python/jc
1
0
Fork 0
mirror of https://github.com/kellyjonbrazil/jc.git synced 2025-06-17 00:07:37 +02:00
Code Issues Releases Activity

rewrite of date parser using datetime library

Browse Source
This commit is contained in:
Kelly Brazil
2021-03-24 20:20:53 -07:00
parent b3eb064b67
commit 5936940532
7 changed files with 135 additions and 185 deletions
Download Patch File Download Diff File Expand all files Collapse all files

74
docs/parsers/date.md
View File

@ -28,32 +28,24 @@ Examples:
$ date | jc --date -p $ date | jc --date -p
{ {
"year": 2021, "year": 2021,
"month": "Mar",
"month_num": 3, "month_num": 3,
"day": 23, "day": 25,
"hour": 8, "weekday": "Thu",
"hour_24": 20, "weekday_num": 4,
"minute": 45, "hour": 2,
"second": 29, "hour_24": 2,
"period": "PM", "minute": 2,
"month": "Mar", "second": 26,
"weekday": "Tue", "period": "AM",
"weekday_num": 2,
"timezone": "UTC", "timezone": "UTC",
"epoch": 1616557529, "utc_offset": "+0000",
"epoch_utc": 1616532329 "day_of_year": 84,
} "week_of_year": 12,
"iso": "2021-03-25T02:02:26+00:00",
$ date | jc --date -p -r "epoch": 1616662946,
{ "epoch_utc": 1616637746,
"year": "2021", "timezone_aware": true
"month": "Mar",
"day": "23",
"weekday": "Tue",
"hour": "08",
"minute": "45",
"second": "29",
"period": "PM",
"timezone": "UTC"
} }
@ -77,22 +69,26 @@ Parameters:
Returns: Returns:
Dictionary. Structured data with the following schema: Dictionary. Structured data with the following schema:
{ {
"year": integer, "year": integer,
"month_num": integer, "month": string,
"day": integer, "month_num": integer,
"hour": integer, # originally parsed hour "day": integer,
"hour_24": integer, # parsed hour converted to 24-hour value "weekday": string,
"minute": integer, "weekday_num": integer,
"second": integer, "hour": integer,
"period": string, # 'AM' or 'PM'. null if 24-hour output "hour_24": integer,
"month": string, "minute": integer,
"weekday": string, "second": integer,
"weekday_num": integer, "period": string,
"timezone": string, "timezone": string,
"epoch": integer, # naive timestamp "utc_offset": string, # null if timezone field is not UTC
"epoch_utc": integer, # timezone-aware timestamp. Only available if timezone field is UTC "day_of_year": integer,
"week_of_year": integer,
"iso": string,
"epoch": integer, # naive timestamp
"epoch_utc": integer, # timezone-aware timestamp. Only available if timezone field is UTC
"timezone_aware": boolean # if true, all fields are correctly based on UTC
} }

236
jc/parsers/date.py
View File

@ -26,32 +26,24 @@ Examples:
$ date | jc --date -p $ date | jc --date -p
{ {
"year": 2021, "year": 2021,
"month": "Mar",
"month_num": 3, "month_num": 3,
"day": 23, "day": 25,
"hour": 8, "weekday": "Thu",
"hour_24": 20, "weekday_num": 4,
"minute": 45, "hour": 2,
"second": 29, "hour_24": 2,
"period": "PM", "minute": 2,
"month": "Mar", "second": 26,
"weekday": "Tue", "period": "AM",
"weekday_num": 2,
"timezone": "UTC", "timezone": "UTC",
"epoch": 1616557529, "utc_offset": "+0000",
"epoch_utc": 1616532329 "day_of_year": 84,
} "week_of_year": 12,
"iso": "2021-03-25T02:02:26+00:00",
$ date | jc --date -p -r "epoch": 1616662946,
{ "epoch_utc": 1616637746,
"year": "2021", "timezone_aware": true
"month": "Mar",
"day": "23",
"weekday": "Tue",
"hour": "08",
"minute": "45",
"second": "29",
"period": "PM",
"timezone": "UTC"
} }
""" """
from datetime import datetime, timezone from datetime import datetime, timezone
@ -59,7 +51,7 @@ import jc.utils
class info(): class info():
version = '1.2' version = '2.0'
description = 'date command parser' description = 'date command parser'
author = 'Kelly Brazil' author = 'Kelly Brazil'
author_email = 'kellyjonbrazil@gmail.com' author_email = 'kellyjonbrazil@gmail.com'
@ -83,98 +75,30 @@ def process(proc_data):
Returns: Returns:
Dictionary. Structured data with the following schema: Dictionary. Structured data with the following schema:
{ {
"year": integer, "year": integer,
"month_num": integer, "month": string,
"day": integer, "month_num": integer,
"hour": integer, # originally parsed hour "day": integer,
"hour_24": integer, # parsed hour converted to 24-hour value "weekday": string,
"minute": integer, "weekday_num": integer,
"second": integer, "hour": integer,
"period": string, # 'AM' or 'PM'. null if 24-hour output "hour_24": integer,
"month": string, "minute": integer,
"weekday": string, "second": integer,
"weekday_num": integer, "period": string,
"timezone": string, "timezone": string,
"epoch": integer, # naive timestamp "utc_offset": string, # null if timezone field is not UTC
"epoch_utc": integer, # timezone-aware timestamp. Only available if timezone field is UTC "day_of_year": integer,
"week_of_year": integer,
"iso": string,
"epoch": integer, # naive timestamp
"epoch_utc": integer, # timezone-aware timestamp. Only available if timezone field is UTC
"timezone_aware": boolean # if true, all fields are correctly based on UTC
} }
""" """
# ISO 8601 month numberings # no further processing
month_map = { return proc_data
"Jan": 1,
"Feb": 2,
"Mar": 3,
"Apr": 4,
"May": 5,
"Jun": 6,
"Jul": 7,
"Aug": 8,
"Sep": 9,
"Oct": 10,
"Nov": 11,
"Dec": 12
}
# ISO 8601 weekday numberings
weekday_map = {
"Mon": 1,
"Tue": 2,
"Wed": 3,
"Thu": 4,
"Fri": 5,
"Sat": 6,
"Sun": 7
}
if proc_data:
dt_year = int(proc_data['year'])
dt_month = month_map[proc_data['month']]
dt_day = int(proc_data['day'])
dt_hour = int(proc_data['hour'])
dt_hour_24 = int(proc_data['hour'])
dt_minute = int(proc_data['minute'])
dt_second = int(proc_data['second'])
# fix for 12 vs. 24 hour output
if 'period' in proc_data:
if proc_data['period']:
if proc_data['period'].lower() == 'pm':
dt_hour_24 = dt_hour + 12
if dt_hour_24 > 23:
dt_hour_24 = 12
if proc_data['period'].lower() == 'am':
if dt_hour_24 == 12:
dt_hour_24 = 0
epoch_dt = datetime(dt_year, dt_month, dt_day, hour=dt_hour_24, minute=dt_minute, second=dt_second)
date_obj = {
'year': dt_year,
'month_num': dt_month,
'day': dt_day,
'hour': dt_hour,
'hour_24': dt_hour_24,
'minute': dt_minute,
'second': dt_second,
'period': proc_data['period'].upper() if 'period' in proc_data else None,
'month': proc_data['month'],
'weekday': proc_data['weekday'],
'weekday_num': weekday_map[proc_data['weekday']],
'timezone': proc_data['timezone'],
'epoch': int(epoch_dt.timestamp())
}
# create aware datetime object only if the timezone is UTC
if proc_data['timezone'] == 'UTC':
utc_epoch_dt = datetime(dt_year, dt_month, dt_day, hour=dt_hour_24, minute=dt_minute, second=dt_second, tzinfo=timezone.utc)
date_obj['epoch_utc'] = int(utc_epoch_dt.timestamp())
return date_obj
else:
return {}
def parse(data, raw=False, quiet=False): def parse(data, raw=False, quiet=False):
@ -197,34 +121,64 @@ def parse(data, raw=False, quiet=False):
raw_output = {} raw_output = {}
if jc.utils.has_data(data): if jc.utils.has_data(data):
data = data.replace(':', ' ')
split_data = data.split()
# date v8.32 uses a different format depending on locale, so need to support LANG=en_US.UTF-8 # find the timezone no matter where it is in the string
if len(split_data) == 9 and ('AM' in split_data or 'am' in split_data or 'PM' in split_data or 'pm' in split_data): # from https://www.timeanddate.com/time/zones/
raw_output = { tz_abbr = ['A', 'ACDT', 'ACST', 'ACT', 'ACWST', 'ADT', 'AEDT', 'AEST', 'AET', 'AFT', 'AKDT', 'AKST', 'ALMT',
"year": split_data[8], 'AMST', 'AMT', 'ANAST', 'ANAT', 'AQTT', 'ART', 'AST', 'AT', 'AWDT', 'AWST', 'AZOST', 'AZOT',
"month": split_data[1], 'AZST', 'AZT', 'AoE', 'B', 'BNT', 'BOT', 'BRST', 'BRT', 'BST', 'BTT', 'C', 'CAST', 'CAT', 'CCT',
"day": split_data[2], 'CDT', 'CEST', 'CET', 'CHADT', 'CHAST', 'CHOST', 'CHOT', 'CHUT', 'CIDST', 'CIST', 'CKT', 'CLST',
"weekday": split_data[0], 'CLT', 'COT', 'CST', 'CT', 'CVT', 'CXT', 'ChST', 'D', 'DAVT', 'DDUT', 'E', 'EASST', 'EAST',
"hour": split_data[3], 'EAT', 'ECT', 'EDT', 'EEST', 'EET', 'EGST', 'EGT', 'EST', 'ET', 'F', 'FET', 'FJST', 'FJT', 'FKST',
"minute": split_data[4], 'FKT', 'FNT', 'G', 'GALT', 'GAMT', 'GET', 'GFT', 'GILT', 'GMT', 'GST', 'GYT', 'H', 'HDT', 'HKT',
"second": split_data[5], 'HOVST', 'HOVT', 'HST', 'I', 'ICT', 'IDT', 'IOT', 'IRDT', 'IRKST', 'IRKT', 'IRST', 'IST', 'JST',
"period": split_data[6], 'K', 'KGT', 'KOST', 'KRAST', 'KRAT', 'KST', 'KUYT', 'L', 'LHDT', 'LHST', 'LINT', 'M', 'MAGST',
"timezone": split_data[7] 'MAGT', 'MART', 'MAWT', 'MDT', 'MHT', 'MMT', 'MSD', 'MSK', 'MST', 'MT', 'MUT', 'MVT', 'MYT', 'N',
} 'NCT', 'NDT', 'NFDT', 'NFT', 'NOVST', 'NOVT', 'NPT', 'NRT', 'NST', 'NUT', 'NZDT', 'NZST', 'O',
else: 'OMSST', 'OMST', 'ORAT', 'P', 'PDT', 'PET', 'PETST', 'PETT', 'PGT', 'PHOT', 'PHT', 'PKT', 'PMDT',
# standard LANG=C date output 'PMST', 'PONT', 'PST', 'PT', 'PWT', 'PYST', 'PYT', 'Q', 'QYZT', 'R', 'RET', 'ROTT', 'S', 'SAKT',
raw_output = { 'SAMT', 'SAST', 'SBT', 'SCT', 'SGT', 'SRET', 'SRT', 'SST', 'SYOT', 'T', 'TAHT', 'TFT', 'TJT', 'TKT',
"year": split_data[7], 'TLT', 'TMT', 'TOST', 'TOT', 'TRT', 'TVT', 'U', 'ULAST', 'ULAT', 'UYST', 'UYT', 'UZT', 'V', 'VET',
"month": split_data[1], 'VLAST', 'VLAT', 'VOST', 'VUT', 'W', 'WAKT', 'WARST', 'WAST', 'WAT', 'WEST', 'WET', 'WFT', 'WGST',
"day": split_data[2], 'WGT', 'WIB', 'WIT', 'WITA', 'WST', 'WT', 'X', 'Y', 'YAKST', 'YAKT', 'YAPT', 'YEKST', 'YEKT', 'Z',
"weekday": split_data[0], 'UTC']
"hour": split_data[3], tz = None
"minute": split_data[4], for term in data.split():
"second": split_data[5], if term in tz_abbr:
"timezone": split_data[6] tz = term
}
dt = None
dt_utc = None
timestamp = jc.utils.parse_datetime_to_timestamp(data)
if timestamp:
dt = datetime.fromtimestamp(timestamp['timestamp_naive'])
if timestamp['timestamp_utc']:
dt_utc = datetime.fromtimestamp(timestamp['timestamp_utc'], timezone.utc)
if dt_utc:
dt = dt_utc
raw_output = {
'year': dt.year,
'month': dt.strftime('%b'),
'month_num': dt.month,
'day': dt.day,
'weekday': dt.strftime('%a'),
'weekday_num': dt.isoweekday(),
'hour': int(dt.strftime('%I')),
'hour_24': dt.hour,
'minute': dt.minute,
'second': dt.second,
'period': dt.strftime('%p'),
'timezone': tz,
'utc_offset': dt.strftime('%z') or None,
'day_of_year': int(dt.strftime('%j')),
'week_of_year': int(dt.strftime('%W')),
'iso': dt.isoformat(),
'epoch': timestamp['timestamp_naive'],
'epoch_utc': timestamp['timestamp_utc'],
'timezone_aware': True if timestamp['timestamp_utc'] else False
}
if raw: if raw:
return raw_output return raw_output

2
tests/fixtures/generic/date-after-midnight.json vendored
View File

@ -1 +1 @@
{"year": 2021, "month_num": 3, "day": 24, "hour": 12, "hour_24": 0, "minute": 54, "second": 47, "period": "AM", "month": "Mar", "weekday": "Wed", "weekday_num": 3, "timezone": "UTC", "epoch": 1616572487, "epoch_utc": 1616547287} {"year": 2021, "month": "Mar", "month_num": 3, "day": 24, "weekday": "Wed", "weekday_num": 3, "hour": 12, "hour_24": 0, "minute": 54, "second": 47, "period": "AM", "timezone": "UTC", "utc_offset": "+0000", "day_of_year": 83, "week_of_year": 12, "iso": "2021-03-24T00:54:47+00:00", "epoch": 1616572487, "epoch_utc": 1616547287, "timezone_aware": true}

2
tests/fixtures/generic/date-before-midnight.json vendored
View File

@ -1 +1 @@
{"year": 2021, "month_num": 3, "day": 24, "hour": 11, "hour_24": 23, "minute": 54, "second": 47, "period": "PM", "month": "Mar", "weekday": "Wed", "weekday_num": 3, "timezone": "UTC", "epoch": 1616655287, "epoch_utc": 1616630087} {"year": 2021, "month": "Mar", "month_num": 3, "day": 24, "weekday": "Wed", "weekday_num": 3, "hour": 11, "hour_24": 23, "minute": 54, "second": 47, "period": "PM", "timezone": "UTC", "utc_offset": "+0000", "day_of_year": 83, "week_of_year": 12, "iso": "2021-03-24T23:54:47+00:00", "epoch": 1616655287, "epoch_utc": 1616630087, "timezone_aware": true}

2
tests/fixtures/generic/date.json vendored
View File

@ -1 +1 @@
{"year": 2020, "month_num": 8, "day": 3, "hour": 9, "hour_24": 9, "minute": 12, "second": 51, "period": null, "month": "Aug", "weekday": "Mon", "weekday_num": 1, "timezone": "PDT", "epoch": 1596471171} {"year": 2020, "month": "Aug", "month_num": 8, "day": 3, "weekday": "Mon", "weekday_num": 1, "hour": 9, "hour_24": 9, "minute": 12, "second": 51, "period": "AM", "timezone": "PDT", "utc_offset": null, "day_of_year": 216, "week_of_year": 31, "iso": "2020-08-03T09:12:51", "epoch": 1596471171, "epoch_utc": null, "timezone_aware": false}

2
tests/fixtures/ubuntu-20.04/date.json vendored
View File

@ -1 +1 @@
{"year": 2021, "month_num": 1, "day": 5, "hour": 1, "hour_24": 1, "minute": 2, "second": 4, "period": "AM", "month": "Jan", "weekday": "Tue", "weekday_num": 2, "timezone": "UTC", "epoch": 1609837324, "epoch_utc": 1609808524} {"year": 2021, "month": "Jan", "month_num": 1, "day": 5, "weekday": "Tue", "weekday_num": 2, "hour": 1, "hour_24": 1, "minute": 2, "second": 4, "period": "AM", "timezone": "UTC", "utc_offset": "+0000", "day_of_year": 5, "week_of_year": 1, "iso": "2021-01-05T01:02:04+00:00", "epoch": 1609837324, "epoch_utc": 1609808524, "timezone_aware": true}

2
tests/fixtures/ubuntu-20.04/date2.json vendored
View File

@ -1 +1 @@
{"year": 2021, "month_num": 3, "day": 23, "hour": 8, "hour_24": 20, "minute": 45, "second": 29, "period": "PM", "month": "Mar", "weekday": "Tue", "weekday_num": 2, "timezone": "UTC", "epoch": 1616557529, "epoch_utc": 1616532329} {"year": 2021, "month": "Mar", "month_num": 3, "day": 23, "weekday": "Tue", "weekday_num": 2, "hour": 8, "hour_24": 20, "minute": 45, "second": 29, "period": "PM", "timezone": "UTC", "utc_offset": "+0000", "day_of_year": 82, "week_of_year": 12, "iso": "2021-03-23T20:45:29+00:00", "epoch": 1616557529, "epoch_utc": 1616532329, "timezone_aware": true}