Search By Time
Nostalgia DataFrames can be queried using pandas-like functionality. Look at the class for all available methods.
Note that time is automatically recognized in data. When there is only 1 datetime column, it is understood that it is a moment rather than a period.
It is important to understand that by default nostalgia tries to be smart about strings, datetime and metadate, especially since it can concern moments and periods.
Let's have a brief look at metadate:
# On 6th of January 2020 from metadate import parse_date mp = parse_date("yesterday") # MetaPeriod(2020-01-05 00:00:00, 2020-01-06 00:00:00, {<Units.DAY: 4>}, ['yesterday'], en) mp.start_date # datetime.datetime(2020, 1, 5, 0, 0) mp = parse_date("november 2019") # MetaPeriod(2019-11-01 00:00:00, 2019-12-01 00:00:00, {<Units.YEAR: 9>, <Units.MONTH: 6>}, ['november', '2019'], en) mp.start_date # datetime.datetime(2019, 11, 1, 0, 0) mp = parse_date("today at 1") # MetaPeriod(2020-01-06 01:00:00, 2020-01-06 02:00:00, {<Units.HOUR: 3>, <Units.DAY: 4>}, ['today', 'at 1'], en)
You can see it has both a starting and ending date, and that it tries to be clever about finding the boundaries. It understood we are interested at the hour level with "today at 1".
Keep in mind that start and end date are important when querying.
Data
Let's consider the following example data:
In[11]: from nostalgia.sources.chrome_history import WebHistory web = WebHistory.load() web.tail(n=2) Out[11]: domain domain_and_suffix time 98477 archlinux archlinux.org 2019-12-23 01:23:38.429000+01:00 98478 stackoverflow stackoverflow.com 2019-12-23 01:27:22.652000+01:00 title 98477 [BUG FILED] pyqt5-common removed from repos; u... 98478 Stack Overflow - Where Developers Learn, Share... url 98477 https://bbs.archlinux.org/viewtopic.php?id=251356 98478 https://stackoverflow.com/ path 98477 ~/nostalgia_data/html/1577060618.4289877_https... 98478 ~/nostalgia_data/html/1577060842.6447568_https...
Time helpers
Simple examples
In [7]: last_year()
Out[7]: datetime.datetime(2019, 1, 1, 0, 0, tzinfo=
In [8]: last_month()
Out[8]: datetime.datetime(2019, 12, 1, 0, 0, tzinfo=
In [11]: last_days(0)
Out[11]: datetime.datetime(2020, 1, 6, 0, 0, tzinfo=
In [12]: last_days(1)
Out[12]: datetime.datetime(2020, 1, 5, 0, 0, tzinfo=
### ndf.last_year, ndf.last_month, ndf.last_day ```python # On 6th of January 2020 web.last_year # returns all rows in 2019 web.last_month # returns all rows of december 2019 web.yesterday
This will filter the current data by last_year, in January 2020 that means the whole of 2019.
ndf.at_time
This is the wrapper for allowing flexibility:
def at_time(self, start, end=None, sort_diff=True, **window_kwargs):
If you give start="yesterday"