231 lines
9.1 KiB
Python
231 lines
9.1 KiB
Python
import random
|
|
import re
|
|
import six
|
|
|
|
from geodata.address_expansions.gazetteers import *
|
|
from geodata.encoding import safe_decode, safe_encode
|
|
from geodata.text.tokenize import tokenize_raw, token_types
|
|
from geodata.text.utils import non_breaking_dash_regex
|
|
|
|
|
|
LOWER, UPPER, TITLE, MIXED = range(4)
|
|
|
|
|
|
def token_capitalization(s):
|
|
if s.istitle():
|
|
return TITLE
|
|
elif s.islower():
|
|
return LOWER
|
|
elif s.isupper():
|
|
return UPPER
|
|
else:
|
|
return MIXED
|
|
|
|
|
|
expansion_token_regex = re.compile('([^ \-\.]+)([\.\- ]+|$)')
|
|
|
|
|
|
def recase_abbreviation(expansion, tokens, space_token=six.u(' ')):
|
|
expansion_tokens = expansion_token_regex.findall(expansion)
|
|
|
|
if len(tokens) > len(expansion_tokens) and all((token_capitalization(t) != LOWER for t, c in tokens)):
|
|
expansion_tokenized = tokenize(expansion)
|
|
is_acronym = len(expansion_tokenized) == 1 and expansion_tokenized[0][1] == token_types.ACRONYM
|
|
if len(expansion) <= 3 or is_acronym:
|
|
return expansion.upper()
|
|
else:
|
|
return expansion.title()
|
|
elif len(tokens) == len(expansion_tokens):
|
|
strings = []
|
|
for (t, c), (e, suf) in zip(tokens, expansion_tokens):
|
|
cap = token_capitalization(t)
|
|
if suf == six.u(' '):
|
|
suf = space_token
|
|
if cap == LOWER:
|
|
strings.append(six.u('').join((e.lower(), suf)))
|
|
elif cap == UPPER:
|
|
strings.append(six.u('').join((e.upper(), suf)))
|
|
elif cap == TITLE:
|
|
strings.append(six.u('').join((e.title(), suf)))
|
|
elif t.lower() == e.lower():
|
|
strings.append(t)
|
|
else:
|
|
strings.append(six.u('').join((e.title(), suf)))
|
|
return six.u('').join(strings)
|
|
else:
|
|
|
|
strings = []
|
|
for e, suf in expansion_tokens:
|
|
strings.append(e.title())
|
|
if suf == six.u(' '):
|
|
strings.append(space_token)
|
|
else:
|
|
strings.append(suf)
|
|
return six.u('').join(strings)
|
|
|
|
|
|
def abbreviate(gazetteer, s, language, abbreviate_prob=0.3, separate_prob=0.2, add_period_hyphen_prob=0.3):
|
|
'''
|
|
Abbreviations
|
|
-------------
|
|
|
|
OSM discourages abbreviations, but to make our training data map better
|
|
to real-world input, we can safely replace the canonical phrase with an
|
|
abbreviated version and retain the meaning of the words
|
|
'''
|
|
raw_tokens = tokenize_raw(s)
|
|
s_utf8 = safe_encode(s)
|
|
tokens = [(safe_decode(s_utf8[o:o + l]), token_types.from_id(c)) for o, l, c in raw_tokens]
|
|
norm_tokens = [(t.lower() if c in token_types.WORD_TOKEN_TYPES else t, c) for t, c in tokens]
|
|
|
|
n = len(tokens)
|
|
|
|
abbreviated = []
|
|
|
|
i = 0
|
|
|
|
def abbreviated_tokens(i, tokens, t, c, length, data, space_token=six.u(' ')):
|
|
data = [d.split(six.b('|')) for d in data]
|
|
|
|
# local copy
|
|
abbreviated = []
|
|
|
|
# Append the original tokens with whitespace if there is any
|
|
if random.random() > abbreviate_prob or not any((int(is_canonical) and lang in (language, 'all') for lang, dictionary, is_canonical, canonical in data)):
|
|
for j, (t_i, c_i) in enumerate(t):
|
|
abbreviated.append(tokens[i + j][0])
|
|
|
|
if j < length - 1:
|
|
abbreviated.append(space_token)
|
|
return abbreviated
|
|
|
|
for lang, dictionary, is_canonical, canonical in data:
|
|
if lang not in (language, 'all'):
|
|
continue
|
|
|
|
is_canonical = int(is_canonical)
|
|
is_stopword = dictionary == 'stopword'
|
|
is_prefix = dictionary.startswith('concatenated_prefixes')
|
|
is_suffix = dictionary.startswith('concatenated_suffixes')
|
|
is_separable = is_prefix or is_suffix and dictionary.endswith('_separable') and len(t[0][0]) > length
|
|
|
|
suffix = None
|
|
prefix = None
|
|
|
|
if not is_canonical:
|
|
continue
|
|
|
|
if not is_prefix and not is_suffix:
|
|
abbreviations = gazetteer.canonicals.get((canonical, lang, dictionary))
|
|
token = random.choice(abbreviations) if abbreviations else canonical
|
|
token = recase_abbreviation(token, tokens[i:i + len(t)], space_token=space_token)
|
|
abbreviated.append(token)
|
|
break
|
|
elif is_prefix:
|
|
token = tokens[i][0]
|
|
prefix, token = token[:length], token[length:]
|
|
|
|
abbreviated.append(prefix)
|
|
if random.random() < separate_prob:
|
|
sub_tokens = tokenize(token)
|
|
if sub_tokens and sub_tokens[0][1] in (token_types.HYPHEN, token_types.DASH):
|
|
token = six.u('').join((t for t, c in sub_tokens[1:]))
|
|
|
|
abbreviated.append(space_token)
|
|
if token.islower():
|
|
abbreviated.append(token.title())
|
|
else:
|
|
abbreviated.append(token)
|
|
abbreviated.append(space_token)
|
|
break
|
|
elif is_suffix:
|
|
token = tokens[i][0]
|
|
|
|
token, suffix = token[:-length], token[-length:]
|
|
|
|
concatenated_abbreviations = gazetteer.canonicals.get((canonical, lang, dictionary), [])
|
|
|
|
separated_abbreviations = []
|
|
phrase = gazetteer.trie.get(suffix.rstrip('.'))
|
|
suffix_data = [safe_decode(d).split(six.u('|')) for d in (phrase or [])]
|
|
for l, d, _, c in suffix_data:
|
|
if l == lang and c == canonical:
|
|
separated_abbreviations.extend(gazetteer.canonicals.get((canonical, lang, d)))
|
|
|
|
separate = random.random() < separate_prob
|
|
|
|
if concatenated_abbreviations and not separate:
|
|
abbreviation = random.choice(concatenated_abbreviations)
|
|
elif separated_abbreviations:
|
|
abbreviation = random.choice(separated_abbreviations)
|
|
else:
|
|
abbreviation = canonical
|
|
|
|
if separate:
|
|
sub_tokens = tokenize(token)
|
|
if sub_tokens and sub_tokens[-1][1] in (token_types.HYPHEN, token_types.DASH):
|
|
token = six.u('').join((t for t, c in sub_tokens[:-1]))
|
|
|
|
abbreviated.append(token)
|
|
if separate:
|
|
abbreviated.append(space_token)
|
|
if suffix.isupper():
|
|
abbreviated.append(abbreviation.upper())
|
|
elif separate:
|
|
abbreviated.append(abbreviation.title())
|
|
else:
|
|
abbreviated.append(abbreviation)
|
|
abbreviated.append(space_token)
|
|
break
|
|
else:
|
|
for j, (t_i, c_i) in enumerate(t):
|
|
abbreviated.append(tokens[i + j][0])
|
|
if j < length - 1:
|
|
abbreviated.append(space_token)
|
|
return abbreviated
|
|
return abbreviated
|
|
|
|
for t, c, length, data in gazetteer.filter(norm_tokens):
|
|
if c == token_types.PHRASE:
|
|
abbrev_tokens = abbreviated_tokens(i, tokens, t, c, length, data)
|
|
abbreviated.extend(abbrev_tokens)
|
|
|
|
if i + len(t) < n and raw_tokens[i + len(t)][0] > sum(raw_tokens[i + len(t) - 1][:2]):
|
|
abbreviated.append(six.u(' '))
|
|
|
|
i += len(t)
|
|
|
|
else:
|
|
token = tokens[i][0]
|
|
if not non_breaking_dash_regex.search(token):
|
|
abbreviated.append(token)
|
|
else:
|
|
sub_tokens = tokenize(non_breaking_dash_regex.sub(six.u(' '), token))
|
|
sub_tokens_norm = [(t.lower() if c in token_types.WORD_TOKEN_TYPES else t, c) for t, c in sub_tokens]
|
|
|
|
sub_token_abbreviated = []
|
|
sub_i = 0
|
|
sub_n = len(sub_tokens)
|
|
for t, c, length, data in gazetteer.filter(sub_tokens_norm):
|
|
if c == token_types.PHRASE:
|
|
abbrev_tokens = abbreviated_tokens(sub_i, sub_tokens, t, c, length, data, space_token=six.u('-'))
|
|
sub_token_abbreviated.extend(abbrev_tokens)
|
|
sub_i += len(t)
|
|
if sub_i < sub_n:
|
|
if abbrev_tokens and random.random() < add_period_hyphen_prob and not abbrev_tokens[-1].endswith(six.u('.')) and not abbrev_tokens[-1].lower().endswith(sub_tokens_norm[sub_i - 1][0]):
|
|
sub_token_abbreviated.append(six.u('.'))
|
|
sub_token_abbreviated.append(six.u('-'))
|
|
else:
|
|
sub_token_abbreviated.append(sub_tokens[sub_i][0])
|
|
sub_i += 1
|
|
if sub_i < sub_n:
|
|
sub_token_abbreviated.append(six.u('-'))
|
|
|
|
abbreviated.append(six.u('').join(sub_token_abbreviated))
|
|
|
|
if i < n - 1 and raw_tokens[i + 1][0] > sum(raw_tokens[i][:2]):
|
|
abbreviated.append(six.u(' '))
|
|
i += 1
|
|
|
|
return six.u('').join(abbreviated).strip()
|