Parsing of ASN1 definitions

Latest on Hackage:0.5

This package is not currently in any snapshots. If you're interested in using it, we recommend adding it to Stackage Nightly. Doing so will make builds more reliable, and allow to host generated Haddocks.

BSD3 licensed and maintained by Dmitry Astapov
ASN.1 definitions parser for Haskell.

Should cover a pretty large subset of standards
ITU-T Rec. X.680 -- X.683 | ISO/IEC 8824-1 -- 8824-4

This is unfinished work, so beware - there be dragons!

Test code could be run like this:

TestASN1Parser LDAP.asn1

Configure with -ftest to build and install test suite (with code
coverage analysis)

Implementation status
X.680-0207: DONE, medium test coverage. Restricted character strings
are done sloppily
X.681-0207: DONE, poor test coverage. "WITH SYNTAX" is not implemented.
X.682-0207: partially done, completeness not evaluated
X.683-0207: partially done, completeness not evaluated
Tests: partially done, see comments in Testsuite.hs for status
Test code coverage:
88% expressions used (4682/5279)
60% boolean coverage (3/5)
60% guards (3/5), 2 always True
100% 'if' conditions (0/0)
100% qualifiers (0/0)
42% alternatives used (29/68)
33% local declarations used (32/96)
28% top-level declarations used (271/958)

Parser uses information about types when it is immediately available.
For example, when parsing value assignments ("var SomeType ::= val"),
parser will accept only valid values of type "SomeType". Such
type-dependency is also used when parsing "DEFAULT value" constructs
and constraints.

Other than that, parser performs no validation/semantical analysis:
* Values of enumerated/choice/sequences are not checked against
respective types.
* References to types are not followed to find out the real type
* Associates sequence types of REAL, EMBEDDED-PDV and others are not
taken into account
* Type constraints are not taken into account when parsing values

In cases when no typing context is available, some values could not be
definitively attributed to particular type. In this case, parser will
produce so-called ambiguous values:
* SomeNumber Double -- for values that could be either Integer or Real
* SomeNamedValueList ComponentValueList -- for SequenceRealValue,
SequenceValue, SequenceOfValue (named values), SetValue,
SetOfValue (named values)
* SomeValueList [Value] -- for IdentifierListBitString,
SequenceOfValue (values only), SetOfValue (values only)
* SomeIdentifiedValue Identifier -- for Integer or Enumerated
* SomeOIDLikeValue OID -- for OID or RELATIVE-OID

Even when module does not import external types, those values are not
resolved to their more definitive forms. In order to do so, one would
have to do a multi-pass parsing, or move to GLR parsing with happy or
uu-parsinglib suite (this is actually under consideration).

Implementation notes
There are 24 ASN.1 types (BuiltinType + ReferencedType in X.680
terminology), represented by the "BuiltinType" Haskell datatype.

Some of the ASN1 types have several distinct variants, which is why
"BuiltinType" has more than 24 constructors.

Parsing of the ASN1 type declaration is handled by the parser

For each type there is a single value parser, used in contexts when
type information is easily available -- for example, when parsing
right side of ValueAssignment one should use parser for values of the
type of the left-hand side (see "valueOfType").

X.68[0-4] have several places that go like this "here is non-terminal
with several alternatives. However, alternatives 1 and 4 are only
applicable when ..., and alternatives 2 and 3 - when ..., otherwise,
alternatives 5 and 6 are used". All those places have to be re-checked
with care because I am not 100% sure that I have not missed one of

Somethimes, I decided to postpone implementation of such complex
rules. Those places are marker in source code as TODO.

All major TODOs are marked with "TODO" in source code.
Depends on 3 packages:
Used by 1 package:
comments powered byDisqus