Spaces:
Sleeping
Sleeping
Added tense correction file for errors coming from pyinflect library database
Browse files- .gitignore +1 -1
- app.py +1 -1
- data/overrides.csv +12 -0
- modules/m_parser.py +81 -49
.gitignore
CHANGED
@@ -1 +1 @@
|
|
1 |
-
__pycache__
|
|
|
1 |
+
__pycache__
|
app.py
CHANGED
@@ -23,7 +23,7 @@ with iface:
|
|
23 |
output = gr.Markdown()
|
24 |
|
25 |
btn_get.click(
|
26 |
-
fn = execute.
|
27 |
inputs = input_verb,
|
28 |
outputs = [error, output],
|
29 |
api_name="get"
|
|
|
23 |
output = gr.Markdown()
|
24 |
|
25 |
btn_get.click(
|
26 |
+
fn = execute.get,
|
27 |
inputs = input_verb,
|
28 |
outputs = [error, output],
|
29 |
api_name="get"
|
data/overrides.csv
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
VB,VBD,VBN
|
2 |
+
work,worked,worked
|
3 |
+
can,could,-
|
4 |
+
wake,woke,woken
|
5 |
+
cost,cost,cost
|
6 |
+
drink,drank,drunk
|
7 |
+
learn,learnt,learnt
|
8 |
+
leave,left,left
|
9 |
+
pay,paid,paid
|
10 |
+
ring,rang,rung
|
11 |
+
speak,spoke,spoken
|
12 |
+
write,wrote,written
|
modules/m_parser.py
CHANGED
@@ -1,14 +1,25 @@
|
|
1 |
import subprocess
|
2 |
import spacy
|
3 |
import pyinflect
|
4 |
-
|
|
|
5 |
|
6 |
class Parser:
|
7 |
def __init__(
|
8 |
self
|
9 |
) -> None:
|
10 |
|
11 |
-
self.parser = self.__init_parser("
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
|
13 |
def __init_parser(
|
14 |
self,
|
@@ -30,63 +41,86 @@ class Parser:
|
|
30 |
|
31 |
return parser
|
32 |
|
33 |
-
def
|
34 |
self,
|
35 |
-
|
36 |
-
) ->
|
37 |
|
38 |
-
|
39 |
-
|
40 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
41 |
self,
|
42 |
-
|
43 |
-
|
|
|
|
|
44 |
|
45 |
-
|
|
|
|
|
|
|
|
|
|
|
46 |
|
47 |
-
def
|
48 |
self,
|
49 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
50 |
) -> Union[str, None]:
|
|
|
|
|
|
|
51 |
|
52 |
-
|
|
|
|
|
|
|
53 |
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
|
59 |
-
return
|
60 |
|
61 |
-
def
|
62 |
-
self,
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
|
69 |
-
| :----: | :----: | :----: |
|
70 |
-
|{} | {}| {}|
|
71 |
-
"""
|
72 |
-
return template.format(infiniteve, past, participle)
|
73 |
|
|
|
74 |
|
75 |
-
def
|
76 |
self,
|
77 |
-
|
78 |
-
) ->
|
79 |
|
80 |
-
|
81 |
-
<center>
|
82 |
-
<div class="alert alert-warning" role="alert">
|
83 |
-
<h6><b>{}</b></h6>
|
84 |
-
</div>
|
85 |
-
</center>
|
86 |
-
"""
|
87 |
-
return template.format(error)
|
88 |
|
89 |
-
def
|
90 |
self,
|
91 |
verb: str
|
92 |
) -> Tuple[str,str]:
|
@@ -95,22 +129,20 @@ class Parser:
|
|
95 |
error, output = "", ""
|
96 |
|
97 |
if verb == "":
|
98 |
-
error = self.
|
99 |
f"Error: The Verb field can not be empty!"
|
100 |
)
|
101 |
return error, output
|
102 |
|
103 |
tk_verb = self.__tokenizer(verb)
|
104 |
-
infinitive = self.
|
105 |
-
past = self.__v2past(tk_verb)
|
106 |
-
participle = self.__v2participle(tk_verb)
|
107 |
|
108 |
if infinitive is None or past is None or participle is None:
|
109 |
-
error = self.
|
110 |
f"Error: The verb '<b>{verb}</b>' has not been found or not spelled correctly!"
|
111 |
)
|
112 |
return error, output
|
113 |
|
114 |
-
output = self.
|
115 |
|
116 |
return error, output
|
|
|
1 |
import subprocess
|
2 |
import spacy
|
3 |
import pyinflect
|
4 |
+
import pandas as pd
|
5 |
+
from typing import Tuple, Union, Any
|
6 |
|
7 |
class Parser:
|
8 |
def __init__(
|
9 |
self
|
10 |
) -> None:
|
11 |
|
12 |
+
self.parser = self.__init_parser("en_core_web_sm")
|
13 |
+
self.__overrides = self.__load_overrides("data/overrides.csv")
|
14 |
+
|
15 |
+
def __load_overrides(
|
16 |
+
self,
|
17 |
+
file_path: str
|
18 |
+
) -> Any:
|
19 |
+
|
20 |
+
dic = pd.read_csv(file_path).set_index('VB').to_dict()
|
21 |
+
fun = lambda verb, tag: dic[tag].get(verb, False)
|
22 |
+
return fun
|
23 |
|
24 |
def __init_parser(
|
25 |
self,
|
|
|
41 |
|
42 |
return parser
|
43 |
|
44 |
+
def __format_error(
|
45 |
self,
|
46 |
+
error: str
|
47 |
+
) -> str:
|
48 |
|
49 |
+
template = """
|
50 |
+
<center>
|
51 |
+
<div class="alert alert-warning" role="alert">
|
52 |
+
<h6><b>{}</b></h6>
|
53 |
+
</div>
|
54 |
+
</center>
|
55 |
+
"""
|
56 |
+
return template.format(error)
|
57 |
+
|
58 |
+
def __format_output(
|
59 |
self,
|
60 |
+
infinitive: str,
|
61 |
+
past: str,
|
62 |
+
participle: str,
|
63 |
+
) -> str:
|
64 |
|
65 |
+
template = """
|
66 |
+
|Infinitive| Simple Past | Past Participle |
|
67 |
+
| :----: | :----: | :----: |
|
68 |
+
|{} | {}| {}|
|
69 |
+
"""
|
70 |
+
return template.format(infinitive, past, participle)
|
71 |
|
72 |
+
def is_in_overrides(
|
73 |
self,
|
74 |
+
verb: str,
|
75 |
+
tense: str
|
76 |
+
) -> Tuple[str, bool]:
|
77 |
+
|
78 |
+
res = self.__overrides(verb, tense)
|
79 |
+
if isinstance(res, bool):
|
80 |
+
return res, ""
|
81 |
+
|
82 |
+
return True, res
|
83 |
+
|
84 |
+
def __get_inflections(
|
85 |
+
self,
|
86 |
+
infinitive: spacy.tokens.token.Token,
|
87 |
+
tense: str
|
88 |
) -> Union[str, None]:
|
89 |
+
|
90 |
+
# Check if verb is in overrides file
|
91 |
+
res, inflections = self.is_in_overrides(infinitive.text, tense)
|
92 |
|
93 |
+
if not res:
|
94 |
+
form1 = infinitive._.inflect(tense, form_num=0)
|
95 |
+
form2 = infinitive._.inflect(tense, form_num=1)
|
96 |
+
inflections = list(set([form1, form2]))
|
97 |
|
98 |
+
if len(inflections) == 1 and inflections[0] == None:
|
99 |
+
return None
|
100 |
+
|
101 |
+
inflections = '/'.join(inflections)
|
102 |
|
103 |
+
return inflections
|
104 |
|
105 |
+
def __getAllTenses(
|
106 |
+
self,
|
107 |
+
tk_verb: spacy.tokens.token.Token
|
108 |
+
) -> Tuple[str,str,str,str]:
|
109 |
+
|
110 |
+
infinitive = tk_verb._.inflect('VB')
|
111 |
+
past = self.__get_inflections(tk_verb, 'VBD')
|
112 |
+
participle = self.__get_inflections(tk_verb, 'VBN')
|
|
|
|
|
|
|
|
|
113 |
|
114 |
+
return infinitive, past, participle
|
115 |
|
116 |
+
def __tokenizer(
|
117 |
self,
|
118 |
+
verb: str
|
119 |
+
) -> spacy.tokens.token.Token:
|
120 |
|
121 |
+
return self.parser(verb)[0]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
122 |
|
123 |
+
def get(
|
124 |
self,
|
125 |
verb: str
|
126 |
) -> Tuple[str,str]:
|
|
|
129 |
error, output = "", ""
|
130 |
|
131 |
if verb == "":
|
132 |
+
error = self.__format_error(
|
133 |
f"Error: The Verb field can not be empty!"
|
134 |
)
|
135 |
return error, output
|
136 |
|
137 |
tk_verb = self.__tokenizer(verb)
|
138 |
+
infinitive, past, participle = self.__getAllTenses(tk_verb)
|
|
|
|
|
139 |
|
140 |
if infinitive is None or past is None or participle is None:
|
141 |
+
error = self.__format_error(
|
142 |
f"Error: The verb '<b>{verb}</b>' has not been found or not spelled correctly!"
|
143 |
)
|
144 |
return error, output
|
145 |
|
146 |
+
output = self.__format_output(infinitive, past, participle)
|
147 |
|
148 |
return error, output
|