DOC: Fix doc redirects

[rhshadrach]

• closes #xxxx (Replace xxxx with the GitHub issue number)
• Tests added and passed if fixing a bug or adding a new feature
• All code checks passed.
• Added type annotations to new arguments/methods/functions.
• Added an entry in the latest doc/source/whatsnew/vX.X.X.rst file if fixing a bug or adding a new feature.

Summary:

• Removed any redirects for methods that have been removed/deprecated
• Added/updated redirects for GroupBy -> DataFrameGroupby
• Added redirects for EWM -> ExponentialMovingWindow
• Added/updated redirects for alias pages we used to generate but the alias was removed/deprecated (e.g. backfill -> bfill, applymap -> map) or just the doc page was removed (e.g. notnull -> notna)

There are many old API pages that no longer exist but are still part of the API (or at least, I couldn't find deprecation/removal notes). These are in the missing_methods list in the script below. I'm also currently excluding pandas.tseries; there are too many potential issues to handle here.

Testing script

from pathlib import Path
import os

# Download each of the versions of docs listed below as zip files, unzip into a
# directory with subdirectories 1.1, 1.2, etc.
base_old = Path(f"/home/richard/Downloads/docs/")
base_pr = Path("/home/richard/dev/pandas/doc/build/html")

# We'll make sure redirects don't exist for any methods that have been removed / deprecated
removed_methods = [
    "pandas.DataFrame.append.html",
    "pandas.DataFrame.iteritems.html",
    "pandas.DataFrame.lookup.html",
    "pandas.DataFrame.mad.html",
    "pandas.DataFrame.slice_shift.html",
    "pandas.DataFrame.tshift.html",
    "pandas.DatetimeIndex.to_perioddelta.html",
    "pandas.DatetimeIndex.weekofyear.html",
    "pandas.Float64Index.html",
    "pandas.Index.asi8.html",
    "pandas.Index.get_value.html",
    "pandas.Index.holds_integer.html",
    "pandas.Index.is_all_dates.html",
    "pandas.Index.is_mixed.html",
    "pandas.Index.is_monotonic.html",
    "pandas.Index.is_type_compatible.html",
    "pandas.Index.set_value.html",
    "pandas.Index.to_native_types.html",
    "pandas.Int64Index.html",
    "pandas.MultiIndex.is_lexsorted.html",
    "pandas.Resampler.pad.html",
    "pandas.Series.append.html",
    "pandas.Series.is_monotonic.html",
    "pandas.Series.iteritems.html",
    "pandas.Series.mad.html",
    "pandas.Series.slice_shift.html",
    "pandas.Series.slice_shift.html",
    "pandas.Series.swapaxes.html",
    "pandas.Series.tshift.html",
    "pandas.Timedelta.delta.html",
    "pandas.Timedelta.is_populated.html",
    "pandas.Timestamp.freq.html",
    "pandas.Timestamp.freqstr.html",
    "pandas.UInt64Index.html",
    "pandas.api.types.is_categorical.html",
    "pandas.api.types.is_extension_type.html",
    "pandas.core.groupby.GroupBy.pad.html",
    "pandas.core.groupby.DataFrameGroupBy.mad.html",
    "pandas.core.groupby.DataFrameGroupBy.pad.html",
    "pandas.core.groupby.DataFrameGroupBy.tshift.html",
    "pandas.core.resample.Resampler.pad.html",
    "pandas.io.formats.style.Styler.hide_columns.html",
    "pandas.io.formats.style.Styler.hide_index.html",
    "pandas.io.formats.style.Styler.set_na_rep.html",
    "pandas.io.formats.style.Styler.set_precision.html",
]
# These methods no longer have doc pages but are still part of the API 
# (or at least, I couldn't find deprecation/removal notes)
missing_methods = [
    "pandas.DataFrame.flags.html",
    "pandas.DataFrame.isetitem.html",
    "pandas.DataFrame.to_numpy.html",
    "pandas.DatetimeIndex.week.html",
    "pandas.ExcelWriter.sheets.html",
    "pandas.ExcelFile.close.html",
    "pandas.ExcelWriter.supported_extensions.html",
    "pandas.ExcelWriter.write_cells.html",
    "pandas.ExcelWriter.handles.html",
    "pandas.ExcelWriter.datetime_format.html",
    "pandas.ExcelWriter.book.html",
    "pandas.ExcelWriter.close.html",
    "pandas.ExcelWriter.check_extension.html",
    "pandas.ExcelWriter.if_sheet_exists.html",
    "pandas.ExcelWriter.path.html",
    "pandas.ExcelWriter.date_format.html",
    "pandas.ExcelWriter.cur_sheet.html",
    "pandas.ExcelWriter.engine.html",
    "pandas.ExcelWriter.save.html",
    "pandas.Flags.allows_duplicate_labels.html",
    "pandas.Index.array.html",
    "pandas.Index.diff.html",
    "pandas.Index.format.html",
    "pandas.Index.groupby.html",
    "pandas.Index.round",
    "pandas.Index.infer_objects.html",
    "pandas.Index.isnull.html",
    "pandas.Index.nlevels.html",
    "pandas.Index.sort.html",
    "pandas.Index.sortlevel.html",
    "pandas.Index.to_flat_index.html",
    "pandas.Index.to_numpy.html",
    "pandas.NamedAgg.index.html",
    "pandas.NamedAgg.aggfunc.html",
    "pandas.NamedAgg.column.html",
    "pandas.NamedAgg.count.html",
    "pandas.PandasArray.html",
    "pandas.Series.axes.html",
    "pandas.Series.divmod.html",
    "pandas.Series.dt.week.html",
    "pandas.Series.dt.weekofyear.html",
    "pandas.Series.info.html",
    "pandas.Series.rdivmod.html",
    "pandas.Timedelta.freq.html",
    "pandas.Timedelta.resolution_string.html",
    "pandas.Timestamp.fromisocalendar.html",
    "pandas.Timestamp.fromisoformat.html",
    "pandas.api.extensions.ExtensionArray._ndarray_values.html",
    "pandas.api.extensions.ExtensionDtype.construct_array_type.html",
    "pandas.api.extensions.ExtensionDtype.construct_from_string.html",
    "pandas.api.extensions.ExtensionDtype.empty.html",
    "pandas.api.extensions.ExtensionDtype.is_dtype.html",
    "pandas.api.extensions.ExtensionDtype.kind.html",
    "pandas.api.extensions.ExtensionDtype.na_value.html",
    "pandas.api.extensions.ExtensionDtype.name.html",
    "pandas.api.extensions.ExtensionDtype.names.html",
    "pandas.api.extensions.ExtensionDtype.type.html",
    "pandas.api.indexers.BaseIndexer.get_window_bounds.html",
    "pandas.api.indexers.FixedForwardWindowIndexer.get_window_bounds.html",
    "pandas.api.indexers.VariableOffsetWindowIndexer.get_window_bounds.html",
    "pandas.core.window.Rolling.html",
    "pandas.errors.AccessorRegistrationWarning",
    "pandas.io.formats.style.Styler.render.html",
    "pandas.io.formats.style.Styler.template.html",
    "pandas.io.formats.style.Styler.where.html",
    "pandas.option_context.__call__",
    "pandas.window.Rolling.html",
]

def strip_str(haystack, needle, before):
    idx = haystack.find(needle)
    assert idx >= 0
    if before:
        result = haystack[idx + len(needle):]
    else:
        result = haystack[:idx]
    return result

for version in ["1.0", "1.1", "1.2", "1.3", "1.4", "1.5", "2.0", "2.1"]:
    base_old_api = base_old / version / "reference" / "api"
    base_pr_api = base_pr / "reference" / "api"

    print('Version:', version)
    for e in base_old_api.rglob("*.html"):
        relative_path = e.relative_to(base_old_api)
        
        if str(relative_path).startswith("pandas.tseries"):
            # Only looking at pandas.core redirects
            continue
        
        target_exists = os.path.exists(base_pr_api / relative_path)
        
        if any(e in str(relative_path) for e in removed_methods + missing_methods):
            if target_exists:
                print(f'  {relative_path}: Removed method has redirect')
            continue
            
        if not target_exists:
            print(f'  {relative_path}: Does not exist')
            continue

        contents = open(base_pr_api / relative_path).read()
        
        if "The page has been moved to" not in contents:
            # There is no redirect
            continue

        # Get the redirect path
        contents = strip_str(contents, "0;URL=", before=True)
        path = Path(strip_str(contents, '"/>\n', before=False))
        
        if not os.path.exists(base_pr_api / path):
            print(f'  {relative_path}: Redirect does not exist')
            continue

        contents = open(base_pr_api / path).read()
        if "The page has been moved to" in contents:
            print(f'  {relative_path}: Redirect is another redirect')
            
    print()
    print('-' * 20)
    print()

[mroeschke]

LGTM. Haven't worked with the redirects, but these redirects are compatible with each version in our version switcher correct?

[rhshadrach]

I believe only the redirects of the version you've chosen will be used. In other words, this PR has no impact on the docs when you look at 2.1 or earlier.

[lithomas1]

thanks @rhshadrach