A simple set of scripts and tools to work with a clang-style
compile_commands.json, re-run the compiler to extract header tree
information, and make decisions based on that information.
To generate *.hdep files alongside the working directories listed in
compile_commands.json:
generate_depfiles.py some/path/to/compile_commands.json
This will read the database, and re-run the compiler for every file with the
options -H (generate header information) and -E (preprocessor only). Then,
process these files with:
read_dep_files.py some/path/to/compile_commands.json
And a file depdata.yaml will be created in the current directory. See the
command --help for more options including restricting to header subsets.
To process a dumped file (in yaml or any other format) load it into a
dictionary and pass it to DepParser.from_dict:
# Given an instance of DepParser, parser
dict_data = parser.asdict()
from deptools import DepParser
newParser = DepParser.fromdict(dict_data)
By default, the read_dep_files.py tool writes a yaml-converted version
of the dictionary.
The class deptools.SourceFile represents files passed directly to the compiler;
they have the properties .name and .object for the source file and output
object file names, and a collection of .includes, an entry for each header file.
The class deptools.Header represents any file that is included by another file.
It has a .name property, an .includes collection (of other header files that it
depends on) and an .included collection mapping back to files (headers or source)
that include it.
After loading a deptools.DepParser file, you can use the attribute .source_files
to access a collection of SourceFile objects, and the attribute .headers to get
a flat list of all header files that are tracked with the object. During
initial processing, it is possible to add a set of filters (.filters attribute)
and this reduced header set can be accessed through the .filtered_headers
attribute, but note that changing the filters will not remove items from the
SourceFile data tree.