o Rŀg @sdZddlZddlZddlZddlmZddlmZddl m Z e dZ dedee fd d Zdedeefd d Zdedefd dZGdddZ dde deee dedefddZedkrmddlmZedSdS)aCode for dealing with assorted UniProt file formats and interacting with the UniProt database. This currently include parsers for the GAF, GPA and GPI formats from UniProt-GOA as the module Bio.UniProt.GOA. See also Bio.SwissProt and the "swiss" support in Bio.SeqIO for the legacy plain text sequence format still used in UniProt. See also Bio.SeqIO.SwissIO for the "uniprot-xml" support in Bio.SeqIO. N) HTTPResponse)Optional)urlopenz<(.+)>; rel="next"responsereturncCs6|j}d|vr|drt|d}|r|dSdS)NLink)headers _re_next_linkmatchgroup)rr r r H/var/www/html/myenv/lib/python3.10/site-packages/Bio/UniProt/__init__.py_get_next_links  rcCst|dS)Nresults)jsonloadsreaddecode)rr r r _get_results(srcCs&|j}d|vr|drt|dSdS)Nzx-total-resultsr)r int)rr r r r_get_search_result_count,s rc@sjeZdZdZdefddZdefddZdd Zde fd d Z d e dd fddZ de fddZ ddZd S)_UniProtSearchResultszA sequence over the results of a UniProt search. Do not use this class directly. Instead, use the :meth:`UniProt.search` method. rcCs\|jdusJt|j}|jt|7_t||_|WdS1s'wYdSN)next_urlr results_cacherr)selfrr r r_fetch_next_batch;s   $z'_UniProtSearchResults._fetch_next_batch first_urlcCs(||_g|_d|_|}t||_dS)Nr)rrnext_result_indexrrsearch_result_count)rrrr r r__init__Bs z_UniProtSearchResults.__init__cCs|Srr rr r r__iter__Isz_UniProtSearchResults.__iter__cCs|jS)zHReturn the total number of search results, regardless of the batch size.)r r"r r r__len__Lsz_UniProtSearchResults.__len__indexNcCs@|tt|vs J|t|jkr||t|jksdSdS)z4Fetch batches until the given index is in the cache.N)rangelenrr)rr%r r r _fetch_forPsz _UniProtSearchResults._fetch_forcCsR|jt|kr'||jz|j|j}|jd7_|WSty&twt)Nr)rr'r(r IndexError StopIteration)r next_resultr r r__next__Vs   z_UniProtSearchResults.__next__cCst|trP|t|\}}}d|krt|kr,nJd|kr+t|ks.JJ|dkrB||krA|dkrA||dn,||krO|dkrO||nt|trn|tt| t|vretd||t||j|S)NrrzIndex out of bounds.) isinstancesliceindicesr'r(rr&r)r)rr%startstopstepr r r __getitem__bs >   z!_UniProtSearchResults.__getitem__)__name__ __module__ __qualname____doc__rrstrr!r#rr$r(dictr,r3r r r rr5s rqueryfields batch_sizecCs8||dd}|rd||d<dtj|}t|S)aaSearch the UniProt database. Consider using `query syntax `_ and `query fields `_ to refine your search. See the API details `here `_. >>> from Bio import UniProt >>> from itertools import islice >>> # Get the first 10 results >>> results = UniProt.search("(organism_id:2697049) AND (reviewed:true)")[:10] :param query: The query string to search UniProt with :type query: str :param fields: The columns to retrieve in the results, defaults to all fields :type fields: List[str], optional :param batch_size: The number of results to retrieve in each batch, defaults to 500 :type batch_size: int :return: An iterator over the search results :rtype: _UniProtSearchResults r)r;sizeformat,r<z*https://rest.uniprot.org/uniprotkb/search?)joinurllibparse urlencoder)r;r<r= parametersurlr r rsearchusrG__main__) run_doctest)Nr:)r7rre urllib.parserB http.clientrtypingrurllib.requestrcompiler r8rlistr9rrrrrGr4 Bio._utilsrIr r r rs4       A  $