path: root/kate/data/prolog.xml

<?xml version="1.0" encoding="UTF-8"?>
<!-- [Purpose & Copyright]

    Highlight ISO-Prolog texts or template/include for Prolog dialects
    based on the ISO standard.

    Includes detection of DCG since it is 'hooked' in the ISO standard.
    Uses text style "Warning/ISO-bogus" where portability is likely to
    fail.

    This file is part of TDE's kate project.

  copyright   : (C) 2012 by Torsten Eichstädt

 **********************************************************************
 * This library is free software; you can redistribute it and/or      *
 * modify it under the terms of the GNU Library General Public        *
 * License as published by the Free Software Foundation; either       *
 * version 2 of the License, or (at your option) any later version.   *
 *                                                                    *
 * This library is distributed in the hope that it will be useful,    *
 * but WITHOUT ANY WARRANTY; without even the implied warranty of     *
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU  *
 * Library General Public License for more details.                   *
 *                                                                    *
 * You should have received a copy of the GNU Library General Public  *
 * License along with this library; if not, write to the              *
 * Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,   *
 * Boston, MA  02110-1301, USA.                                       *
 **********************************************************************
-->
<!DOCTYPE language SYSTEM "language.dtd"
[<!-- [Regular Expressions, Character Classes & Important Tokens]
E.g. the dot "." has so many different meanings (i.e. in an RegExpr 'any char',
'fullstop' of a clause, etc. pp.) that I decided it's usefull to write these out
to make the rules more expressive. -->

<!-- RegExpr for graphic tokens.  Comment start "/*" is catched earlier. -->
    <!ENTITY gr_char_iso    	"[#$&amp;*+\-./:&lt;=>?@^~\\]">

<!-- RegExpr: The highlighting engine always strips newline, so this ok: -->
    <!ENTITY any_nw         "[^ \t]">
    <!ENTITY any            ".">

<!-- RegExpr: we want to highlight some graphic tokens: Logic & Control
    fullstop shall not be falsely detected if it is actually a list functor
    comma is catched extra to highlight it as "Normal Text" in an arg_seq
    (and the cut "!" is NOT an op) -->
    <!ENTITY fullstop_iso          "\.(?!(\(|&gr_char_iso;))">
    <!ENTITY logic_control_ops_iso "(;|->|\\\+|:-|=|\\=)(?!&gr_char_iso;)">

    <!-- RegExpr: DCG (not my idea it is not reversed and named 'reduction') -->
    <!ENTITY dcg_production_iso    "-->(?!&gr_char_iso;)">

    <!-- RegExpr: partitioning of arithmetic operators:
    - the 1st "arith_compare" always demand arithmetic evaluation
      => highlight as "Arithmetics".
    - the 2nd only at the right side of "is" or when an arith_compare applies.
      They are used in other contexts as well, e.g. traditionally A-B is used
      for 'a pair of something', the "^" in bagof/3 and setof/3.
      => highlight as "Arithmetics" only in an arith_expr.
    - the 3rd are true ISO bogus: it is not specified whether it is arithmetic
      or logical shift, and for bitwise AND, OR and NEG the integer format is
      unspecified => not portable, depend on Prolog implementation. -->
    <!ENTITY arith_compare_iso  "(=:=|=\\=|=&lt;|&lt;|>=|>)(?!&gr_char_iso;)">
    <!ENTITY arith_ops_mixed_iso      "(\+|-|\*|\^)(?!&gr_char_iso;)">
    <!ENTITY arith_ops_int_iso      "//(?!&gr_char_iso;)">
    <!ENTITY arith_ops_float_iso      "(/|\*\*)(?!&gr_char_iso;)">
    <!ENTITY arith_bogus_iso  "(/\\|\\/|\\|&lt;&lt;|>>)(?!&gr_char_iso;)">

    <!-- RegExpr: other predefined operators -->
    <!ENTITY predefined_ops_iso "(\?-|==|\\==|@=&lt;|@&lt;|@>=|@>|=\.\.|@|:)(?!&gr_char_iso;)">

    <!-- We treat a number as float if it includes a decimal dot or exponent -->
    <!ENTITY float3_iso "[0-9]+E[+\-]?[0-9]+">
    <!ENTITY float2_iso "[0-9]+\.[0-9]+">
    <!ENTITY float1_iso "[0-9]+\.[0-9]+E[+\-]?[0-9]+">

<!-- RegExpr for esc'd char code in quoted 'strings' e.g. "\007\" -->
    <!ENTITY esc_oct_iso	"\\[0-7]+\\">
    <!ENTITY esc_hex_iso	"\\x[a-fA-F0-9]+\\">

<!-- Char classes for AnyChar, NOT RegExpr: no esc seqs, ranges, etc. pp. here!
     Used to avoid expensive RegExpr whenever possible. -->
    <!--  Not needed: any_alnum_under_iso:  use "DetectIdentifier" instead -->
    <!ENTITY any_lower_iso         "abcdefghijklmnopqrstuvwxyz">
    <!ENTITY any_upper_under_iso   "ABCDEFGHIJKLMNOPQRSTUVWXYZ_">
    <!ENTITY any_bin_iso           "01">
    <!ENTITY any_oct_iso           "01234567">
    <!ENTITY any_dec_iso           "0123456789">
    <!ENTITY any_hex_iso           "0123456789abcdefABCDEF">
    <!ENTITY any_graphic_iso       "#$&amp;*+-./:&lt;=>?@^~\">

    <!-- Common esc seq in quoted & char code (after "0'").
    Tab is handled extra.  Esc'd newline is catched earlier. -->
    <!ENTITY any_esc_iso		"abfnrtv\'`&quot;]">

    <!ENTITY dot        ".">

    <!-- Some solo chars for DetectChar -->
    <!ENTITY comma	",">
    <!ENTITY cut        "!">
    <!ENTITY bar        "|">
    <!ENTITY tab	"	">
    <!ENTITY bs         "\">
]>
<language name="Prolog" section="Sources"
	  version="1.1-1" kateversion="2.3"
	  mimetype="text/x-prolog"
	  extensions="*.prolog;*.dcg;*.pro"
	  author="Torsten Eichstädt ([email protected])"
	  license="LGPL">

<!-- [Terminology & Abbreviations In The Comments]
    - Literal text is in double quotes "like this", single quotes sometimes used
      for a 'terminus technicus'.
    - "w/", "w/o", "s/t" mean 'with', 'without', and "sometimes" resp.
    - Use "brace" instead of 'bracket'/'parenthesis' for any of "()[]{}"
    - Specify "round", "list/squared", "curly", "opening" and "closing" brace.
    - "quote/quoted/string" mean any of single-, double-, or back-quote(d)
    - "seq" sequence, "nl" newline, "bs" backslash, "q" quote(d), "cc" char code
    - "KISS" Keep it simple, stupid! (be pragmatic)
    - "style" = "attribute" (the applied text style)
    - Keep comments as short as possible and as descriptive as neccessary LOL
-->
<!-- [Documentation, Bugs, Limitations & Solutions, TODOs]
################################################################################
All bugs except three by the author. !FOLD THIS AWAY if long comments annoy you.

BUG - Nobody's perfect - If I knew about the bug you found, it would have been
      fixed already.
 one- Takes a "shebang" at the start of text on other than the 1st line (Kate).
 two- Indentation-based folding is not reliable.
 three- Folding is sometimes broken (Kate?).  See SOLUTIONS below.
  4 - An optional leading sign is not highlighted as part of the number.
  5 - A float w/o fractional part and exponent is highlighted as an integer.
  6 - Detection of whitespace may not be 100% correct, only space/tab/newline.
  7 - Handling inside arithmetic expressions might not be 100%.  See "AL GUSTO".
 inf- see TODOs below.

LIMITS of this parser
    - Highlighting of syntax errors is as intuitive as possible, i.e. for an
      invalid newline highlight the last token before and leading white of the
      next line - but that's not always possible.  Use your built-in problem
      solver (between your ears) when you do not know what the exact error is.
    - The parser knows nothing about currently defined dynamic ops, but has to
      respect possible definitions of dynamic ops.  That's why:
    - Any sequence of 'graphic chars' (_incl. those containing a dot_) is taken
      as _one_ graphic token, and not - as you might expect - your user-defined
      operator followed by fullstop (or sequence of ops).
    - A single dot in the outmost term of a clause is taken as a fullstop, even
      if it is actually a user-defined operator.
      - Else, a single dot is highlighted as "Warning" because it might be a
        user-def'd op (usually an error, but can not decide w/o current_op/3).
    - Detection of the fullstop relies upon the usual semantics of braces.
    - Axioms of folding capabilities are: the literal fullstop as stop symbol
      and the usual semantics of braces w/ pairs of opening and closing brace.
   ?- period.
uncaught exception: error(existence_error(procedure,current_op/3),highlight/2)

SOLUTIONS
    - Help the parser by puting whitespace (space/tab/nl) here and there, espc.
      after the fullstop, or use quotes; e.g. '/*' does not begin a comment,
      neither does '%', and ***'.' can be two operators.
      Functored syntax should always do, e.g. */*(X,Y) is valid (and does not
      begin a comment).
    - Folding: s/t moving the mouse slightly below the marker shows what you
      want, s/t an empty line helps (or removing an empty line), when it's
      removed regions stay ok, or break auto-folding w/ non-white @begin of
      line.  Try a newer kate version.

AL GUSTO (simple changes):
    - Change classification of built-in predicates: adjust the <list> below.
    - Reminder for other changes: solo chars - DetectChar, graphic - RegExpr.
    - bar "|" (solo) is highlighted as "other built-in operator" in lists, else
      "Normal Text".  You may safely remove/comment out  the rules in contexts
      "term/list/curly" below if you add in in context "operator" or "atomic".
      If it shall be part of user-defined ops, it has to be added to "gr_char",
      and a few rules need adjustment.  Search for "&bar;".  You may then safely
      add it to "predefined_ops" above (don't forget to escape it).
    - cut "!" (solo) can be adjusted like the bar.
    - comma "," (solo) is "Normal Text" in lists, else "Logic & Control".  You
      may safely remove/comment out the rules in "term/nested/list/curly" below
      and handle it in the context "operator".
    - Disable arith_expr: replace "context="arith_expr"" w/ "context="#stay""
      in the rules in contexts "atomic" and "operator" (two or three occurrences).
    - Disable indentation-based folding: in the section "general"@EOF:"folding".
    - General guidelines: Be nice to slow/old hardware: avoid RegExpr (start w/
      a RegExpr, test, then find a replacement), else match a RegExpr as late as
      possible - it's expensive, a context switch is relatively cheap.
      Eat as much text as possible in one run, e.g. in comments and quoted we
      use "DetectIdentifier", else the engine would try all rules of the context
      (w/ failure) and eat only one char on each run.
    If your modification is kind of general, upload it and drop me a note.

  The simplified syntax below should highlight 99.999% of ISO-Prolog texts
correctly, as well as many non-ISO Prolog texts.
  Else modify this file.  Save with new name in your HOME directory to
.kde/share/apps/katepart/syntax/prolog-xy.xml You may read the NOTE at the end
of this file.  WARNING: Be careful, though: the highlighting works according to
it's program code, not neccessarily it's documentation...

DONE- Parse (m)any (!) conforming Prolog texts w/ correct highlighting (LIMITS)
    - Parse any erratic text w/o crash or endless loops and highlight errors
    - Folding for multiline comments, clauses, nested terms, and quoted (LIMITS)
    - Folding inside comments and of 1-line comments and clauses: indentation-
      based like in Python, e.g.
      my_pred(a).   % w/ marker@left (western style; depends on your settings)
	my_pred(b). % you can fold
	my_pred(c). % these lines
      Inside clauses (i.e. context "term" and below) only explicit folding for
      braces and quotes applies, except comments.  I did not find a better
      solution, since the engine does not provide lookahead to the next line.
    - Additionally, folding with %BEGIN ... %END (iff 1st nonwhite of the line).
    - Classify built-ins (YMMV) for different highlighting, mainly:
	- w/ side effects from 'outside world' (arithmetics, stream I/O,...)
	- w/o side effects, only depend on user's program ('internal world')
	- program logic & control constructs
	- altering the clause knowledge base or Prolog state
	- integer/float/mixed arithmetics, so one can see the type of the result
	    - Iff it's an arithmetic expression; else arith ops and built-ins
	      are highlighted normal.
    - Incl. Sep2012 DRAFT Technical Corrigendum 2: Prolog - Part 1: General Core
    - Incl. honourable Richard O'Keefe's remarks as warnings as "ISO bogus".
    - Detect 'shebang' shell scripts, see BUGs
    - Includes alert_indent.xml to highlight alert keywords in comments.
    - Dot "." may be an _unquoted_ user-defined operator in nested terms (but is
      highlighted as "Warning" 'cause it's more likely a typo than an user-def'd
      op) and is only taken as fullstop and highlighted as "Logic & Control" in
      the outmost term of a clause.
    - Über-correct handling of quoted, escape seqs and "0'" (char code integer):
	- detection of valid/invalid newline in quoted 'strings'.
	- highlight whitespace after esc'd newline in quoted so it differs from
	  whitespace contained in the quoted 'string'.
	- Highlight esc'd tab and tab in quoted as "Warning" because some text
	editors may convert them to space and it is easy to use "\t" instead.
    - Preliminary: most names end w/ ISO; prepared to be included in syntax
      files adapted for Prolog dialects.

FIXME	When katepart can supply line numbers, fix the "shebang" rule.

TODO sorted by priority
  1 - [always] Cruise over all FIXMEs.
  2 - Check if comma, cut & bar are really solo chars (where's the red book?).
  3  - Check final Technical Corrigendum 2: Prolog - Part 1: General Core
  4 - Buy ISO Prolog standard - Part 2: Modules.  Likely only keywords.  Need it
      anyway.
  5 - Make this strict-ISO and a "Prolog (portable)" that includes this file.
  6 - Likewise, adopt for GNU & SWI Prolog and others.
  8 - keywords 'phrase', streams depricated, etc.pp. ==> Prolog (portable).
  9 - Likewise: shebang ==> Prolog (portable).
 10 - Check if shebang can span multiple (escaped) lines.
 11 - Check: all KB altering preds can have side effects (via resource_error)?
 12 - For next two TODOs: ask katepart dev to provide lookahead to the next line
    13 - Folding for consecutive one-line comments [easy? HARD! Impossible.]
    14 - Likewise, folding for consecutive clauses [HARD! Impossible.]
    I can live w/ indentation-based folding as a workaround.  Not 100%, but ok.
 15 - +arg_seq (and list): "," as normal text: Detect name( [no layout; easy]
 16 - +Classify meta predicates.
 18 - The dollar "$" is commonly used as the 1st char of (unquoted?) atoms to
      indicate 'internal mode', e.g. to declare a predicate as compile-in.
      Does this mean the dollar is used as if it were a prefix op or is it part
      of the atom?  Would be easy to implement.
 19 - Bug #4: could be partly fixed, some built-ins take only float not integer.
 20 - Bug them to better document weakDeliminators (and rename to delimiter ;)
 21 - Adjust doxygen syntax for Prolog (e.g. adapt doxygen-lua), and include it.
 22 - Check if vertical tab is white/layout; are there more non-printables?
################################################################################
-->
<!-- [Facts & Implications Of The ISO Standard, as a reminder]
  - 'layout' is whitespace (space/tab/newline) and/or comments.
  - Braces (any of "(){}[]"), exclamation "!" (aka 'cut'), 'bar' "|", degree "°",
    paragraph "§" and percent "%" are NOT 'graphic' chars, but 'solo' chars.
  - The 'bar' "|" may be a predefined operator, and then a user-def'd op, too.
  - There can only be (at most) one bar "|" in a list (written inside "[]"), and
    no commata after it (after the bar).
  - Graphic tokens are atoms (and solo chars are one-char atoms?).
  - "{}" and "[]" are 'special atoms' (likewise "()"?).
  - "{}" and "[]" shall not be operators (but round or single braces may be?).
  - The fullstop "." is not a predefined postfix operator, and
  - the list functor "." is not a predefined prefix operator.
  - An unquoted graphic token shall not begin with "/*".
  - Multiline comments shall not be nested.
  - Floats are NOT written with lowercase "e" (i.e. not neccessarily an error if
    e.g. "e" is a pre- or user-defined expr or op and the whole term is valid),
  - and the fractional part and exponent of floats are optional (any and both).
  - A leading minus immediately preceding a number is optional for all numbers
    and then part of the number and not a prefix operator (currently beeing
    clarified by the standard commitee if "immediately" includes layout, only
    comments, or none of these at all ;) but
  - the minus "-" is a predefined prefix- and infix operator...
  - Terms are rewritten when read in, extendable by user-def'd rules (and these
    may be dynamic), i.e. clauses are not restricted to the standard forms
    (to the extent that the fullstop ending a clause may be omited at all) and
  - the dot "." may be (solo or part of) an user-defined operator.
  - A Prolog processor may use any character set, e.g. (multi-byte) Unicode.
  - Virtually every token may be an (user- or predefined) operator, except a few
    specifically excluded tokens (i.e. NOT just graphic and 'standard' atoms).
  - All operators except the comma are dynamic (although the standard denotes
    atoms as 'constants' ;), and
  - (the comma may appear in a user-defined operator (beeing part of it)?).
  (Now try to write syntax highlighting for Prolog w/o Prolog HAHAHA ;)
-->
    <highlighting>
	<!-- aka "guru meditation trigger" -->
	<list name="error term ISO">
	    <item> error </item>
	</list>
	<list name="guru meditation terms ISO">
	    <item> instantiation_error </item>
	    <item> uninstantiation_error </item>
	    <item> type_error </item>
	    <item> domain_error </item>
	    <item> existence_error </item>
	    <item> permission_error </item>
	    <item> representation_error </item>
	    <item> evaluation_error </item>
	    <item> resource_error </item>
	    <item> syntax_error </item>
	    <item> system_error </item>
	</list>

	<!-- These are in fact bogus, DO NOT USE them and bug the ISO standard
	commitee to abandon char_conversion/2 and to apply common sense to the
	semantics of include/1 and ensure_loaded/1, i.e. to handle recursion.
	Bomb them w/ e-mails! It's not the purpose of a standard to fixate
	errors till eternity. -->
	<list name="bogus ISO">
	    <item> char_conversion </item>
	    <item> current_char_conversion </item>
	    <item> include </item>
	    <item> ensure_loaded </item>
	    <!-- Use atan2 instead -->
	    <item> atan </item>
	    <!-- Unspecified integer representation/format -->
	    <item> xor </item>
	</list>
	<list name="directives non-ISO">
	    <!-- FIXME is this a top_level goal only? (consult :- include) -->
	    <item> consult </item>
	    <item> built_in </item>
<!--GNU            <item> foreign </item> -->
<!--GNU            <item> ensure_linked </item> -->
<!--GNU            <item> built_in_fd </item> -->
<!--             <item> if </item> -->
<!-- 	    <item> else </item> -->
<!-- 	    <item> endif </item> -->
<!--             <item> elif </item> -->
        </list>

        <!-- including directive "initialization" -->
        <list name="logic+control ISO">
	    <item> initialization </item>
<!-- 	    <item> ! </item> -->
	    <item> fail </item>
	    <item> repeat </item>
	    <item> call </item>
            <item> catch </item>
	    <item> throw </item>
	    <item> true </item>
	    <item> false </item>
	    <item> once </item>
	</list>

	<!-- including directive "dynamic" -->
        <list name="dyn clause mgmt ISO">
	    <item> dynamic </item>
	    <item> asserta </item>
            <item> assertz </item>
            <item> retractall </item>
	    <item> retract </item>
	    <item> abolish </item>
	    <item> clause </item>
<!-- 	    <item> numbervars </item> -->
<!-- 	    <item> current_predicate </item> -->
	</list>
	<list name="listing non-ISO">
	    <!-- FIXME is this a top_level goal? -->
	    <item> listing </item>
<!-- standard/compatible	    <item> public </item> -->
	</list>
	<list name="terms ISO">
	    <!-- keep tokens with same prefix sorted by length? -->
	    <item> atom_concat </item>
	    <item> atom_length </item>
	    <item> atom_chars </item>
	    <item> atom_codes </item>
	    <item> arg </item>
	    <item> subsumes_term </item>
	    <item> acyclic_term </item>
	    <item> char_code </item>
	    <item> compare </item>
	    <item> copy_term </item>
	    <item> functor </item>
	    <item> number_chars </item>
	    <item> number_codes </item>
<!--FIXME check if GNU or ISO 	    <item> sub_atom </item> -->
	    <item> term_variables </item>
	    <item> unify_with_occurs_check </item>
	</list>
	<list name="terms non-ISO">
	    <!-- keep tokens with same prefix sorted by length? -->
	    <item> number_atom </item>
	    <item> expand_term </item>
	    <item> term_expansion </item>

	    <!--FIXME These go to streams non-ISO: -->
	    <item> display </item>
	    <item> print </item>
	    <item> format </item>
	    <item> portray_clause </item>
	    <item> portray </item>
	    <item> term_expansion </item>
	</list>
	<list name="DCG non-ISO">
	    <item> phrase </item>
	</list>
	<list name="lists ISO">
	    <item> sort </item>
	    <item> keysort </item>
	</list>
	<list name="lists non-ISO">
	    <item> append </item>
	    <item> delete </item>
	    <item> length </item>
	    <item> last </item>
	    <item> map_list </item>
	    <item> min_list </item>
	    <item> maplist </item>
	    <item> msort </item>
	    <item> memberchk </item>
	    <item> member </item>
	    <item> nth </item>
	    <item> permutation </item>
	    <item> reverse </item>
	    <item> select </item>
	    <item> prefix </item>
	    <item> suffix </item>
	    <item> sublist </item>
	    <item> sum_list </item>
	</list>
	<list name="streams ISO">
	    <item> open </item>
<!--GNU	    <item> current_stream </item>-->
	    <item> set_stream_position </item>
	    <item> get_char </item>
	    <item> get_code </item>
	    <item> peek_char </item>
	    <item> peek_code </item>
	    <item> get_byte </item>
	    <item> peek_byte </item>

	    <item> put_char </item>
	    <item> put_code </item>
	    <item> put_byte </item>
	    <item> nl </item>

	    <item> read_term </item>
	    <item> read </item>
	    <item> write_canonical </item>
	    <item> writeq </item>
	    <item> write </item>
	</list>
	<list name="streams deprecated">
	    <item> append </item>
	    <item> seeing </item>
	    <item> seen </item>
	    <item> see </item>
	    <item> telling </item>
	    <item> tell </item>
	    <item> told </item>
	    <item> get0 </item>
	    <item> get </item>
	    <item> skip </item>
	    <item> put </item>
	    <item> tab </item>
	</list>
	<list name="arith eval ISO">
	    <item> is </item>
	</list>
	<list name="arith ops int ISO">
	    <item> rem </item>
	    <item> mod </item>
	    <item> div </item>
	</list>
        <list name="arith expr mixed ISO">
<!--GNU	arith IF expr    <item> inc </item> -->
<!--GNU	arith IF expr    <item> dec </item> -->
	    <item> abs </item>
	    <item> sign </item>
	    <item> min </item>
	    <item> max </item>
        </list>
        <list name="arith expr int ISO">
<!--GNU	arith IF expr    <item> inc </item> -->
<!--GNU	arith IF expr    <item> dec </item> -->
	    <item> ceiling </item>
	    <item> floor </item>
	    <item> round </item>
            <item> truncate </item>
<!-- ISO bogus	    <item> xor </item> -->
        </list>
        <list name="arith expr float ISO">
	    <item> pi </item>
<!--GNU	(see pi)    <item> e </item> -->
<!--GNU	(see pi)    <item> epsilon </item> -->
	    <item> sqrt </item>
	    <item> tan </item>
	    <item> cos </item>
	    <item> sin </item>
	    <item> atan2 </item>
	    <item> acos </item>
	    <item> asin </item>
	    <item> exp </item>
            <item> log </item>
	    <item> float </item>
	    <item> float_fractional_part </item>
            <item> float_integer_part </item>
        </list>

	<!-- including directives "op/multifile/discontigous" -->
	<list name="prolog state ISO">
	    <item> multifile </item>
	    <item> discontigous </item>
	    <item> op </item>
	    <item> set_prolog_flag </item>
	</list>
	<list name="types ISO">
	    <item> var </item>
	    <item> nonvar </item>
	    <item> atom </item>
	    <item> integer </item>
	    <item> float </item>
	    <item> number </item>
	    <item> atomic </item>
	    <item> compound </item>
	    <item> callable </item>
	    <item> ground </item>
	</list>
	<!-- FIXME rename to "types non-ISO"? -->
	<list name="list+is_list non-ISO">
	    <item> list </item>
	    <item> is_list </item>
	<!--GNU            <item> partial_list </item> -->
	    <!--GNU            <item> list_or_partial_list </item> -->
	</list>
	<list name="built-ins ISO">
	    <item> current_op </item>
	    <item> current_prolog_flag </item>
	    <item> current_input </item>
	    <item> current_output </item>

	    <!-- These act on streams, but the permission_error is only thrown
	    if the programmer tries to set_input(S) when S is alread an output
	    stream and vice versa; so: not affected by 'outside world' -->
	    <item> set_input </item>
	    <item> set_output </item>

	    <!-- may throw a system_error in GNU -->
	    <item> close </item>

	    <item> flush_output </item>

	    <!-- may throw permission_error in GNU if S is an output stream -->
	    <item> at_end_of_stream </item>

	    <item> stream_property </item>
	</list>

	<!-- !FOLD the <context> rules & comments for an overview, then it's no magic. -->
	<contexts>
	<!-- Catch typos: style="Syntax Error" for all rules except comments and
	quoted.  This helps to catch bugs in the rules itself, too.  Nice
	debuging aid.  Note that (naturally) some of the syntax rules given in
	EBNF are simplified and do not describe 100% valid Prolog.

	    prolog text	:= shebang clause_seq | clause_seq
	    clause_seq	:= clause clause_seq | empty
	    clause      := layout term_seq fullstop | term_seq fullstop
	-->
	    <!-- shebang: ("#!") Detect shebang and fallthrough to clause_seq -->
	    <context name="shebang" lineEndContext="clause" attribute="Syntax Error" >
		<!-- no way: fallthrough="1" fallthroughContext="clause_seq" > -->
		<Detect2Chars column="0" char="#" char1="!" context="1-comment" attribute="% italic predicates: w/ side effects" />
		<!-- else fallthrough (workaround broken fallthrough) -->
		<RegExpr String="&any;" lookAhead="true" context="clause" attribute="Syntax Error" />
	    </context>

	    <!-- syntax error (Test & Debug Aid, too): it's enough to highlight next token -->
	    <context name="syntax_error" lineEndContext="#stay" attribute="Syntax Error" noIndentationBasedFolding="true" >
		<DetectSpaces context="#pop" attribute="Syntax Error" />
		<DetectIdentifier context="#pop" attribute="Syntax Error" />
		<RegExpr String="&any;" context="#pop" attribute="Syntax Error" />
	    </context>

	    <!-- clause: Start a region for code folding and switch to term -->
<!-- 	    <context name="clause" lineEndContext="#stay" attribute="Syntax Error" noIndentationBasedFolding="true" > -->
	    <context name="clause" lineEndContext="#stay" attribute="Syntax Error" >
		<IncludeRules context="layout" />
		<!-- KISS: do not force term to detect an empty clause as syntax error -->
		<RegExpr String="&fullstop_iso;" context="#stay" attribute="Warning (!use background)" />
<!-- 		<RegExpr lookAhead="true" String="&any;" context="term" attribute="Syntax Error" /> -->
		<RegExpr lookAhead="true" String="&any;" context="term" beginRegion="clause" attribute="Syntax Error" />
	    </context>

	    <!--term/nested/list/curly: the master/dispatcher -->
	    <!-- (abbreviated): term := var | atomic | compound | "(" term ")"

	    (term_seq := term "," term_seq | term    Not needed: comma is an op)
	    term      := layout solo layout | layout solo | solo layout | solo
	    solo      := var | atomic | compound | "(" term ")"
	    compound  := op term | term op | term op term | functored | list
	    functored := atom "(" arg_seq ")" | op "(" arg_seq ")"
	    list      := "[" l_arg_seq "]" | double_quoted | "." "(" arg_seq ")"
	    (if semantics of double_quoted not changed by set_prolog_flag/2)
	    sorry I forgot curly: the DCG term but can not contain the fullstop.

	    Comments in 'term' apply to the others as well.
	    Note that eating layout once is enough, since we come back here.
	    Only end the clause in the outmost term, which can only be term. -->
	    <context name="term" lineEndContext="#stay" attribute="Syntax Error" noIndentationBasedFolding="true" >
		<IncludeRules context="layout" />
		<DetectChar char="(" context="nested" beginRegion="nested" attribute="( ) [ ]" />
		<DetectChar char="[" context="list" beginRegion="list" attribute="( ) [ ]" />
		<DetectChar char="{" context="curly" beginRegion="curly" attribute="{ DCG }" />
		<DetectChar char="&comma;" context="#stay" attribute="Logic &amp; Control" />
		<DetectChar char="&cut;" context="#stay" attribute="Logic &amp; Control" />
		<DetectChar char="&bar;" context="#stay" attribute="Normal Text" />
		<!-- lookAhead to give different style to the dot and the brace -->
		<Detect2Chars lookAhead="true" char="&dot;" char1="(" context="list_functor" attribute="Syntax Error" />
<!-- 		<RegExpr String="&fullstop_iso;" context="#pop" attribute="Logic &amp; Control" /> -->
		<RegExpr String="&fullstop_iso;" context="#pop" endRegion="clause" attribute="Logic &amp; Control" />
		<IncludeRules context="atomic" />
		<!-- Default: anything not eaten here is a syntax error -->
	    </context>
	    <context name="nested" lineEndContext="#stay" attribute="Syntax Error" noIndentationBasedFolding="true" >
		<IncludeRules context="layout" />
		<DetectChar char=")" context="#pop" endRegion="nested" attribute="( ) [ ]" />
		<DetectChar char="(" context="nested" beginRegion="nested" attribute="( ) [ ]" />
		<DetectChar char="[" context="list" beginRegion="list" attribute="( ) [ ]" />
		<DetectChar char="{" context="curly" beginRegion="curly" attribute="{ DCG }" />
		<DetectChar char="&cut;" context="#stay" attribute="Logic &amp; Control" />
		<DetectChar char="&comma;" context="#stay" attribute="Logic &amp; Control" />
		<DetectChar char="&bar;" context="#stay" attribute="Normal Text" />
		<Detect2Chars lookAhead="true" char="&dot;" char1="(" context="list_functor" attribute="Normal Text" />
		<RegExpr String="&fullstop_iso;" context="#stay" attribute="Warning (!use background)" />
		<IncludeRules context="atomic" />
	    </context>
	    <context name="list" lineEndContext="#stay" attribute="Syntax Error" noIndentationBasedFolding="true" >
		<IncludeRules context="layout" />
		<DetectChar char="(" context="nested" beginRegion="nested" attribute="( ) [ ]" />
		<DetectChar char="]" context="#pop" endRegion="list" attribute="( ) [ ]" />
		<DetectChar char="[" context="list" beginRegion="list" attribute="( ) [ ]" />
		<DetectChar char="{" context="curly" beginRegion="curly" attribute="( ) [ ]" />
		<DetectChar char="&comma;" context="#stay" attribute="Normal Text" />
		<DetectChar char="&cut;" context="#stay" attribute="Normal Text" />
		<DetectChar char="&bar;" context="#stay" attribute="other built-in operator" />
		<Detect2Chars lookAhead="true" char="&dot;" char1="(" context="list_functor" attribute="Normal Text" />
		<RegExpr String="&fullstop_iso;" context="#stay" attribute="Warning (!use background)" />
		<IncludeRules context="atomic" />
	    </context>
	    <context name="curly" lineEndContext="#stay" attribute="Syntax Error" noIndentationBasedFolding="true" >
		<IncludeRules context="layout" />
		<DetectChar char="(" context="nested" beginRegion="nested" attribute="( ) [ ]" />
		<DetectChar char="[" context="list" beginRegion="list" attribute="( ) [ ]" />
		<DetectChar char="}" context="#pop" endRegion="curly" attribute="{ DCG }" />
		<DetectChar char="{" context="curly" beginRegion="curly" attribute="{ DCG }" />
		<DetectChar char="&comma;" context="#stay" attribute="Logic &amp; Control" />
		<DetectChar char="&cut;" context="#stay" attribute="Logic &amp; Control" />
		<DetectChar char="&bar;" context="#stay" attribute="Normal Text" />
		<Detect2Chars lookAhead="true" char="&dot;" char1="(" context="list_functor" attribute="Normal Text" />
		<RegExpr String="&fullstop_iso;" context="#stay" attribute="Warning (!use background)" />
		<IncludeRules context="atomic" />
	    </context>

	    <!--  arith_expr := expr op expr | op expr | expr op | "(" expr ")"
			expr := number_expr | var | arith_expr

	    This is only a goody to highlight arith ops and detect syntax errors.
	    If it causes problems, disable it: exchange context="arith_expr" in
	    "atomic" and "operator" below with context="#stay" (two or three occurrences) -->
	    <context name="arith_expr" lineEndContext="#stay" attribute="Syntax Error" noIndentationBasedFolding="true" >
		<DetectChar char="(" context="nested_expr" beginRegion="nested" attribute="( ) [ ]" />
		<DetectChar lookAhead="true" char=")" context="#pop" attribute="( ) [ ]" />
		<DetectChar lookAhead="true" char="}" context="#pop" attribute="{ DCG }" />
		<DetectChar lookAhead="true" char="]" context="#pop" attribute="( ) [ ]" />
		<!-- FIXME check if cut may be an op, else (and in any case 99.9% likely) it's a usual cut here -->
		<DetectChar lookAhead="true" char="&cut;" context="#pop" attribute="Logic &amp; Control" />
		<DetectChar lookAhead="true" char="&comma;" context="#pop" attribute="Logic &amp; Control" />
		<!-- bar & dot could be a user-def'd op, pre-def'd ops could be
		redef'd; but let's assume the default and just end the expr -->
		<DetectChar lookAhead="true" char="&bar;" context="#pop" attribute="other built-in operator" />
		<RegExpr lookAhead="true" String="&fullstop_iso;" context="#pop" attribute="Logic &amp; Control" />
		<RegExpr lookAhead="true" String="&logic_control_ops_iso;" context="#pop" attribute="Logic &amp; Control" />
		<IncludeRules context="arith_expr_common" />
	    </context>
	    <context name="nested_expr" lineEndContext="#stay" attribute="Syntax Error" noIndentationBasedFolding="true" >
		<DetectChar char="(" context="nested_expr" beginRegion="nested" attribute="( ) [ ]" />
		<DetectChar char=")" context="#pop" endRegion="nested" attribute="( ) [ ]" />
		<!-- FIXME check if cut may be an op, else it's a syntax error here -->
		<DetectChar char="&cut;" context="#stay" attribute="Normal Text" />
		<DetectChar char="&comma;" context="#stay" attribute="Syntax Error" />
		<!-- bar & dot could be a user-def'd op, pre-def'd ops could be redef'd; else these were errors -->
		<DetectChar char="&bar;" context="#stay" attribute="Normal Text" />
		<RegExpr String="&fullstop_iso;" context="#stay" attribute="Warning (!use background)" />
		<RegExpr String="&logic_control_ops_iso;" context="#stay" attribute="other built-in operator" />
		<IncludeRules context="arith_expr_common" />
	    </context>

	    <!-- list functor: assign style to the dot and let term/list/curly take the brace
	    (We have no style "built-in term", so take "Normal text" or "built-in predicate")
	    ASSERT 1st is the dot, else "Syntax Error" INTENTIONALLY -->
	    <context name="list_functor" lineEndContext="syntax_error" attribute="Syntax Error" >
		<DetectChar char="&dot;" context="#pop" attribute="other built-in predicate" />
	    </context>

	    <!-- single/double/back-quoted: handle esc seq and closing quote
	    Comments in single-quoted apply to the others as well.
	    Entry point is single/double/back-quoted (only to handle syntax error:
	    nl after opening quote), sq/dq/bq is inside the string.
	    lineEndContext should be "syntax_error", but then we couldn't handle
	    a valid esc'd line continuation.  So we do that 'in vitro' (took me
	    2 days w/ a plethora of useless rules to solve that puzzle LOL).
	    On error must not #stay, else the error isn't shown iff white or empty -->
	    <context name="sq" lineEndContext="#stay" attribute="'quo Ted'" noIndentationBasedFolding="true">
<!-- 	    <context name="single-quoted" lineEndContext="#stay" attribute="'quo Ted'" > -->
		<IncludeRules context="quoted_1st"/>
		<!-- un-esc'd nl is a syntax error; match max seq (+), else the
		error is only shown in the next line -->
		<RegExpr String="(''|&esc_oct_iso;|&esc_hex_iso;|\\&any;|[^'\\]+)$"
			context="syntax_error_sq" attribute="Syntax Error" />
		<!-- Usual handling: -->
		<Detect2Chars char="'" char1="'" context="#stay" attribute="escaped (!use background)" />
		<DetectChar char="'" context="#pop#pop" endRegion="quoted" attribute="'quo Ted'" />
		<IncludeRules context="quoted_last"/>
		<!-- Default: literal content of the quoted string, context's style applies -->
	    </context>
	    <context name="dq" lineEndContext="#stay" attribute="&quot;double-quoted&quot;" noIndentationBasedFolding="true" >
<!-- 	    <context name="double-quoted" lineEndContext="#stay" attribute="&quot;double-quoted&quot;" > -->
		<IncludeRules context="quoted_1st"/>
		<RegExpr String="(&quot;&quot;|&esc_oct_iso;|&esc_hex_iso;|\\&any;|[^&quot;\\]+)$"
			context="syntax_error_dq" attribute="Syntax Error" />
		<Detect2Chars char="&quot;" char1="&quot;" context="#stay" attribute="escaped (!use background)" />
		<DetectChar char="&quot;" context="#pop#pop" endRegion="quoted" attribute="&quot;double-quoted&quot;" />
		<IncludeRules context="quoted_last"/>
	    </context>
	    <context name="bq" lineEndContext="#stay" attribute="`back-quoted`" noIndentationBasedFolding="true" >
<!-- 	    <context name="back-quoted" lineEndContext="#stay" attribute="`back-quoted`" > -->
		<IncludeRules context="quoted_1st"/>
		<RegExpr String="(``|&esc_oct_iso;|&esc_hex_iso;|\\&any;|[^`\\]+)$"
			context="syntax_error_bq" attribute="Syntax Error" />
		<Detect2Chars char="`" char1="`" attribute="escaped (!use background)" />
		<DetectChar char="`" context="#pop#pop" endRegion="quoted" attribute="`back-quoted`" />
		<IncludeRules context="quoted_last"/>
	    </context>
	    <context name="single-quoted" lineEndContext="#stay" attribute="'quo Ted'" >
		<RegExpr String="'$" context="sq" beginRegion="quoted" attribute="Syntax Error" />
		<DetectChar char="'" context="sq" beginRegion="quoted" attribute="'quo Ted'" />
	    </context>
	    <context name="double-quoted" lineEndContext="#stay" attribute="&quot;double-quoted&quot;" >
		<RegExpr String="&quot;$" context="dq" beginRegion="quoted" attribute="Syntax Error" />
		<DetectChar char="&quot;" context="dq" beginRegion="quoted" attribute="&quot;double-quoted&quot;" />
	    </context>
	    <context name="back-quoted" lineEndContext="#stay" attribute="`back-quoted`" >
		<RegExpr String="`$" context="bq" beginRegion="quoted" attribute="Syntax Error" />
		<DetectChar char="`" context="bq" beginRegion="quoted" attribute="`back-quoted`" />
	    </context>

	    <!-- syntax_error_q: highlight next token and proceed normal afterwards -->
	    <context name="syntax_error_sq" lineEndContext="#stay" attribute="Syntax Error" noIndentationBasedFolding="true" >
		<Detect2Chars char="\" char1="'" context="#pop" attribute="Syntax Error" />
		<Detect2Chars char="'" char1="'" context="#pop" attribute="Syntax Error" />
		<DetectChar char="'" context="#pop#pop#pop" endRegion="quoted" attribute="Syntax Error" />
		<IncludeRules context="syntax_error"/>
	    </context>
	    <context name="syntax_error_dq" lineEndContext="#stay" attribute="Syntax Error" noIndentationBasedFolding="true" >
		<Detect2Chars char="\" char1="&quot;" context="#pop" attribute="Syntax Error" />
		<Detect2Chars char="&quot;" char1="&quot;" context="#pop" attribute="Syntax Error" />
		<DetectChar char="&quot;" context="#pop#pop#pop" endRegion="quoted" attribute="Syntax Error" />
		<IncludeRules context="syntax_error"/>
	    </context>
	    <context name="syntax_error_bq" lineEndContext="#stay" attribute="Syntax Error" noIndentationBasedFolding="true" >
		<Detect2Chars char="\" char1="`" context="#pop" attribute="Syntax Error" />
		<Detect2Chars char="`" char1="`" context="#pop" attribute="Syntax Error" />
		<DetectChar char="`" context="#pop#pop#pop" endRegion="quoted" attribute="Syntax Error" />
		<IncludeRules context="syntax_error"/>
	    </context>

	    <!-- char_code (after "0'"): esc seq, singleq twice, or any other -->
	    <context name="char_code" lineEndContext="#pop" attribute="Syntax Error" >
		<Detect2Chars char="'" char1="'" context="#pop" attribute="escaped (!use background)" />
		<DetectChar char="'" context="#pop" attribute="Syntax Error" />
		<DetectChar char="&bs;" context="esc_seq_cc" attribute="escaped (!use background)" />
		<DetectChar char="&tab;" context="#pop" attribute="Warning (!use background)" />
		<!-- Default _AND IF_ esc_seq #pops back here; we need this to #pop out -->
		<RegExpr String="&any;" context="#pop" attribute="0'a (!use background)" />
	    </context>
	    <!-- "0'" or "0'\" at EOL is a syntax error, catched below.  If
	    possible, highlight white @next line to make clear error is the nl -->
	    <context name="syntax_error_cc" lineEndContext="#stay" attribute="Syntax Error"
		    fallthrough="true" fallthroughContext="#pop" >
		<DetectSpaces context="#pop" attribute="Syntax Error" />
	    </context>

	    <!-- esc_seq: it's not worth it to handle common for quoted and cc
	    esc_seq_q: handle only cc in bs (e.g. "\007\"), else -> esc_seq_q2
	    ASSERT esc'd newline is handled by the calling context
	    ASSERT we get the leading bs here to highlight the whole thing -->
	    <context name="esc_seq_q" lineEndContext="syntax_error" attribute="Syntax Error" >
		<RegExpr String="&esc_oct_iso;" context="#pop" attribute="0'a (!use background)" />
		<RegExpr String="&esc_hex_iso;" context="#pop" attribute="0'a (!use background)" />
		<DetectChar char="&bs;" context="esc_seq_q2" attribute="escaped (!use background)" />
	    </context>
	    <!-- esc_seq_cc: #pop#pop out of cc on std esc seq, else pass char back
	    ASSERT we do NOT need the leading bs here anymore -->
	    <context name="esc_seq_cc" fallthrough="true" fallthroughContext="#pop"
		    lineEndContext="#pop#pop" attribute="Syntax Error" >
		<DetectChar char="&tab;" context="#pop#pop" attribute="Warning (!use background)" />
		<AnyChar String="&any_esc_iso;" context="#pop#pop" attribute="escaped (!use background)" />
	    </context>
	    <!-- esc_seq_q2: Handle standard esc seq in quoted else pass char back
	    ASSERT we do NOT need the leading bs here anymore -->
	    <context name="esc_seq_q2" fallthrough="true" fallthroughContext="#pop#pop"
		    lineEndContext="syntax_error" attribute="Syntax Error" >
		<DetectChar char="&tab;" context="#pop#pop" attribute="Warning (!use background)" />
		<AnyChar String="&any_esc_iso;" context="#pop#pop" attribute="escaped (!use background)" />
	    </context>

	    <!-- id,var,graphic: ASSERT calling context ashured 1st char is ok
				 ASSERT we get the 1st char for id & var -->
	    <context name="id" attribute="Syntax Error" >
		<DetectIdentifier context="#pop" attribute="Normal Text" />
	    </context>
	    <context name="var" attribute="Syntax Error" >
		<DetectIdentifier context="#pop" attribute="_VARIABLE" />
	    </context>
	    <context name="graphic" lineEndContext="#pop" attribute="Syntax Error" fallthrough="true" fallthroughContext="#pop" >
		<AnyChar String="&any_graphic_iso;" context="#stay" attribute="Normal Text" />
	    </context>

	    <!-- numbers (after "0[box]") -->
	    <context name="bin" lineEndContext="#pop" attribute="Syntax Error" fallthrough="true" fallthroughContext="#pop" >
		<AnyChar String="&any_bin_iso;" context="#stay" attribute="0b1001 0o007 0xF1" />
	    </context>
	    <context name="oct" lineEndContext="#pop" attribute="Syntax Error" fallthrough="true" fallthroughContext="#pop" >
		<AnyChar String="&any_oct_iso;" context="#stay" attribute="0b1001 0o007 0xF1" />
	    </context>
	    <context name="hex" lineEndContext="#pop" attribute="Syntax Error" fallthrough="true" fallthroughContext="#pop" >
		<AnyChar String="&any_hex_iso;" context="#stay" attribute="0b1001 0o007 0xF1" />
	    </context>

	    <!-- comment-iso: multi-line comment, handle closing "*/" -->
	    <context name="comment-iso" lineEndContext="#stay" attribute="% italic predicates: w/ side effects" >
	    <!-- Nested comments are not allowed in strict ISO-Prolog - - >
		<Detect2Chars char="/" char1="*" context="comment" beginRegion="comment"
			    attribute="% italic predicates: w/ side effects" />-->
		<Detect2Chars char="*" char1="/" context="#pop" endRegion="comment"
			    attribute="% italic predicates: w/ side effects" />
		<DetectSpaces />
		<IncludeRules context="##Alerts_indent" />
		<DetectIdentifier />
	    </context>

	    <!-- 1-line comment: #pop@EOL  -->
	    <!-- Folding for consecutive 1-line comments: let indentation-based auto-folding work -->
	    <context name="1-comment" lineEndContext="#pop" attribute="% italic predicates: w/ side effects" >
		<DetectSpaces />
		<IncludeRules context="##Alerts_indent" />
		<DetectIdentifier />
	    </context>
	    <context name="region_marker" lineEndContext="#pop" attribute="%BEGIN folding region" noIndentationBasedFolding="true" >
		<IncludeRules context="1-comment" />
	    </context>
	    <context name="layout_fold" lineEndContext="#stay" attribute="Syntax Error"
		    fallthrough="true" fallthroughContext="#pop" >
		<DetectSpaces attribute="Normal Text" />
		<StringDetect String="%BEGIN" context="region_marker" attribute="%BEGIN folding region"
			    firstNonSpace="true" beginRegion="user_region" />
		<StringDetect String="%END" context="region_marker" attribute="%BEGIN folding region"
			    firstNonSpace="true" endRegion="user_region" />
		<DetectChar firstNonSpace="true" char="%" context="1-comment" attribute="% italic predicates: w/ side effects" />
	    </context>

<!--####### BEGIN sub rules to be included - <context> never taken -->

	    <!-- Common for all quoted: Handle line continuation, esc seq, and most other input efficiently -->
	    <context name="quoted_1st" lineEndContext="#stay" attribute="Syntax Error" >
		<LineContinue attribute="escaped (!use background)" />
		<DetectSpaces column="0" attribute="escaped (!use background)" />
	    </context>
	    <context name="quoted_last" lineEndContext="#stay" attribute="Syntax Error" >
		<DetectChar lookAhead="true" char="&bs;" context="esc_seq_q" attribute="Syntax Error" />
		<DetectChar char="&tab;" attribute="Warning (!use background)" />
		<DetectIdentifier />
		<DetectSpaces />
	    </context>

	    <!-- layout_seq := layout layout_seq
		     layout := " " | tab | nl | comment -->
	    <context name="layout" lineEndContext="#stay" attribute="Syntax Error" >
		<DetectSpaces attribute="Normal Text" />
		<Detect2Chars char="/" char1="*" context="comment-iso" attribute="% italic predicates: w/ side effects"
			    beginRegion="comment" />
		<StringDetect String="%BEGIN" context="region_marker" attribute="%BEGIN folding region"
			    firstNonSpace="true" beginRegion="user_region" />
		<StringDetect String="%END" context="region_marker" attribute="%BEGIN folding region"
			    firstNonSpace="true" endRegion="user_region" />
		<DetectChar firstNonSpace="true" char="%" context="layout_fold" attribute="% italic predicates: w/ side effects"
			    lookAhead="true" />
		<DetectChar firstNonSpace="false" char="%" context="1-comment" attribute="% italic predicates: w/ side effects" />
	    </context>

	    <!--atomic: the work horse -->
	    <!-- (simplified) atomic := atom | number | op | var -->
	    <context name="atomic" lineEndContext="#stay" attribute="Syntax Error" >
		<!-- NOTE the order of rules is important... (as always)
		Predefined 1st, unknown (user defined) last -->
		<keyword String="logic+control ISO" context="#stay" attribute="Logic &amp; Control" />
		<keyword String="types ISO" context="#stay" attribute="Type Checking" />
		<keyword String="dyn clause mgmt ISO" context="#stay" attribute="Dynamic Clause Management" />
		<keyword String="streams ISO" context="#stay" attribute="Stream I/O" />
		<keyword String="terms ISO" context="#stay" attribute="other built-in predicate" />
		<keyword String="prolog state ISO" context="#stay" attribute="Prolog State" />
		<keyword String="DCG non-ISO" context="#stay" attribute="{ DCG }" />
		<keyword String="arith eval ISO" context="arith_expr" attribute="Arithmetics" />
		<!-- These are user predicates outside arit_expr -->
<!-- 		<keyword String="arith expr mixed ISO" context="#stay" attribute="other built-in predicate" /> -->
<!-- 		<keyword String="arith expr int ISO" context="#stay" attribute="other built-in predicate" /> -->
<!-- 		<keyword String="arith expr float ISO" context="#stay" attribute="other built-in predicate" /> -->
		<keyword String="built-ins ISO" context="#stay" attribute="other built-in predicate" />
		<keyword String="error term ISO" context="#stay" attribute="Guru Meditation" />
		<keyword String="guru meditation terms ISO" context="#stay" attribute="Guru Meditation Terms" />
		<keyword String="bogus ISO" context="#stay" attribute="ISO Bogus" />

		<!-- Match char code before quoted and ops... -->
 		<IncludeRules context="number" />

		<!-- these depend on current_prolog_flag/2 so can all return a term i.e. a list (of char codes) -->
		<DetectChar lookAhead="true" char="'" context="single-quoted" attribute="'quo Ted'" />
		<DetectChar lookAhead="true" char="`" context="back-quoted" attribute="`back-quoted`" />
		<DetectChar lookAhead="true" char="&quot;" context="double-quoted" attribute="&quot;double-quoted&quot;" />

		<!-- ...and ops before atoms, else "div", "is" etc. are usual atoms -->
 		<IncludeRules context="operator" />
		<AnyChar lookAhead="true" String="&any_lower_iso;" context="id" attribute="Normal Text" />
		<AnyChar lookAhead="true" String="&any_upper_under_iso;" context="var" attribute="_VARIABLE" />
		<AnyChar String="&any_graphic_iso;" context="graphic" attribute="Normal Text" />
	    </context> <!-- atomic -->

	    <!-- [(mostly) "atomic" for] arith_expr := arith_op | number_expr | var
				      number_expr := number | built-in | user-def

	    This is only a goody to highlight arith ops and detect syntax errors.
	    If it causes problems, disable it: exchange context="arith_expr" in
	    contexts "atomic" and "operator" with context="#stay" -->
	    <context name="arith_expr_common" lineEndContext="#stay" attribute="Syntax Error" >
		<IncludeRules context="layout" />
		<IncludeRules context="number" />
		<keyword String="arith eval ISO" context="#pop" attribute="Syntax Error" />
		<keyword String="bogus ISO" context="#stay" attribute="ISO bogus" />
		<keyword String="arith expr mixed ISO" context="#stay" attribute="Arithmetics" />
		<keyword String="arith expr int ISO" context="#stay" attribute="Integer Arithmetics" />
		<keyword String="arith expr float ISO" context="#stay" attribute="Float Arithmetics" />
		<keyword String="arith ops int ISO" context="#stay" attribute="Integer Arithmetics" />
		<RegExpr String="&arith_compare_iso;" context="#pop" attribute="Syntax Error" />
		<RegExpr String="&arith_ops_mixed_iso;" context="#stay" attribute="Arithmetics" />
		<RegExpr String="&arith_ops_int_iso;" context="#stay" attribute="Integer Arithmetics" />
		<RegExpr String="&arith_ops_float_iso;" context="#stay" attribute="Float Arithmetics" />
		<RegExpr String="&arith_bogus_iso;" context="#stay" attribute="ISO bogus" />
		<!-- do not miss user-def'd arithmetic expr and number expr-->
 		<IncludeRules context="operator" />
		<AnyChar lookAhead="true" String="&any_lower_iso;" context="id" attribute="Normal Text" />
		<AnyChar lookAhead="true" String="&any_upper_under_iso;" context="var" attribute="_VARIABLE" />
		<AnyChar String="&any_graphic_iso;" context="graphic" attribute="Normal Text" />
	    </context>

	    <!-- number := float | integer
		integer := decimal | baseN | "0'" char_code
		baseN   := "0b" [01]+ | "0o" [0-7]+ | "0x" [0-9a-fA-F]+ -->
	    <context name="number" lineEndContext="#pop" attribute="Syntax Error" >
		<!-- Match integers after other numbers -->
		<!-- FIXME a minus may precede all numbers and is part of the number! -->
		<RegExpr String="0'\\?$" context="syntax_error_cc" attribute="Syntax Error" />
		<Detect2Chars char="0" char1="'" context="char_code" attribute="0'a (!use background)" />

<!-- NOT: [box] could be a postfix op <RegExpr String="0[box]$" context="#stay" attribute="Syntax Error" /> -->
		<Detect2Chars char="0" char1="b" context="bin" attribute="0b1001 0o007 0xF1" />
		<Detect2Chars char="0" char1="o" context="oct" attribute="0b1001 0o007 0xF1" />
		<Detect2Chars char="0" char1="x" context="hex" attribute="0b1001 0o007 0xF1" />

		<RegExpr String="&float1_iso;" context="#stay" attribute="2.718281E-9" />
		<RegExpr String="&float2_iso;" context="#stay" attribute="2.718281E-9" />
		<RegExpr String="&float3_iso;" context="#stay" attribute="2.718281E-9" />
		<Int context="#stay" attribute="1  2  3  42" />
	    </context>

	    <!-- operator := predefined_op | dynamic_op (which we can not detect)
	    comma, single dot & bar are catched earlier above in term/list/curly -->
	    <context name="operator" lineEndContext="#pop" attribute="Syntax Error" >
		<keyword String="arith eval ISO" context="arith_expr" attribute="Arithmetics" />
		<keyword String="arith ops int ISO" context="#stay" attribute="other built-in operator" />
		<RegExpr String="&logic_control_ops_iso;" context="#stay" attribute="Logic &amp; Control" />
		<RegExpr String="&predefined_ops_iso;" context="#stay" attribute="other built-in operator" />
		<RegExpr String="&dcg_production_iso;" context="#stay" attribute="{ DCG }" />
		<RegExpr String="&arith_compare_iso;" context="arith_expr" attribute="Arithmetics" />
		<RegExpr String="&arith_ops_mixed_iso;" context="#stay" attribute="other built-in operator" />
		<RegExpr String="&arith_ops_int_iso;" context="#stay" attribute="other built-in operator" />
		<RegExpr String="&arith_ops_float_iso;" context="#stay" attribute="other built-in operator" />
		<RegExpr String="&arith_bogus_iso;" context="#stay" attribute="ISO bogus" />
	    </context>

<!--####### END sub rules to be included - <context> never taken -->
	</contexts>

	<!-- The idea is to give all predicates that may be affected by the
	"outside world" a common style (italic).  I.e. these can throw a
	permission, evaluation, representation or resource_error, even though
	the program itself is 100% correct.  E.g. you get_char/2 a char code 1
	from a text stream (representation_error): not your mistake, but you
	have to deal with that, too. -->
	<itemDatas>
	    <itemData name="Normal Text" defStyleNum="dsNormal" spellChecking="false" />
	    <itemData name="_VARIABLE" defStyleNum="dsDataType" bold="false" spellChecking="false" />
	    <itemData name="'quo  Ted'" defStyleNum="dsString" />
	    <itemData name="`back-quoted`" defStyleNum="dsString" italic="true" />
	    <!-- The standard wants double-quoted to be a list of char codes, but
	    e.g. GNU Prolog allows that to be set with set_prolog_flag/2 -->
	    <itemData name="&quot;double-quoted&quot;" defStyleNum="dsOthers" />
	    <itemData name="escaped (!use background)" defStyleNum="dsChar" spellChecking="false"
			 backgroundColor="#0C0C0C" selBackgroundColor="#FFFFFF" />

	    <!-- Numbers -->
	    <itemData name="2.718281E-9" defStyleNum="dsFloat" spellChecking="false" />
	    <itemData name="1  2  3  42" defStyleNum="dsDecVal" spellChecking="false" />
	    <itemData name="0b1001 0o007 0xF1" defStyleNum="dsBaseN" spellChecking="false" />
	    <itemData name="0'a (!use background)" defStyleNum="dsBaseN" spellChecking="false"
			 backgroundColor="#0C0C0C" selBackgroundColor="#FFFFFF" />

	    <itemData name="( ) [ ]" defStyleNum="dsNormal" spellChecking="false" />
	    <itemData name="{ DCG }" defStyleNum="dsKeyword" spellChecking="false" />
	    <itemData name="Prolog State" defStyleNum="dsFunction" bold="true" spellChecking="false" />
	    <itemData name="Type Checking" defStyleNum="dsDataType" spellChecking="false" />
	    <itemData name="Logic &amp; Control" defStyleNum="dsKeyword" spellChecking="false" />
	    <itemData name="other built-in operator" defStyleNum="dsFunction" spellChecking="false" />
	    <itemData name="other built-in predicate" defStyleNum="dsFunction" spellChecking="false" />

	    <!-- Comments (2nd not used, just a hint) -->
	    <itemData name="% italic predicates: w/ side effects" defStyleNum="dsComment" />
	    <itemData name="/* !avoid leading * to have folding */" defStyleNum="dsComment" />

	    <!-- Predicates affected by 'outside world' -->
	    <itemData name="Dynamic Clause Management" defStyleNum="dsKeyword" italic="true" spellChecking="false" />
	    <itemData name="Stream I/O" defStyleNum="dsFunction" italic="true" spellChecking="false" />

	    <!-- arithmetics: get the color of ints/floats; includes ops & expressions like pi -->
	    <itemData name="Arithmetics" defStyleNum="dsDataType" bold="true" italic="true" spellChecking="false" />
	    <itemData name="Integer Arithmetics" defStyleNum="dsDecVal" bold="true" italic="true" spellChecking="false" />
	    <itemData name="Float Arithmetics" defStyleNum="dsFloat" bold="true" italic="true" spellChecking="false" />
	    <itemData name="ISO Bogus" defStyleNum="dsAlert" bold="true" italic="true" spellChecking="false" />

	    <!-- no way w/ buggy kate 3.8.5: get the color of default style dsError for error and terms -->
	    <itemData name="Guru Meditation" defStyleNum="dsNormal" spellChecking="false"
		    bold="true" italic="false" underline="false" color="#FF0000" selColor="#00FFFF" />
	    <itemData name="Guru Meditation Terms" defStyleNum="dsNormal" spellChecking="false"
		    bold="false" italic="true" underline="false" color="#FF0000" selColor="#00FFFF" />
	    <itemData name="Syntax Error" defStyleNum="dsError" spellChecking="false" />
	    <itemData name="Warning (!use background)" defStyleNum="dsAlert" spellChecking="false"
			backgroundColor="#FC000C" selBackgroundColor="#03FFF3" />
	    <itemData name="%BEGIN folding region" defStyleNum="dsRegionMarker" spellChecking="false" />
	</itemDatas>
    </highlighting>
    <general>
	<keywords casesensitive="true" additionalDeliminator="$#'&quot;`" />
	<!-- let indentationsensitive folding provide it's magic -->
	<folding indentationsensitive="true" />
<!-- 	<EmptyLines> -->
<!--        <EmptyLine regexpr="^[ /t]*(?!%).*$" /> -->
<!-- 	</EmptyLines> -->
	<comments>
	    <comment name="singleLine" start="%" position="afterwhitespace" />
	    <comment name="multiLine" start="/*" end="*/" region="comment" />
	</comments>
    </general>
</language>
<!--##### NOTE [HOWTO check your syntax file, extracted from language.dtd] #####

	Copyright (c) 2001 Joseph Wenninger <[email protected]>
        modified (c) 2002 Anders Lund <[email protected]>
        modified (c) 2003 Simon Huerlimann <[email protected]>
        modified (c) 2005 Dominik Haumann <[email protected]>
        modified (c) 2008 Wilbert Berendsen <[email protected]>

    You can validate your syntax files using checkXML from the development
    package of tdelibs [author: i.e. tdelibs-dev]:

	checkXML yourSyntax.xml [author: | grep -v 'no template matches']

    If you see any 'validity error' lines, you should fix them. If you get
    a lot of 'No template matches' lines, everything's just fine. You've
    produced a valid syntax file!

    It's also possible to use the (much faster) xmllint which comes with the
    GNOME (oops:-) XML Library libxml2:

	xmllint - -dtdvalid language.dtd yourSyntax.xml
	(don't use a space between the two - [author: dashes/minus]
	That's just because XML comments don't allow that:-(

    To use your syntax file, copy it to .kde/share/apps/katepart/syntax/ in
    your home directory. You have to open a new instance of kwrite/kate to use
    the new syntax file.
-->