From f84ac489752d306b545d18bf8690100e8a99ae63 Mon Sep 17 00:00:00 2001 From: Patrick Jentsch Date: Thu, 22 Jun 2023 14:19:14 +0200 Subject: [PATCH] Add test snippet for fast cpos boundary calculation for s_attrs --- .../cqi_over_socketio/cqi_corpora_corpus.py | 15 +++++++++++++++ 1 file changed, 15 insertions(+) diff --git a/app/corpora/cqi_over_socketio/cqi_corpora_corpus.py b/app/corpora/cqi_over_socketio/cqi_corpora_corpus.py index d3503b66..c5125da2 100644 --- a/app/corpora/cqi_over_socketio/cqi_corpora_corpus.py +++ b/app/corpora/cqi_over_socketio/cqi_corpora_corpus.py @@ -50,6 +50,21 @@ def cqi_corpora_corpus_update_db(cqi_client: cqi.CQiClient, corpus_name: str): @cqi_over_socketio def cqi_corpora_corpus_get_visualization_data(cqi_client: cqi.CQiClient, corpus_name: str): cqi_corpus = cqi_client.corpora.get(corpus_name) + ########################################################################## + # A faster way to get cpos boundaries for smaller s_attrs # + ########################################################################## + # cqi_corpus.query('Last', ' []* ;') + # cqi_subcorpus = cqi_corpus.subcorpora.get('Last') + # print(cqi_subcorpus.size) + # first_match = 0 + # last_match = cqi_subcorpus.attrs['size'] - 1 + # match_boundaries = zip( + # list(range(first_match, last_match + 1)), + # cqi_subcorpus.dump(cqi_subcorpus.attrs['fields']['match'], first_match, last_match), + # cqi_subcorpus.dump(cqi_subcorpus.attrs['fields']['matchend'], first_match, last_match) + # ) + # for x in match_boundaries: + # print(x) cqi_p_attrs = { p_attr.name: p_attr for p_attr in cqi_corpus.positional_attributes.list()