When clean_names and make_clean_names appear to produce different results

[sfirke]

On a recent project where I was processing thousands of spreadsheets, I started with the workflow of readxl::read_excel("data.xlsx") %>% clean_names(). I decided to change it to a single line, readxl::read_excel("data.xlsx", .name_repair = janitor::make_clean_names) and it broke my subsequent code. This baffled me for longer than it should have, so I'm sharing the result here in case it helps others. Why would those produce different results, when all that clean_names does is call make_clean_names?

[sfirke]

In the first instance, read_excel() is performing its own minimal name repair while ingesting data. Due to the malformed and missing column names in the target spreadsheet, this caused a change in names from the raw data before it reached clean_names. Switching to .name_repair = janitor::make_clean_names avoids this initial name repair performed by read_excel.

In particular, with the clean_names in a subsequent step I got x2, x3, x4 for three missing column names, while with .name_repair = make_clean_names I got the column names x, x_2, x_3.