nomsu.1.peg - nomsu

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