def __init__(self):
self.client = Client()
class VUNLPParser(AnalysisScript):
"""Analysisscript subclass for parsers bound to a specific ('home') folder"""
def __init__(self):
self.client = Client()
def submit_article(self, article):
plugin = self.get_plugin()
handle = self.client.upload(self.parse_command, self._get_text_to_submit(article))
log.info("Submitted article {article.id}, handle {handle}".format(**locals()))
if isinstance(article, AnalysedArticle):
# (re)set done, error, info
article.done = False
article.error = False
article.info = handle
article.save()
else:
# Create AnalysedArticle object
article = AnalysedArticle.objects.create(article=article, plugin=plugin,
done=False, error=False, info=handle)
return article
def _get_sentences(self, article):
"""
Return the sentences in the article, creating them if needed.
@param article: an amcat.Article or AnalysedArticle model instance.
@return: a sequence of Sentence model instances
"""
if isinstance(article, AnalysedArticle):
article = article.article
return sbd.get_or_create_sentences(article)
def _get_text_to_submit(self, article):
"""
Return the text to be submitted to the VUNLP web service
@param article: an amcat.Article or AnalysedArticle model instance.
"""
return "\n".join(s.sentence for s in self._get_sentences(article))
def _do_retrieve_article(self, analysed_article):
if self.check_article(analysed_article):
parse = Client().download(analysed_article.info)
open("/tmp/aa_%i.xml" % analysed_article.id, "w").write(parse)
if len(parse) == 0:
raise Exception("Empty article: downloading %i / %s yielded %r"
% (analysed_article.id, analysed_article.info, parse))
self.store_parse(analysed_article, parse)
log.info("Stored article {analysed_article.id}".format(**locals()))
return True
def _do_check_article(self, analysed_article):
status = Client().check(analysed_article.info)
log.info("Article {analysed_article.id} has parse status {status}".format(**locals()))
if status == "unknown":
raise Exception("Article {analysed_article.id} has status 'unknown'"
.format(**locals()))
return (status == "ready")
