code / nomsu

   1 -- Nomsu version 3
   2 file <-
   3     {:curr_indent: ' '* :}
   4     (((action / expression / inline_block / indented_block) eol !.)
   5      / file_chunks / empty_block)
   6     {:curr_indent: %nil :}
   7     !.
   8 
   9 shebang <- "#!" (!"nomsu" [^%nl])* "nomsu" ws+ "-V" ws* {:version: [0-9.]+ :} [^%nl]*
  10 
  11 file_chunks (FileChunks) <-
  12     shebang? comment? blank_lines?
  13     (top_block (nl_nodent section_division top_block)*)
  14     blank_lines?
  15     unexpected_indent? unexpected_chunk?
  16 
  17 top_block (Block) <-
  18     ((blank_lines nodent) / (comment nl_nodent))? statement (nl_nodent statement)*
  19 
  20 empty_block (Block) <-
  21     comment? blank_lines?
  22 
  23 nodent <- (unexpected_indent [^%nl]* / =curr_indent)
  24 indent <- =curr_indent "    "
  25 blank_lines <- %nl ((nodent comment / ws*) %nl)*
  26 eol <- ws* eol_comment? (!. / &%nl)
  27 
  28 nl_nodent <- blank_lines nodent
  29 nl_indent <- blank_lines {:curr_indent: indent :} (comment nl_nodent)*
  30 
  31 comment (Comment) <-
  32     "#" {~ [^%nl]* (%nl+ (indent -> '') [^%nl]*)* ~}
  33 eol_comment (Comment) <-
  34     "#" {[^%nl]*}
  35 
  36 unexpected_code <- ws* _unexpected_code
  37 _unexpected_code (Error) <-
  38     {:error: {~ [^%nl]+ -> "Couldn't parse this code." ~} :}
  39 unexpected_chunk (Error) <-
  40     {:error: {~ .+ -> "Couldn't parse this chunk of code." ~} :}
  41 unexpected_indent (Error) <-
  42     {:error: {~ (=curr_indent ws+) -> "This indentation is messed up." ~} :}
  43     {:hint: {~ '' -> 'This line should either have the same indentation as the line above it, or exactly 4 spaces more.' ~} :}
  44 missing_paren_err (Error) <-
  45     {:error: {~ eol -> 'Line ended without finding a closing )-parenthesis' ~} :}
  46     {:hint: {~ '' -> 'Put a ")" here' ~} :}
  47 missing_quote_err (Error) <-
  48     {:error: {~ eol -> 'Line ended before finding a closing double quotation mark' ~} :}
  49     {:hint: {~ "" -> "Put a quotation mark here" ~} :}
  50 missing_bracket_error (Error) <-
  51     {:error: {~ eol -> "Line ended before finding a closing ]-bracket" ~} :}
  52     {:hint: {~ '' -> 'Put a "]" here' ~} :}
  53 missing_brace_error (Error) <-
  54     {:error: {~ eol -> "Line ended before finding a closing }-brace" ~} :}
  55     {:hint: {~ '' -> 'Put a "}" here' ~} :}
  56 
  57 section_division <- ("~")^+3 eol
  58 
  59 inline_block <-
  60     "(" ws* inline_block  ws* ")" / raw_inline_block
  61 raw_inline_block (Block) <-
  62     (!"::") ":" ws* ((inline_statement (ws* ";" ws* inline_statement)*) / !(eol nl_indent))
  63 indented_block (Block) <-
  64     ":" eol nl_indent statement (nl_nodent statement)*
  65     (%nl (ws* %nl)* nodent (comment / eol / unexpected_code))*
  66     {:curr_indent: %nil :}
  67 
  68 statement <-
  69     (action / expression) (eol / unexpected_code)
  70 
  71 inline_statement <- (inline_action / inline_expression)
  72 
  73 noindex_inline_expression <-
  74     number / variable / inline_text / inline_list / inline_dict / inline_nomsu
  75     / ( "("
  76             ws* (inline_action / inline_expression) ws*
  77             (ws* ',' ws* (inline_action / inline_expression) ws*)*
  78         (")" / missing_paren_err / unexpected_code)
  79       )
  80 inline_expression <- index_chain / noindex_inline_expression 
  81 indented_expression <-
  82     indented_text / indented_nomsu / indented_list / indented_dict / ({|
  83         "(..)" nl_indent
  84             (action / expression) (eol / unexpected_code)
  85             (%nl (ws* %nl)* nodent (comment / eol / unexpected_code))*
  86         {:curr_indent: %nil :}
  87     |} -> unpack)
  88 expression <-
  89     inline_expression / indented_expression
  90 
  91 inline_nomsu (EscapedNomsu) <- "\" (inline_expression / inline_block)
  92 indented_nomsu (EscapedNomsu) <-
  93     "\" (noindex_inline_expression / inline_block / indented_expression / indented_block)
  94 
  95 index_chain (IndexChain) <-
  96     noindex_inline_expression ("." (text_word / noindex_inline_expression))+
  97 
  98 inline_action <- inline_methodcall / _inline_action
  99 inline_methodcall <- 
 100     inline_arg ws* "::" ws* _inline_action
 101 -- Actions need either at least 1 word, or at least 2 tokens
 102 _inline_action (Action) <-
 103     !section_division
 104     (   (inline_arg (ws* (inline_arg / word))+)
 105       / (word       (ws* (inline_arg / word))*))
 106     (ws* inline_block)?
 107 inline_arg <- inline_expression / inline_block
 108 
 109 action <- methodcall / _action
 110 methodcall <-
 111     arg (nl_nodent "..")? ws* "::" (nl_nodent "..")? ws* _action
 112 _action (Action) <-
 113     !section_division
 114     (  (arg  ((nl_nodent "..")? ws* (arg / word))+)
 115      / (word ((nl_nodent "..")? ws* (arg / word))*))
 116 arg <- expression / inline_block / indented_block
 117 
 118 word <- !number { operator_char+ / ident_char+ }
 119 
 120 text_word (Text) <- word
 121 
 122 inline_text (Text) <-
 123     !(indented_text)
 124     ('"' _inline_text* ('"' / missing_quote_err / unexpected_code))
 125 _inline_text <-
 126     {~ (('\"' -> '"') / ('\\' -> '\') / escaped_char / [^%nl\"]+)+ ~}
 127     / inline_text_interpolation
 128 inline_text_interpolation <-
 129     "\" (
 130         variable / inline_list / inline_dict / inline_text
 131         / ("("
 132             ws* (inline_action / inline_expression) ws*
 133             (ws* ',' ws* (inline_action / inline_expression) ws*)*
 134         (")" / missing_paren_err / unexpected_code))
 135     )
 136 
 137 indented_text (Text) <-
 138     '".."' eol %nl {%nl+}? {:curr_indent: indent :}
 139         (indented_plain_text / text_interpolation / {~ %nl+ (=curr_indent -> "") ~})*
 140     unexpected_code?
 141     {:curr_indent: %nil :}
 142 -- Tracking text-lines-within-indented-text as separate objects allows for better debugging line info
 143 indented_plain_text (Text) <-
 144     {~ (("\\" -> "\") / (("\" blank_lines =curr_indent "..") -> "") / (!text_interpolation "\") / [^%nl\]+)+
 145         (%nl+ (=curr_indent -> ""))* ~}
 146 text_interpolation <-
 147     inline_text_interpolation / ("\" indented_expression (blank_lines =curr_indent "..")?)
 148 
 149 number (Number) <- (("-"? (([0-9]+ "." [0-9]+) / ("." [0-9]+) / "0x" [0-9a-fA-F]+ / ([0-9]+)))-> tonumber)
 150 
 151 -- Variables can be nameless (i.e. just %) and can only contain identifier chars.
 152 -- This ensures you don't get weird parsings of `%x+%y` or `%'s thing`.
 153 variable (Var) <- "%" {ident_char*}
 154 
 155 inline_list (List) <-
 156     !('[..]')
 157     "[" ws*
 158         (inline_list_item (ws* ',' ws* inline_list_item)* (ws* ',')?)? ws*
 159     ("]" / (","? (missing_bracket_error / unexpected_code)))
 160 indented_list (List) <-
 161     "[..]" eol nl_indent
 162         list_line (nl_nodent list_line)*
 163         (%nl (ws* %nl)* nodent (comment / eol / unexpected_code))*
 164     (","? unexpected_code)?
 165 list_line <-
 166       (inline_list_item ws* "," ws*)+ eol
 167     / (inline_list_item ws* "," ws*)* (action / expression) eol
 168 inline_list_item <- inline_action / inline_expression
 169 
 170 inline_dict (Dict) <-
 171     !('{..}')
 172     "{" ws*
 173         (inline_dict_entry (ws* ',' ws* inline_dict_entry)*)? ws*
 174     ("}" / (","? (missing_brace_error / unexpected_code)))
 175 indented_dict (Dict) <-
 176     "{..}" eol nl_indent
 177         dict_line (nl_nodent dict_line)*
 178         (%nl (ws* %nl)* nodent (comment / eol / unexpected_code))*
 179     (","? unexpected_code)?
 180 dict_line <-
 181       (inline_dict_entry ws* "," ws*)+ eol
 182     / (inline_dict_entry ws* "," ws*)* dict_entry eol
 183 dict_entry(DictEntry) <-
 184     dict_key (ws* ":" ws* (action / expression))?
 185 inline_dict_entry(DictEntry) <-
 186     dict_key (ws* ":" ws* (inline_action / inline_expression)?)?
 187 dict_key <-
 188     text_word / inline_expression
 189 
 190 operator_char <- ['`~!@$^&*+=|<>?/-]
 191 ident_char <- [a-zA-Z0-9_] / %utf8_char
 192 ws <- [ %tab]
 193 
 194 escaped_char <-
 195     ("\"->'') (
 196         (([xX]->'') ((({[0-9a-fA-F]^2} %number_16) -> tonumber) -> tochar))
 197       / ((([0-9] [0-9]^-2) -> tonumber) -> tochar)
 198       / ("a"->ascii_7)  / ("b"->ascii_8)  / ("t"->ascii_9)  / ("n"->ascii_10)
 199       / ("v"->ascii_11) / ("f"->ascii_12) / ("r"->ascii_13)
 200     )