Frequently used DTD/XML schema expressions ?

[Razvan]

Hi !




Where can I find the definition of the most used DTD/Schema
expressions ? Something like:


<!-- exclusive or: AAA or BBB but not both -->
<!ELEMENT TestXOR (AAA | BBB)>

<!-- or: AAA or BBB (could be one, both (any order) or none) -->
<!ELEMENT TestOR1 (AAA | BBB | (AAA, BBB) | (BBB, AAA))?>
<!ELEMENT TestOR2 ((AAA?, BBB?) | (BBB, AAA))>

<!-- and: AAA and BBB (in this order) -->
<!ELEMENT TestAND1 (AAA, BBB)>

<!-- and: AAA and BBB (in this order) or none -->
<!ELEMENT TestAND2 (AAA, BBB)?>



Thanks,
Razvan

 

[Stanimir Stamenkov]

/Razvan/:

Where can I find the definition of the most used DTD/Schema
expressions ? Something like:

<!-- exclusive or: AAA or BBB but not both -->
<!ELEMENT TestXOR (AAA | BBB)>

[...]

You find it in the XML specification <http://www.w3.org/TR/REC-xml/>.

"Element Type Declarations" <http://www.w3.org/TR/REC-xml/#elemdecls>

 

[Razvan]

Hi



You did not understand me. I am looking for a web page where
I can find all sort of expression written according to DTD rules or XML
Schema rules, like:


(a or b or c) xor d
(a and b) or c
(a or b) xor c

a.s.o

In the official docs I can only find a few random examples.
Practically I am searching for a DTD/XML schema repository. I
hope that I am more clear this time.



Regards,
Razvan

 

[Stanimir Stamenkov]

/Razvan/:

You did not understand me. I am looking for a web page where
I can find all sort of expression written according to DTD rules or XML
Schema rules, like: [...]
In the official docs I can only find a few random examples.

The official document specifies all the possibilities, but seems
you're too lazy to read. How a specification could not specify
something - it won't be specified and it won't be recognized, then.
Just follow the links in the BNF notations:

Element Type Declaration

[45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>'
[46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children

Element-content Models

[47] children ::= (choice | seq) ('?' | '*' | '+')?
[48] cp ::= (Name | choice | seq) ('?' | '*' | '+')?
[49] choice ::= '(' S? cp ( S? '|' S? cp )+ S? ')'
[50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'

etc., etc. Here are some more:

"Attribute-List Declarations" <http://www.w3.org/TR/REC-xml/#attdecls>

Read the whole document. For the XML Schema language:

http://www.w3.org/XML/Schema#dev


/Razvan/:

Practically I am searching for a DTD/XML schema repository. I
hope that I am more clear this time.

Can't help you here.

 

[Razvan]

You are too laizy to read. I have already read the damn document before
you pinpointed it to me. I am looking for best practices for a lot of
expresions that must be very frequently used, like:

<!-- or: AAA or BBB (could be one, both (any order) or none) -->
<!ELEMENT TestOR1 (AAA | BBB | (AAA, BBB) | (BBB, AAA))?>
<!ELEMENT TestOR2 ((AAA?, BBB?) | (BBB, AAA))>

As you can see there are 2 ways to do the same thing.

The best way to write DTDs from scratch is to have a set of "bricks" to
help you build it faster and in a proper way. It is a waste of time to
think each time how to implement the above relation => it is useful to
write a list of frequently-used expressions.

About the "lazy" stuff: use a civilized language. I don't take crap
from you or anybody else.



Razvan

 

[Richard Tobin]

You did not understand me. I am looking for a web page where
I can find all sort of expression written according to DTD rules or XML
Schema rules, like:

I don't think you're likely to find this, because it's supposed to be
obvious. I'm sure are there are FAQs with examples of some well known
tricky cases, such as N elements in any order.

-- Richard

 

[Razvan]

Hi


Can you pinpoint me to such FAQ ? Yesterday I spent half a day
trying to find something like this but I gave up. The amount of
information is tremendous. I probably need to spend one full week just
to parse the most important sites.



Thanks,
Razvan

 

[Stanimir Stamenkov]

/Razvan/:

You are too laizy to read. I have already read the damn document before
you pinpointed it to me. I am looking for best practices for a lot of
expresions that must be very frequently used, like:

Hm. May be I haven't understood you completely, still I think
you're... whatever you are.

<!-- or: AAA or BBB (could be one, both (any order) or none) -->
<!ELEMENT TestOR1 (AAA | BBB | (AAA, BBB) | (BBB, AAA))?>
<!ELEMENT TestOR2 ((AAA?, BBB?) | (BBB, AAA))>

As you can see there are 2 ways to do the same thing.

In the XML specification there's a section "Deterministic Content

The best way to write DTDs from scratch is to have a set of "bricks" to
help you build it faster and in a proper way. It is a waste of time to
think each time how to implement the above relation => it is useful to
write a list of frequently-used expressions.

I write a DTD according to what an XML structure must look like. As
to best practices in representing data as XML, I can't give you
concrete resources, but I'm sure there are many articles regarding
different situations - "Elements vs. attributes", etc.

About the "lazy" stuff: use a civilized language. I don't take crap
from you or anybody else.

I don't take crap from you, too. Use more suitable explanation next
time... "web page where I can find all sort of expression written
according to DTD rules..." - that's what you get from the spec.

 

[Razvan]

Hi


Since you mention it: how do you write N elements in any order
? These are my solutions for 2 elements: (DTD)

<!ELEMENT TestOR1 (AAA | BBB | (AAA, BBB) | (BBB, AAA))?>
<!ELEMENT TestOR2 ((AAA?, BBB?) | (BBB, AAA))>




Thanks,
Razvan

 

[David Carlisle]

Since you mention it: how do you write N elements in any order
? These are my solutions for 2 elements: (DTD)

<!ELEMENT TestOR1 (AAA | BBB | (AAA, BBB) | (BBB, AAA))?>
<!ELEMENT TestOR2 ((AAA?, BBB?) | (BBB, AAA))>

TestOR1 isn't allowed in XML as it fails this test:
http://www.w3.org/TR/xml11/#determinism

When you come to an AAA element

<!ELEMENT TestOR1 (AAA | BBB | (AAA, BBB) | (BBB, AAA))?>
^^^^ ^^^

you can't tell which of the two indicated AAA you have without looking
ahead to see if there is a BBB and such lookahead isn't allowed.

Similarly your TestOR2 fails as it's OK for AAA but given a BBB
you can't tell whether it matches (AAA?, BBB?) or (BBB, AAA)

I think to allow any of
empty
AAA
BBB
AAA,BBB
BBB,AAA

you need something like:

(AAA,BBB?)|(BBB,AAA?)?

David

 

[Razvan]

That means XML SPY is broken ? It allowed without any problems the
above expressions (*both* of them). What if your DTD contain some
undeterministic expressions - like the above mentioned ones. What are
the risks ? Some validating parsers will accept them while others might
not ?



Regards,
Razvan

 

[David Carlisle]

checking for non determinism is classified as "an error" that means that
http://www.w3.org/TR/xml11/#dt-error
Conforming software MAY detect and report an error and MAY recover from it.]

so it means strictly speaking XML Spy is conforming not to report this
error, but your file is still in error.


I see Richard has already commented on this thread, personally I always
check things with his excellent rxp parser which says given:

<!DOCTYPE TestOR1 [
<!ELEMENT TestOR1 (AAA | BBB | (AAA, BBB) | (BBB, AAA))?>
<!ELEMENT TestOR2 ((AAA?, BBB?) | (BBB, AAA))>
]>
<TestOR1/>



$ rxp -sxV d.xml
Warning: Content model for TestOR1 is not deterministic.
At start of content there are multiple choices when the next element is AAA.
(detected at end of prolog of document file:///e:/tmp/d.xml)
Warning: Content model for TestOR2 is not deterministic.
At start of content there are multiple choices when the next element is BBB.
(detected at end of prolog of document file:///e:/tmp/d.xml)


Note this error is only optionally reported when using a dtd. The
equivalent thing in W3C schema is called the unique particle constraint
http://www.w3.org/TR/xmlschema-1/#cos-nonambig
and as far as I can see, a conforming schema validator has to report
this.

One way to avoid all this nonsense is to use a different schema language
such as Relax NG.

David

 

[Razvan]

Thanks a lot for the tip with the rxp parser. I just install it. I
really needed a parser to check for valid DTDs / XML Schemas. I am new
to this XML stuff and without your mention I would have started on the
wrong path.

One way to avoid all this nonsense is to use a different schema language
such as Relax NG.

This is not possible. I am preparing to take IBM 141 certification
exam - "XML and related technologies". For the moment I will
concentrate on the exam's objectives. For whatever reason Relax NG is
not on their objective list. Later, I might look at alternative
technologies.



Regards,
Razvan