better collocation export

This commit is contained in:
Luka Kavčič 2020-09-03 10:29:24 +02:00
parent 8802a09c9c
commit 19d3e38dcb
2 changed files with 64 additions and 20 deletions

View File

@ -24,6 +24,7 @@ def build_structure_conversions():
if line[1] == "struktura":
continue
vto_structure = line[1].strip().split(">")[1].split("<")[0]
vto_name = line[2].strip()
vto_id = line[4].strip()
@ -31,14 +32,14 @@ def build_structure_conversions():
continue
vfrom = "^" + line[0].replace("?", "\?").replace("%s", "([a-zA-Z螚ȎŠ-]+)") + "$"
structure_conversions.append((__new__(RegExp(vfrom, 'u')), vto_name, vto_id))
structure_conversions.append((__new__(RegExp(vfrom, 'u')), vto_name, vto_structure, vto_id))
def convert_structure(structure, type):
if structure_conversions is None:
build_structure_conversions()
for vfrom, vto_name, vto_id in structure_conversions:
for vfrom, vto_name, vto_structure, vto_id in structure_conversions:
match = structure.match(vfrom)
# fix for ids 65, 66, 67 which instead matched with 64
if match and vto_id == '64' and '-s' in type:
@ -57,7 +58,7 @@ def convert_structure(structure, type):
result = structure.replace(vfrom, vto_name).strip()
__pragma__('alias', 'replace', "py_replace")
return result, vto_id
return result, vto_structure, vto_id
window.console.log("Unknown structure: ", structure)
return 'N/A', '/'

View File

@ -59,7 +59,8 @@ class SkeCollocation:
def __init__(self, data):
self.word = data.word
self.frequency = data.count
self.structure_name, self.structure_id = convert_structure(data.gramrel, data.lempos)
self.gramrel = data.gramrel
self.structure_name, self.structure, self.structure_id = convert_structure(data.gramrel, data.lempos)
self.other = {"score": data.score, "cm": data.cm}
@ -145,9 +146,21 @@ class SkeModal(ClickMessage):
# we get the data, we have to match it with available data on our gf2 examples API
match_gf2_examples(data, page_num, search_term, ske_index)
elif type(data[0]) is SkeCollocation:
data.sort(key= lambda x: x.structure_name)
data = [x for x in data if x.structure_name is not "N/A"]
# filtering, grouping and sorting data
data.sort(key= lambda x: float(x.other["score"]), reverse=True)
_data = []
while len(data) > 0:
max_item = data.pop(0) # max(data, key= lambda x: x.other["score"])
_data.append(max_item)
for item in data:
if "N/A" in item.structure_name:
data.remove(item)
elif item.structure_name.strip() == max_item.structure_name.strip():
_data.append(item)
for delete_item in _data:
if delete_item in data:
data.remove(delete_item)
data = _data
model.modal_set(lambda: modals.ske_list(
search_term, data, page_num, model.entry.senses, model.ske.request_kinds))
@ -260,17 +273,47 @@ class SkeInsert(DataChgClickMessage):
new_collocation.inner.other_attributes["frequency"] = example.frequency
new_collocation.inner.type = "collocation"
lex_left = ComponentLexeme()
lex_left.text = ""
lex_left.role = None
lex_mid = ComponentLexeme()
lex_mid.text = example.word
lex_mid.role = "collocate"
lex_right = ComponentLexeme()
lex_right.text = ""
lex_right.role = None
new_collocation.components.extend([lex_left, lex_mid, lex_right])
headword = document.getElementById("ske-search").value
lexemes = []
structure_name = example.structure_name.split("-")
gramrel = example.gramrel.split("_")
structure = example.structure.split(" ")
structure.append("") # Bad fix: we have to add something for structure l-gg-ggn
for i in range(len(structure_name)):
console.log("current:", i, "\ngramrel:", gramrel[i], "\nstructure:", structure[i], "\nstruc_name:", structure_name[i])
lex = ComponentLexeme()
structure[i] = structure[i].replace("Inf-", "")
# take care of negations "ne"
if "Neg-" in structure[i]:
structure[i] = structure[i].replace("Neg-", "")
negation_flag = True
n_lex = ComponentLexeme()
n_lex.text = "ne"
n_lex.role = "other"
lexemes.append(n_lex)
if structure[i] is "":
continue # skipping bcs of fix
elif "Vez-gbz" in structure[i]:
lex.text = "je"
lex.role = "other"
elif structure_name[i] in ["d", "vd", "zp"]:
lex.text = gramrel[i]
lex.text = lex.text.replace("-d", "").replace("%", "")
lex.role = "other"
elif structure_name[i] is "vp":
lex.text = structure[i]
lex.role = "other"
elif structure[i][0] in ["S", "G", "P", "R"]:
lex.text = headword
lex.role = "headword"
else:
lex.text = example.word
lex.role = "collocate"
lexemes.append(lex)
new_collocation.components.extend(lexemes)
return new_collocation