-
Notifications
You must be signed in to change notification settings - Fork 0
/
stanza_test.py
134 lines (131 loc) · 5.39 KB
/
stanza_test.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
""" minimal Stanford STANZA demo
runs under the stanza environment
conda activate stanza
Will read a text from the IN_TXT constant in this code
Will generate the corresponding BRAT files to display the results
"""
# import libraries
import pandas as pd
import stanza
# declare constants and other variables
TOKLET = "T" # used by BRAT to prefix Token unique IDs
RELLET = "R" # used by BRAT to prefix Relationship unique IDs
LANG = "it"
UPOSCOLSDICT = {
"ADJ": "pink",
"ADP": "#FF99FF",
"ADV": "beige",
"AUX": "#FF6699",
"CCONJ": "lavender",
"DET": "orange",
"INTJ": "lime",
"NOUN": "yellow",
"NUM": "green",
"PART": "cyan",
"PRON": "orchid",
"PROPN": "aqua",
"PUNCT": "lemonchiffon",
"SCONJ": "grey",
"SYM": "gold",
"VERB": "#99FFFF",
"X": "orange",
}
uposcolsdf = pd.DataFrame(
UPOSCOLSDICT.items(), columns=["pos", "col"]
) # upos tag and corresponding color dataframe
# input text
IN_TXT = """
Egregio collega, dimettiamo in data odierna, il sig. Disney Eolo nato il 17/01/1941 a MORDOR residente in VIA ELFI VERDI 17 ricoverato presso il reparto di CHIRURGIA VASCOLARE OSE in data 12/11/2022
Motivo del Ricovero
Aneurisma arteria iliaca comune sinistra
Diagnosi alla dimissione
Aneurisma arteria iliaca comune sinistra sottoposto ad esclusione endovascolare
Dati Anamnestici
stenosi aortica lieve, ipoacusia, ipetrofia prostatica benigna
Interventi e procedure eseguite
In data 13/11/22 intervento di esclusione endovascolare aneurisma arteria iliaca comune sinistra tramite kissing aorto-iliaco (VBX Gore 1 1x79) mediante accesso inguinale bilaterale.
Decorso Clinico e condizioni del paziente alla Dimissione
Regolare
Terapia Farmacologica consigliata
APROVEL* 150 MG 1 CP alle ore 07:00
REXTAT 20MG 1 CP alle ore 22:00
AVODART 0,5MG 1 volta al di alle ore: 20:00 1 CP
PLAVIX*75 MG 1 CP alle ore 18:00
OMNIC*0,4MG I CP alle ore 20:00
CARDIOASPIRIN*IOOMG 1 CP alle ore 12:00
AUGMENTIN* 1 G. 1 CP alle ore 07:00, 1 CP alle ore 18:00 fino al 21/10 FOLIFILL 5 MG 1 CP alle ore 07:00 per via orale
PANTORC 20 MG I CP alle ore 07:00
Controlli e consigli
II paziente è atteso il giorno 20/12/22 alle ore 11.00 presso l'ambulatorio di Chirurgia vascolare (piano 0 stanza 10 edificio vecchio) peril controllo degli accessi inguinali.
II paziente dovrà effettuare Angio-TC di controllo ad I mese dalla dimissione (presentarsi a digiuno da almeno 6 ore e con dosaggio creatininemia, 6 piano, Chirurgia Generale).
"""
# get Stanford's STANZA do its magic
# download appropriate language model
stanza.download(LANG, verbose=False)
# instantiate a analysis pipeline
nlp = stanza.Pipeline(
LANG, processors="tokenize,lemma,pos,depparse", verbose=False, use_gpu=False
)
# finally generate a STANZA document from the given text via the pipeline
doc = nlp(IN_TXT)
# initialize empty dataframes to hold token and syntactic relations
t_df = pd.DataFrame(columns=["text", "start", "end", "upos"])
r_df = pd.DataFrame(columns=["srcupos", "deprel", "destid"])
# cycle through STANZA's Document, Sentence, Token, Words and add dataframe rows
for i, sentence in enumerate(doc.sentences):
for t in sentence.tokens:
for w in t.words:
# fill a token spans and a relationships dataframe
eid = f"{i+1}_{t.id[0]}" # build a unique ID
text = f"{t.text}"
start = f"{t.start_char}"
end = f"{t.end_char}"
upos = f"{w.upos}"
tlist = [text, start, end, upos]
t_df.loc[
f"{TOKLET}{eid}"
] = tlist # add list for a token as new to dataframe row and use tid as the index value
srcupos = f"{w.upos}"
deprel = f"{w.deprel}"
destid = f"{i+1}_{w.head}"
rlist = [srcupos, deprel, destid]
r_df.loc[
f"{eid}"
] = rlist # add list of relations as row to dataframe with id index value
### now reading from the dataframes write all BRAT Standoff files
## These are the corresponding txt and ann anallysis files
# txt file
with open("output/test.txt", "w", encoding="utf-8") as f:
f.write(IN_TXT)
# ann file
with open("output/test.ann", "w", encoding="utf-8") as f:
for row in t_df.itertuples():
f.write(
f"{row.Index}\t{row.upos} {row.start} {row.end}\t{row.text}\n"
) # entity annotations
for row in r_df.itertuples():
if row.deprel != "root":
f.write(
f"{RELLET}{row.Index}\t{row.deprel.replace(':','_')} Arg1:T{row.Index} Arg2:{TOKLET}{row.destid}\n"
) # relation annotations
## These are the annotation.con and visual.conf configuration files
# annotation.conf file
with open("output/annotation.conf", "w", encoding="utf-8") as f:
f.write("[entities]\n")
for row in uposcolsdf.itertuples():
f.write(row.pos + "\n")
f.write("[relations]\n")
for row in r_df.itertuples():
if row.deprel != "root":
destupos = t_df.loc[f"{TOKLET}{row.destid}"].upos
f.write(
f"{row.deprel.replace(':','_')}\tArg1:{row.srcupos}, Arg2:{destupos}\n"
)
f.write("[events]\n")
f.write("[attributes]\n")
# visual.conf file
with open("output/visual.conf", "w", encoding="utf-8") as f:
f.write("[labels]\n")
f.write("[drawing]\n") # write UPOS tags and corresponding chosen colors
for row in uposcolsdf.itertuples():
f.write(f"{row.pos}\tbgColor:{row.col}\n")