%% This OTP is part of the Makor package (version2) for typesetting
%% Hebrew with Omega (Dec, 2002).  Its purpose is to mark word
%% beginnings and endings with special flags---sequences of
%% characters---that mark these boundaries.  Subsequent filters can
%% then strip off these markers, and do whatever else at these
%% boundaries.  Most notably, we'll substitute certain glyphs by word
%% final or word initial forms.
%%
%% It's not the purpose of this OTP to perform strict lexical analysis
%% as to what constitutes a valid input of a Hebrew word; other
%% filters do that at subsequent stages.  This current filter shall
%% immediately precede `m2context', so that any contextual analysis
%% done before this needs to be done separately by a foregoing OTP.
%% This will not be a problem, except for Yiddish (and possibly other
%% Hebrew-whatever dialects).
%%
%% For our purposes, this OTP examines input before the Makor
%% transcription rules have been changed.  Therefore, at this point,
%% Hebrew words can consist of sequences of Ascii chars drawn from
%% this set:
%%      lowercase consonants;
%%      lowercase vowels;
%%      uppercase `O';
%%      special chars + | * ^ ` ' _
%%      special special chars : .
%% Some comments on the special specials!  The colon when followed by
%% certain vowels (+, e, a) is a hataf-vowel; otherwise, it is a colon.  The
%% period is a period unless it is immediately followed by `s' or `t'.

input: 
        1;
output: 
        1;
states:
        IN,             % are we inside a word?
        OUT             % are we between words?
        ;
aliases:
        OKALPH  = (`b'|`c'|`d'|`i'|`m'|`n'|`p'|`s'|`t')
                ; % only these needed to specify units (bp, cm, pt, etc)
        ARG     = (43|45|46|48-57|{OKALPH}
                        ) % digits, -, +, OK letters are valid
                ;
        U       = 95 % underscore
                ;
        C       = 94 % circumflex
                ;
        COMMA   = 44
                ;
        SEMI    = 59 % semicolon
                ;
        % 39=right apostrophe; 96=left apostrophe
        LETTER  = (`a'-`z'|`O'|`+'|`|'|`*'|`^'|`_'|`"'|
                        `['|`]'|39|96| % reg Hebrew + trope
                        128-135|197) % Yiddish, Ladino glyphs
                ;
        SPECIAL = (`:'|46) % 46=period
                ;
        AUGMENTED       = ({LETTER}|{SPECIAL})
                        ;
        L       = ({LETTER})
                ;
        NL      = ^({LETTER})
                ;
        A       = ({AUGMENTED})
                ;
        NA      = ^({AUGMENTED})
                ;
%% (As of late December, 2002, version 1.15 of Omega will not work
%% when both categories of letters---that is, LETTER and AUGMENTED---
%% are used.  We therefore use the
%% less restrictive version, and hope that we can use explicit pattern
%% matching in m2context.otp to fix the lacks.  We also that Omega will
%% soon be fixed!

expressions:
%% Before we do some word parsing, note that _ and ^ are also used to
%% delimit fine tunign and various adjustments.  Thus, we need to
%% match this stuff first...

%% First let's do the accent fine-tuning.  Fine Tuning consists of a
%% pair of dimens, which are used as coordinates to adjust the
%% placement of an accent or cantillation mark.  Tunings can consist
%% of a PAIR of coordinates.  In case a letter has both an accent and
%% a trope, the first pair applies to the vowel+trope unit, while the
%% second applies to the displacement of the second diacrit from the
%% first.
%% 
        {U} {ARG}<1,9>{COMMA}{ARG}<1,9>{SEMI}{ARG}<1,9>{COMMA}{ARG}<1,9> {U}
                => "{\clearocplists\lowertuning[" \(* + 1 - 1) "]}"
              <= <push: OUT>
                ; 
        {C} {ARG}<1,9>{COMMA}{ARG}<1,9>{SEMI}{ARG}<1,9>{COMMA}{ARG}<1,9> {C}
                => "{\clearocplists\uppertuning[" \(* + 1 - 1) "]}"
              <= <push: OUT>
                ;
        {U} {ARG}<1,9>{COMMA}{ARG}<1,9> {U}
                => "{\clearocplists\lowertune[" \(* + 1 - 1) "]}"
              <= <push: OUT>
                ; 
        {C} {ARG}<1,9>{COMMA}{ARG}<1,9> {C}
                => "{\clearocplists\uppertune[" \(* + 1 - 1) "]}"
              <= <push: OUT>
                ;
%% An adjustment is a single number, with or without a decimal, which
%% represents the fraction of the width of the glyph by which the
%% accent is displaced.  It is much less powerful than a tuning, but
%% is sometimes useful.
%% 
        {U} {ARG}<1,9> {U}      => 
                "{\clearocplists\loweradj{" \(* + 1 - 1) "}}"
              <= <push: OUT>
                ;
        {C} {ARG}<1,9> {C}      => 
                "{\clearocplists\upperadj{" \(* + 1 - 1) "}}"
              <= <push: OUT>
                ;

%% A `word' is a sequence of zero or more ` augmented letters' ended by
%% at least one regular letter (and terminated by a space, punct, etc). 
%%
%% We use the sequence `)nullchar' and `nullchar(' as the end/start of
%% word markers.  (Nullchar=char254)
%% 
        <IN> {A} end:   => \* `)' 254 % eow
                        ;
        <OUT>{A}        => 254 `(' % sow
                        <= \* <push: IN>  
                        ;
        <IN>{NA}        => `)' 254 % eow 
                        <= \* <push: OUT> 
                        ;
        {A}             => 254 `(' % sow 
                        <= \* <push: IN>
                        ;
        {NA}            => 
                        <= \* <push: OUT>
                        ;