BUG: Incorrect parsing of ISO 8601 durations strings in Timedelta constructor

[rosstitmarsh]

Pandas version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

import pandas as pd

# Raises an error for >2 digit hours or minutes
pd.Timedelta('PT100H')
pd.Timedelta('PT100M')

# Incorrectly parses any decimal component apart from seconds
pd.Timedelta('PT3.5H') == pd.Timedelta('PT5H3S')

# Raises an error for comma as decimal separator
pd.Timedelta('PT3,5H')

# Does not raise an error for plusses in numbers
pd.Timedelta('P+1S') == pd.Timedelta('PT1S')

# Does not raise an error for unsupported month component, instead parses it as minutes
pd.Timedelta('P1M') == pd.Timedelta('PT1M')

# Does not raise an error for hour and second components before the 'T' seperator
pd.Timedelta('P1H') == pd.Timedelta('PT1H')
pd.Timedelta('P1S') == pd.Timedelta('PT1S')

# Does not raise an error for day components after the 'T' seperator
pd.Timedelta('PT1D') == pd.Timedelta('P1D')

# Does not raise an error for duplicate components (and produces unexpected result)
pd.Timedelta('PT1S1S1S1S') == pd.Timedelta('PT20M34S')

# Does not raise an error for components in the incorrect order
pd.Timedelta('PT1M1H') == pd.Timedelta('PT1H1M')

# Does not raise an error for trailing 'T' separator
pd.Timedelta('P1DT') == pd.Timedelta('P1D')

# Does not raise an error for duplicate 'T' separators
pd.Timedelta('PTTT1H') == pd.Timedelta('PT1H')

# Does not raise an error for 'P' or 'T' in any position
pd.Timedelta('PTP1TPTH') == pd.Timedelta('PT1H')

Issue Description

Numbers

• Raises a ValueError when the hours or minutes part is longer than 2 digits. There is nothing in the standard that requires this. There is an explicit check for the length of the hours and minute parts in the parser. BUG: Timedelta does not accept ISO 8601 duration strings with 3 digit hours or minutes #51820

  pandas/pandas/_libs/tslibs/timedeltas.pyx

  Line 883 in f9be5d9

  if c in ["H", "M"] and len(number) > 2:

• Incorrectly parses all decimal components aparts from seconds.
• Does not allow the comma , as a decimal separator
• Allows plusses (+) to prefix numbers, this is outwith the standard.

  pandas/pandas/_libs/tslibs/timedeltas.pyx

  Line 880 in f9be5d9

  elif c == "+":

Components

• Parses month (M) components as minutes.

  pandas/pandas/_libs/tslibs/timedeltas.pyx

  Line 885 in f9be5d9

  if c == "M":

• Parses hour (H) or second (S) components before the T
• Parses day components after the T
• Parses duplicate components with weird results
• Parses components in the incorrect order

P and T symbols

• Ignores P and T symbols anywhere in the string.

  pandas/pandas/_libs/tslibs/timedeltas.pyx

  Line 873 in f9be5d9

  if c == "P" or c == "T":

Expected Behavior

The follwing should all parse without errors.

pd.Timedelta("PT100H") == pd.Timedelta(hours=100)
pd.Timedelta("PT100M") == pd.Timedelta(minutes=100)
pd.Timedelta("PT3,5H") == pd.Timedelta(hours=3.5)

The follwing should all raise errors.

# Bad numbers
pd.Timedelta("PT.1S")
pd.Timedelta("PT+1S")

# Bad components
pd.Timedelta("P1M") 
pd.Timedelta("P1H")
pd.Timedelta("P1S")
pd.Timedelta("PT1D")
pd.Timedelta("PT1M1H")
pd.Timedelta("P1S1S")

# Misplaced P or T
pd.Timedelta("P1PD")
pd.Timedelta("P1TD")
pd.Timedelta("P1DT")

Installed Versions

INSTALLED VERSIONS

commit : 2e218d1
python : 3.11.1.final.0
python-bits : 64

pandas : 1.5.3
numpy : 1.24.2
pytz : 2022.7.1
dateutil : 2.8.2
setuptools : 65.5.0
pip : 22.3.1
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : None
IPython : None
pandas_datareader: None
bs4 : None
bottleneck : None
brotli : None
fastparquet : None
fsspec : None
gcsfs : None
matplotlib : None
numba : None
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : None
pyreadstat : None
pyxlsb : None
s3fs : None
scipy : None
snappy : None
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlwt : None
zstandard : None
tzdata : None

[rosstitmarsh]

Also covers #48122