Upgraded nomsu.2.peg and fixed minor bug in Source serialization in

trees.

nomsu.2.peg

@@ -3,69 +3,94 @@ file:
     {:curr_indent: ' '* :}
     (((action / expression / inline_block / indented_block) eol !.)
      / file_chunks / empty_block)
-    %ws* (!! .+ -> "Parse error" !!)?
+    {:curr_indent: %nil :}
+    !.
 
-shebang: "#!" (!"nomsu" [^%nl])* "nomsu" %ws+ "-V" %ws* {:version: [0-9.]+ :} [^%nl]*
+shebang: "#!" (!"nomsu" [^%nl])* "nomsu" ws+ "-V" ws* {:version: [0-9.]+ :} [^%nl]*
 
 file_chunks (FileChunks):
     {:curr_indent: ' '* :}
     shebang? comment? blank_lines?
     (top_block (nl_nodent section_division top_block)*)
     blank_lines?
+    ws* unexpected_chunk?
+    {:curr_indent: %nil :}
 
 top_block (Block):
     {:curr_indent: ' '* :}
     comment? blank_lines? statement (nl_nodent statement)*
+    {:curr_indent: %nil :}
 
 empty_block (Block):
     {:curr_indent: ' '* :}
     comment? blank_lines?
+    {:curr_indent: %nil :}
 
-nodent: =curr_indent !(" ")
+nodent: (unexpected_indent [^%nl]* / =curr_indent)
 indent: =curr_indent "    "
-blank_lines: %nl ((nodent comment / %ws*) %nl)*
+blank_lines: %nl ((nodent comment / ws*) %nl)*
-eol: %ws* eol_comment? (!. / &%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:
+comment (Comment):
-    "#" (({} {~ [^%nl]* (%nl+ (indent -> '') [^%nl]*)* ~} %userdata) => add_comment)
+    "#" {~ [^%nl]* (%nl+ (indent -> '') [^%nl]*)* ~}
-eol_comment:
+eol_comment (Comment):
-    "#" (({} {[^%nl]*} %userdata) => add_comment)
+    "#" {[^%nl]*}
+
+unexpected_code: ws* _unexpected_code
+_unexpected_code (Error):
+    {:error: {~ [^%nl]+ -> "Couldn't parse this code" ~} :}
+unexpected_chunk (Error):
+    {:error: {~ .+ -> "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' ~} :}
 
 section_division: ("~")^+3 eol
 
 inline_block:
-    "(" %ws* inline_block  %ws* ")" / raw_inline_block
+    "(" ws* inline_block  ws* ")" / raw_inline_block
 raw_inline_block (Block):
-    ":" %ws* ((inline_statement (%ws* ";" %ws* inline_statement)*) / !(eol nl_indent))
+    ":" ws* ((inline_statement (ws* ";" ws* inline_statement)*) / !(eol nl_indent))
 indented_block (Block):
-    ":" eol nl_indent statement (nl_nodent statement)* (%nl (%ws* %nl)* nodent comment)*
+    ":" eol nl_indent statement (nl_nodent statement)*
+    (%nl (ws* %nl)* nodent (comment / eol / unexpected_code))*
+    {:curr_indent: %nil :}
 
 statement:
-    (action / expression) (eol / (!! [^%nl]+ -> "Unexpected code while parsing line" !!))
+    (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_action / inline_expression) %ws*
+            ws* (inline_action / inline_expression) ws*
-            (%ws* ',' %ws* (inline_action / inline_expression) %ws*)*
+            (ws* ',' ws* (inline_action / inline_expression) ws*)*
-        (")"
+        (")" / missing_paren_err / unexpected_code)
-            / (!! eol -> 'Line ended without finding a closing )-parenthesis' !!)
-            / (!! [^%nl]+ -> 'Unexpected code 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)*
+            (action / expression) (eol / unexpected_code)
-            (eol / (!! [^%nl]+ -> "Unexpected code while parsing indented expression" !!))
+            (%nl (ws* %nl)* nodent (comment / eol / unexpected_code))*
+        {:curr_indent: %nil :}
     |} -> unpack)
-    / (nl_indent (!! [^%nl]* -> "Unexpected indentation. Perhaps you meant to put a ':' or '(..)' on the previous line?" !!) (nl_nodent [^%nl]*)*)
 expression:
     inline_expression / indented_expression
 
@@ -79,43 +104,40 @@ index_chain (IndexChain):
 -- Actions need either at least 1 word, or at least 2 tokens
 inline_action (Action):
     !section_division
-    (   (inline_arg (%ws* (inline_arg / word))+)
+    (   (inline_arg (ws* (inline_arg / word))+)
-      / (word       (%ws* (inline_arg / word))*))
+      / (word       (ws* (inline_arg / word))*))
-    (%ws* inline_block)?
+    (ws* inline_block)?
 inline_arg: inline_expression / inline_block
 action (Action):
     !section_division
-    (  (arg  ((nl_nodent "..")? %ws* (arg / word))+)
+    (  (arg  ((nl_nodent "..")? ws* (arg / word))+)
-     / (word ((nl_nodent "..")? %ws* (arg / word))*))
+     / (word ((nl_nodent "..")? ws* (arg / word))*))
 arg: expression / inline_block / indented_block
 
-word: !number { %operator_char+ / %ident_char+ }
+word: !number { operator_char+ / ident_char+ }
 
 text_word (Text): word
 
 inline_text (Text):
-    !('".."' eol)
+    !(indented_text)
-    '"'
+    ('"' _inline_text* ('"' / missing_quote_err / unexpected_code))
-        ({~ (('\"' -> '"') / ('\\' -> '\') / %escaped_char / [^%nl\"])+ ~}
+_inline_text:
-        / inline_text_interpolation)*
+    {~ (('\"' -> '"') / ('\\' -> '\') / escaped_char / [^%nl\"]+)+ ~}
-    ('"'
+    / inline_text_interpolation
-     / (!! eol -> 'Line ended before finding a closing double quotation mark' !!)
-     / (!! [^%nl]+ -> 'Unexpected code while parsing Text' !!))
 inline_text_interpolation:
     "\" (
         variable / inline_list / inline_dict / inline_text
         / ("("
-            %ws* (inline_action / inline_expression) %ws*
+            ws* (inline_action / inline_expression) ws*
-            (%ws* ',' %ws* (inline_action / inline_expression) %ws*)*
+            (ws* ',' ws* (inline_action / inline_expression) ws*)*
-        (")"
+        (")" / missing_paren_err / unexpected_code))
-         / (!! eol -> 'Line ended without finding a closing )-parenthesis' !!)
-         / (!! [^%nl]+ -> 'Unexpected code while parsing Text interpolation' !!)))
     )
 
 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 code while parsing Text" !!)?
+    unexpected_code?
+    {:curr_indent: %nil :}
 -- Tracking text-lines-within-indented-text as separate objects allows for better debugging line info
 indented_plain_text (Text):
     {~ (("\\" -> "\") / (("\" blank_lines =curr_indent "..") -> "") / (!text_interpolation "\") / [^%nl\]+)+
@@ -127,43 +149,51 @@ number (Number): (("-"? (([0-9]+ "." [0-9]+) / ("." [0-9]+) / "0x" [0-9a-fA-F]+
 
 -- 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*}
+variable (Var): "%" {ident_char*}
 
 inline_list (List):
     !('[..]')
-    "[" %ws*
+    "[" ws*
-        (inline_list_item (%ws* ',' %ws* inline_list_item)* (%ws* ',')?)? %ws*
+        (inline_list_item (ws* ',' ws* inline_list_item)* (ws* ',')?)? ws*
-    ("]" / (","? (
+    ("]" / (","? (missing_bracket_error / unexpected_code)))
-        (!! eol -> "Line ended before finding a closing ]-bracket" !!)
-       /(!! [^%nl]+ -> "Unexpected code while parsing List" !!)
-    )))
 indented_list (List):
     "[..]" eol nl_indent
-        list_line (nl_nodent list_line)* (nl_nodent comment)*
+        list_line (nl_nodent list_line)*
-    (","? (!! [^%nl]+ -> "Unexpected code while parsing List" !!))?
+        (%nl (ws* %nl)* nodent (comment / eol / unexpected_code))*
+    (","? unexpected_code)?
 list_line:
-      (inline_list_item %ws* "," %ws*)+ eol
+      (inline_list_item ws* "," ws*)+ eol
-    / (inline_list_item %ws* "," %ws*)* (action / expression) eol
+    / (inline_list_item ws* "," ws*)* (action / expression) eol
 inline_list_item: inline_action / inline_expression
 
 inline_dict (Dict):
     !('{..}')
-    "{" %ws*
+    "{" ws*
-        (inline_dict_entry (%ws* ',' %ws* inline_dict_entry)*)? %ws*
+        (inline_dict_entry (ws* ',' ws* inline_dict_entry)*)? ws*
-    ("}" / (","? (
+    ("}" / (","? (missing_brace_error / unexpected_code)))
-        (!! eol -> "Line ended before finding a closing }-brace" !!)
-      / (!! [^%nl]* -> "Unexpected code while parsing Dictionary" !!)
-    )))
 indented_dict (Dict):
     "{..}" eol nl_indent
-        dict_line (nl_nodent dict_line)* (nl_nodent comment)*
+        dict_line (nl_nodent dict_line)*
-    (","? (!! [^%nl]+ -> "Unexpected code while parsing Dictionary" !!))?
+        (%nl (ws* %nl)* nodent (comment / eol / unexpected_code))*
+    (","? unexpected_code)?
 dict_line:
-      (inline_dict_entry %ws* "," %ws*)+ eol
+      (inline_dict_entry ws* "," ws*)+ eol
-    / (inline_dict_entry %ws* "," %ws*)* dict_entry 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_action / inline_expression)?)?
+    dict_key (ws* ":" ws* (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)
+    )

nomsu_compiler.lua

@@ -100,7 +100,7 @@ local max_parser_version = 0
 for version = 1, 999 do
   local _continue_0 = false
   repeat
-    if not (version == 4 or version == 3) then
+    if not (version == 4 or version == 3 or version == 2) then
       _continue_0 = true
       break
     end

nomsu_compiler.moon

@@ -56,7 +56,7 @@ make_tree = (tree, userdata)->
 Parsers = {}
 max_parser_version = 0
 for version=1,999
-    continue unless version == 4 or version == 3 -- TODO: remove
+    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("[^;]+")

syntax_tree.lua

@@ -43,12 +43,12 @@ for _index_0 = 1, #types do
     end
     cls.__tostring = function(self)
       return tostring(self.type) .. tostring(repr(self, (function(x)
-        return Source:is_instance(x) and tostring(x) or nil
+        return Source:is_instance(x) and repr(tostring(x)) or nil
       end)))
     end
     cls.__repr = function(self)
       return tostring(self.type) .. tostring(repr(self, (function(x)
-        return Source:is_instance(x) and tostring(x) or nil
+        return Source:is_instance(x) and repr(tostring(x)) or nil
       end)))
     end
     cls.source_code_for_tree = { }

syntax_tree.moon

@@ -19,8 +19,8 @@ for name in *types
         .__name = name
         .type = name
         .is_instance = (x)=> getmetatable(x) == @
-        .__tostring = => "#{@type}#{repr @, ((x)-> Source\is_instance(x) and tostring(x) or nil)}"
+        .__tostring = => "#{@type}#{repr @, ((x)-> Source\is_instance(x) and repr(tostring(x)) or nil)}"
-        .__repr = => "#{@type}#{repr @, ((x)-> Source\is_instance(x) and tostring(x) or nil)}"
+        .__repr = => "#{@type}#{repr @, ((x)-> Source\is_instance(x) and repr(tostring(x)) or nil)}"
         .source_code_for_tree = {}
         .get_source_code = => @source_code_for_tree[@]
         .map = (fn)=>