def db(ctx, propbank_db):
conn = get_connection()
proc_results(
propbank_db,
conn.execute(headword_grouped(typ=ctx.obj["typ"])),
outf=ctx.obj["out"],
)
def single(ctx, propbank_db, headword):
conn = get_connection()
proc_results(
propbank_db,
conn.execute(headword_grouped((headword, ), typ=ctx.obj["typ"])),
outf=ctx.obj["out"],
)
def propbank(ctx, propbank_db, propbank_tsv):
headwords = {
key: len(list(group))
for key, group in groupby((
row[0] for row in csv.reader(propbank_tsv, delimiter="\t")))
}
conn = get_connection()
proc_results(
propbank_db,
conn.execute(headword_grouped(headwords.keys(), typ=ctx.obj["typ"])),
outf=ctx.obj["out"],
frame_counts=headwords,
)
def confmat(ctx, propbank_db, headword, figout):
conn = get_connection()
group = conn.execute(headword_grouped((headword, ), typ=ctx.obj["typ"]))
group = list(group)
gapped_mwes = [tpl[4] for tpl in group]
print("gapped_mwes", gapped_mwes)
args = objectview(
query_dir="/tmp/",
case=False,
database=propbank_db,
)
with dep_searcher(args) as searcher:
queries = extract_queries(group)
(
counters,
frame_dict,
bitvec_dict,
frame_marginal,
query_marginal,
hits,
) = cont_mat(searcher, queries)
contingency = array(counters)
print("contingency", contingency)
# XXX: Here
df = pd.DataFrame(contingency)
df.columns = list(frame_dict.keys())
print("gapped_mwes", gapped_mwes)
bitvec_labels = list(
bitvecs_to_labels(
bitvec_dict.keys(),
gapped_mwes,
headword,
"\\~{}" if figout and figout.endswith(".pgf") else "~",
))
print("bitvec_labels", bitvec_labels)
df.index = bitvec_labels
print(df)
if figout:
plt.figure(figsize=(5, 3.4))
sns.set(font_scale=0.5)
sns.heatmap(df, annot=True, fmt="d", cmap="YlGnBu")
plt.xticks(rotation=90)
plt.yticks(rotation=0, ha="right")
if figout:
plt.tight_layout()
plt.savefig(figout)
else:
plt.show()
def mwesize(insert):
conn = get_connection()
setup_dist()
joined = mk_joined()
print(BEGIN)
cnts = {}
total_cnt = conn.execute(
select([func.count(tables["ud_mwe"].c.id)
]).select_from(tables["ud_mwe"])).scalar()
fmt_row(0, "Total multiwords", total_cnt)
cnts["total"] = total_cnt
link_cnts = conn.execute(
select([tables["link"].c.name,
func.count(tables["ud_mwe"].c.id)
]).select_from(joined).group_by(tables["link"].c.name))
for idx, (link_name, src_cnt) in enumerate(link_cnts):
cnts[link_name] = src_cnt
print("\\midrule\n")
fmt_row(1, LINK_NAME_MAP[link_name], src_cnt, total_cnt)
fmt_row(2, "have head", have_head_cnt(conn, joined, link_name),
src_cnt)
if link_name != "wikidefn":
for mwe_typ, typ_cnt in mwe_typ_group(conn, joined, link_name):
fmt_row(2, TYP_MAP[mwe_typ], typ_cnt, src_cnt)
if link_name == "wikihw":
if mwe_typ == MweType.inflection:
for has_sense, cnt in wiki_hw_group(
conn,
joined,
mwe_typ,
tables["wiktionary_hw_link"].c.has_senses,
):
fmt_row(
3,
"from a page with definitions" if has_sense
else "from a page without definitions",
cnt,
typ_cnt,
)
if has_sense:
pospos_case = postpositional_case(
conn,
joined,
tables["ud_mwe"].c.typ ==
MweType.inflection,
tables["wiktionary_hw_link"].c.has_senses,
)
fmt_row(
4,
"postpositional case",
pospos_case,
cnt,
)
elif mwe_typ == MweType.multiword:
for page_exists, cnt in wiki_hw_group(
conn,
joined,
mwe_typ,
tables["wiktionary_hw_link"].c.page_exists,
):
fmt_row(
3,
"have a Wiktionary page"
if page_exists else "are a redlink",
cnt,
typ_cnt,
)
elif link_name == "wnhw" and mwe_typ == MweType.inflection:
pospos_case = postpositional_case(
conn,
joined,
tables["link"].c.name == "wnhw",
tables["ud_mwe"].c.typ == MweType.inflection,
)
fmt_row(
3,
"postpositional case",
pospos_case,
typ_cnt,
)
print("\\midrule\n")
# fmt_row(1, "syntactic frames", src_cnt, total_cnt)
# fmt_row(2, "with more than one token", src_cnt, total_cnt)
if insert:
insert_metadata(conn, {"mweproc_" + k: v
for k, v in cnts.items()},
table=tables["meta"])
print(END)
def main():
conn = get_connection()
setup_dist()
conn.execute(DropView("joined", cascade=False))
metadata.create_all(conn)