Bug summary

When plotting by default at the end of a month, the date locator seems to jump a day, creating non regularly spaced ticks.

Code for reproduction

To create the plot which cause problem


import matplotlib.pyplot as plt
import pandas as pd

x = pd.to_datetime(['2024-07-29 00:12:00', '2024-08-10 19:13:00'])
y = [0, 1]
fig, ax = plt.subplots(figsize=(10, 6))
ax.plot(x, y)
ax.grid()
plt.show()


If we isolate the `AutoDateLocator` which causes the problem:


from matplotlib.dates import AutoDateLocator, DayLocator
import pandas as pd

x = pd.to_datetime(['2024-07-29 00:12:00', '2024-08-10 19:13:00'])
locator = AutoDateLocator()
best_locator = locator.get_locator(x.min(), x.max())
print(best_locator.tick_values(x.min(), x.max()))

Actual outcome

The locator object gives this output:

[19935. 19936. 19938. 19940. 19942. 19944.]

Which has a frequency of 2, excepted of 1 between the first points.

Expected outcome

I would expect to have a regularly spaced output.

Additional information

I had a look to the code (AutoDateLocator.get_locator in matplotlib.dates ) and it seems that the problem is the range for 'weekly' frequency is defined as range(1, 32) in the AutoDateLocator._by_ranges.

Then below in the method, an interval is added to the range

byranges[i] = self._byranges[i][::interval]

But in the case of an end of the month, this can create an irregular range.

It would be better do define the range manually in this case.

I tried to apply the following fix which seems to work:
(in the get_locator method, after the if i in (DAILY, WEEKLY): statement)

  if i == WEEKLY:
      # Set the range with the correct number of days in the current month from dmin
      year, month = dmin.year, dmin.month
      # Calculate the number of days in the current month
      next_year, next_month = (year + 1, 1) if month == 12 else (year, month + 1)
      n_days_in_month = (datetime.date(next_year, next_month, 1) - datetime.date(year, month, 1)).days
      start_day = dmin.day
      total_days = (dmax - dmin).days + interval
      byranges[i] = [
          (d - 1) % n_days_in_month + 1
          for d in range(start_day, start_day + total_days, interval)
      ]

Operating system

Windows

Matplotlib Version

3.10.3

Matplotlib Backend

No response

Python version

Python 3.13.5

Jupyter version

No response

Installation

None