From 54e1f99e90764570666d692dd97be888f98ee2fc Mon Sep 17 00:00:00 2001 From: Bruce Hill Date: Fri, 14 Sep 2018 14:25:55 -0700 Subject: [PATCH] Upgraded nomsu.1.peg to tidier version. --- nomsu.1.peg | 161 ++++++++++++++++++++++++++------------------ nomsu_compiler.lua | 35 ++++------ nomsu_compiler.moon | 1 - 3 files changed, 106 insertions(+), 91 deletions(-) diff --git a/nomsu.1.peg b/nomsu.1.peg index 10a7715..5d881b3 100644 --- a/nomsu.1.peg +++ b/nomsu.1.peg @@ -1,65 +1,87 @@ --- Nomsu version 2 +-- Nomsu version 1 file (FileChunks): {:curr_indent: ' '* :} - ("#!" (!"nomsu" [^%nl])* "nomsu" %ws+ "-V" %ws* {:version: [0-9.]+ :} [^%nl]*)? + ("#!" (!"nomsu" [^%nl])* "nomsu" ws+ "-V" ws* {:version: [0-9.]+ :} [^%nl]*)? comment? blank_lines? (chunk (nl_nodent chunk_delimeter nl_nodent chunk)*)? blank_lines? - %ws* (!! .+ -> "Parse error" !!)? + {:curr_indent: %nil :} + !. -nodent: =curr_indent !(" ") +nodent: (unexpected_indent [^%nl]* / =curr_indent) indent: =curr_indent " " -blank_lines: %nl ((nodent comment / %ws*) %nl)* -eol: %ws* eol_comment? (!. / &%nl) +blank_lines: %nl ((nodent comment / ws*) %nl)* +eol: ws* eol_comment? (!. / &%nl) nl_nodent: blank_lines nodent -nl_indent: blank_lines {:curr_indent: indent :} (comment nl_nodent)? +nl_indent: blank_lines {:curr_indent: indent :} (comment nl_nodent)* -comment: - "#" (({} {~ [^%nl]* (%nl+ (indent -> '') [^%nl]*)* ~} %userdata) => add_comment) -eol_comment: - "#" (({} {[^%nl]*} %userdata) => add_comment) +comment (Comment): + "#" {~ [^%nl]* (%nl+ (indent -> '') [^%nl]*)* ~} +eol_comment (Comment): + "#" {[^%nl]*} + +unexpected_code: ws* _unexpected_code +_unexpected_code (Error): + {:error: {~ [^%nl]+ -> "Couldn't parse this code" ~} :} +unexpected_indent (Error): + {:error: {~ (=curr_indent ws+) -> "Messed up indentation" ~} :} + {:hint: {~ '' -> 'Either make sure this line is aligned with the one above it, or make sure the previous line ends with something that uses indentation, like ":" or "(..)"' ~} :} +missing_paren_err (Error): + {:error: {~ eol -> 'Line ended without finding a closing )-parenthesis' ~} :} + {:hint: {~ '' -> 'Put a ")" here' ~} :} +missing_quote_err (Error): + {:error: {~ eol -> 'Line ended before finding a closing double quotation mark' ~} :} + {:hint: {~ "" -> "Put a quotation mark here" ~} :} +missing_bracket_error (Error): + {:error: {~ eol -> "Line ended before finding a closing ]-bracket" ~} :} + {:hint: {~ '' -> 'Put a "]" here' ~} :} +missing_brace_error (Error): + {:error: {~ eol -> "Line ended before finding a closing }-brace" ~} :} + {:hint: {~ '' -> 'Put a "}" here' ~} :} +missing_block_expr_error (Error): + {:error: '' -> "Missing expression after the ':'" :} chunk: block / action / expression chunk_delimeter: ("~")^+3 eol inline_block (Block): - inline_statement (%ws* ";" %ws* inline_statement)+ + inline_statement (ws* ";" ws* inline_statement)+ block (Block): statement (nl_nodent statement)+ + (%nl (ws* %nl)* nodent (comment / eol / unexpected_code))* + {:curr_indent: %nil :} -statement: (action / expression) (eol / (!! [^%nl]+ -> "Unexpected character while parsing line" !!)) +statement: (action / expression) (eol / unexpected_code) inline_statement: (inline_action / inline_expression) noindex_inline_expression: number / variable / inline_text / inline_list / inline_dict / inline_nomsu / ( "(" - %ws* (inline_block / inline_action / inline_expression) %ws* - (%ws* ',' %ws* (inline_block / inline_action / inline_expression) %ws*)* - (")" - / (!! eol -> 'Line ended without finding a closing )-parenthesis' !!) - / (!! [^%nl]+ -> 'Unexpected character while parsing subexpression' !!) - ) + ws* (inline_block / inline_action / inline_expression) ws* + (ws* ',' ws* (inline_block / inline_action / inline_expression) ws*)* + (")" / missing_paren_err / unexpected_code) ) inline_expression: index_chain / noindex_inline_expression indented_expression: indented_text / indented_nomsu / indented_list / indented_dict / ({| ("(..)")? nl_indent - (block / action / expression) (nl_nodent comment)* - (eol / (!! [^%nl]+ -> "Unexpected character while parsing indented expression" !!)) + (block / action / expression) (eol / unexpected_code) + (%nl (ws* %nl)* nodent (comment / eol / unexpected_code))* + {:curr_indent: %nil :} |} -> unpack) expression: inline_expression - / (":" %ws* - (inline_block / inline_action / inline_expression / (!! '' -> "Missing expression after the ':'" !!))) + / (":" ws* + (inline_block / inline_action / inline_expression / missing_block_expr_error)) / indented_expression inline_nomsu (EscapedNomsu): "\" inline_expression indented_nomsu (EscapedNomsu): "\" ( noindex_inline_expression - / (":" %ws* - (inline_block / inline_action / inline_expression) (!! '' -> "Missing expression after the '\:'" !!)) + / (":" ws* + (inline_block / inline_action / inline_expression / missing_block_expr_error)) / indented_expression) index_chain (IndexChain): @@ -68,42 +90,39 @@ index_chain (IndexChain): -- Actions need either at least 1 word, or at least 2 tokens inline_action (Action): !chunk_delimeter - ( (inline_expression (%ws* (inline_expression / word))+) - / (word (%ws* (inline_expression / word))*)) - (%ws* ":" %ws* (inline_block / inline_action / inline_expression - / (!! '' -> "Missing expression after the ':'" !!)))? + ( (inline_expression (ws* (inline_expression / word))+) + / (word (ws* (inline_expression / word))*)) + (ws* ":" ws* (inline_block / inline_action / inline_expression + / missing_block_expr_error))? action (Action): !chunk_delimeter - ( (expression ((nl_nodent "..")? %ws* (expression / word))+) - / (word ((nl_nodent "..")? %ws* (expression / word))*)) + ( (expression ((nl_nodent "..")? ws* (expression / word))+) + / (word ((nl_nodent "..")? ws* (expression / word))*)) -word: !number { %operator_char+ / %ident_char+ } +word: !number { operator_char+ / ident_char+ } text_word (Text): word inline_text (Text): - !('".."' eol) + !(indented_text) '"' - ({~ (('\"' -> '"') / ('\\' -> '\') / %escaped_char / [^%nl\"])+ ~} + ({~ (('\"' -> '"') / ('\\' -> '\') / escaped_char / [^%nl\"])+ ~} / inline_text_interpolation)* - ('"' - / (!! eol -> 'Line ended before finding a closing double quotation mark' !!) - / (!! [^%nl]+ -> 'Unexpected character while parsing Text' !!)) + ('"' / missing_quote_err / unexpected_code) inline_text_interpolation: "\" ( variable / inline_list / inline_dict / inline_text / ("(" - %ws* (inline_block / inline_action / inline_expression) %ws* - (%ws* ',' %ws* (inline_block / inline_action / inline_expression) %ws*)* - (")" - / (!! eol -> 'Line ended without finding a closing )-parenthesis' !!) - / (!! [^%nl]+ -> 'Unexpected character while parsing Text interpolation' !!))) + ws* (inline_block / inline_action / inline_expression) ws* + (ws* ',' ws* (inline_block / inline_action / inline_expression) ws*)* + (")" / missing_paren_err / unexpected_code)) ) indented_text (Text): - '".."' eol %nl {%nl*} {:curr_indent: indent :} + '".."' eol %nl {%nl+}? {:curr_indent: indent :} (indented_plain_text / text_interpolation / {~ %nl+ (=curr_indent -> "") ~})* - (!! [^%nl]+ -> "Unexpected character while parsing Text" !!)? + unexpected_code? + {:curr_indent: %nil :} indented_plain_text (Text): {~ (("\\" -> "\") / (("\" blank_lines =curr_indent "..") -> "") / (!text_interpolation "\") / [^%nl\]+)+ (%nl+ (=curr_indent -> ""))* ~} @@ -114,43 +133,51 @@ number (Number): (("-"? (([0-9]+ "." [0-9]+) / ("." [0-9]+) / ([0-9]+)))-> tonum -- Variables can be nameless (i.e. just %) and can't contain operators like apostrophe -- which is a hack to allow %'s to parse as "%" and "' s" separately -variable (Var): "%" {(%ident_char+ ((!"'" %operator_char+) / %ident_char+)*)?} +variable (Var): "%" {(ident_char+ ((!"'" operator_char+) / ident_char+)*)?} inline_list (List): !('[..]') - "[" %ws* - (inline_list_item (%ws* ',' %ws* inline_list_item)* (%ws* ',')?)? %ws* - ("]" / (","? ( - (!! eol -> "Line ended before finding a closing ]-bracket" !!) - /(!! [^%nl]+ -> "Unexpected character while parsing List" !!) - ))) + "[" ws* + (inline_list_item (ws* ',' ws* inline_list_item)* (ws* ',')?)? ws* + ("]" / (","? (missing_bracket_error / unexpected_code))) indented_list (List): "[..]" eol nl_indent - list_line (nl_nodent list_line)* (nl_nodent comment)* - (","? (!! [^%nl]+ -> "Unexpected character while parsing List" !!))? + list_line (nl_nodent list_line)* + (%nl (ws* %nl)* nodent (comment / eol / unexpected_code))* + (","? unexpected_code)? list_line: - (inline_list_item %ws* "," %ws*)+ eol - / (inline_list_item %ws* "," %ws*)* (action / expression) eol + (inline_list_item ws* "," ws*)+ eol + / (inline_list_item ws* "," ws*)* (action / expression) eol inline_list_item: inline_block / inline_action / inline_expression inline_dict (Dict): !('{..}') - "{" %ws* - (inline_dict_entry (%ws* ',' %ws* inline_dict_entry)*)? %ws* - ("}" / (","? ( - (!! eol -> "Line ended before finding a closing }-brace" !!) - / (!! [^%nl]* -> "Unexpected character while parsing Dictionary" !!) - ))) + "{" ws* + (inline_dict_entry (ws* ',' ws* inline_dict_entry)*)? ws* + ("}" / (","? (missing_brace_error / unexpected_code))) indented_dict (Dict): "{..}" eol nl_indent - dict_line (nl_nodent dict_line)* (nl_nodent comment)* - (","? (!! [^%nl]+ -> "Unexpected character while parsing Dictionary" !!))? + dict_line (nl_nodent dict_line)* + (%nl (ws* %nl)* nodent (comment / eol / unexpected_code))* + (","? unexpected_code)? dict_line: - (inline_dict_entry %ws* "," %ws*)+ eol - / (inline_dict_entry %ws* "," %ws*)* dict_entry eol + (inline_dict_entry ws* "," ws*)+ eol + / (inline_dict_entry ws* "," ws*)* dict_entry eol dict_entry(DictEntry): - dict_key (%ws* ":" %ws* (action / expression))? + dict_key (ws* ":" ws* (action / expression))? inline_dict_entry(DictEntry): - dict_key (%ws* ":" %ws* (inline_block / inline_action / inline_expression)?)? + dict_key (ws* ":" ws* (inline_block / inline_action / inline_expression)?)? dict_key: text_word / inline_expression + +operator_char: ['`~!@$^&*+=|<>?/-] +ident_char: [a-zA-Z0-9_] / %utf8_char +ws: [ %tab] + +escaped_char: + ("\"->'') ( + (([xX]->'') ((({[0-9a-fA-F]^2} %number_16) -> tonumber) -> tochar)) + / ((([0-9] [0-9]^-2) -> tonumber) -> tochar) + / ("a"->ascii_7) / ("b"->ascii_8) / ("t"->ascii_9) / ("n"->ascii_10) + / ("v"->ascii_11) / ("f"->ascii_12) / ("r"->ascii_13) + ) diff --git a/nomsu_compiler.lua b/nomsu_compiler.lua index 9572640..e655759 100644 --- a/nomsu_compiler.lua +++ b/nomsu_compiler.lua @@ -98,33 +98,22 @@ end local Parsers = { } local max_parser_version = 0 for version = 1, 999 do - local _continue_0 = false - repeat - if not (version == 4 or version == 3 or version == 2) then - _continue_0 = true - break - end - local peg_file = io.open("nomsu." .. tostring(version) .. ".peg") - if not peg_file and package.nomsupath then - for path in package.nomsupath:gmatch("[^;]+") do - peg_file = io.open(path .. "/nomsu." .. tostring(version) .. ".peg") - if peg_file then - break - end + local peg_file = io.open("nomsu." .. tostring(version) .. ".peg") + if not peg_file and package.nomsupath then + for path in package.nomsupath:gmatch("[^;]+") do + peg_file = io.open(path .. "/nomsu." .. tostring(version) .. ".peg") + if peg_file then + break end end - if not (peg_file) then - break - end - max_parser_version = version - local peg_contents = peg_file:read("*a") - Parsers[version] = make_parser(peg_contents, make_tree) - peg_file:close() - _continue_0 = true - until true - if not _continue_0 then + end + if not (peg_file) then break end + max_parser_version = version + local peg_contents = peg_file:read("*a") + Parsers[version] = make_parser(peg_contents, make_tree) + peg_file:close() end local MAX_LINE = 80 local NomsuCompiler = setmetatable({ diff --git a/nomsu_compiler.moon b/nomsu_compiler.moon index 8711839..fb3dd4c 100644 --- a/nomsu_compiler.moon +++ b/nomsu_compiler.moon @@ -56,7 +56,6 @@ make_tree = (tree, userdata)-> Parsers = {} max_parser_version = 0 for version=1,999 - continue unless version == 4 or version == 3 or version == 2 -- TODO: remove peg_file = io.open("nomsu.#{version}.peg") if not peg_file and package.nomsupath for path in package.nomsupath\gmatch("[^;]+")