75 lines
8.5 KiB
Plaintext
75 lines
8.5 KiB
Plaintext
|
Ë
|
|||
|
|
>û g ã ó¬ — d dl mZmZmZmZmZ d dlmZ ddlm Z ddl
|
|||
|
|
m Z ddl m
|
|||
|
|
Z
|
|||
|
|
ddlmZ ddlmZ d d d
|
|||
|
|
œZ G d „ d e « Zd
|
|||
|
|
edeeeeef f fd„Zy)é )ÚCallableÚDictÚListÚOptionalÚTuple)ÚModelé )Ú
|
|||
|
|
Lemmatizer)Úlemmatizer_score)ÚPOS)ÚToken)ÚVocabú")õ «õ »c óÚ ‡ — e Zd Z dddedœdedee dededed ee d
|
|||
|
|
d fˆ fd „Z
|
|||
|
|
d
|
|||
|
|
e d
|
|||
|
|
e e fd„Z
|
|||
|
|
d
|
|||
|
|
e d
|
|||
|
|
e e fd„Zd
|
|||
|
|
e d
|
|||
|
|
e e fd„Zd
|
|||
|
|
e d
|
|||
|
|
e e fd„Zd
|
|||
|
|
e d
|
|||
|
|
e e fd„Zd
|
|||
|
|
e d
|
|||
|
|
e e fd„Zˆ xZS )ÚRussianLemmatizerÚ pymorphy3F©ÚmodeÚ overwriteÚscorerÚvocabÚmodelÚnamer r r ÚreturnNc ó. •— |dv r# ddl m} t | dd « €5 |d¬« | _ n&|dv r" ddlm} t | dd « € |d¬« | _ t ‰| � ||||||¬
|
|||
|
|
« y # t $ r
|
|||
|
|
t d« d ‚w xY w# t $ r
|
|||
|
|
t d « d ‚w xY w) N> Ú pymorphy2Úpymorphy2_lookupr )Ú
|
|||
|
|
MorphAnalyzerz´The lemmatizer mode 'pymorphy2' requires the pymorphy2 library and dictionaries. Install them with: pip install pymorphy2# for Ukrainian dictionaries:pip install pymorphy2-dicts-ukÚ_morphÚru)Úlang> r Úpymorphy3_lookupz´The lemmatizer mode 'pymorphy3' requires the pymorphy3 library and dictionaries. Install them with: pip install pymorphy3# for Ukrainian dictionaries:pip install pymorphy3-dicts-ukr )r r Ú ImportErrorÚgetattrr! r ÚsuperÚ__init__) Úselfr r r r r r r Ú __class__s €úYC:\Users\garci\AppData\Roaming\Python\Python312\site-packages\spacy/lang/ru/lemmatizer.pyr( zRussianLemmatizer.__init__ sÏ ø€ ð Ð4Ñ 4ð
|
|||
|
|
Ý3ô �t˜X tÓ,Ð4Ù +°Ô 6�• Ø
|
|||
|
|
Ð6Ñ
|
|||
|
|
6ð
|
|||
|
|
Ý3ô �t˜X tÓ,Ð4Ù +°Ô 6�” Ü
|
|||
|
|
‰ÑØ �5˜$ T°YÀvð õ
|
|||
|
|
øô/ ò
|
|||
|
|
Ü!ð5óð ð
|
|||
|
|
ð
|
|||
|
|
ûô ò
|
|||
|
|
Ü!ð5óð ð
|
|||
|
|
ð
|
|||
|
|
ús ‡A% ®A> Á%A;Á>BÚtokenc ó| — |j }|j }|j j « }|dk( rt j ||« gS |dvr| j
|
|||
|
|
|« S | j j |« }g }|D ]Z }|j sŒt t |j « « \ }} ||k( s|dv r|dv s |dk( sŒD|dk( sŒJ|j |« Œ\ t |« s|j « gS |�t |« dk( r>t |v r6t# t$ j' |D �cg c] }|j( ‘Œ c}« « S |dv rg d¢}
|
|||
|
|
n|d k( rd
|
|||
|
|
d g}
|
|||
|
|
n|dk( rg d ¢}
|
|||
|
|
ng d
|
|||
|
|
¢}
|
|||
|
|
|g }}|D ]n }t t |j « « \ } } |
|
|||
|
|
D ]5 } | |v sŒ| | v sŒ
|
|||
|
|
|| j « | | j « k7 sŒ5 Œ] |j |« Œp t |« s|j « gS t# t$ j' |D �cg c] }|j( ‘Œ c}« « S c c}w c c}w )NÚPUNCT)ÚADJÚDETÚNOUNÚNUMÚPRONÚPROPNÚVERB)r1 r4 r3 r0 é )r/ r0 r1 r4 )ÚCaseÚNumberÚGenderr2 r7 r9 )r7 r8 r9 ÚPerson)ÚAspectr9 ÚMoodr8 ÚTenseÚVerbFormÚVoice)ÚtextÚpos_ÚmorphÚto_dictÚ PUNCT_RULESÚgetÚ_pymorphy_lookup_lemmatizer! ÚparseÚis_knownÚoc2udÚstrÚtagÚappendÚlenÚlowerr ÚlistÚdictÚfromkeysÚ normal_form)
|
|||
|
|
r) r, ÚstringÚuniv_posÚ
|
|||
|
|
morphologyÚanalysesÚfiltered_analysesÚanalysisÚ analysis_posÚ_Úfeatures_to_compareÚanalysis_morphÚfeatures
|
|||
|
|
r+ Ú_pymorphy_lemmatizez%RussianLemmatizer._pymorphy_lemmatize7 sN € Ø—‘ˆØ—:‘:ˆØ—[‘[×(Ñ(Ó*ˆ
|
|||
|
|
Ø �wÒ Ü —O‘O F¨FÓ3Ð4Ð 4Ø ÐQÑ QØ×2Ñ2°5Ó9Ð 9Ø—;‘;×$Ñ$ VÓ,ˆØ ÐÛ ˆHØ×$Ò$àÜ #¤C¨¯ © Ó$5Ó 6‰OˆL˜!à Ò(Ø Ð$5Ñ5¸(ÐFWÑ:WØ! VÓ+°(¸eÓ2Cà!×(Ñ(¨Õ2ð !ô Ð$Ô%Ø—L‘L“NÐ#Ð #Ø Ð ¤# j£/°QÒ"6¼3À*Ñ;LÜÜ—
|
|||
|
|
‘
|
|||
|
|
ÑDUÓ VÑDU¸˜x× 3Ó 3ÐDUÑ VÓWóð
|
|||
|
|
ð Ð6Ñ 6Ú">Ñ Ø
|
|||
|
|
˜Ò
|
|||
|
|
Ø#)¨8Ð"4Ñ Ø
|
|||
|
|
˜Ò
|
|||
|
|
Ú"HÑ ò#Ð ð '8¸Ð#ˆÛ ˆHÜ %¤c¨(¯,©,Ó&7Ó 8Ñ ˆAˆ~Û.�à˜zÒ)Ø >Ò1Ø" 7Ñ+×1Ñ1Ó3°~ÀgÑ7N×7TÑ7TÓ7VÓVáð
|
|||
|
|
/ð "×(Ñ(¨Õ2ð !ô Ð$Ô%Ø—L‘L“NÐ#Ð #ÜÜ �M‰MÑ@QÓRÑ@Q°H˜8×/Ó/Ð@QÑRÓ Só
|
|||
|
|
ð
|
|||
|
|
ùòA WùòB Ss Ä,H4 ÈH9 c óì — |j }| j j |« }t |D �cg c] }|j ‘Œ c}« }t |« dk( rt
|
|||
|
|
t |« « gS |gS c c}w )Nr6 )r@ r! rG ÚsetrR rM ÚnextÚiter)r) r, rS rV ÚanÚ normal_formss r+ rF z,RussianLemmatizer._pymorphy_lookup_lemmatizet sj € Ø—‘ˆØ—;‘;×$Ñ$ VÓ,ˆô ±XÓ>±X¨r˜BŸN›N°XÑ>Ó?ˆ Ü ˆ|Ó Ò !Üœ˜lÓ+Ó,Ð-Ð -؈xˆùò ?s ±A1c ó$ — | j |« S ©N©r^ ©r) r, s r+ Úpymorphy2_lemmatizez%RussianLemmatizer.pymorphy2_lemmatize~ ó € Ø×'Ñ'¨Ó.Ð.ó c ó$ — | j |« S rf ©rF rh s r+ Úpymorphy2_lookup_lemmatizez,RussianLemmatizer.pymorphy2_lookup_lemmatize� ó € Ø×.Ñ.¨uÓ5Ð5rk c ó$ — | j |« S rf rg rh s r+ Úpymorphy3_lemmatizez%RussianLemmatizer.pymorphy3_lemmatize„ rj rk c ó$ — | j |« S rf rm rh s r+ Úpymorphy3_lookup_lemmatizez,RussianLemmatizer.pymorphy3_lookup_lemmatize‡ ro rk )Ú
|
|||
|
|
lemmatizer)Ú__name__Ú
|
|||
|
|
__module__Ú __qualname__r r r r rJ Úboolr r( r
|
|||
|
|
r r^ rF ri rn rq rs Ú
|
|||
|
|
__classcell__)r* s @r+ r r sö ø„ ð
|
|||
|
|
!ð &
|
|||
|
|
ð Ø Ø%5ò&
|
|||
|
|
àð&
|
|||
|
|
𠘉ð&
|
|||
|
|
ð ð &
|
|||
|
|
ð ð
|
|||
|
|
&
|
|||
|
|
ð ð&
|