Spaces:
Sleeping
Sleeping
Refactor and error handling
Browse files- eventbrite_summarizer.py +41 -14
eventbrite_summarizer.py
CHANGED
@@ -22,13 +22,19 @@ def get_title_subtitle_from_event(event_url):
|
|
22 |
tag_classes = {'title': 'event-title',
|
23 |
'subtitle': 'summary',
|
24 |
'details': 'has-user-generated-content'}
|
25 |
-
|
26 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
return title, subtitle
|
28 |
|
29 |
-
def
|
30 |
-
# Now get details:
|
31 |
-
|
32 |
event_id = get_event_id(event_url)
|
33 |
headers = {'Authorization': 'Bearer {}'.format(EVENTBRITE_API_KEY)}
|
34 |
params = {}
|
@@ -39,9 +45,12 @@ def get_event_details(event_url):
|
|
39 |
effective_encoding = 'utf-8-sig' #r.apparent_encoding #
|
40 |
r.encoding = effective_encoding
|
41 |
|
42 |
-
|
43 |
|
44 |
-
|
|
|
|
|
|
|
45 |
content = content.replace('\ufeff', '') # Remove byte order mark for utf-8-sig decoding.
|
46 |
|
47 |
soup = BeautifulSoup(content, 'html.parser')
|
@@ -49,13 +58,31 @@ def get_event_details(event_url):
|
|
49 |
details = soup.get_text(separator=' ')
|
50 |
return details
|
51 |
|
52 |
-
def get_eventbrite_summary(event_url, top_level_prompt_stub):
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
59 |
|
60 |
top_level_prompt = top_level_prompt_stub + """
|
61 |
Event Title:{title}
|
|
|
22 |
tag_classes = {'title': 'event-title',
|
23 |
'subtitle': 'summary',
|
24 |
'details': 'has-user-generated-content'}
|
25 |
+
try:
|
26 |
+
title = soup.find(class_ = tag_classes['title']).text
|
27 |
+
except AttributeError as e:
|
28 |
+
raise AttributeError("No event title found on page; has it expired?")
|
29 |
+
|
30 |
+
try:
|
31 |
+
subtitle = soup.find(class_ = tag_classes['subtitle']).text
|
32 |
+
except AttributeError as e:
|
33 |
+
raise AttributeError("No subtitle found on page; has it expired?")
|
34 |
+
|
35 |
return title, subtitle
|
36 |
|
37 |
+
def get_eventbrite_json(event_url):
|
|
|
|
|
38 |
event_id = get_event_id(event_url)
|
39 |
headers = {'Authorization': 'Bearer {}'.format(EVENTBRITE_API_KEY)}
|
40 |
params = {}
|
|
|
45 |
effective_encoding = 'utf-8-sig' #r.apparent_encoding #
|
46 |
r.encoding = effective_encoding
|
47 |
|
48 |
+
return json.loads(r.text)
|
49 |
|
50 |
+
def get_event_details(event_json):
|
51 |
+
# Now get details:
|
52 |
+
|
53 |
+
content = event_json['modules'][0]['data']['body']['text']
|
54 |
content = content.replace('\ufeff', '') # Remove byte order mark for utf-8-sig decoding.
|
55 |
|
56 |
soup = BeautifulSoup(content, 'html.parser')
|
|
|
58 |
details = soup.get_text(separator=' ')
|
59 |
return details
|
60 |
|
61 |
+
def get_eventbrite_summary(event_url, top_level_prompt_stub, error_handling='attempt', modeltype="text-davinci-003", temperature=0.3):
|
62 |
+
"""
|
63 |
+
error_handling in "skip", "attempt",
|
64 |
+
"""
|
65 |
+
errors = []
|
66 |
+
|
67 |
+
try:
|
68 |
+
title, subtitle = get_title_subtitle_from_event(event_url)
|
69 |
+
except AttributeError as e:
|
70 |
+
errors.append(e)
|
71 |
+
print(e)
|
72 |
+
title, subtitle = '', ''
|
73 |
+
|
74 |
+
try:
|
75 |
+
event_json = get_eventbrite_json(event_url)
|
76 |
+
details = get_event_details(event_json)
|
77 |
+
except AttributeError as e:
|
78 |
+
errors.append(e)
|
79 |
+
print(e)
|
80 |
+
details = ''
|
81 |
+
|
82 |
+
if errors and error_handling=='skip':
|
83 |
+
return 'Error'
|
84 |
+
|
85 |
+
openai_modeltype = modeltype
|
86 |
|
87 |
top_level_prompt = top_level_prompt_stub + """
|
88 |
Event Title:{title}
|