2] = '$') and (in_record[3] = '}')
then begin
digest := false;
remove_characters(3);
end
else write_print_chars(1);
end

@ If the next three are `\{\$\}' then we {\it might} have a sortas.
and check the next two
characters to see if they
are `{\tt sa}'. If so, we {\bf do} have a sortas.
Otherwise write those 6 characters out. In any case, if we don't have
a `\{\$\}' to begin with, then write the first character out
(which was a
`\$' if you remember from earlier.)

@<Check for Sortas@>=
if (in_record[2] = '{') and
(in_record[3] = '$') and (in_record[4] = '}')
then begin
if (in_record[5] = 's') and (in_record[6] = 'a')
then do_sort_as
else write_print_chars(6);
end
else write_print_chars(1)

@
The  {\it equal\_arrays}
function returns a true if the two arrays being passed are
identical. False if not.

@p function equal_arrays(fieldy:field_array;fieldz:field_array):boolean;
var i :integer;
still_checking :boolean;
begin i := 1;
still_checking := true;
repeat
if fieldy[i] <> fieldz[i] {if arrays not equal}
then still_checking := false
else incr(i);
until (i>max_field) or not still_checking;
equal_arrays := still_checking;
end;

@ The {\it plsort} procedure is written in PL/1 to call the CMS sort program
Syncsort. This was necessary since you cannot call Syncsort from PASCAL/VS.
The fields to be sorted are defined internally in the {\it plsort}
program. This should be the same as the three sort fields in {\it sort\_type}.
@:sort_type@>
@:plsort@> @^PL/1@>
@^Syncsort@>

@p procedure plsort(var sort_rc:integer); fortran;

@ This function converts to all uppercase.
Notice that this is an EBCIDIC conversion, not an ASCII conversion
to uppercase.

@p function upper_case(x:char):char;
var temp:char;
begin
  if (ord(x)>=129)and(ord(x)<=169) then
      temp:=chr(ord(x)+64)
  else temp:=x;
upper_case:=temp;
end;


@* Read Each Index Entry.
Each line in the {\it ix\_file} is a separate index entry generated with the
\bs{\it index markup}. This markup has the format
\par{\tt\bs{}index\it n\tt\{\it entry\rm/\it markup\tt\}}
\par
\nin where ``{\it entry}''
is the textual material of the primary index. The
``{\it markup}'' may be one of the following submarkup
which may {\bf only}
appear within the \bs{\it index markup}. Never outside of it.
\item{1.} {\tt\bs{}subentry{\it n\/}}: where {\it n} is ``1'' or ``2'',
denoting the subentry level.
You would use this submarkup in the following fashion:
\par{\lin{+.5in}\tt\bs{}index1\{\rm entry
\bs\rm subentry1\tt\{\it submarkup\tt\}\}\par}
\indent where {\it submarkup} here can only be \bs{\it sortas}.
\item{2.}{\tt\bs{}sortas}: This markup is used to indicate text for either the
primary sort level, or one of the two subentry sort levels that is to be
sorted, but not printed in the resulting index. For example, someone may
want ``10 Downing Street''
to sort as ``ten Downing Street''. This would be accomplished by entering
``{\tt\bs{}index1\{10 Downing Street\bs{}sortas\{ten Downing Street\}\}}''.
You can also use \bs{\it sortas} within the \bs{\it subentry} submarkup.
\item{3.}
{\tt\bs{}itpn}: This markup indicates this index number
is to be printed in italics in the sorted index.
To use it, enter:
\par{\lin{+.5in}\tt\bs{}index1\{\rm entry \tt
\bs{}itpn\}\par}
\item{4.} {\tt\bs{}bdpn}: This markup indicates this
index number is to be
printed in bold in the sorted index.
To use it, enter:
\par{\lin{+.5in}\tt\bs{}index1\{\rm entry \tt
\bs{}bdpn\}\par}
\item{5.} {\tt\bs{}uspn}: This markup indicates
this index number is to
be underscored in the sorted index.
To use it, enter:
\par{\lin{+.5in}\tt\bs{}index1\{\rm entry \tt
\bs{}uspn\}\par}
\item{6.} {\tt\bs{}blindentry}: This markup indicates this index is to be
a blind entry or ``cross reference''. It may appear inside of a
standard entry or subentry. There is no page number associated with
this submarkup.
\par

@ @<Global Var...@>=
@!respect_case :boolean;
@!dot_leadering :boolean;
@!print_style:char;
@!sort_record :sort_type;
@!in_record :string_type(max_lrecl);
@!temp_strvalue:pass_pn_alpha;

@^System dependent code@>
@ This procedure will read the {\it ix\_file} in until all index requests
have been read.

@p procedure read_all_entries;
var @!i,j,k:integer;
begin
@<Get Default Index Values@>
while not eof(ix_file) do begin
@<Initialize Sort Records@>@/
@<Read Next Record@>@/
@<Move Sort Type to Sort Record@>@/
@<Move Page Number to Sort Record@>@/
@<Digest the Input Record@>@/
@<Fill in Sort Fields@>@/
@<Write Sort Record@>@/
end;end;

@ The first index default variable passed is ``case=". A ``yes" value will
respect the case in an index sort field. ``no" (the default) will convert
everything to uppercase before sorting.

The second index default ``style=" will be one of ``1", ``2'', or ``3''
depending on whether the style is ``paragraph'', ``dash'', or ``indented''
(which is the default).

The third default read in here is the ``leadering=". A ``yes'' here will give
dot leadering. Guess what ``no'' does.
Dot leadering in the paragraph style ({\it print\_style}=1). If that
has been requested, change {\it dot\_leadering} to false.

@<Get Default Index Values@>=
readln(ix_file,in_record);
if dosubstr(in_record,doindex(in_record,'=')+1) = 'no'
then respect_case := false
else respect_case := true;
readln(ix_file,in_record);@/
print_style:=in_record[doindex(in_record,'=')+1];@/
readln(ix_file,in_record);@/
if dosubstr(in_record,doindex(in_record,'=')+1) = 'no'
then dot_leadering := false
else dot_leadering := true;@/
if dot_leadering and (print_style = '1')
then begin writeln(messages,
'<TEXT1> Error! Dot leadering not permitted with the paragraph style.');
writeln(messages,'        Dot leadering will be disabled.');
dot_leadering := false;
end;

@ The {\it all\_blanks} variable is filled with all blanks. Other arrays
of the same dimension of {\it all\_blanks} can be set to blank themselves
by saying $array\leftarrow all\_blanks$.

@<Global Var...@>=
@!all_blanks:field_array;

@ @<Initialize Sort Records@>=
with sort_record do
for i := 1 to max_levels do begin
sort_part[i].field_level:= all_blanks;
sort_part[i].field_lngth := 0;
print[i].field_level:= all_blanks;
print[i].field_lngth := 0;
end;
curr_level := 0;

@ Read the next index request to be processed.
Get rid of trailing blanks.

@<Read Next Record@>=
readln(ix_file,in_record);
in_record:=trim(in_record);

@ The {\it sort\_type} is always in column 2. Column 1, 3 and 4 will always
be curly braces. (`\{{\it s}\}\{' where ``{\it s}'' is the sort type.)
After moving the {\it sort\_type} to the sort record, then
delete the {\it sort\_type} and the three curly braces.

@<Move Sort Type to Sort Record@>=
sort_record.record_type := in_record[2];
remove_characters(4);

@ The end of the page number is the next right curly brace (`\}').
Move all of the digits from the input record to the sort record,
starting with the last digit and going forward. If a minus sign (`-')
is in
the page number (preface material), then put a minus sign in
byte one of the sort record's page number.

@<Global Var...@>=
@!in_page_number:pn_type;

@ Find and move the page number to the sort record. If it is negative,
then its a roman numeral. In that case, move the page number to the
{\it abs\_page\_number} to be sorted descending, as negative numbers
should be. If its not negative, then move zero to {\it abs\_page\_number}
so it will move to the bottom of that sort field.
After it is moved, delete the appropriate number of characters to
finish digesting the entry.

@<Move Page Number to Sort Record@>=
k := doindex(in_record,'}'); {find the end of the page number}
i := k - 1;
for j := 1 to max_pnum do in_page_number[j] := '0';
j := max_pnum;
repeat
if in_record[i] = '-'
then begin
in_page_number[1] := '-';
i := 0;
end
else begin
in_page_number[j] := in_record[i];
decr(i);
decr(j);
end;
until i <= 0;
remove_characters(k+1);@/
sort_record.page_number:=numeric(in_page_number);@/
if sort_record.page_number < 0
then sort_record.abs_page_number:=abs(sort_record.page_number)
else sort_record.abs_page_number:=0;
k := doindex(in_record,'}'); {find the end of the page number string}
for i := 1 to max_pn_alpha do sort_record.page_string[i] := ' ';
for i := 1 to k-1 do
sort_record.page_string[i] := in_record[i];
remove_characters(k);

@^System dependent code@>
@ Process the rest of the input record. Each time you see a dollar sign,
there could be a submarkup coming, so examine the next three characters.

@<Digest the Input Record@>=
repeat
if (in_record[1] = '$') and (length(in_record) >= 4)
then @<Examine Next 3 Chars@>
else write_print_chars(1);
until length(in_record) < 1;

@ Once the index request has been digested, each sort field is checked
to see if it is blank. If so, then {\it sort\_as} was not used, so copy the
{\it print} field to the {\it sort} field.

@<Fill in Sort Fields@>=
with sort_record do begin
for i := 1 to max_levels do
if sort_part[i].field_lngth=0
then begin
sort_part[i].field_level := print[i].field_level;
sort_part[i].field_lngth := print[i].field_lngth;
check_case; end
else check_case;
end;

@^System dependent code@>
@ Ok, the record has been processed, so write it to the output file.

@<Write Sort Record@>=
sort_file@@:=sort_record;
put(sort_file);

@ Ok, a dollar sign signals the beginning of some sort of action code if the
next three characters are `\{\$\}'. So lets look at them and see.

@<Examine Next 3 Chars@>=
begin if (in_record[2] = '{') and
(in_record[3] = '$') and (in_record[4] = '}')
then @<Examine Next 2 Chars@>
else write_print_chars(1);{Well then, the dollar sign must be part of the index}
end

@ If a {\it blindentry} (`{\tt be}') or {\it subentry} (`{\tt se}'),
then perform the appropriate sections.

@<Examine Next 2 Chars@>=
begin if (in_record[5] = 'b') and (in_record[6] = 'e')
then @<Do Blind Entry@>
else if (in_record[5] = 's') and (in_record[6] = 'e')
then @<Do Sub Entries@>
else if (in_record[5] = 's') and (in_record[6] = 'a')
then do_sort_as
else write_print_chars(6);{Might as well write all 6 characters examined}
end

@ Everything up to the next `\{\$\}' is the blind entry.
The page number for a blind entry will always be 999999999.

@<Do Blind Entry@>=
begin remove_characters(6);{Throw away the `\$\{\$\}be'}@/
i := doindex(in_record,'{$}');
curr_level:=2;
write_print_chars(i-1);
curr_level:=0;
remove_characters(3);{Throw away the `\{\$\}'}
sort_record.page_number := 999999999;
end

@ Process the subentry, depending on whether or not it is
{\it subentry}1 or {\it subentry}2. If not, ignore the whole thing.

@<Do Sub Entries@>=
begin remove_characters(6); {Throw away the `\$\{\$\}se'}
if in_record[1] = '1'
then process_subentry(1)
else if in_record[1] = '2'
then process_subentry(2)
else write_print_chars(1);{Perhaps someone entered \bs{}subentry3??}
end

@* Write Formated Index. This section of code takes the sorted index
entries, merges them together, and writes them back out to the index file
to be read in by \bs{}printindex in the user's program. This section consists
of three procedures for accomplishing this feat: {\it add\_page\_number},
{\it digest\_the\_line}, and {\it read\_sorted\_records}.
{\it numeric} is a function to convert the character string page number into
an integer.

@^System dependent code@>
@ The procedure {\it add\_page\_number} is used to add the page number of the
current sorted record to the {\it string\_bold} if a
\bs{\it bdpn}, or to
the {\it string\_underscore} if a \bs{\it uspn},
or otherwise to
the {\it str\_build}.

@p procedure add_page_number;
var i:integer;

@ @<Global Var...@>=
@!prev_page_number:integer;
@!prev_pg_string:string_type(max_pn_alpha);
@!prev_record_type:char;

@ First check to see if we have a new page number. If so, then add the page
number to the string. Otherwise, only add the page number if the sort type
is new and is not equal to ``6'' (italic).

@p
begin
with sort_record do if prev_page_number = page_number
then if (prev_record_type = record_type) and (record_type <> '9')
 then getout
else if record_type = '6' then getout;
@<Add Page Number to the String@>
end;

@ {\it str\_build} is used to accumulate all of the page numbers.
{\it str\_blind\_entry} is used to accumulate the blind entry information
(typically there should be no more than one, but allowance is made for more.)

@<Global Var...@>=
@!str_build:string_type(max_lrecl);
@!str_blind_entry:string_type(max_lrecl);
{come on, who's going to have one that long?}

@ This module will take the page number of the current sorted index
record and add it to the build string, unless the record type is
9, which is a blind entry, in which case it is added to the blind entry
string. Note that if the record type is 0 (bold), 4 (underscore) or
6 (italic), then the page number is prefaced with ``B'', ``U'' or
``I'' respectively. This identifies that the page number is to be
emphasized when the build string is processed through the
{\it digest\_the\_line} procedure.

@<Add Page Number to the String@>=
if sort_record.record_type = '9' then @<Do Sorted Blind Entry@>
else begin @<Do Sorted Non Blind Entry@>
end;
prev_page_number := sort_record.page_number;@/
prev_pg_string := strconv(sort_record.page_string);@/
prev_record_type := sort_record.record_type;

@ @<Do Sorted Blind Entry@>=
with sort_record.print[3] do begin
if length(str_blind_entry) > 0 then add_cmma_blnk(str_blind_entry);
for i := 1 to field_lngth do
str_blind_entry := ccat(str_blind_entry,str(field_level[i]));
end

@ If the {\it page\_string[1]} is less than zero, its alphabetic and
roman numeral processing is necessary.

@<Do Sorted Non Blind Entry@>=
with sort_record do begin
if ord(page_string[1]) < ord('0')
then @<Add Roman Page Number@>
else @<Add Arabic Page Number@>;
end;

@ @<Add Roman Page Number@>=
begin
if length(str_build) >= 1
then str_build:=ccat(',',str_build);
strvalue(page_number,temp_strvalue);
str_build := ccat(ccat(ccat(strconv(page_string),
','),temp_strvalue),str_build);
if record_type = '0'
then str_build := ccat('B',str_build)
else if record_type = '4'
then str_build := ccat('U',str_build)
else if record_type = '6'
then str_build := ccat('I',str_build)
end

@ @<Add Arabic Page Number@>=
begin
if length(str_build) >= 1
then add_comma(str_build);
if record_type = '0'
then str_build := ccat(str_build,'B')
else if record_type = '4'
then str_build := ccat(str_build,'U')
else if record_type = '6'
then str_build := ccat(str_build,'I');
str_build := ccat(str_build,strconv(page_string));
end

@ {\it output\_string} is used to collect all of the page numbers
before writing them back out to the index file;
{\it bold\_string} collects the bold page numbers;
{\it us\_string} collects the underscored page numbers;
{\it rm\_string} collects everything not bold or underscored.
As you might expect, {\it italic\_last\_page} is true when the last
page number of a `roll' is to be italic.

@<Global Var...@>=
@!output_string:string_type(max_lrecl);
@!bold_string:string_type(max_lrecl);
@!us_string:string_type(max_lrecl);
@!rm_string:string_type(max_lrecl);
@!italic_last_page:boolean;

@^System dependent code@>
@ {\it finish\_the\_process} is a procedure that is used to
first check to see if we are {\it on\_a\_roll}, which means we are formatting
something like pages 1-3.
After doing that, if there were any bold page numbers, they are concatenated
to the front of the {\it output\_string}; underscored page numbers (if any)
are concatenated to the end of the {\it output\_string}.

@p procedure finish_the_process;
begin
@<Concatenate Bold Pages to Output@>;@/
@<Concatenate Roman Pages to Output@>;@/
@<Concatenate Underscored Pages to Output@>;
end;

@ If there were any bold page numbers, then add them to the {\it
output\_string}.

@<Concatenate Bold...@>=
if length(bold_string) > 0 then begin
if length(output_string) > 0 then add_cmma_blnk(output_string);
output_string := ccat(output_string,bold_string);
end

@ If we were on a roll (a consecutive series of page numbers) then finish
the roll. Then if there were any roman or italic page numbers, add them to
the {\it output\_string}.

@<Concatenate Roman...@>=
if on_a_roll then if italic_last_page
then rm_string :=
ccat(ccat(ccat(rm_string,'{\it '),trim(prev_pg_string)),'}')
else rm_string := ccat(rm_string,prev_pg_string);
@.\bs{}it@>
if length(rm_string) > 0 then begin if length(output_string) > 0
then add_cmma_blnk(output_string);
output_string := ccat(output_string,rm_string);
end

@ If there were any underscored page numbers, add them to the {\it
output\_string} here.

@<Concatenate Under...@>=
if length(us_string) > 0 then begin if length(output_string) > 0
then add_cmma_blnk(output_string);
output_string := ccat(output_string,us_string);
end

@ The procedure {\it start\_digesting} will initialize a few variables,
then add the {\it current\_page} to the appropriate list.

@p procedure start_digesting;
begin @<Initialize Some Global Variables@>;@/
prev_pg_string := curr_str_page;@/
@<Do Appropriate Highlighting@>;
end;

@ @<Initialize Some Global Variables@>=
italic_last_page := false;
bold_string := ''; us_string :=''; rm_string:='';
starting_to_process:=false;
on_a_roll:=false

@ @<Get the Current Print Type of this Page Number@>=
print_type:= curr_str_page[1];
if (print_type='B') or (print_type='I') or (print_type='U')
then curr_str_page:=dosubstr(curr_str_page,2)
else print_type := ' '

@ @<Do Appropriate Highlighting@>=
if print_type = 'B' then do_when_bold
else if print_type = 'I' then do_when_italic
else if print_type = 'U' then do_when_underscore
else do_when_roman

@^System dependent code@>
@ The procedure {\it digest\_the\_line} will take the full {\it str\_build}
and add the dashes when there is a run of page numbers (i.e. ``{\tt 1,2,3}'' bec
``1-3''); add the bold page numbers in front of the string (i.e.
``{\tt 1,2,B3,3,4}''
prints as ``{\bf 3},1-4''. Note that page 3 had to appear twice
``{\tt B3,3}''. If it had only appeard as a bold page number and not the
default, ``{\tt 1,2,B3,4}'', you would get ``1-2,{\bf 3},4'');
add the underscored
page numbers to the end of the string (similar to bold page numbers,
``1,2,U3,3,4'' prints as ``1-4,\us{3}''); and print in italics when
necessary.

@p procedure digest_the_line;
var
i:integer;
@!temp_roman:string_type(max_pn_alpha);
@!no_page_numbers:boolean;
begin
@<Process Each Page Number@>;@/
@<Write the Line Out@>;
end;

@ @<Global Var...@>=
@!starting_to_process :boolean;
@!print_type:char;
@!curr_str_page :string_type(max_pn_alpha);
@!curr_num_page :integer;

@ Initialize the variables, then read each page number until the whole
string is digested, then finish off the page number string before returning.
If {\it str\_build} is null, then we are doing a blind entry with no
page numbers associated. In that case, initialize the highlighting
strings.
@<Process Each Page Number@>=
output_string := '';
starting_to_process := true;
no_page_numbers:=false;
if length(str_build) > 0
then repeat @<Digesting Each Page Number@>
        until length(str_build) < 1
else if starting_to_process then begin
@<Initialize Some Global Variables@>;
no_page_numbers:=true;
end;
finish_the_process;

@ First get the next page number. Then if we are starting, initialize
everything. If we have already started, then check to see if we have a
consecutive page number sequence (i.e. pages 1,2,3,4 etc.) which is called
a `roll'.
@<Digesting Each Page Number@>=
@<Get Next Page Number@>;@/
if starting_to_process
then start_digesting
else @<Check for a Roll@>;
prev_page_number:=curr_num_page;

@ @<Get Next Page Number@>=
i := doindex(str_build,',');
if i < 1 then begin curr_str_page:=str_build; str_build:='';end
else begin
curr_str_page:=dosubstr(str_build,1,i-1);
str_build := dosubstr(str_build,i+1);
end;
@<Get the Current Print Type of this Page Number@>;@/
if ord(curr_str_page[1]) < ord('0') {Then its alphabetic}
then @<Get Numeric for Roman Page Number@>
else curr_num_page:=get_numeric(curr_str_page);

@ @<Get Numeric for Roman Page Number@>=
begin
i := doindex(str_build,',');
if i < 1 then begin temp_roman:=str_build; str_build:='';end
else begin
temp_roman:=dosubstr(str_build,1,i-1);
str_build := dosubstr(str_build,i+1);
end;
curr_num_page:=get_numeric(temp_roman);
end

@ If we are on a roll, then the previous page number will be one less than
the current page number. If this is not the case, then {\it
finish\_the\_process}
and {\it start\_digesting} all over again.

@<Check for a Roll@>=
if ((prev_page_number = curr_num_page) and (print_type <> ' ')) or
   ((prev_page_number>=0) and (prev_page_number = curr_num_page - 1)) or
   ((prev_page_number<0) and (prev_page_number = curr_num_page + 1))
then begin
@<Got a Roll Going@>
prev_pg_string:=curr_str_page;
end
else begin finish_the_process; start_digesting; end

@ Ok, so we're on a roll, right? If the {\it print\_type} is ``B'' (bold) or
``U'' (underscored) then do those and return. Otherwise, check to se if we
were previously {\it on\_a\_roll}. If we weren't, then add the hyphen to the
starting page number. In either case, move the {\it current\_page} to the
{\it prev\_pg\_string}. Lastly, if the page number is italic, then set
{\it italic\_last\_page} to true, in case it ends up being the last page in
the string.

@<Got a Roll Going@>=
if print_type = 'B' then do_when_bold
else if print_type = 'U' then do_when_underscore
else begin if not on_a_roll then begin on_a_roll := true;
rm_string:=ccat(rm_string,'--'); end;
if print_type = 'I' then italic_last_page := true
else italic_last_page := false;
end;

@ Before the page numbers are actually written, we need to write out
the ID depending on which level we are formatting. Level 1 is the
primary level, level 2 is the subentry1, and level 3 is subentry2.
Then, we need to do a couple of things depending on whether there is
dot leadering going on or not.
Finally, write the output string to the formatted index file.

@<Write the Line Out@>=
write_prev_header;
if dot_leadering then begin @<Do Dot Leadering@> end
else begin @<Do not do Dot Leadering@> end;
do_lnwrite(ix_file,output_string);

@ If there is dot leadering, then add the blind entry first, before adding
the leadering.

@<Do Dot Leadering@>=
if length(str_blind_entry) > 0
then begin
if length(output_string) > 0 then add_cmma_blnk(output_string);
output_string := ccat(output_string,str_blind_entry);
end;
output_string := ccat(ccat('\leader{}',output_string),'\par');
@.\bs{}leader@>
@.\bs{}par@>

@ @<Global Var...@>=
@!new_level_1:boolean;

@ If there is not dot leadering, then the blind entry goes at the end of the
page, except for
the paragraph style ({\it print\_style}=1). Also, if the paragraph style,
then do not add the {\tt \bs{}par} at the end. Instead, add a comma
if there is a new level 1 entry.

@<Do not do Dot Leadering@>=
if (print_style <> '1') and (length(str_blind_entry) > 0)@/
then begin
if length(output_string) > 0 then add_cmma_blnk(output_string);
output_string := ccat(output_string,str_blind_entry);
end;
if print_style = '1'
then if new_level_1 and ((length(str_blind_entry)=0) or no_page_numbers)
then do_nothing
else add_cmma_blnk(output_string)
else output_string := ccat(output_string,'\par');
@.\bs{}par@>

@ The {\it build\_sorted\_index} procedure is the driving force behind
formatting the index. The basic process is to read all of the sorted records in
and create a record called {\it str\_build}. There will be one of these for
each primary and subentry level index. After doing the {\it str\_build},
this record is then read through again to add the {\it\TeX T1} markup
necessary to
print out the sorted index line.

@p procedure build_sorted_index;
var @!first_time_through:boolean;

@ @<Global Var...@>=
@!prev_sr1:field_array;
@!prev_sr2:field_array;
@!prev_sr3:field_array;
@!prev_pn1:field_array;
@!prev_ln1:integer;{the length of pn1}
@!prev_pn2:field_array;
@!prev_ln2:integer;{the length of pn2}
@!prev_pn3:field_array;
@!prev_ln3:integer;{the length of pn3}
@!id1_is_missing:boolean;
@!id2_is_missing:boolean;

@ The first thing we need to do is read the next sorted record. If this is the
first time ever through this procedure, then do some quick initializing. Then,
once we come to a new ID, we should write out the previous index page numbers,
etc. If the ID has not changed from the previous record, then add the page
number given here on the sorted record to the {\it str\_build}.

@p begin first_time_through:=true;
while not eof(sort_file) do begin
sort_record:=sort_file@@;
get(sort_file);
if first_time_through then begin @<Do First Time Inits@> end;
if not equal_arrays(prev_sr1,sort_record.sort_part[1].field_level)
then begin
@<Start a New Level 1@> end
else if not equal_arrays(prev_sr2,sort_record.sort_part[2].field_level)
then begin
@<Start a New Level 2@> end
else if not equal_arrays(prev_sr3,sort_record.sort_part[3].field_level)
and (sort_record.record_type <> '9')
then begin
@<Start a New Level 3@> end
else add_page_number;
end;
@<Digest and Print Level 1@>
end;

@ First write out the {\it \bs{}everyindex} record at the beginning
of the index file.

@<Do First Time Inits@>=
do_lnwrite(ix_file,'\everyindex{',ix,'}');
@.\bs{}everyindex@>

@ These initializations only get done at the beginning of the
first sort record.

@<Do First Time Inits@>=
first_time_through:=false;
id1_is_missing:=false;
id2_is_missing:=false;
new_level_1:=false;
with sort_record do begin
prev_sr1:=sort_part[1].field_level;
prev_sr2:=sort_part[2].field_level;
prev_sr3:=sort_part[3].field_level;
prev_pn1:=print[1].field_level;
prev_ln1:=print[1].field_lngth;
prev_pn2:=print[2].field_level;
prev_ln2:=print[2].field_lngth;
prev_pn3:=print[3].field_level;
prev_ln3:=print[3].field_lngth;
end;
prev_page_number:=0;
str_blind_entry:='';@/
str_build:='';@/

@ If this is not a blind entry, then check to see if sort records for
the first level id are missing. If they are, then the id name itself
will be printed in {\it id1\_missing\_check}.

@<Do First Time Inits@>=
if sort_record.record_type <> '9'
then id1_missing_check;

@ Only alpha (a--z and A--Z) and numeric (0--9) characters get an
{\it \bs{}everyletterbreak}. Punctuation, for example, will not get
an {\it \bs{}everyletterbreak}.

@<Do First Time Inits@>=
with sort_record.sort_part[1] do begin
if ((field_level[1]>='a') and (field_level[1]<='i')) or @/
((field_level[1]>='j') and (field_level[1]<='r')) or @/
((field_level[1]>='s') and (field_level[1]<='z')) or @/
((field_level[1]>='A') and (field_level[1]<='I')) or @/
((field_level[1]>='J') and (field_level[1]<='R')) or @/
((field_level[1]>='S') and (field_level[1]<='Z')) or @/
((field_level[1]>='0') and (field_level[1]<='9'))
 then
do_lnwrite(ix_file,'\everyletterbreak{',ix,'}{',field_level[1],'}')
 else
do_lnwrite(ix_file,'\everyletterbreak{',ix,'}{}');
end
@.\bs{}everyletterbreak@>

@
First we need to {\it digest\_the\_line} which is the current index's
entry. {\it digest\_the\_line} will write out the formatted index unless
we are in the paragraph
style of formatted index
({\it print\_style}=1). If we are in the paragraph style,
then we need to write out any {\it str\_blind\_entry} pending.

@<Digest and Print Level 1@>=
new_level_1:=true;
digest_the_line;
if print_style = '1' then
do_lnwrite(ix_file,str_blind_entry,'\par');
new_level_1:=false;
@.\bs{}par@>

@ We have just found the start of a new primary index level.
First we need to process the previous index's
entry.
If there is a blind entry, and it is style `1', then add it now.
Next, if we are at a letter break (i.e., going from the sorted ``A''
primary index letters to the ``B''s) then write the ``{\tt
\bs{}everyletterbreak}'' to the formatted index file.

@<Start a New Level 1@>=
@<Digest and Print Level 1@>
with sort_record.sort_part[1] do begin
if (field_level[1] <> prev_sr1[1])
then if
(((field_level[1]>='a') and (field_level[1]<='i')) or @/
((field_level[1]>='j') and (field_level[1]<='r')) or @/
((field_level[1]>='s') and (field_level[1]<='z')) or @/
((field_level[1]>='A') and (field_level[1]<='I')) or @/
((field_level[1]>='J') and (field_level[1]<='R')) or @/
((field_level[1]>='S') and (field_level[1]<='Z')) or @/
((field_level[1]>='0') and (field_level[1]<='9'))) @/
then do_lnwrite(ix_file,'\everyletterbreak{',ix,'}{',field_level[1],'}')
 else
do_lnwrite(ix_file,'\everyletterbreak{',ix,'}{}');
end;
@.\bs{}everyletterbreak@>

@ The next thing to be done when starting a new primary index is to
check the subentries. If they exist, then you can assume that this
primary level index has no page numbers associated with it. The macro
{\it id1\_missing\_check} accomplishes this.

@<Start a New Level 1@>=
curr_level:=1;@/
if sort_record.record_type <> '9'
then id1_missing_check;

@ The complicated stuff is done. So now we can actually do some simple
initializations to get this new level 1 (primary level) sort rolling.

@<Start a New Level 1@>=
str_build:='';
str_blind_entry:='';
with sort_record do begin
prev_sr1:=sort_part[1].field_level;
prev_sr2:=sort_part[2].field_level;
prev_sr3:=sort_part[3].field_level;
prev_pn1:=print[1].field_level;
prev_ln1:=print[1].field_lngth;
prev_pn2:=print[2].field_level;
prev_ln2:=print[2].field_lngth;
prev_pn3:=print[3].field_level;
prev_ln3:=print[3].field_lngth;
prev_page_number:=0;
end;
add_page_number;

@ We have just found the start of a new subentry1 index level.
These modules are similar to {\it Start a New Level 1}, but not as
detailed since it is dealing with fewer levels.
First we need to {\it digest\_the\_line} which is the previous index's
entry. {\it digest\_the\_line} will write out the formatted index unless
we are in the paragraph
style of formatted index
({\it print\_style}=1).

@<Start a New Level 2@>=
digest_the_line;

@ The next thing to be done when starting a new subentry1 index is to
check the subentry2. If it exists, then you can assume that this
subentry index has no page numbers associated with it.
This is done in {\it id2\_missing\_check}.

@<Start a New Level 2@>=
id2_missing_check;

@ The complicated stuff is done. So now we can actually do some simple
initializations to get this new level 2 (subentry2 level) sort rolling.

@<Start a New Level 2@>=
str_build:='';
if print_style <> '1' then str_blind_entry:='';
with sort_record do begin
prev_sr2:=sort_part[2].field_level;
prev_sr3:=sort_part[3].field_level;
prev_pn2:=print[2].field_level;
prev_ln2:=print[2].field_lngth;
prev_pn3:=print[3].field_level;
prev_ln3:=print[3].field_lngth;
prev_page_number:=0;
end;
add_page_number;

@ We have just found the start of a new subentry2 index level.
These modules are similar to {\it Start a New Level 2}, but not as
detailed since it is dealing with the lowest level.
First we need to {\it digest\_the\_line} which is the previous index's
entry. {\it digest\_the\_line} will write out the formatted index unless
we are in the paragraph
style of formatted index
({\it print\_style}=1).

@<Start a New Level 3@>=
digest_the_line;
curr_level:=3;

@ Do some
initializations to get this new level 3 (subentry2) sort rolling.

@<Start a New Level 3@>=
str_build:='';
if print_style <> '1' then str_blind_entry:='';
with sort_record do begin
prev_sr3:=sort_part[3].field_level;
prev_pn3:=print[3].field_level;
prev_ln3:=print[3].field_lngth;
prev_page_number:=0;
end;
add_page_number;

@* Main Program.
Ok, here is the main program. First we initialize (all\_blanks); then
set the ix\_file for input and the sort\_file for output; read all of the
entries, processing each one; close the files; sort; read the sorted file
in and build the entries, writing them back to the ix\_file to be read in
by the index markup. WHEW!!

@p
begin @<Initialize Main@>@/
termout(messages);
reset_file(ix);@/
file_rewrite(s_file_number); {Should always be file 9}@/
read_all_entries;@/
close(ix_file);@/
close(sort_file);@/
@<Sort the Index@>@/
reset_file(s_file_number);@/
file_rewrite(ix);@/
build_sorted_index;@/
end;

@ @<Initialize Main@>=
for i := 1 to max_field do all_blanks[i] := ' ';

@ As mentioned earlier, {\it plsort} is an external PL/1 subroutine
@^PL/1@>
@^Syncsort@>
which is used to call Syncsort to sort the file.
The sort fields are as follows:
\halign{\hskip3em\hfill#\hfill&&\hskip3em\hfill#\hfill\cr
\bf Starting Column&\bf Length&\bf Field Description&\bf Order\cr
38&300&character&ascending\cr
342&300&character&ascending\cr
646&300&character&ascending\cr
10&4&binary&descending\cr
6&4&binary&ascending\cr}

@<Global Var...@>=
sort_rc:integer;

@ @<Sort the Index@>=
plsort(sort_rc);
if sort_rc = 0
   then writeln(messages,'Index Successfully Completed')
   else writeln(messages,'Index Failed');

@* Index.
All modules in which an identifier is
used are listed with that identifier, except that reserved words are
indexed only when they appear in format definitions, and the appearances
of identifiers in module names are not indexed. Underlined entries
correspond to where the identifier was declared. Error messages, control
sequences put into the output, and a few
other things like ``Syncsort'' are indexed here too.