On Fri, 13 Jun 2025 at 13:40, Mauro Carvalho Chehab <mchehab+huawei@xxxxxxxxxx> wrote: > > Em Fri, 13 Jun 2025 12:20:33 +0100 > Donald Hunter <donald.hunter@xxxxxxxxx> escreveu: > > > Mauro Carvalho Chehab <mchehab+huawei@xxxxxxxxxx> writes: > > > > > As we'll be importing netlink parser into a Sphinx extension, > > > move all functions and global variables inside two classes: > > > > > > - RstFormatters, containing ReST formatter logic, which are > > > YAML independent; > > > - NetlinkYamlParser: contains the actual parser classes. That's > > > the only class that needs to be imported by the script or by > > > a Sphinx extension. > > > > I suggest a third class for the doc generator that is separate from the > > yaml parsing. > > Do you mean moving those two (or three? [*]) methods to a new class? > > def parse_yaml(self, obj: Dict[str, Any]) -> str: > def parse_yaml_file(self, filename: str) -> str: > def generate_main_index_rst(self, output: str, index_dir: str) -> None: > > Also, how should I name it to avoid confusion with NetlinkYamlParser? > Maybe YnlParser? On second thoughts, I see that the rst generation is actually spread through all the parse_* methods so they are all related to doc generation. I suggest putting all the parse_* methods into a class called YnlDocGenerator, so just the 2 classes. And I'm hoping that generate_main_index_rst can be removed. > [*] generate_main_index_rst is probably deprecated. eventually > we may drop it or keep it just at the command line stript. > > > The yaml parsing should really be refactored to reuse > > tools/net/ynl/pyynl/lib/nlspec.py at some point. > > Makes sense, but such change is out of the scope of this series. Agreed Thanks, Donald.
