entsoe-py query_imbalance_(prices|volumes) fails with ValueError: invalid literal for int(): '1,346' in parser — best fix?

I’m fetching ENTSO-E imbalance prices/volumes with entsoe-py and hit a parser crash because the <position> field contains a thousands separator comma (e.g. "1,346"), which int() can’t parse.

Environment:

• Windows 10, Python 3.11.9

• pandas 2.2.x

• entsoe-py 0.6.10 (also repro’d on latest as of Nov 2025)

• Locale is en-GB; requests made from the official Transparency API via EntsoePandasClient

Minimal repro:

import keyring
import pandas as pd
from entsoe import EntsoePandasClient

ENTSOE_TOKEN = keyring.get_password("baringa-entsoe", "token")
client = EntsoePandasClient(api_key=ENTSOE_TOKEN)

start = pd.Timestamp('2024-01-01 00:00:00', tz='UTC')
end   = pd.Timestamp('2024-12-31 23:59:59', tz='UTC')

# France example (happens on other countries/years too)
df = client.query_imbalance_volumes(country_code='FR', start=start, end=end)
print(df.shape)

Traceback (excerpt):

File ...\entsoe\parsers.py", line 665, in _parse_imbalance_volumes_timeseries
    position = int(point.find('position').text)
ValueError: invalid literal for int() with base 10: '1,346'

I also occasionally see a follow-on error when the above doesn’t happen:

ValueError: Index contains duplicate entries, cannot reshape
# from df.set_index(['position','category']).unstack()

What I’ve tried / Notes

• Cleaning Quantity post-hoc doesn’t help (crash occurs inside the parser before I get a dataframe).

• Timestamps are tz='UTC'; switching to Etc/UTC doesn’t change the behavior.

• Looks like the XML returned by the API sometimes includes <position> with commas (1,346) rather than a plain integer. I can’t see an option in entsoe-py to sanitize this or request a different number format.

• The duplicate-index error seems to come from multiple <TimeSeries> sharing the same (timestamp, position, category) combo in the ZIP payload (not my main blocker, but mentioning for completeness).

Questions

1. Is there a recommended way in entsoe-py to handle locale/thousands separators in <position>?

   • e.g., a documented flag, or a known version that doesn’t parse <position> with int() directly?

2. If not, what’s the cleanest workaround?

   • Monkey-patch the parser to strip commas before int()?

   • Pre-download the ZIP, sanitize XML (replace ,<digit> in <position>), then call the internal parser?

   • Another approach I’m missing?

3. Any guidance on the “Index contains duplicate entries” when unstacking on ['position','category']?

   • Is deduping by (['timestamp','position','category']) with first the right approach, or is there a better semantic grouping?