andreas-kupries / marpa

Mirror of marpa @ core.tcl.tk/akupries -- Tcl binding for Marpa

Setup

All the various blocks of related files are collected in their own directories.

• Low-level wrapper around libmarpa, bringing the API into the script level as is. See c/.

• A parsing runtime implemented in Tcl, on top of the low-level wrapper. See runtime-tcl/.

• A parsing runtime implemented in C, on top of libmarpa. This and the other runtime share the general architecture, making problems in one a problem for the other, and fixes apply to both too. See runtime-c/.

• A parser implementing the SLIF meta grammar. See slif-parser. Generated from the SLIF meta grammar. Currently based on the C runtime. Basing it on the Tcl runtime works as well, just runs slower. See slif-parser/.

• A container to hold grammars specified via SLIF. See slif-container/.

• The semantics to translate a SLIF AST coming out of the slif-parser into a series of instructions for a slif-container which stores the grammar encoded in the AST in the container. See slif-semantics/.

• Support for generators, technically still part of the SLIF support, a package to rewrite a set of precedenced priority rules into a new set where the precedences are encoded in the structure of the new rules. See slif-precedence.tcl.

• Support for generators, technically still part of the SLIF support, a package to rewrite literals into simpler forms (normalization and reduction). Former keeps literal-ness, latter can break literals apart into sets of rules. See slif-literal.tcl.

• Configuration and dispatcher core for code generators. See gen.tcl.

• Several code generator packages accessible from the generator core. See gen-formats/. The code generators are able to generate parsers and lexers using either Tcl or C runtime, and pre-loaded containers, readable and compact.

• Several packages shared between code generators. See gen-common/.

• General utility package. See util/.

• Utility package specific to unicode functionality. See unicode/.

• Supporting code to ingest unicode data tables, and the tables we use. See tools/ and unidata/.

• The SLIF meta grammar and tools to regenerate the marpa-slif-parser/. See bootstrap/.

• A generator application using all the pieces. See bin/.

• Tests. See tests/.

• Documentation. See doc/ and doc.1/. This part is in need of cleanup and proper writeup.

• Manual benchmarks in bench/ (Not based on tclbench, tough they should be made to).

After building the marpa package a directory generated will contain the output of the tools on the unidata tables. This is a cache directory which can be blown away, the build will regenerate it as needed. The information in the files becomes part of the unicode package.