aboutsummaryrefslogtreecommitdiffstats
AgeCommit message (Collapse)AuthorFilesLines
2023-12-02Be precise about the meaning of node "length" and "size"David Robillard4-39/+36
2023-12-02Align nodes on the reader stackDavid Robillard3-14/+13
2023-12-02Replace serd_reader_set_strict() with SerdReaderFlagsDavid Robillard6-33/+26
2023-12-02Handle read errors more preciselyDavid Robillard3-20/+39
2023-12-02Simplify reader stack pushing codeDavid Robillard1-6/+6
2023-12-02Split out simple reader unit testsDavid Robillard3-317/+404
2023-12-02Split up serd_internal.hDavid Robillard9-15/+19
2023-12-02Reduce complexity of blank node reading functionsDavid Robillard1-18/+28
2023-12-02Reduce complexity of string writing functionsDavid Robillard1-77/+97
2023-12-02Reduce complexity of URI node writing functionsDavid Robillard1-21/+38
2023-12-02Reduce complexity of statement writing functionsDavid Robillard2-79/+175
2023-12-02Add empty syntax type for suppressing outputDavid Robillard8-21/+107
2023-12-02Factor out test runner script argument parsingDavid Robillard4-46/+41
2023-12-02Use a simple type-safe stack in writerDavid Robillard7-51/+94
2023-12-02Add a set of limits to the worldDavid Robillard8-15/+74
The idea here is to remove the burden of passing things around like stack sizes (where most users don't care and will be happy with a reasonably large default) and keeping the call sites to things like serd_reader_new() clean. The cost is a bit more state, so it's both more powerful and more potentially flaky, since changing the limits has action at a distance that isn't clear from the call site. I think it's worth it for the cleaner API in the common case, and the much better forward compatibility.
2023-12-02Replace SERD_WRITE_STRICT flag with SERD_WRITE_LAXDavid Robillard5-17/+16
The unset value for flags should represent the best default, which in this case is strict parsing. Lax parsing is the riskier opt-in option.
2023-12-02Add support for writing terse output with minimal newlinesDavid Robillard17-17/+225
2023-12-02Pass context flags to write_sep()David Robillard1-17/+20
2023-12-02Add initial flags to write contextDavid Robillard1-12/+17
2023-12-02Simplify statement flagsDavid Robillard6-49/+47
2023-12-02Replace multiple stream callbacks with SerdEventDavid Robillard17-233/+353
This makes plumbing easier since everything goes through the same "stream" and only one callback is required to handling everything. It's also more easily extensible in case more event types need to be added in the future.
2023-12-02Simplify writer style options and write UTF-8 by defaultDavid Robillard9-75/+40
2023-12-02Add SerdStatementDavid Robillard14-54/+463
2023-12-02Add SerdCaretDavid Robillard18-64/+310
2023-12-02Add SerdField to public APIDavid Robillard2-41/+41
A statement field enum was previously only used by the writer internally, but will be used in the public API (without a sentinel value) in following commits.
2023-12-02Add serd_world_get_blank()David Robillard5-2/+86
2023-12-02Remove serd_reader_set_default_graph()David Robillard4-49/+14
2023-12-02Use thread-safe strerror_r() if availableDavid Robillard6-7/+45
2023-12-02Move fopen wrapper to worldDavid Robillard5-27/+34
2023-12-02Move error handling to worldDavid Robillard5-30/+69
2023-12-02Add SerdWorld for shared library stateDavid Robillard21-112/+219
2023-12-02Add serd_node_compare()David Robillard4-0/+106
2023-12-02Clean up base64 node construction and access APIDavid Robillard6-20/+162
2023-12-02Clean up numeric node construction and access APIDavid Robillard9-93/+450
2023-12-02[WIP] Use exess for reading and writing numeric and binary literalsDavid Robillard14-359/+219
2023-12-02Clean up nodes in reader/writer testsDavid Robillard1-12/+20
2023-12-02Move syntax name/extension utilities to public APIDavid Robillard6-55/+158
2023-12-02Improve reader error handlingDavid Robillard5-137/+320
2023-12-02Set flags directly when reading literal nodesDavid Robillard1-64/+43
2023-12-02Add debug checks for node paddingDavid Robillard2-13/+37
2023-12-02Zero node padding before passing to reader sinksDavid Robillard5-2/+41
2023-12-02Remove datatype and language from reader contextDavid Robillard2-40/+32
2023-12-02Simplify stack management by popping in bulk at higher levelsDavid Robillard4-139/+55
Since all memory used by the reader is POD in the stack, there is no benefit to forcing code to explicitly pop everything pushed to the stack, since any function can record an offset and pop back down to it regardless of what its callers pushed if it knows that it does not need those items. This is slightly more efficient (due to avoiding many pop calls), but also more resilient since "leaks" at deeper levels of recursion get nuked by some caller regardless of what was pushed. This should help prevent future regressions like f6437f606 (Fix memory consumption when reading documents).
2023-12-02Use a fixed-size reader stackDavid Robillard13-208/+301
2023-12-02Simplify reader interfaceDavid Robillard7-266/+179
2023-12-02Add SerdSink interface and hide implementationsDavid Robillard17-196/+480
2023-12-02Bring read/write interface closer to the C standardDavid Robillard16-105/+130
2023-12-02Add serd_new_boolean()David Robillard3-0/+35
2023-12-02Simplify literal construction APIDavid Robillard5-70/+123
2023-12-02Set datatypes on integer, decimal, and base64 nodesDavid Robillard3-41/+111