Spaces:
Sleeping
Sleeping
Commit
·
eb0f4d2
1
Parent(s):
902d4bf
Update helper.py
Browse files
helper.py
CHANGED
@@ -222,4 +222,89 @@ def numeric_number_dot_freetext(text):
|
|
222 |
return 0
|
223 |
|
224 |
except:
|
225 |
-
return 0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
222 |
return 0
|
223 |
|
224 |
except:
|
225 |
+
return 0
|
226 |
+
|
227 |
+
|
228 |
+
def convert_into_numeric(num_list):
|
229 |
+
'''
|
230 |
+
This is a function to convert the identified numbers into a numeric form
|
231 |
+
'''
|
232 |
+
|
233 |
+
if num_list:
|
234 |
+
|
235 |
+
# at first we examine how many numbers were captured. Only one number should exist
|
236 |
+
if len(num_list) > 1:
|
237 |
+
return (0,'MAGNITUDE','more_magnitudes')
|
238 |
+
|
239 |
+
else:
|
240 |
+
target_num = num_list[0]
|
241 |
+
|
242 |
+
# case it is an integer or float, convert it, otherwise move to following cases
|
243 |
+
try:
|
244 |
+
target_num_float = float(target_num)
|
245 |
+
return {'Number' : target_num}
|
246 |
+
|
247 |
+
except:
|
248 |
+
# case that it belongs to one of the patterns of freetext number followed by numeric form etc (all the combinations)
|
249 |
+
if "$pattern" in target_num:
|
250 |
+
num, _ = target_num.split("$")
|
251 |
+
|
252 |
+
# Try with this function for all the rest of cases (6 point 5, 6 point five, six point 5)
|
253 |
+
num_conversion = numeric_number_dot_freetext(num)
|
254 |
+
|
255 |
+
if num_conversion:
|
256 |
+
return {'Number' : num_conversion}
|
257 |
+
|
258 |
+
# if none of the above has worked, then examine the case of freetext numbers without patterns (e.g. two, million, twenty three, etc)
|
259 |
+
else:
|
260 |
+
try:
|
261 |
+
num_conversion = w2n.word_to_num(target_num)
|
262 |
+
return {'Number' : num_conversion}
|
263 |
+
|
264 |
+
# if none of the above try to handle cases of "million and two" or "a million and two". In such cases, we delete any 'a' reference
|
265 |
+
# and we insert the word 'one' at the beginning. In that way the w2n library can handle them besides immediately throw an error
|
266 |
+
except:
|
267 |
+
|
268 |
+
try:
|
269 |
+
target_num = target_num.replace(" a ", " ")
|
270 |
+
new_target_num = "one " + target_num
|
271 |
+
num_conversion = w2n.word_to_num(new_target_num)
|
272 |
+
return {'Number' : num_conversion}
|
273 |
+
|
274 |
+
except:
|
275 |
+
return (0,'MAGNITUDE','unknown_error')
|
276 |
+
|
277 |
+
else:
|
278 |
+
return (0,'MAGNITUDE','no_magnitude')
|
279 |
+
|
280 |
+
|
281 |
+
def magnitude_binding(input_text):
|
282 |
+
'''
|
283 |
+
This is a function that binds together all the subcomponents of the magnitude number identification, while also controlling for multiple, or zero magnitude references
|
284 |
+
'''
|
285 |
+
|
286 |
+
try:
|
287 |
+
|
288 |
+
# capture the referred magnitudes
|
289 |
+
target_numbers = capture_numbers(input_text)
|
290 |
+
|
291 |
+
# we only accept for one magnitude reference
|
292 |
+
if len(target_numbers) == 1:
|
293 |
+
numeric_target_numbers = convert_into_numeric(target_numbers)
|
294 |
+
|
295 |
+
return numeric_target_numbers
|
296 |
+
|
297 |
+
# in case of zero references return the appropriate code (to aid returning the correct prompt)
|
298 |
+
elif len(target_numbers) == 0:
|
299 |
+
return (0,'MAGNITUDE','no_magnitude')
|
300 |
+
|
301 |
+
# in case of more than one references return the appropriate code (to aid returning the correct prompt)
|
302 |
+
elif len(target_numbers) > 1:
|
303 |
+
return (0,'MAGNITUDE','more_magnitudes')
|
304 |
+
|
305 |
+
# in case of unexpected error return the appropriate code (to aid returning the correct prompt)
|
306 |
+
else:
|
307 |
+
return (0,'MAGNITUDE','unknown_error')
|
308 |
+
|
309 |
+
except:
|
310 |
+
return (0,'MAGNITUDE','unknown_error')
|