def proc(method, inf, outf, head=None, no_zstd_out=False):
from plumbum.cmd import zstdcat, zstdmt
if os.environ.get("TRACE_PIPELINE"):
print(method)
pipeline = add_head(filter_py, zstdcat["-D", "zstd-compression-dictionary",
inf], head)
pipeline = (pipeline
| python[filter_py, "fold-support", "fi", "-", "-"]
| python[filter_py, "lang", "fi", "-", "-"])
method_stages = METHODS[method]
for stage in method_stages:
long_stage = lookup_stage(stage)
args = [filter_py] + long_stage.split(" ") + ["-", "-"]
pipeline = pipeline | python[args]
if not no_zstd_out:
pipeline = (pipeline | zstdmt["-D", "zstd-compression-dictionary", "-",
"-o", outf])
else:
pipeline = pipeline > outf
exec_pipeline(pipeline, retcode=[-13, 0])
def eurosense2stifflike(inf, outf, head, babel2wn_map):
pipeline = add_head(filter_py, add_zstd(inf), head)
pipeline = (mk_eurosense2stifflike_pipeline(pipeline, babel2wn_map)
| python[munge_py, "eurosense-add-anchor-positions", "-",
outf])
print(pipeline)
pipeline(retcode=[-13, 0], stderr=sys.stderr)
def stiff2unified(inf, outf, keyout, head, input_fmt):
pipeline = (add_head(filter_py, add_zstd(inf), head)
| python[munge_py, "stiff-select-wn", "--wn", "qf2", "-", "-"]
| python[filter_py, "tok-span-dom", "-", "-"]
| python[munge_py, "lemma-to-synset", "-", "-"]
| python[munge_py, "stiff-to-unified", "--input-fmt",
input_fmt, "-", "-"]
| python[munge_py, "unified-split", "-", outf, keyout])
pipeline(retcode=[-13, 0], stderr=sys.stderr)
def eurosense2unified(inf, outf, keyout, head, babel2wn_map):
"""
Convert from the Eurosense format to the Unified format so that Eurosense
tagged data can be compared with STIFF.
"""
pipeline = add_head(filter_py, add_zstd(inf), head)
pipeline = (mk_eurosense2stifflike_pipeline(pipeline, babel2wn_map)
| python[munge_py, "eurosense-to-unified", "-", "-"]
| python[munge_py, "unified-split", "-", outf, keyout])
pipeline(retcode=[-13, 0], stderr=sys.stderr)