def expression_valid_windows(windows, candidate):
valid_window = []
valid_windows = []
buff = []
for window in windows:
if (ana_useful.exists_linkword(window) == True
and ana_useful.count_cand(window) == 2):
if ana_useful.is_cand(
window[-1]): #list[-1] returns last item of the list
cand2 = ana_useful.which_cand([window[-1]])
if cand2[2] not in candidate.split(
): #to avoid building expression like "bâtiment de cet ensemble de bâtiments" -> "batiment de bâtiment"
buff.append(
window
) #dans ce cas la fenetre valide est de type (CAND1 + "aword" + CAND2) avec un mot de schéma quelque part.
# in the buffer because we need to know if the aword in center is not the same 3 times or more (this case it would be better to build an expansion first, then an expression)
else:
short_window = ana_useful.cut_window(window, 2)
#Puisqu'on a 2 CAND et que la fenetre fait 3 mots et que le dernier mot n'est pas un CAND alors la fenetre était de type CAND + CAND + mot quelconque
if ana_useful.exists_linkword(short_window) == True:
valid_windows.append(
short_window
) #dans ce cas la fenetre valide est de type (CAND1 + CAND2) avec un mot de schéma entre eux .
# check if the aword in center is not the same 3 times or more (this case it would be better to build an expansion first, then an expression)
valid_windows.extend(not_expa_inside_expre(buff))
return valid_windows
def nucleus_valid_window(window):
if ana_useful.exists_linkword(window):
for occurrence in window:
index_cand = 0
if ana_useful.is_cand(occurrence):
index_cand = window.index(occurrence)
break
right_window = window[index_cand:]
if ana_useful.count_cand(right_window) < 2 and ana_useful.exists_linkword(right_window):
return right_window
def nucleus_valid_window(window):
if ana_useful.exists_linkword(window):
for occurrence in window:
index_cand = 0
if ana_useful.is_cand(occurrence):
index_cand = window.index(occurrence)
break
right_window = window[index_cand:]
if ana_useful.count_cand(
right_window) < 2 and ana_useful.exists_linkword(right_window):
return right_window
def expression_find_cand(valid_windows, expression_threshold):
shortshape_list = []
dict_cand_windows = {}
i = 0
for window in valid_windows:
shortshape = ''
#créer une shortshape pour chaque fenetre. une shortshape est 'CANDCAND'
#apriori toutes les shortshapes commenceront par le même cand (celui en argument de la fonction `recherche_expression`)
for occurrence in window:
if ana_useful.is_cand(occurrence):
shortshape += occurrence[2]
shortshape_list.append(shortshape) # l'ordre des shortshapes dans shortshape_list conserve l'ordre des fenetres in valid_windows
for shortshape in shortshape_list:
occ_count = shortshape_list.count(shortshape)
if occ_count >= expression_threshold:
dict_cand_windows.setdefault(shortshape,[]).append(valid_windows[i])
i += 1
return dict_cand_windows
def expansion_valid_window(windows):
valid_windows = []
for window in windows:
for occurrence in window:
if ana_useful.is_cand(occurrence):
pos_cand = window.index(occurrence)
left_window = window[:pos_cand + 1]
right_window = window[pos_cand:]
exists_linkword_R = ana_useful.exists_linkword(right_window)
exists_linkword_L = ana_useful.exists_linkword(left_window)
clean_window = ana_useful.window_wo_fword(window)
#Les expansions ne doivent pas contenir de mot de schéma
# Le CAND est forcément en position 2 par construction et suppression des mots v
if clean_window[2][2] == 't' and not exists_linkword_R:
valid_windows.append(window[pos_cand:]) #RightWindow
if clean_window[0][2] == 't' and not exists_linkword_L:
valid_windows.append(window[:pos_cand + 1]) #LeftWindow
return valid_windows
def expansion_valid_window(windows):
valid_windows = []
for window in windows:
for occurrence in window:
if ana_useful.is_cand(occurrence):
pos_cand = window.index(occurrence)
left_window = window[:pos_cand+1]
right_window = window[pos_cand:]
exists_linkword_R = ana_useful.exists_linkword(right_window)
exists_linkword_L = ana_useful.exists_linkword(left_window)
clean_window = ana_useful.window_wo_fword(window)
#Les expansions ne doivent pas contenir de mot de schéma
# Le CAND est forcément en position 2 par construction et suppression des mots v
if clean_window[2][2] == 't' and not exists_linkword_R:
valid_windows.append(window[pos_cand:])#RightWindow
if clean_window[0][2] == 't' and not exists_linkword_L:
valid_windows.append(window[:pos_cand+1])#LeftWindow
return valid_windows
def expression_valid_windows(windows, candidate):
valid_window = []
valid_windows = []
buff = []
for window in windows:
if (ana_useful.exists_linkword(window) == True and ana_useful.count_cand(window) == 2):
if ana_useful.is_cand(window[-1]): #list[-1] returns last item of the list
cand2 = ana_useful.which_cand([window[-1]])
if cand2[2] not in candidate.split(): #to avoid building expression like "bâtiment de cet ensemble de bâtiments" -> "batiment de bâtiment"
buff.append(window) #dans ce cas la fenetre valide est de type (CAND1 + "aword" + CAND2) avec un mot de schéma quelque part.
# in the buffer because we need to know if the aword in center is not the same 3 times or more (this case it would be better to build an expansion first, then an expression)
else:
short_window = ana_useful.cut_window(window, 2)
#Puisqu'on a 2 CAND et que la fenetre fait 3 mots et que le dernier mot n'est pas un CAND alors la fenetre était de type CAND + CAND + mot quelconque
if ana_useful.exists_linkword(short_window) == True:
valid_windows.append(short_window) #dans ce cas la fenetre valide est de type (CAND1 + CAND2) avec un mot de schéma entre eux .
# check if the aword in center is not the same 3 times or more (this case it would be better to build an expansion first, then an expression)
valid_windows.extend(not_expa_inside_expre(buff))
return valid_windows
def expression_find_cand(valid_windows, expression_threshold):
shortshape_list = []
dict_cand_windows = {}
i = 0
for window in valid_windows:
shortshape = ''
#créer une shortshape pour chaque fenetre. une shortshape est 'CANDCAND'
#apriori toutes les shortshapes commenceront par le même cand (celui en argument de la fonction `recherche_expression`)
for occurrence in window:
if ana_useful.is_cand(occurrence):
shortshape += occurrence[2]
shortshape_list.append(
shortshape
) # l'ordre des shortshapes dans shortshape_list conserve l'ordre des fenetres in valid_windows
for shortshape in shortshape_list:
occ_count = shortshape_list.count(shortshape)
if occ_count >= expression_threshold:
dict_cand_windows.setdefault(shortshape,
[]).append(valid_windows[i])
i += 1
return dict_cand_windows
