26 lines
3.7 KiB
Plaintext
26 lines
3.7 KiB
Plaintext
|
Ë
|
|||
|
|
=û gl ã óB — d dl mZmZ ddlmZ ddlmZ G d„ de« Zy)é )ÚListÚTupleé )Ú
|
|||
|
|
Lemmatizer)ÚTokenc ób ‡ — e Zd ZdZededeee ee f fˆ fd„ « Zde dee fd„Z
|
|||
|
|
ˆ xZ S )ÚCatalanLemmatizerar
|
|||
|
|
Copied from French Lemmatizer
|
|||
|
|
Catalan language lemmatizer applies the default rule based lemmatization
|
|||
|
|
procedure with some modifications for better Catalan language support.
|
|||
|
|
|
|||
|
|
The parts of speech 'ADV', 'PRON', 'DET', 'ADP' and 'AUX' are added to use
|
|||
|
|
the rule-based lemmatization. As a last resort, the lemmatizer checks in
|
|||
|
|
the lookup table.
|
|||
|
|
ÚmodeÚreturnc ó< •— |dk( rg d¢}|g fS t ‰| � |« S )NÚrule)Ú lemma_lookupÚ lemma_rulesÚ lemma_excÚ lemma_index)ÚsuperÚget_lookups_config)Úclsr
|
|||
|
|
ÚrequiredÚ __class__s €úYC:\Users\garci\AppData\Roaming\Python\Python312\site-packages\spacy/lang/ca/lemmatizer.pyr z$CatalanLemmatizer.get_lookups_config s) ø€ à �6Š>ÚRˆHØ ˜b�>Ð !ä‘7Ñ-¨dÓ3Ð 3ó Útokenc óæ — |j |j f}|| j v r| j | S |j }|j j « }|dv r|j « gS d| j vs|dvr| j |« S | j j di « }| j j di « }| j j di « }| j j di « }|j |i « } |j |i « }
|
|||
|
|
|j |g « } |j « }g } || v r"| j |« | | j |< | S | j |
|
|||
|
|
j |g « « g }
|
|||
|
|
| ss| D ]n \ }}|j |« sŒ|d t |« t |« z
|
|||
|
|
|z }|sŒ8|| v s|j « s| j |« Œ^|
|
|||
|
|
j |« Œp | s| j |
|
|||
|
|
« | s%| j |j ||g« d « t t j# | « « } | | j |< | S )N)Ú ÚeolÚspacer ) ÚnounÚverbÚadjÚadpÚadvÚauxÚcconjÚdetÚpronÚpunctÚsconjr r r r )ÚorthÚposÚcacheÚtextÚpos_ÚlowerÚlookupsÚlookup_lemmatizeÚ get_tableÚgetÚappendÚextendÚendswithÚlenÚisalphaÚlistÚdictÚfromkeys)Úselfr Ú cache_keyÚstringÚuniv_posÚ index_tableÚ exc_tableÚ rules_tableÚ lookup_tableÚindexÚ
|
|||
|
|
exceptionsÚrulesÚformsÚ oov_formsÚoldÚnewÚforms r Úrule_lemmatizez CatalanLemmatizer.rule_lemmatize s/ € Ø—Z‘Z §¡Ð+ˆ Ø ˜Ÿ
|
|||
|
|
™
|
|||
|
|
Ñ "Ø—:‘:˜iÑ(Ð (Ø—‘ˆØ—:‘:×#Ñ#Ó%ˆØ Ð+Ñ +Ø—L‘L“NÐ#Ð #Ø
|
|||
|
|
$§,¡,Ñ
|
|||
|
|
.°(ð C
|
|||
|
|
ñ 3
|
|||
|
|
ð ×(Ñ(¨Ó/Ð /Ø—l‘l×,Ñ,¨]¸BÓ?ˆ Ø—L‘L×*Ñ*¨;¸Ó;ˆ Ø—l‘l×,Ñ,¨]¸BÓ?ˆ Ø—|‘|×-Ñ-¨n¸bÓAˆ Ø—‘ ¨"Ó-ˆØ —]‘] 8¨RÓ0ˆ
|