From b997a41a9b9f2ed27706e9e79be85db828b5c428 Mon Sep 17 00:00:00 2001
From: mitchell <70453897+orbitalquark@users.noreply.github.com>
Date: Sun, 17 Sep 2023 10:24:17 -0400
Subject: [PATCH] Migrated Matlab lexer. Dropped Octave constants. Thanks to
 Randy Palamar.

---
 lexers/matlab.lua | 92 ++++++++++++++++++++++++-----------------------
 1 file changed, 47 insertions(+), 45 deletions(-)

diff --git a/lexers/matlab.lua b/lexers/matlab.lua
index 6e09d4f..681ddbb 100644
--- a/lexers/matlab.lua
+++ b/lexers/matlab.lua
@@ -2,72 +2,42 @@
 -- Matlab LPeg lexer.
 -- Based off of lexer code by Mitchell.
 
-local lexer = require('lexer')
-local token, word_match = lexer.token, lexer.word_match
-local P, S = lpeg.P, lpeg.S
+local lexer = lexer
+local P, B, S = lpeg.P, lpeg.B, lpeg.S
 
-local lex = lexer.new('matlab')
-
--- Whitespace.
-lex:add_rule('whitespace', token(lexer.WHITESPACE, lexer.space^1))
+local lex = lexer.new(...)
 
 -- Keywords.
-lex:add_rule('keyword', token(lexer.KEYWORD, word_match({
-  'break', 'case', 'catch', 'continue', 'do', 'else', 'elseif', 'end', 'end_try_catch',
-  'end_unwind_protect', 'endfor', 'endif', 'endswitch', 'endwhile', 'for', 'function',
-  'endfunction', 'global', 'if', 'otherwise', 'persistent', 'replot', 'return', 'static', 'switch',
-  'try', 'until', 'unwind_protect', 'unwind_protect_cleanup', 'varargin', 'varargout', 'while'
-}, true)))
+lex:add_rule('keyword', lex:tag(lexer.KEYWORD, lex:word_match(lexer.KEYWORD)))
 
 -- Functions.
-lex:add_rule('function', token(lexer.FUNCTION, word_match{
-  'abs', 'any', 'argvatan2', 'axes', 'axis', 'ceil', 'cla', 'clear', 'clf', 'columns', 'cos',
-  'delete', 'diff', 'disp', 'doc', 'double', 'drawnow', 'exp', 'figure', 'find', 'fix', 'floor',
-  'fprintf', 'gca', 'gcf', 'get', 'grid', 'help', 'hist', 'hold', 'isempty', 'isnull', 'length',
-  'load', 'log', 'log10', 'loglog', 'max', 'mean', 'median', 'min', 'mod', 'ndims', 'numel',
-  'num2str', 'ones', 'pause', 'plot', 'printf', 'quit', 'rand', 'randn', 'rectangle', 'rem',
-  'repmat', 'reshape', 'round', 'rows', 'save', 'semilogx', 'semilogy', 'set', 'sign', 'sin',
-  'size', 'sizeof', 'size_equal', 'sort', 'sprintf', 'squeeze', 'sqrt', 'std', 'strcmp', 'subplot',
-  'sum', 'tan', 'tic', 'title', 'toc', 'uicontrol', 'who', 'xlabel', 'ylabel', 'zeros'
-}))
-
--- Constants.
-lex:add_rule('constant', token(lexer.CONSTANT, word_match(
-  'EDITOR I IMAGEPATH INFO_FILE J LOADPATH OCTAVE_VERSION PAGER PS1 PS2 PS4 PWD')))
+local builtin_func = lex:tag(lexer.FUNCTION_BUILTIN, lex:word_match(lexer.FUNCTION_BUILTIN))
+local func = lex:tag(lexer.FUNCTION, lexer.word)
+local method = B('.') * lex:tag(lexer.FUNCTION_METHOD, lexer.word)
+lex:add_rule('function', (method + builtin_func + func) * #(lexer.space^0 * S('(')))
 
 -- Variable.
-lex:add_rule('variable', token(lexer.VARIABLE, word_match{
-  'ans', 'automatic_replot', 'default_return_value', 'do_fortran_indexing',
-  'define_all_return_values', 'empty_list_elements_ok', 'eps', 'false', 'gnuplot_binary',
-  'ignore_function_time_stamp', 'implicit_str_to_num_ok', 'Inf', 'inf', 'NaN', 'nan',
-  'ok_to_lose_imaginary_part', 'output_max_field_width', 'output_precision', 'page_screen_output',
-  'pi', 'prefer_column_vectors', 'prefer_zero_one_indexing', 'print_answer_id_name',
-  'print_empty_dimensions', 'realmax', 'realmin', 'resize_on_range_error',
-  'return_last_computed_value', 'save_precision', 'silent_functions', 'split_long_rows',
-  'suppress_verbose_help_message', 'treat_neg_dim_as_zero', 'true', 'warn_assign_as_truth_value',
-  'warn_comma_in_global_decl', 'warn_divide_by_zero', 'warn_function_name_clash',
-  'whitespace_in_literal_matrix'
-}))
+lex:add_rule('variable', lex:tag(lexer.VARIABLE_BUILTIN, lex:word_match(lexer.VARIABLE_BUILTIN)))
 
 -- Identifiers.
-lex:add_rule('identifier', token(lexer.IDENTIFIER, lexer.word))
+lex:add_rule('identifier', lex:tag(lexer.IDENTIFIER, lexer.word))
 
 -- Strings.
 local sq_str = lexer.range("'", true)
 local dq_str = lexer.range('"')
 local bq_str = lexer.range('`')
-lex:add_rule('string', token(lexer.STRING, sq_str + dq_str + bq_str))
+lex:add_rule('string', lex:tag(lexer.STRING, sq_str + dq_str + bq_str))
 
 -- Comments.
 local line_comment = lexer.to_eol(S('%#'))
 local block_comment = lexer.range('%{', '%}')
-lex:add_rule('comment', token(lexer.COMMENT, block_comment + line_comment))
+lex:add_rule('comment', lex:tag(lexer.COMMENT, block_comment + line_comment))
 
 -- Numbers.
-lex:add_rule('number', token(lexer.NUMBER, lexer.number))
+lex:add_rule('number', lex:tag(lexer.NUMBER, lexer.number))
 
 -- Operators.
-lex:add_rule('operator', token(lexer.OPERATOR, S('!%^&*()[]{}-=+/\\|:;.,?<>~`´')))
+lex:add_rule('operator', lex:tag(lexer.OPERATOR, S('!%^&*()[]{}-=+/\\|:;.,?<>~`´')))
 
 -- Fold points.
 lex:add_fold_point(lexer.KEYWORD, 'if', 'end')
@@ -78,6 +48,38 @@ lex:add_fold_point(lexer.OPERATOR, '(', ')')
 lex:add_fold_point(lexer.OPERATOR, '[', ']')
 lex:add_fold_point(lexer.COMMENT, '%{', '%}')
 
-lexer.property['scintillua.comment'] = '#'
+-- Word lists
+lex:set_word_list(lexer.KEYWORD, {
+  'break', 'case', 'catch', 'continue', 'do', 'else', 'elseif', 'end', 'end_try_catch',
+  'end_unwind_protect', 'endfor', 'endif', 'endswitch', 'endwhile', 'for', 'function',
+  'endfunction', 'global', 'if', 'otherwise', 'persistent', 'replot', 'return', 'static', 'switch',
+  'try', 'until', 'unwind_protect', 'unwind_protect_cleanup', 'varargin', 'varargout', 'while'
+})
+
+lex:set_word_list(lexer.FUNCTION_BUILTIN, {
+  'abs', 'any', 'argvatan2', 'axes', 'axis', 'ceil', 'cla', 'clear', 'clf', 'columns', 'cos',
+  'delete', 'diff', 'disp', 'doc', 'double', 'drawnow', 'exp', 'figure', 'find', 'fix', 'floor',
+  'fprintf', 'gca', 'gcf', 'get', 'grid', 'help', 'hist', 'hold', 'isempty', 'isnull', 'length',
+  'load', 'log', 'log10', 'loglog', 'max', 'mean', 'median', 'min', 'mod', 'ndims', 'numel',
+  'num2str', 'ones', 'pause', 'plot', 'printf', 'quit', 'rand', 'randn', 'rectangle', 'rem',
+  'repmat', 'reshape', 'round', 'rows', 'save', 'semilogx', 'semilogy', 'set', 'sign', 'sin',
+  'size', 'sizeof', 'size_equal', 'sort', 'sprintf', 'squeeze', 'sqrt', 'std', 'strcmp', 'subplot',
+  'sum', 'tan', 'tic', 'title', 'toc', 'uicontrol', 'who', 'xlabel', 'ylabel', 'zeros'
+})
+
+lex:set_word_list(lexer.VARIABLE_BUILTIN, {
+  'ans', 'automatic_replot', 'default_return_value', 'do_fortran_indexing',
+  'define_all_return_values', 'empty_list_elements_ok', 'eps', 'false', 'gnuplot_binary',
+  'ignore_function_time_stamp', 'implicit_str_to_num_ok', 'Inf', 'inf', 'NaN', 'nan',
+  'ok_to_lose_imaginary_part', 'output_max_field_width', 'output_precision', 'page_screen_output',
+  'pi', 'prefer_column_vectors', 'prefer_zero_one_indexing', 'print_answer_id_name',
+  'print_empty_dimensions', 'realmax', 'realmin', 'resize_on_range_error',
+  'return_last_computed_value', 'save_precision', 'silent_functions', 'split_long_rows',
+  'suppress_verbose_help_message', 'treat_neg_dim_as_zero', 'true', 'warn_assign_as_truth_value',
+  'warn_comma_in_global_decl', 'warn_divide_by_zero', 'warn_function_name_clash',
+  'whitespace_in_literal_matrix'
+})
+
+lexer.property['scintillua.comment'] = '%'
 
 return lex