nomsu/nomsu.2.peg

-- Nomsu version 2
file (FileChunks):
    {:curr_indent: ' '* :}
    ("#!" (!"nomsu" [^%nl])* "nomsu" %ws+ "-V" %ws* {:version: [0-9.]+ :} [^%nl]*)?
    comment? blank_lines?
    (chunk (nl_nodent section_division nl_nodent chunk)*)?
    blank_lines?
    %ws* (!! .+ -> "Parse error" !!)?

nodent: =curr_indent !(" ")
indent: =curr_indent "    "
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)?

comment:
    "#" (({} {~ [^%nl]* (%nl+ (indent -> '') [^%nl]*)* ~} %userdata) => add_comment)
eol_comment:
    "#" (({} {[^%nl]*} %userdata) => add_comment)

section_division: ("~")^+3 eol

inline_block (Block):
    "(" %ws* inline_block  %ws* ")"
    / ":" %ws* ((inline_statement (%ws* ";" %ws* inline_statement)*) / !(eol nl_indent))
chunk (Block):
    statement (nl_nodent statement)*
indented_block (Block):
    ":" eol nl_indent statement (nl_nodent statement)* (%nl (%ws* %nl)* nodent comment)*

statement: (action / expression) (eol / (!! [^%nl]+ -> "Unexpected character while parsing line" !!))
inline_statement: (inline_action / inline_expression)

noindex_inline_expression:
    number / variable / inline_text / inline_list / inline_dict / inline_nomsu
    / ( "("
            %ws* (inline_action / inline_expression) %ws*
            (%ws* ',' %ws* (inline_action / inline_expression) %ws*)*
        (")"
            / (!! eol -> 'Line ended without finding a closing )-parenthesis' !!)
            / (!! [^%nl]+ -> 'Unexpected character while parsing subexpression' !!)
        )
      )
inline_expression: index_chain / noindex_inline_expression 
indented_expression:
    indented_text / indented_nomsu / indented_list / indented_dict / ({|
        "(..)" nl_indent
            (action / expression) (nl_nodent comment)*
            (eol / (!! [^%nl]+ -> "Unexpected character while parsing indented expression" !!))
    |} -> unpack)
expression:
    inline_expression / indented_expression

inline_nomsu (EscapedNomsu): "\" inline_expression
indented_nomsu (EscapedNomsu):
    "\" (noindex_inline_expression / indented_expression)

index_chain (IndexChain):
    noindex_inline_expression ("." (text_word / noindex_inline_expression))+

-- Actions need either at least 1 word, or at least 2 tokens
inline_action (Action):
    !section_division
    (   (inline_arg (%ws* (inline_arg / word))+)
      / (word       (%ws* (inline_arg / word))*))
    (%ws* inline_block)?
inline_arg: inline_expression / inline_block
action (Action):
    !section_division
    (  (arg  ((nl_nodent "..")? %ws* (arg / word))+)
     / (word ((nl_nodent "..")? %ws* (arg / word))*))
arg: expression / inline_block / indented_block

word: !number { %operator_char+ / %ident_char+ }

text_word (Text): word

inline_text (Text):
    !('".."' eol)
    '"'
        ({~ (('\"' -> '"') / ('\\' -> '\') / %escaped_char / [^%nl\"])+ ~}
        / inline_text_interpolation)*
    ('"'
     / (!! eol -> 'Line ended before finding a closing double quotation mark' !!)
     / (!! [^%nl]+ -> 'Unexpected character while parsing Text' !!))
inline_text_interpolation:
    "\" (
        variable / inline_list / inline_dict / inline_text
        / ("("
            %ws* (inline_action / inline_expression) %ws*
            (%ws* ',' %ws* (inline_action / inline_expression) %ws*)*
        (")"
         / (!! eol -> 'Line ended without finding a closing )-parenthesis' !!)
         / (!! [^%nl]+ -> 'Unexpected character while parsing Text interpolation' !!)))
    )

indented_text (Text):
    '".."' eol %nl {%nl*} {:curr_indent: indent :}
        (indented_plain_text / text_interpolation / {~ %nl+ (=curr_indent -> "") ~})*
    (!! [^%nl]+ -> "Unexpected character while parsing Text" !!)?
indented_plain_text (Text):
    {~ (("\\" -> "\") / (("\" blank_lines =curr_indent "..") -> "") / (!text_interpolation "\") / [^%nl\]+)+
        (%nl+ (=curr_indent -> ""))* ~}
text_interpolation:
    inline_text_interpolation / ("\" indented_expression blank_lines =curr_indent "..")

number (Number): (("-"? (([0-9]+ "." [0-9]+) / ("." [0-9]+) / ([0-9]+)))-> tonumber)

-- Variables can be nameless (i.e. just %) and can only contain identifier chars.
-- This ensures you don't get weird parsings of `%x+%y` or `%'s thing`.
variable (Var): "%" {%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" !!)
    )))
indented_list (List):
    "[..]" eol nl_indent
        list_line (nl_nodent list_line)* (nl_nodent comment)*
    (","? (!! [^%nl]+ -> "Unexpected character while parsing List" !!))?
list_line:
      (inline_list_item %ws* "," %ws*)+ eol
    / (inline_list_item %ws* "," %ws*)* (action / expression) eol
inline_list_item: 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" !!)
    )))
indented_dict (Dict):
    "{..}" eol nl_indent
        dict_line (nl_nodent dict_line)* (nl_nodent comment)*
    (","? (!! [^%nl]+ -> "Unexpected character while parsing Dictionary" !!))?
dict_line:
      (inline_dict_entry %ws* "," %ws*)+ eol
    / (inline_dict_entry %ws* "," %ws*)* dict_entry eol
dict_entry(DictEntry):
    dict_key (%ws* ":" %ws* (action / expression))?
inline_dict_entry(DictEntry):
    dict_key (%ws* ":" %ws* (inline_action / inline_expression)?)?
dict_key:
    text_word / inline_expression