BNF (Backus-Naur Form) of basicDTD:
dtddocument ::= declaration {declaration} .
declaration ::= attrdecl | elemdecl .
elemdecl ::= '<!ELEMENT' name ('EMPTY' | 'ANY' | '(#PCDATA)' | elemchild) '>' .
elemchild ::= '(' (choice | seq)['?' | '*' | '+'] ')' .
choice ::= '(' cp ['|' cp] ')' .
seq ::= '(' cp {',' cp} ')' .
cp ::= (name | choice | seq) ['?' | '*' | '+'] .
attrdecl ::= '<!ATTLIST' name {name attrtype defaultdecl} '>' .
attrtype ::= 'CDATA' | 'NMTOKEN' | 'IDREF' | '(' word ['|' word] ')' . defaultdecl ::= '#REQUIRED' | '#IMPLIED' | (['#FIXED'] '"' word {word} '"' ) .
name ::= (letter | '_' | ':') {namechar} .
namechar ::= letter | digit | '.' | '-' | '_' | ':' .
letter ::='A'|..|'Z'|'a'|..|'z'.
number ::= digit {digit} .
digit ::= '0' | .. | '9' .
word ::= char {char} .
char ::= letter | digit | '%' | '&' | '^' .
BNF grammar transformed into LL(1) grammar (rules form):
DTDOC -> '<' DECLARATION L
L -> DTDOC | ε
DECLARATION -> ATTLIST WORD Z '>' | ELEMENT WORD X '>'
X -> EMPTY| ANY | PCDATA | '(' F ')'
F -> '(' CP K Y
Y -> '?' | '*' | '+' | ε
H -> ',' CP H | ε
K -> ')' | '|' CP ')' | ',' CP H ')'
CP -> WORD Y | '(' CP K Y
Z -> WORD ATTRTYPE DEFAULTDECL Z | ε
ATTRTYPE -> CDATA | NMTOKEN | IDREF | '(' WORD E ')'
E -> '|' WORD | ε
DEFAULTDECL -> REQUIRED | IMPLIED | J '"' WORD B '"'
J -> FIXED | ε
B -> WORD B | ε
Note that the grammar (in BNF) had to be slightly modified in order to transform it into a LL(1) grammar (e.g., merging rules name and word together).
##Usage
- Single sentence:
python parser.py '<!ELEMENT integer ((bool,string,float)*)>' - Multiple sentences:
python parser.py '<!ELEMENT integer ((bool,string,float)*)> <!ELEMENT integer ((bool,string,float)*)>' - Help:
python parser.py -h