def juxta_basicstats(u):
ns = u.note_owner
nvals = u.note_owner.values()
stats = (
("N", ns.count()), ## number of notes
("keep/del", "%s/%s" % (ns.filter(deleted=0).count(), ns.filter(deleted=1).count())),
("s:", exporter.makedate_usec(min([float(x["created"]) for x in nvals]))),
("l:", exporter.makedate_usec(max([float(x["edited"]) for x in nvals] + [float(x["created"]) for x in nvals]))),
("|words|", mean([ca.note_words(nvs)[1] for nvs in nvals])),
("|chars|", mean([len(nvs["contents"].strip()) for nvs in nvals])),
("|urls%|", sum([1 for nvs in nvals if ca.note_urls(nvs)[1] > 0]) / (1.0 * ns.count()) * 100),
("|#vers|", mean([float(nvs["version"]) for nvs in nvals])),
)
return '<span class="userstats">%s</span>' % "; ".join(
['%s: <span class="val">%s</span>' % (k, v) for k, v in stats]
)
def juxta_basicstats(u):
ns = u.note_owner
nvals = u.note_owner.values()
stats = (
("N", ns.count()), ## number of notes
("keep/del", "%s/%s" %
(ns.filter(deleted=0).count(), ns.filter(deleted=1).count())),
("s:", exporter.makedate_usec(min([float(x['created'])
for x in nvals]))),
("l:",
exporter.makedate_usec(
max([float(x['edited'])
for x in nvals] + [float(x['created']) for x in nvals]))),
("|words|", mean([ca.note_words(nvs)[1] for nvs in nvals])),
("|chars|", mean([len(nvs['contents'].strip()) for nvs in nvals])),
("|urls%|", sum([1 for nvs in nvals if ca.note_urls(nvs)[1] > 0]) /
(1.0 * ns.count()) * 100),
("|#vers|", mean([float(nvs['version']) for nvs in nvals])))
return '<span class="userstats">%s</span>' % "; ".join(
['%s: <span class="val">%s</span>' % (k, v) for k, v in stats])
first_word_action = lambda notevals, words: ("first_word_symbol", words[0] in actionWords)
first_words_action = lambda notevals, words: ("first_2words_action", words[0] in actionWords or (len(words) > 1 and words[1] in actionWords))#['WDT','WP','WRB','WDT','VBZ'])
first_3words_action = lambda notevals, words: ("first_3words_action", words[0] in actionWords or (len(words) > 1 and words[1] in actionWords) or (len(words) > 2 and words[2] in actionWords))
word_features = [first_word_action, first_words_action, first_word_stop, first_3words_action]
#count_verbs = lambda notevals, words: ("count_verbs", ca.note_verbs(notevals)['note_verbs'])
#count_urls = lambda notevals, words: ("count_urls", ca.note_urls(notevals)['note_urls'])
#count_numbers = lambda notevals, words: ("count_numbers", ca.numbers(notevals)['numbers'])
#count_todos = lambda notevals, words: ("count_todos", ca.note_todos(notevals)['note_todos'])
#count_names = lambda notevals, words: ("count_names", ca.note_names(notevals)["names"])
# Counts seem to be overfitting majorly, but also increasing overall accuracy...
count_features = [] ## Not using counts does better!
#count_features = [count_verbs,count_urls,count_numbers,count_todos,count_names]
contains_url = lambda notevals, words: ("contains_url", ca.note_urls(notevals)['note_urls'] > 0)
contains_verbs = lambda notevals, words: ("3+_verbs", ca.note_verbs(notevals)['note_verbs'] >= 3)
contains_dets = lambda notevals, words: ("1+_det", count_pos(words, ['DT']) > 0)
contains_3_dets = lambda notevals, words: ("3+_det", count_pos(words, ['DT']) >=3) ## 6 / 17 split
contains_adj = lambda notevals, words: ("1+_adj", count_pos(words, ['JJ', 'JJR', 'JJS']) > 0)
contains_adv = lambda notevals, words: ("1+_adv", count_pos(words, ['RB','RBR','RBS']) > 0)
#contains_linesZ = lambda notevals, words: ("contains_1_line", notevals['contents'].count('\n') == 0)
contains_linesA = lambda notevals, words: ("contains_2+_lines", notevals['contents'].count('\n') >= 1)
contains_linesB = lambda notevals, words: ("3+_lines", notevals['contents'].count('\n') >= 2)
#contains_pronoun = lambda notevals, words: ("1+_pronouns", count_pos(words, ['PRP', 'PRP$']) >= 1)
# contains_verbs .28 # contains_dets .5 # contains_3_dets .2187 # contains_adj .52 # contains_adv .49 # contains_linesB .35
# Revision 1
(len(words) > 2 and words[2] in actionWords))
word_features = [
first_word_action, first_words_action, first_word_stop, first_3words_action
]
#count_verbs = lambda notevals, words: ("count_verbs", ca.note_verbs(notevals)['note_verbs'])
#count_urls = lambda notevals, words: ("count_urls", ca.note_urls(notevals)['note_urls'])
#count_numbers = lambda notevals, words: ("count_numbers", ca.numbers(notevals)['numbers'])
#count_todos = lambda notevals, words: ("count_todos", ca.note_todos(notevals)['note_todos'])
#count_names = lambda notevals, words: ("count_names", ca.note_names(notevals)["names"])
# Counts seem to be overfitting majorly, but also increasing overall accuracy...
count_features = [] ## Not using counts does better!
#count_features = [count_verbs,count_urls,count_numbers,count_todos,count_names]
contains_url = lambda notevals, words: ("contains_url", ca.note_urls(notevals)[
'note_urls'] > 0)
contains_verbs = lambda notevals, words: ("3+_verbs", ca.note_verbs(notevals)[
'note_verbs'] >= 3)
contains_dets = lambda notevals, words: ("1+_det", count_pos(words, ['DT']) > 0
)
contains_3_dets = lambda notevals, words: ("3+_det", count_pos(words, ['DT'])
>= 3) ## 6 / 17 split
contains_adj = lambda notevals, words: (
"1+_adj", count_pos(words, ['JJ', 'JJR', 'JJS']) > 0)
contains_adv = lambda notevals, words: (
"1+_adv", count_pos(words, ['RB', 'RBR', 'RBS']) > 0)
#contains_linesZ = lambda notevals, words: ("contains_1_line", notevals['contents'].count('\n') == 0)
contains_linesA = lambda notevals, words: ("contains_2+_lines", notevals[
'contents'].count('\n') >= 1)
def one_or_no_url_redblk(note):
note = n2vals(note)
urls = ca.note_urls(note)
if type(urls) == dict:
return 'red' if urls['note_urls'] > 0 else 'black'
return 'red' if urls[1] > 0 else 'black'
def one_or_no_url_redblk(note):
note = n2vals(note)
urls = ca.note_urls(note)
if type(urls) == dict:
return 'red' if urls['note_urls'] > 0 else 'black'
return 'red' if urls[1] > 0 else 'black'