summaryrefslogtreecommitdiff
path: root/src/vyos1x_lexer.mll
diff options
context:
space:
mode:
Diffstat (limited to 'src/vyos1x_lexer.mll')
-rw-r--r--src/vyos1x_lexer.mll91
1 files changed, 45 insertions, 46 deletions
diff --git a/src/vyos1x_lexer.mll b/src/vyos1x_lexer.mll
index f5d17af..a363c65 100644
--- a/src/vyos1x_lexer.mll
+++ b/src/vyos1x_lexer.mll
@@ -43,93 +43,92 @@ when they are significant, so that the parser can use them as terminators.
The informal idea is that a newline is only significant if it follows a leaf node.
So we need rules for finding out if we are inside a leaf node or not.
-These are the formal rules. A newline is significant if and only if
-1. Preceding token is an identifier
-2. Preceding token is a quoted string
+These are the formal rules. A newline is significant if and only if either:
+1. its preceding token is an identifier
+2. OR its preceding token is a quoted string
We set the vy_inside_node flag to true when we enter a leaf node and reset it when
we reach the end of it.
*)
-let vy_inside_node = ref false
-
}
-rule token = parse
+rule token vy_inside_node = parse
| [' ' '\t' '\r']
- { token lexbuf }
+ { token vy_inside_node lexbuf }
| '\n'
- { Lexing.new_line lexbuf; if !vy_inside_node then (vy_inside_node := false; NEWLINE) else token lexbuf }
+ { Lexing.new_line lexbuf; if vy_inside_node then (false, NEWLINE) else token vy_inside_node lexbuf }
| '"'
- { vy_inside_node := true; read_double_quoted_string (Buffer.create 16) lexbuf }
+ { read_double_quoted_string true (Buffer.create 16) lexbuf }
| '''
- { vy_inside_node := true; read_single_quoted_string (Buffer.create 16) lexbuf }
+ { read_single_quoted_string true (Buffer.create 16) lexbuf }
| "/*"
- { vy_inside_node := false; read_comment (Buffer.create 16) lexbuf }
+ { read_comment false (Buffer.create 16) lexbuf }
| '{'
- { vy_inside_node := false; LEFT_BRACE }
+ { (false, LEFT_BRACE) }
| '}'
- { vy_inside_node := false; RIGHT_BRACE }
+ { (false, RIGHT_BRACE) }
| "//" [^ '\n']*
- { token lexbuf }
+ { token vy_inside_node lexbuf }
| [^ ' ' '\t' '\n' '\r' '{' '}' '"' ''' ]+ as s
- { vy_inside_node := true; IDENTIFIER s}
+ { (true, IDENTIFIER s) }
| eof
- { EOF }
+ { (vy_inside_node, EOF) }
| _ as bad_char
{ lexing_error lexbuf (Printf.sprintf "unexpected character \'%c\'" bad_char) }
-and read_double_quoted_string buf =
+and read_double_quoted_string vy_inside_node buf =
parse
- | '"' { STRING (Buffer.contents buf) }
- | '\\' '/' { Buffer.add_char buf '/'; read_double_quoted_string buf lexbuf }
- | '\\' '\\' { Buffer.add_char buf '\\'; read_double_quoted_string buf lexbuf }
- | '\\' 'b' { Buffer.add_char buf '\b'; read_double_quoted_string buf lexbuf }
- | '\\' 'f' { Buffer.add_char buf '\012'; read_double_quoted_string buf lexbuf }
- | '\\' 'n' { Buffer.add_char buf '\n'; read_double_quoted_string buf lexbuf }
- | '\\' 'r' { Buffer.add_char buf '\r'; read_double_quoted_string buf lexbuf }
- | '\\' 't' { Buffer.add_char buf '\t'; read_double_quoted_string buf lexbuf }
- | '\\' '\'' { Buffer.add_char buf '\''; read_double_quoted_string buf lexbuf }
- | '\\' '"' { Buffer.add_char buf '"'; read_double_quoted_string buf lexbuf }
- | '\n' { Lexing.new_line lexbuf; Buffer.add_char buf '\n'; read_double_quoted_string buf lexbuf }
+ | '"' { (vy_inside_node, STRING (Buffer.contents buf)) }
+ | '\\' '/' { Buffer.add_char buf '/'; read_double_quoted_string vy_inside_node buf lexbuf }
+ | '\\' '\\' { Buffer.add_char buf '\\'; read_double_quoted_string vy_inside_node buf lexbuf }
+ | '\\' 'b' { Buffer.add_char buf '\b'; read_double_quoted_string vy_inside_node buf lexbuf }
+ | '\\' 'f' { Buffer.add_char buf '\012'; read_double_quoted_string vy_inside_node buf lexbuf }
+ | '\\' 'n' { Buffer.add_char buf '\n'; read_double_quoted_string vy_inside_node buf lexbuf }
+ | '\\' 'r' { Buffer.add_char buf '\r'; read_double_quoted_string vy_inside_node buf lexbuf }
+ | '\\' 't' { Buffer.add_char buf '\t'; read_double_quoted_string vy_inside_node buf lexbuf }
+ | '\\' '\'' { Buffer.add_char buf '\''; read_double_quoted_string vy_inside_node buf lexbuf }
+ | '\\' '"' { Buffer.add_char buf '"'; read_double_quoted_string vy_inside_node buf lexbuf }
+ | '\n' { Lexing.new_line lexbuf; Buffer.add_char buf '\n'; read_double_quoted_string vy_inside_node buf lexbuf }
| [^ '"' '\\']+
{ Buffer.add_string buf (Lexing.lexeme lexbuf);
- read_double_quoted_string buf lexbuf
+ read_double_quoted_string vy_inside_node buf lexbuf
}
| eof { lexing_error lexbuf "Quoted string is missing the closing double quote" }
-and read_single_quoted_string buf =
+and read_single_quoted_string vy_inside_node buf =
parse
- | ''' { STRING (Buffer.contents buf) }
- | '\\' '/' { Buffer.add_char buf '/'; read_single_quoted_string buf lexbuf }
- | '\\' '\\' { Buffer.add_char buf '\\'; read_single_quoted_string buf lexbuf }
- | '\\' 'b' { Buffer.add_char buf '\b'; read_single_quoted_string buf lexbuf }
- | '\\' 'f' { Buffer.add_char buf '\012'; read_single_quoted_string buf lexbuf }
- | '\\' 'n' { Buffer.add_char buf '\n'; read_single_quoted_string buf lexbuf }
- | '\\' 'r' { Buffer.add_char buf '\r'; read_single_quoted_string buf lexbuf }
- | '\\' 't' { Buffer.add_char buf '\t'; read_single_quoted_string buf lexbuf }
- | '\\' '\'' { Buffer.add_char buf '\''; read_single_quoted_string buf lexbuf }
- | '\\' '"' { Buffer.add_char buf '"'; read_single_quoted_string buf lexbuf }
- | '\n' { Lexing.new_line lexbuf; Buffer.add_char buf '\n'; read_single_quoted_string buf lexbuf }
+ | ''' { (vy_inside_node, STRING (Buffer.contents buf)) }
+ | '\\' '/' { Buffer.add_char buf '/'; read_single_quoted_string vy_inside_node buf lexbuf }
+ | '\\' '\\' { Buffer.add_char buf '\\'; read_single_quoted_string vy_inside_node buf lexbuf }
+ | '\\' 'b' { Buffer.add_char buf '\b'; read_single_quoted_string vy_inside_node buf lexbuf }
+ | '\\' 'f' { Buffer.add_char buf '\012'; read_single_quoted_string vy_inside_node buf lexbuf }
+ | '\\' 'n' { Buffer.add_char buf '\n'; read_single_quoted_string vy_inside_node buf lexbuf }
+ | '\\' 'r' { Buffer.add_char buf '\r'; read_single_quoted_string vy_inside_node buf lexbuf }
+ | '\\' 't' { Buffer.add_char buf '\t'; read_single_quoted_string vy_inside_node buf lexbuf }
+ | '\\' '\'' { Buffer.add_char buf '\''; read_single_quoted_string vy_inside_node buf lexbuf }
+ | '\\' '"' { Buffer.add_char buf '"'; read_single_quoted_string vy_inside_node buf lexbuf }
+ | '\n' { Lexing.new_line lexbuf; Buffer.add_char buf '\n'; read_single_quoted_string vy_inside_node buf lexbuf }
| [^ ''' '\\']+
{ Buffer.add_string buf (Lexing.lexeme lexbuf);
- read_single_quoted_string buf lexbuf
+ read_single_quoted_string vy_inside_node buf lexbuf
}
| eof { lexing_error lexbuf "Quoted string is missing the closing single quote" }
-and read_comment buf =
+and read_comment vy_inside_node buf =
parse
| "*/"
- { COMMENT (Buffer.contents buf) }
+ { (vy_inside_node, COMMENT (Buffer.contents buf)) }
| _
{ Buffer.add_string buf (Lexing.lexeme lexbuf);
- read_comment buf lexbuf
+ read_comment vy_inside_node buf lexbuf
}
(*
-If you are curious how the original parsers handled the issue: they did not.
+If you are curious how the original parsers from Vyatta handled the issue: they did not.
+
The CStore parser cheated by reading data from command definitions to resolve
the ambiguities, which made it impossible to use in standalone config
manipulation programs like migration scripts.