def test_dateoutput(self):
for date, iso, isotime, yw, ym, yq in ((datetime.datetime(
1990, 1, 10, 13, 1, 0), "1990-01-10", "1990-01-10 13:01:00",
1990.02, 1990.01, 1990.1), ):
self.assertEqual(toolkit.writeDate(date), iso)
self.assertEqual(toolkit.writeDateTime(date), isotime)
self.assertEqual(toolkit.getYW(date), yw)
self.assertEqual(toolkit.getYM(date), ym)
self.assertEqual(toolkit.getYQ(date), yq)
def createArticle(db,
headline,
date,
source,
batchid,
text,
texttype=2,
length=None,
byline=None,
section=None,
pagenr=None,
fullmeta=None,
url=None,
externalid=None,
parentUrl=None,
retrieveArticle=1):
"""
Writes the article object to the database
"""
# TODO link to parent if parentUrl is not None
if toolkit.isDate(date): date = toolkit.writeDateTime(date, 1)
if type(source) != int: source = source.id
if type(fullmeta) == dict: fullmeta = ` fullmeta `
if url and len(url) > 490: url = url[:490] + "..."
(headline, byline, fullmeta, section), encoding = encodeAndLimitLength(
[headline, byline, fullmeta, section], [740, 999999, 999999, 90])
if pagenr and type(pagenr) in (types.StringTypes): pagenr = pagenr.strip()
if text: text = text.strip()
if length == None and text: length = len(text.split())
q = {
'date': date,
'length': length,
'metastring': fullmeta,
'headline': headline,
'byline': byline,
'section': section,
'pagenr': pagenr,
'batchid': batchid,
'mediumid': source,
'url': url,
'externalid': externalid,
'encoding': encoding,
# We don't store the parentUrl. Instead, we use the articles_postsings
# table to store this information. This is done by the scraper class.
}
aid = db.insert('articles', q)
text, encoding = dbtoolkit.encodeText(text)
q = {
'articleid': aid,
'type': texttype,
'encoding': encoding,
'text': text
}
db.insert('texts', q, retrieveIdent=0)
if retrieveArticle:
return article.Article(db, aid)
def _default_json(obj):
if not isinstance(obj, datetime.datetime): return unicode(obj)
return writeDateTime(obj, year=True, seconds=False, time=False)
def test_dateoutput(self):
for date, iso, isotime, yw, ym, yq in (
(datetime.datetime(1990, 1, 10, 13,1,0), "1990-01-10", "1990-01-10 13:01:00", 1990.02, 1990.01, 1990.1),
):
self.assertEqual(toolkit.writeDate(date), iso)
self.assertEqual(toolkit.writeDateTime(date), isotime)
