Some minor tweaks for convert() #217
Conversation
1. Use `requireNamespace()` to test if stringi is available. 2. Reduce one `ifelse()` call. 3. Use the safer `seq_along(x)` instead of `1:length(x)`.
R/conversion.R
Outdated
| ) | ||
| unicode_int <- as.integer(as.hexmode(unicode_latex$unicode)) | ||
| char_latex <- ifelse(unicode_int <= 255 & unicode_int != 177, unicode_latex$chr, | ||
| paste0("\\uc1\\u", unicode_int - if (unicode_int < 32768) 0 else 65536, "*") |
There was a problem hiding this comment.
Here is another way that I often use by myself, i.e., "ninja code":
| paste0("\\uc1\\u", unicode_int - if (unicode_int < 32768) 0 else 65536, "*") | |
| paste0("\\uc1\\u", unicode_int - (unicode_int >= 32768) * 65536, "*") |
but I understand that most people don't like obscure ninja code.
Co-authored-by: Yihui Xie <[email protected]>
Co-authored-by: yilong zhang <[email protected]>
There was a problem hiding this comment.
I think these change are exactly what we need to improve code quality so I think we should merge this. Thanks for the sprintf() refactor!
(The R CMD check error was due to the secondary dependency of emmeans (estimability) requires R >= 4.3.0 while the current r-oldrel is only 4.2.x. The issue will go away once R 4.4.0 is released.)
If we use emmeans conditionally in the package, it should pass |
I'm proposing some minor changes while looking at #216 (these changes are irrelevant to #216, though):
requireNamespace()to test if stringi is available.ifelse()call.seq_along(x)instead of1:length(x).