Raptor RDF Parser Library - To Do List

Bugs and Features

Bugs should be reported at the Redland Issue Tracker.

Most of these should be migrated to the issue tracker above.

Handle bad RDF URI references somehow - such as those containing dissallowed URI octets (RFC 2396 as updated by RFC 2732) - silently convert on input or noisly convert on input. For example, > in a URI in RDF/XML would get turned into \u003E in an N-Triples output.
Update the rdf/rdfs schema documents. Convert the turtle labels from unicode character form into UTF-8 (DATA, DOCS)
Ensure there is support to allow Rasqal to register a parser outside raptor that provides RDF query results as triples. Alternatively, provide a skeleton wrapper parser that allows the same by delegating all triple generation to callbacks. (FEATURE)
Add a N3 parser (FEATURE)
Add a PNG parser, like pngmeta and parse the embedded RDF/XML into triples (FEATURE)

Done

The most recent changes are at the top, with the first release version where appropriate.

Describe in libraptor.3 relevant raptor.h structures - describe important ones such as raptor_statement, raptor_locator and raptor_feature as well as reference internal ones such as raptor_identifier, raptor_identifier_type, raptor_uri_source, raptor_ntriples_term_type, raptor_genid_type and raptor_uri_handler for the raptor_uri class. (DOCS) (1.4.8)
Add namespace/prefix declaration user level callback (FEATURE) (1.4.8)
Add a guessing parser that sends an Accept: HTTP header for all supported mime types and uses the returned headers to select a parser. (FEATURE) (1.4.8)
Add an XSLT 'parser' based on libxslt (FEATURE) (1.4.6)
Escape outputing the > delimiter for N-Triples URI-references (1.4.3)
Allow the supported parsers to be selected by configure (1.3.3)
Errors which happened when fetching WWW content were always printed to stderr. They are now passed to the main error routines which allows applications to retrieve them. (1.3.2)
In lax mode, warns when unknown rdf:parseType values are seen to prevent things like 'owl:collection' and 'collection' passing through (1.3.1)
Turtle parser: a bare ':' and qnames such as 'rdf:_1' now work (1.3.1)
Describe in libraptor.3 the use of UTF-8 for strings and URIs. (1.3.1)
Send an HTTP Accept: header with WWW requests corresponding to the mime type of the parser selected, accepting all others at lower q (1.3.0)
Guess parser from a mime type, content fragment and/or content name such as a filename or URI (1.3.0)
Turtle parser: use raptor_generate_id for blank node identifiers (1.3.0)
Added --enable-xml-1-1-names to enable XML 1.1 name checking instead of XML 1.0 (1.1.0)
Updated XML 1.1 name checking for ranges in the XML/Namespaces in XML 1.1 proposed recommendations (1.1.0)
Added --disable-nfc-check to disable the NFC linking/checking with GNOME glib, even if that library is present (1.1.0)
Made the N-Triples parser use raptor_generate_id for blank nodes identifiers (1.1.0)
Updated the RDF/XML parser to handle libxml 2.6.0 SAX2 API which changes the names of all of the SAX1 calls (1.1.0).
Added an N-Triples Plus parser (1.1.0)
Correct line counting for N-Triples with \r\n line (DOS) files and when the line crosses a chunk. (1.1.0)
Handle WIN32 file URIs starting file://c:... (1.1.0)
Scanning (rapper --scan) for rdf:RDF in embedded XML does not work (0.9.13)
URI retrieval - make sure it chops off the fragment before fetching. (0.9.12)
Make 'make check' not die if NFC tests fail, possible if no GNOME glib2 is present. (0.9.12)
Unicode character normalization NFC checks not implemented (0.9.11)
XML (Exclusive) Canonicalization for XML Literals not implemented (0.9.11)
libxml2 currently does not do XML attribute normalization i.e. removing whitespace around attribute content. Added a fix (0.9.11)
Added raptor_www_no_www_library_init_finish to allow once-only www library startup/shutdown to be prevented. Default to performed so that most higher level apps do not need to know or care (0.9.10)
Docs updated for 0.9.7 to 0.9.10 API changes in the libraptor.3 manual page
Escape XML attribute values in parseType literal content generation (0.9.10)
Passing NULL to raptor_start_parse base URI for rdfxml parser crashed it. Now the RDF/XML parser fails. (0.9.9)
rdf:parseType="Literal" content with &, <, > and unicode characters did not get escaped in the encoded string into entities/character entities. (0.9.9)
XML Namespaces declared with a prefix and no namespace name (URI) were accepted - this is illegal (0.9.8)
Empty files made rdf/xml parser crash with libxml2 and expat; now both return failure since an empty doc is not allowed (0.9.8)
rdf:bagID handling added (0.9.7)
Can now configure on system which has expat alone using ./configure --with-xml-parser=expat (0.9.7)
Fix compiling libxml 2.3.5 and nearby versions failing on FreeBSD (which is 20 months old) caused by a change to the xmlSAXHandler structure (0.9.7)
file: URIs were not correctly handled (0.9.6)
Resilience to XML parser errors, RDF/XML grammar errors (0.9.6)
Manual pages (0.9.6)
Made CDATA section work with libxml (0.9.6)
daml:collection is generating some wrong triples (BUG) (0.9.6)
Compiling on OSX fails on most systems since it requires libtool 1.4.2 or CVS version which requires automake 2.50 and a newer version of autoconf. This needs significant changes to the autoconfigure system. See Fink porting libtool to OSX. The fix for now - for the packaged sources only, not CVS - is to use a patched libtool 1.4.2 (from Debian) that can generate a libtool that knows OSX. (0.9.6)
URI resolving to a base URI now working (0.9.5)
Handle <prop:Elt rdf:ID="id" rdf:resource="http://example.org/obj"/> (RDF Core WG syntax change) (0.9.5)
Add xml:base support - RDF Core WG syntax change (0.9.4)
Perform xml:lang processing and pass to application (0.9.4)
parseType literal broken (0.9.4)
Tracking of user IDs/generated IDs available to user code (0.9.3)
daml:collection parseType support (0.9.3)
rdf:li used as a propertyElt does not work (0.9.3)
parseType literal support complete (0.9.3)
Fixed many crashes (0.9.3)

Could not duplicate problem list

Turtle parser: the lexer lval->string values get overwritten on errors (BUG?)
RSS tag soup parser junks/overwrites element content if it is delivered in chunks rather than as one big CDATA, such as when libxml2 sees entities, it emits the content in bits (BUG)

Decided not to do list

Add a gzip2/bzip2 content reading interface, using libxml2 to do the hard work (FEATURE): users can do this if they need it with the chunk API, or using curl with HTTP content compression
Provide a perl interface (FEATURE): use Redland bindings
LSID URN support (FEATURE)
Record the xml parser used and make available from API (FEATURE)
Other rdf:parseType support (FEATURE)

No need to do list

After decisions from the RDF Core WG as recorded in the attention developers area of the RDF Issue Tracking document.

aboutEach support - removed from syntax
aboutEachPrefix support - removed from syntax
Special container support - not special anymore, just typed nodes