68 lines
2.7 KiB
Python
68 lines
2.7 KiB
Python
from collections import defaultdict
|
|
from dataclasses import dataclass, field
|
|
from datetime import datetime
|
|
from operator import itemgetter
|
|
from typing import Any, Dict, List
|
|
|
|
from stats_model import StatsModel
|
|
|
|
|
|
@dataclass(kw_only=True)
|
|
class BuildDeepStatsModel(StatsModel):
|
|
"""Stats model built around calculating stats for original post notes, with additional data dimensions."""
|
|
operation: str = 'build_deep_stats'
|
|
|
|
# Note count by interaction month and year for original posts.
|
|
notes_by_interaction_month_and_year: Dict[str, Any] = field(init=False)
|
|
|
|
# Top blogs by original post note count.
|
|
top_interactors: List[Dict[str, Any]] = field(init=False)
|
|
|
|
def __post_init__(self):
|
|
super().__post_init__()
|
|
self.notes_by_interaction_month_and_year = self.calculate_notes_by_interaction_month_and_year()
|
|
self.top_interactors = self.calculate_top_interactors()
|
|
|
|
def calculate_notes_by_interaction_month_and_year(self) -> Dict[str, Any]:
|
|
# https://docs.python.org/3/library/collections.html#defaultdict-objects
|
|
date_map: Dict[str, Any] = {}
|
|
date_map = defaultdict(lambda: {'note_count': 0},
|
|
date_map)
|
|
|
|
# Gathering the results.
|
|
for post_key in self.original_post_map:
|
|
post = self.original_post_map[post_key]
|
|
if 'notes' not in post:
|
|
continue
|
|
notes = post['notes']
|
|
for note in notes:
|
|
note_date_key = datetime.fromtimestamp(note['timestamp']).strftime('%Y-%m-%d')
|
|
sts = date_map[note_date_key]
|
|
sts['date'] = note_date_key
|
|
if note['type'] != 'posted':
|
|
sts['note_count'] += 1
|
|
|
|
return date_map
|
|
|
|
def calculate_top_interactors(self) -> List[Dict[str, Any]]:
|
|
top_interactors: defaultdict[str, Dict[str, Any]] = defaultdict(lambda: {'note_count': 0},
|
|
{})
|
|
|
|
# Gathering the results.
|
|
for post_key in self.original_post_map:
|
|
post = self.original_post_map[post_key]
|
|
if 'notes' not in post:
|
|
continue
|
|
notes = post['notes']
|
|
for note in notes:
|
|
person_key = note['blog_name']
|
|
sts = top_interactors[person_key]
|
|
if 'blog_name' not in sts: sts['blog_name'] = person_key
|
|
if 'url' not in sts: sts['url'] = f"https://{person_key}.tumblr.com"
|
|
if note['type'] != 'posted':
|
|
sts['note_count'] += 1
|
|
|
|
# https://stackoverflow.com/a/73050
|
|
return sorted(top_interactors.values(), key=itemgetter('note_count'),
|
|
reverse=True)
|