nomsu/nomsu.peg

-- Nomsu version 2
file (File):
    {:curr_indent: '' :}
    blank_line*
    (chunk (nl_nodent chunk_delimeter nl_nodent chunk)*)?
    blank_line*
    (!. / (!! .* -> "Parse error" !!))

nodent: =curr_indent !(" ")
indent: =curr_indent "    " !(" ")
dedent: !(=curr_indent) (" ")*
eol: %ws* (!. / &%nl)

comment (Comment):
    "#" {~ [^%nl]* (%nl+ ({:curr_indent: indent :} -> '') [^%nl]* (%nl+ (=curr_indent -> '') [^%nl]*)* (!. / nl_dedent))? ~}
inline_comment (Comment):
    "(#" {~ (inline_comment / [^%nl])* ~} "#)"

blank_line: %nl eol
nl_nodent: blank_line* %nl nodent
nl_indent: blank_line* %nl {:curr_indent: indent :}
nl_dedent: blank_line* %nl &dedent

chunk: !chunk_delimeter (block/action/expression)
chunk_delimeter: ("~")^+3

inline_block (Block):
    (inline_comment / inline_statement) (%ws* ";" %ws* (inline_comment / inline_statement))+
block (Block):
    block_line (nl_nodent block_line)+
block_line:
    comment / inline_comment / statement / (!! [^%nl]* -> "Unexpected character while parsing block line" !!)

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_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' !!)
        )
      )
inline_expression: index_chain / noindex_inline_expression 
indented_expression:
    indented_text / indented_nomsu / indented_list / indented_dict
    / (("(..)" / ":")? nl_indent
        (block / action / expression)
        (!. / &nl_dedent / (!! (!nl_dedent .)* -> "Unexpected character while parsing indented expression" !!))
      )
expression:
    inline_expression
    / (":" %ws* ((inline_block / inline_action / inline_expression) eol
        / (!! eol -> "Missing expression after the ':'" !!)))
    / indented_expression

inline_nomsu (EscapedNomsu): "\" inline_expression
indented_nomsu (EscapedNomsu):
    "\" (
        noindex_inline_expression
        / (":" %ws* ((inline_block / inline_action / inline_expression) eol
             / (!! eol -> "Missing expression after the ':'" !!)))
        / 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):
    (   (inline_expression (%ws* (inline_expression / word))+)
      / (word              (%ws* (inline_expression / word))*))
    (%ws* ":" %ws* (inline_block / inline_action / inline_expression
        / (!! '' -> "Missing expression after the ':'" !!)))?
action (Action):
       (expression ((nl_nodent "..")? %ws* (expression / word))+)
     / (word       ((nl_nodent "..")? %ws* (expression / word))*)

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_block / inline_action / inline_expression) %ws*
            (%ws* ',' %ws* (inline_block / inline_action / inline_expression) %ws*)*
        (")"
         / (!! &%nl -> 'Line ended without finding a closing )-parenthesis' !!)
         / (!! [^%nl]* -> 'Unexpected character while parsing Text interpolation' !!)))
    )

-- Have to use "%indent" instead of "indent" etc. to avoid messing up text lines that start with "#"
indented_text (Text):
    '".."' eol %nl
        {~ (%nl*) ({:curr_indent: indent :} -> "") ~}
        (indented_plain_text / text_interpolation / {~ %nl+ (nodent -> "") ~})*
    (!. / &nl_dedent / (!! (!nl_dedent .)* -> "Unexpected character while parsing Text" !!))
indented_plain_text (Text):
    {~ (("\\" -> "\") / (("\" nl_nodent "..") -> "") / (!text_interpolation "\") / [^%nl\]+)+
        (%nl+ (nodent -> ""))* ~}
text_interpolation:
    inline_text_interpolation / ("\" indented_expression nl_nodent "..")

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

-- 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+)*)?}

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):
    "[..]" nl_indent
        list_line (nl_nodent list_line)*
    (&nl_dedent / (","? (!! (!nl_dedent .)* -> "Unexpected character while parsing List" !!)))
list_line:
      (inline_list_item %ws* "," %ws*)+ eol
    / (inline_list_item %ws* "," %ws*)* (action / expression)
inline_list_item: inline_block / inline_action / inline_expression

inline_dict (Dict):
    !('{..}')
    "{" %ws*
        (inline_dict_entry (%ws* ',' %ws* inline_dict_entry)*)? %ws*
    ("}" / (","? (
        (!! %ws* eol -> "Line ended before finding a closing }-brace" !!)
      / (!! [^%nl]* -> "Unexpected character while parsing Dictionary" !!)
    )))
indented_dict (Dict):
    "{..}" nl_indent
        dict_line (nl_nodent dict_line)*
    (&nl_dedent / (","? (!! (!nl_dedent .)* -> "Unexpected character while parsing Dictionary" !!)))
dict_line:
      (inline_dict_entry %ws* "," %ws*)+ eol
    / (inline_dict_entry %ws* "," %ws*)* dict_entry
dict_entry(DictEntry):
    dict_key (%ws* ":" %ws* (action / expression))?
inline_dict_entry(DictEntry):
    dict_key (%ws* ":" %ws* (inline_block / inline_action / inline_expression)?)?
dict_key:
    text_word / inline_expression