mirror of https://github.com/explosion/spaCy.git
Partial Hungarian number tokenization is added.
This commit is contained in:
Gyorgy Orosz
parent
0cf2144d24
commit
c035928156
|
|
@ -24,7 +24,7 @@ STOP_WORDS = set(_load_txt_data(_STOPWORDS_PATH))
|
|||
|
||||
HYPHENS = [six.unichr(cp) for cp in [173, 8211, 8212, 8213, 8722, 9472]]
|
||||
|
||||
TOKENIZER_PREFIXES = r''', " ( [ { * < $ £ “ ' `` ` # US$ C$ A$ a- ‘ .... ...'''.split()
|
||||
TOKENIZER_PREFIXES = r'''+ , " ( [ { * < $ £ “ ' `` ` # US$ C$ A$ a- ‘ .... ...'''.split()
|
||||
|
||||
TOKENIZER_SUFFIXES = r'''
|
||||
,
|
||||
|
|
@ -53,10 +53,11 @@ _
|
|||
\.\.
|
||||
\.\.\.
|
||||
\.\.\.\.
|
||||
(?<=[a-züóőúéáűí)\]"'´«‘’%\)²“”-])\.
|
||||
(?<=[a-züóőúéáűí)\]"'´«‘’%\)²“”+-])\.
|
||||
(?<=[a-züóőúéáűí)])-e
|
||||
\-\-
|
||||
´
|
||||
(?<=[0-9])\.
|
||||
(?<=[0-9])km²
|
||||
(?<=[0-9])m²
|
||||
(?<=[0-9])cm²
|
||||
|
|
@ -97,17 +98,17 @@ _
|
|||
(?<=[0-9])M
|
||||
(?<=[0-9])K
|
||||
(?<=[0-9])kb
|
||||
(?<=[0-9])\+
|
||||
'''.strip().split('\n')
|
||||
|
||||
TOKENIZER_INFIXES = (r'''\.\.+ (?<=[a-züóőúéáűí])\.(?=[A-ZÜÓŐÚÉÁŰÍ]) (?<=[a-zA-ZüóőúéáűíÜÓŐÚÉÁŰÍ0-9])"(?=[\-a-zA-ZüóőúéáűíÜÓŐÚÉÁŰÍ]) '''
|
||||
r'''(?<=[a-zA-ZüóőúéáűíÜÓŐÚÉÁŰÍ])--(?=[a-zA-ZüóőúéáűíÜÓŐÚÉÁŰÍ]) (?<=[0-9])-(?=[0-9]) '''
|
||||
r'''(?<=[a-zA-ZüóőúéáűíÜÓŐÚÉÁŰÍ])--(?=[a-zA-ZüóőúéáűíÜÓŐÚÉÁŰÍ]) (?<=[0-9])[+\-\*/^](?=[0-9]) '''
|
||||
r'''(?<=[a-zA-ZüóőúéáűíÜÓŐÚÉÁŰÍ]),(?=[a-zA-ZüóőúéáűíÜÓŐÚÉÁŰÍ])''').split()
|
||||
|
||||
ABBREVIATIONS = {abbrev: [{"ORTH": abbrev}] for abbrev in
|
||||
_load_txt_data(_ABBREVIATIONS_ORIG_PATH, _ABBREVIATIONS_NYTUD_PATH)}
|
||||
|
||||
TOKENIZER_EXCEPTIONS = {
|
||||
"\w*\(\w+\)\w*": [{"F": "???"}],
|
||||
"-e": [{"F": "-e"}],
|
||||
|
||||
"vs.": [{"F": "vs."}],
|
||||
|
|
|
|||
|
|
@ -0,0 +1,618 @@
|
|||
# TOKEN numbers
|
||||
|
||||
szam es betu
|
||||
IN : A 2b van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2b</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2b-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2b-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2b.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2b</w><c>.</c></s>
|
||||
IN : A 2b-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2b-ben</w><c>.</c></s>
|
||||
IN : A 3.b van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>3.b</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 3.b-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>3.b-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 3.b.
|
||||
OUT: <s><w>A</w><ws> </ws><w>3.b</w><c>.</c></s>
|
||||
IN : A 3.b-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>3.b-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 1:20:36.7 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1:20:36.7</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1:20:36.7-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1:20:36.7-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1:20:36.7.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1:20:36.7</w><c>.</c></s>
|
||||
IN : A 1:20:36.7-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1:20:36.7-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 1:35 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1:35</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1:35-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1:35-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1:35.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1:35</w><c>.</c></s>
|
||||
IN : A 1:35-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1:35-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 1.35 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1.35</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1.35-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1.35-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1.35.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1.35</w><c>.</c></s>
|
||||
IN : A 1.35-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1.35-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 4:01,95 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>4:01,95</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 4:01,95-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>4:01,95-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 4:01,95.
|
||||
OUT: <s><w>A</w><ws> </ws><w>4:01,95</w><c>.</c></s>
|
||||
IN : A 4:01,95-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>4:01,95-ben</w><c>.</c></s>
|
||||
|
||||
DASH
|
||||
hyphen minus
|
||||
IN : A 10-12 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-12</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10-12-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-12-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10-12.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-12</w><c>.</c></s>
|
||||
IN : A 10-12-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-12-ben</w><c>.</c></s>
|
||||
hyphen
|
||||
IN : A 10‐12 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10‐12</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10‐12-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10‐12-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10‐12.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10‐12</w><c>.</c></s>
|
||||
IN : A 10‐12-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10‐12-ben</w><c>.</c></s>
|
||||
non-breaking hyphen
|
||||
IN : A 10‑12 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10‑12</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10‑12-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10‑12-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10‑12.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10‑12</w><c>.</c></s>
|
||||
IN : A 10‑12-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10‑12-ben</w><c>.</c></s>
|
||||
figure dash
|
||||
IN : A 10‒12 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10‒12</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10‒12-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10‒12-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10‒12.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10‒12</w><c>.</c></s>
|
||||
IN : A 10‒12-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10‒12-ben</w><c>.</c></s>
|
||||
en dash
|
||||
IN : A 10–12 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10–12</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10–12-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10–12-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10–12.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10–12</w><c>.</c></s>
|
||||
IN : A 10–12-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10–12-ben</w><c>.</c></s>
|
||||
em dash
|
||||
IN : A 10—12 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10—12</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10—12-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10—12-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10—12.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10—12</w><c>.</c></s>
|
||||
IN : A 10—12-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10—12-ben</w><c>.</c></s>
|
||||
horizontal bar
|
||||
IN : A 10―12 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10―12</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10―12-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10―12-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10―12.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10―12</w><c>.</c></s>
|
||||
IN : A 10―12-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10―12-ben</w><c>.</c></s>
|
||||
|
||||
tagolt szamok
|
||||
IN : A -23,12 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>-23,12</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A -23,12-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>-23,12-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A -23,12.
|
||||
OUT: <s><w>A</w><ws> </ws><w>-23,12</w><c>.</c></s>
|
||||
IN : A -23,12-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>-23,12-ben</w><c>.</c></s>
|
||||
|
||||
#TODO: not supported transformations in spaCy
|
||||
# IN : A +12 500 van.
|
||||
# OUT: <s><w>A</w><ws> </ws><w>+12 500</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
# IN : A +12 500-ben van.
|
||||
# OUT: <s><w>A</w><ws> </ws><w>+12 500-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
# IN : A +12 500.
|
||||
# OUT: <s><w>A</w><ws> </ws><w>+12 500</w><c>.</c></s>
|
||||
# IN : A +12 500-ben.
|
||||
# OUT: <s><w>A</w><ws> </ws><w>+12 500-ben</w><c>.</c></s>
|
||||
|
||||
# IN : A +12 500,99 van.
|
||||
# OUT: <s><w>A</w><ws> </ws><w>+12 500,99</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
# IN : A +12 500,99-ben van.
|
||||
# OUT: <s><w>A</w><ws> </ws><w>+12 500,99-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
# IN : A +12 500,99.
|
||||
# OUT: <s><w>A</w><ws> </ws><w>+12 500,99</w><c>.</c></s>
|
||||
# IN : A +12 500,99-ben.
|
||||
# OUT: <s><w>A</w><ws> </ws><w>+12 500,99-ben</w><c>.</c></s>
|
||||
|
||||
IN : A +12.500.000 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>+12.500.000</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A +12.500.000-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>+12.500.000-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A +12.500.000.
|
||||
OUT: <s><w>A</w><ws> </ws><w>+12.500.000</w><c>.</c></s>
|
||||
IN : A +12.500.000-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>+12.500.000-ben</w><c>.</c></s>
|
||||
|
||||
A muveletek (+, -, *, /) jelentette kihivasok:
|
||||
IN : A 2+3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>+</c><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2+3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>+</c><w>3</w><c>.</c></s>
|
||||
IN : A 2 +3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>+</c><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2 +3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>+</c><w>3</w><c>.</c></s>
|
||||
IN : A 2+ 3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>+</c><ws> </ws><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2+ 3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>+</c><ws> </ws><w>3</w><c>.</c></s>
|
||||
IN : A 2 + 3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>+</c><ws> </ws><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2 + 3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>+</c><ws> </ws><w>3</w><c>.</c></s>
|
||||
|
||||
IN : A 2-3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>-</c><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2-3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>-</c><w>3</w><c>.</c></s>
|
||||
IN : A 2 -3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>-</c><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2 -3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>-</c><w>3</w><c>.</c></s>
|
||||
IN : A 2- 3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>-</c><ws> </ws><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2- 3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>-</c><ws> </ws><w>3</w><c>.</c></s>
|
||||
IN : A 2 - 3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>-</c><ws> </ws><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2 - 3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>-</c><ws> </ws><w>3</w><c>.</c></s>
|
||||
|
||||
IN : A 2*3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>*</c><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2*3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>*</c><w>3</w><c>.</c></s>
|
||||
IN : A 2 *3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>*</c><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2 *3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>*</c><w>3</w><c>.</c></s>
|
||||
IN : A 2* 3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>*</c><ws> </ws><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2* 3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>*</c><ws> </ws><w>3</w><c>.</c></s>
|
||||
IN : A 2 * 3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>*</c><ws> </ws><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2 * 3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>*</c><ws> </ws><w>3</w><c>.</c></s>
|
||||
|
||||
IN : A 2/3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>/</c><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2/3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>/</c><w>3</w><c>.</c></s>
|
||||
IN : A 2 /3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>/</c><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2 /3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>/</c><w>3</w><c>.</c></s>
|
||||
IN : A 2/ 3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>/</c><ws> </ws><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2/ 3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><c>/</c><ws> </ws><w>3</w><c>.</c></s>
|
||||
IN : A 2 / 3 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>/</c><ws> </ws><w>3</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2 / 3.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2</w><ws> </ws><c>/</c><ws> </ws><w>3</w><c>.</c></s>
|
||||
|
||||
IN : A C++ van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>C++</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A C++-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>C++-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A C++.
|
||||
OUT: <s><w>A</w><ws> </ws><w>C++</w><c>.</c></s>
|
||||
IN : A C++-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>C++-ben</w><c>.</c></s>
|
||||
|
||||
|
||||
|
||||
datumok
|
||||
IN : A 2003. I. 06. van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003.</w><ws> </ws><w>I.</w><ws> </ws><w>06.</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2003. I. 06-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003.</w><ws> </ws><w>I.</w><ws> </ws><w>06-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2003. I. 06.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003.</w><ws> </ws><w>I.</w><ws> </ws><w>06</w><c>.</c></s>
|
||||
IN : A 2003. I. 06-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003.</w><ws> </ws><w>I.</w><ws> </ws><w>06-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 2003. 01. 06. van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003.</w><ws> </ws><w>01.</w><ws> </ws><w>06.</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2003. 01. 06-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003.</w><ws> </ws><w>01.</w><ws> </ws><w>06-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2003. 01. 06.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003.</w><ws> </ws><w>01.</w><ws> </ws><w>06</w><c>.</c></s>
|
||||
IN : A 2003. 01. 06-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003.</w><ws> </ws><w>01.</w><ws> </ws><w>06-ben</w><c>.</c></s>
|
||||
|
||||
IN : A IV. 12. van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>IV.</w><ws> </ws><w>12.</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A IV. 12-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>IV.</w><ws> </ws><w>12-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A IV. 12.
|
||||
OUT: <s><w>A</w><ws> </ws><w>IV.</w><ws> </ws><w>12</w><c>.</c></s>
|
||||
IN : A IV. 12-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>IV.</w><ws> </ws><w>12-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 2003.01.06. van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003.01.06.</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2003.01.06-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003.01.06-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2003.01.06.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003.01.06</w><c>.</c></s>
|
||||
IN : A 2003.01.06-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003.01.06-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 2003/01/06 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003/01/06</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2003/01/06-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003/01/06-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2003/01/06.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003/01/06</w><c>.</c></s>
|
||||
IN : A 2003/01/06-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2003/01/06-ben</w><c>.</c></s>
|
||||
|
||||
IN : A IV.12. van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>IV.12.</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A IV.12-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>IV.12-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A IV.12.
|
||||
OUT: <s><w>A</w><ws> </ws><w>IV.12</w><c>.</c></s>
|
||||
IN : A IV.12-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>IV.12-ben</w><c>.</c></s>
|
||||
|
||||
fejezetazonosito
|
||||
IN : A 1.1.2. van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1.1.2.</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1.1.2-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1.1.2-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1.1.2.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1.1.2</w><c>.</c></s>
|
||||
IN : A 1.1.2-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1.1.2-ben</w><c>.</c></s>
|
||||
|
||||
egyeb pontot tartalmazo szamok
|
||||
IN : A 123.45/67. van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>123.45/67.</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 123.45/67-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>123.45/67-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 123.45/67.
|
||||
OUT: <s><w>A</w><ws> </ws><w>123.45/67</w><c>.</c></s>
|
||||
IN : A 123.45/67-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>123.45/67-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 1,5-2,5 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1,5-2,5</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1,5-2,5-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1,5-2,5-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1,5-2,5.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1,5-2,5</w><c>.</c></s>
|
||||
IN : A 1,5-2,5-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1,5-2,5-ben</w><c>.</c></s>
|
||||
|
||||
tizedestortek vesszovel
|
||||
IN : A 3,14 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>3,14</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 3,14-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>3,14-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 3,14.
|
||||
OUT: <s><w>A</w><ws> </ws><w>3,14</w><c>.</c></s>
|
||||
IN : A 3,14-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>3,14-ben</w><c>.</c></s>
|
||||
|
||||
tizedestortek ponttal
|
||||
IN : A 3.14 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>3.14</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 3.14-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>3.14-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 3.14.
|
||||
OUT: <s><w>A</w><ws> </ws><w>3.14</w><c>.</c></s>
|
||||
IN : A 3.14-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>3.14-ben</w><c>.</c></s>
|
||||
|
||||
sorszamnevek
|
||||
IN : A 15. van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>15.</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 15-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>15-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 15.
|
||||
OUT: <s><w>A</w><ws> </ws><w>15</w><c>.</c></s>
|
||||
IN : A 15-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>15-ben</w><c>.</c></s>
|
||||
IN : A 15.-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>15.-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 15.-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>15.-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 2002-2003. van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2002-2003.</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2002-2003-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2002-2003-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 2002-2003.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2002-2003</w><c>.</c></s>
|
||||
IN : A 2002-2003-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>2002-2003-ben</w><c>.</c></s>
|
||||
|
||||
%-ot tartalmazo szamok
|
||||
IN : A -0,99% van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>-0,99%</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A -0,99%-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>-0,99%-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A -0,99%.
|
||||
OUT: <s><w>A</w><ws> </ws><w>-0,99%</w><c>.</c></s>
|
||||
IN : A -0,99%-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>-0,99%-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 10-20% van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-20%</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10-20%-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-20%-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10-20%.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-20%</w><c>.</c></s>
|
||||
IN : A 10-20%-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-20%-ben</w><c>.</c></s>
|
||||
|
||||
§-t tartalmazo szamok
|
||||
IN : A 99§ van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>99§</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 99§-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>99§-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 99§.
|
||||
OUT: <s><w>A</w><ws> </ws><w>99§</w><c>.</c></s>
|
||||
IN : A 99§-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>99§-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 10-20§ van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-20§</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10-20§-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-20§-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10-20§.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-20§</w><c>.</c></s>
|
||||
IN : A 10-20§-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-20§-ben</w><c>.</c></s>
|
||||
|
||||
°-t tartalmazo szamok
|
||||
IN : A 99° van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>99°</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 99°-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>99°-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 99°.
|
||||
OUT: <s><w>A</w><ws> </ws><w>99°</w><c>.</c></s>
|
||||
IN : A 99°-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>99°-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 10-20° van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-20°</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10-20°-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-20°-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 10-20°.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-20°</w><c>.</c></s>
|
||||
IN : A 10-20°-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>10-20°-ben</w><c>.</c></s>
|
||||
|
||||
(ez nem szam, de ide jon)
|
||||
IN : A °C van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>°C</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A °C-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>°C-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A °C.
|
||||
OUT: <s><w>A</w><ws> </ws><w>°C</w><c>.</c></s>
|
||||
IN : A °C-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>°C-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 100°C van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>100°C</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 100°C-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>100°C-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 100°C.
|
||||
OUT: <s><w>A</w><ws> </ws><w>100°C</w><c>.</c></s>
|
||||
IN : A 100°C-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>100°C-ben</w><c>.</c></s>
|
||||
|
||||
euroz, dollaz, yeniz
|
||||
IN : A $1,000 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>$1,000</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A $1,000-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>$1,000-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A $1,000.
|
||||
OUT: <s><w>A</w><ws> </ws><w>$1,000</w><c>.</c></s>
|
||||
IN : A $1,000-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>$1,000-ben</w><c>.</c></s>
|
||||
|
||||
IN : A €1,000 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>€1,000</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A €1,000-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>€1,000-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A €1,000.
|
||||
OUT: <s><w>A</w><ws> </ws><w>€1,000</w><c>.</c></s>
|
||||
IN : A €1,000-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>€1,000-ben</w><c>.</c></s>
|
||||
|
||||
IN : A ¥1,000 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>¥1,000</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A ¥1,000-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>¥1,000-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A ¥1,000.
|
||||
OUT: <s><w>A</w><ws> </ws><w>¥1,000</w><c>.</c></s>
|
||||
IN : A ¥1,000-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>¥1,000-ben</w><c>.</c></s>
|
||||
|
||||
IN : A £1,000 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>£1,000</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A £1,000-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>£1,000-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A £1,000.
|
||||
OUT: <s><w>A</w><ws> </ws><w>£1,000</w><c>.</c></s>
|
||||
IN : A £1,000-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>£1,000-ben</w><c>.</c></s>
|
||||
|
||||
aranyok
|
||||
IN : A 1645kJ/1000g van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1645kJ/1000g</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1645kJ/1000g-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1645kJ/1000g-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1645kJ/1000g.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1645kJ/1000g</w><c>.</c></s>
|
||||
IN : A 1645kJ/1000g-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1645kJ/1000g-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 1645 kJ/1000g van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1645</w><ws> </ws><w>kJ/1000g</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1645 kJ/1000g-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1645</w><ws> </ws><w>kJ/1000g-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1645 kJ/1000g.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1645</w><ws> </ws><w>kJ/1000g</w><c>.</c></s>
|
||||
IN : A 1645 kJ/1000g-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1645</w><ws> </ws><w>kJ/1000g-ben</w><c>.</c></s>
|
||||
|
||||
meretek
|
||||
IN : A 800x600 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>800x600</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 800x600-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>800x600-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 800x600.
|
||||
OUT: <s><w>A</w><ws> </ws><w>800x600</w><c>.</c></s>
|
||||
IN : A 800x600-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>800x600-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 1x2x3x4 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1x2x3x4</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1x2x3x4-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1x2x3x4-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 1x2x3x4.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1x2x3x4</w><c>.</c></s>
|
||||
IN : A 1x2x3x4-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>1x2x3x4-ben</w><c>.</c></s>
|
||||
|
||||
szabvany azonositok (nem torodunk veluk. annyi token, amennyi)
|
||||
IN : A MSZ EN ISO 30011-2:1993 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>MSZ</w><ws> </ws><w>EN</w><ws> </ws><w>ISO</w><ws> </ws><w>30011-2:1993</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A MSZ EN ISO 30011-2:1993-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>MSZ</w><ws> </ws><w>EN</w><ws> </ws><w>ISO</w><ws> </ws><w>30011-2:1993-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A MSZ EN ISO 30011-2:1993.
|
||||
OUT: <s><w>A</w><ws> </ws><w>MSZ</w><ws> </ws><w>EN</w><ws> </ws><w>ISO</w><ws> </ws><w>30011-2:1993</w><c>.</c></s>
|
||||
IN : A MSZ EN ISO 30011-2:1993-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>MSZ</w><ws> </ws><w>EN</w><ws> </ws><w>ISO</w><ws> </ws><w>30011-2:1993-ben</w><c>.</c></s>
|
||||
|
||||
tortszamok
|
||||
IN : A 5/4 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/4</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 5/4-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/4-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 5/4.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/4</w><c>.</c></s>
|
||||
IN : A 5/4-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/4-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 5/4. van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/4.</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 5/4.-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/4.-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 5/4.-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/4.-ben</w><c>.</c></s>
|
||||
|
||||
egyeb szamok perjellel
|
||||
IN : A 5/J van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/J</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 5/J-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/J-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 5/J.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/J</w><c>.</c></s>
|
||||
IN : A 5/J-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/J-ben</w><c>.</c></s>
|
||||
|
||||
IN : A 5/J. van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/J.</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 5/J.-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/J.-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A 5/J.-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>5/J.-ben</w><c>.</c></s>
|
||||
|
||||
romai szamok perjellel
|
||||
IN : A III/1 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>III/1</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A III/1-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>III/1-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A III/1.
|
||||
OUT: <s><w>A</w><ws> </ws><w>III/1</w><c>.</c></s>
|
||||
IN : A III/1-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>III/1-ben</w><c>.</c></s>
|
||||
|
||||
IN : A III/1. van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>III/1.</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A III/1.-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>III/1.-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A III/1.-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>III/1.-ben</w><c>.</c></s>
|
||||
|
||||
IN : A III/c van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>III/c</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A III/c-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>III/c-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A III/c.
|
||||
OUT: <s><w>A</w><ws> </ws><w>III/c</w><c>.</c></s>
|
||||
IN : A III/c-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>III/c-ben</w><c>.</c></s>
|
||||
|
||||
egyeb azonositok
|
||||
IN : A TU–154 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>TU–154</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A TU–154-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>TU–154-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A TU–154.
|
||||
OUT: <s><w>A</w><ws> </ws><w>TU–154</w><c>.</c></s>
|
||||
IN : A TU–154-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>TU–154-ben</w><c>.</c></s>
|
||||
|
||||
szamok es nagybetuk legalabb ket perjellel
|
||||
IN : A BDE/2000/01/0983/0010 van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>BDE/2000/01/0983/0010</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A BDE/2000/01/0983/0010-ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>BDE/2000/01/0983/0010-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A BDE/2000/01/0983/0010.
|
||||
OUT: <s><w>A</w><ws> </ws><w>BDE/2000/01/0983/0010</w><c>.</c></s>
|
||||
IN : A BDE/2000/01/0983/0010-ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>BDE/2000/01/0983/0010-ben</w><c>.</c></s>
|
||||
|
||||
|
||||
MINTA
|
||||
IN : A van.
|
||||
OUT: <s><w>A</w><ws> </ws><w></w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A -ben van.
|
||||
OUT: <s><w>A</w><ws> </ws><w>-ben</w><ws> </ws><w>van</w><c>.</c></s>
|
||||
IN : A .
|
||||
OUT: <s><w>A</w><ws> </ws><w></w><c>.</c></s>
|
||||
IN : A -ben.
|
||||
OUT: <s><w>A</w><ws> </ws><w>-ben</w><c>.</c></s>
|
||||
|
||||
|
|
@ -57,7 +57,8 @@ class TokenizerTestCase(object):
|
|||
_DOTS_CASES = list(TokenizerTestCase.read_from_file(_MODULE_PATH + "/test_default_token_dots.txt"))
|
||||
_HYPHEN_CASES = list(TokenizerTestCase.read_from_file(_MODULE_PATH + "/test_default_token_hyphen.txt"))
|
||||
_QUOTE_CASES = list(TokenizerTestCase.read_from_file(_MODULE_PATH + "/test_default_token_quote.txt"))
|
||||
ALL_TESTCASES = _DOTS_CASES + _HYPHEN_CASES
|
||||
_NUMBER_CASES = list(TokenizerTestCase.read_from_file(_MODULE_PATH + "/test_default_token_numbers.txt"))
|
||||
ALL_TESTCASES = _DOTS_CASES + _HYPHEN_CASES + _QUOTE_CASES # + _NUMBER_CASES
|
||||
|
||||
|
||||
@pytest.fixture(scope="session")
|
||||
|
|
@ -71,7 +72,7 @@ def hu_tokenizer(HU):
|
|||
|
||||
|
||||
@pytest.mark.parametrize(("test_case"), ALL_TESTCASES)
|
||||
def test_abbreviations(hu_tokenizer, test_case):
|
||||
def test_testcases(hu_tokenizer, test_case):
|
||||
tokens = hu_tokenizer(test_case.input)
|
||||
token_list = [token.orth_ for token in tokens if not token.is_space]
|
||||
assert test_case.expected_tokens == token_list#, "{} was erronously tokenized as {}".format(test_case, token_list)
|
||||
assert test_case.expected_tokens == token_list # , "{} was erronously tokenized as {}".format(test_case, token_list)
|
||||
|
|
|
|||
Loading…
Reference in New Issue