[lxml] Re: reconstruct a dtd content model

20 Mar 2022


      On 19/03/2022 09:43, Robin Becker wrote:
...
I wonder if anybody has code to reconstruct the content definition of a DTD element. It's needed for some automatic 
documentation process.
I inherited a recursive code, but it's clear that it's doing the wrong thing; it's fairly obvious that the parentheses 
are being done wrongly. I think that parens ought to go before the occur processing at
"if node.occur == 'plus':", but it's also obvious that they are not always required.
.......
I made some changes to the alogorithm and it seems to work better it now produces these results
...
content reconstruction test
'<!ELEMENT a (b?)>' content=(b?)
'<!ELEMENT a (a|b)+>' content=(a | b)+
'<!ELEMENT a (a|b|c)+>' content=(a | b | c)+
'<!ELEMENT a (a|b?|c)+>' content=(a | b | c)*
'<!ELEMENT a (a|(b?,c)+)+>' content=(a | (b?, c)+)+
'<!ELEMENT a (a|(b+|c)+)+>' content=(a | b+ | c+)+
'<!ELEMENT a (z)>' content=(z)
'<!ELEMENT a (#PCDATA)>' content=(#PCDATA)
'<!ELEMENT a (#PCDATA|b)*>' content=(#PCDATA | b)*
'<!ELEMENT a (a,b,c)*>' content=(a, b, c)*
'<!ELEMENT a ANY>' content=ANY
'<!ELEMENT a EMPTY>' content=EMPTY
def elementContent(node):
     return node.name

def _contentRecur(node, parentType, elFmt):
     """
     node.type: ("element" | "pcdata" | "seq" | "or" )
     node.occur: ("once" | "opt" | "plus" | "mult")
     none.name: (str | None)
     """
     s = ""
     if node is  None:
         return s
     t = node.type
     occur = node.occur
     if t == "element":
         s = f"{elFmt(node)}"
     elif t=='pcdata':
         s = '#PCDATA'
     else:
         right = node.right
         left = node.left
         if t == 'or':
             if left and right:
                 s = f"{_contentRecur(left,t,elFmt)} | {_contentRecur(right,t,elFmt)}"
                 if parentType!=t: s = f"({s})"
             elif left is not None:
                 s = _contentRecur(left,t,elFmt)
             elif right is not None:
                 s = _contentRecur(right,t,elFmt)
         elif t == 'seq':
             if left and right:
                 s = f"{_contentRecur(left,t,elFmt)}, {_contentRecur(right,t,elFmt)}"
                 if parentType!=t: s = f"({s})"
             elif left is not None:
                 s = _contentRecur(left,t,elFmt)
             elif right is not None:
                 s = _contentRecur(right,t,elFmt)
     if occur == 'plus':
         s += "+"
     elif occur == 'opt':
         s += '?'
     elif occur=='mult':
         s += '*'
     return s

def content(el):
     if el.type=='pcdata':
         return '(#PCDATA)'
     elif el.type=='empty':
         return 'EMPTY'
     elif el.type=='any':
         return 'ANY'
     s = f"{_contentRecur(el.content,None,elementContent)}"
     if s[0]!='(':
         s = f"({s})"
     return s

-- 
Robin Becker