[Feature request] More case-conversion functions

[Rangi42]

We already have STRLWR and STRUPR, for lowercase and UPPERCASE. However, I'd like a couple more functions, for CamelCaseand snake_case (and maybe Title_Case). The goal would be to allow transforming between conventional CONSTANT_NAMES and LabelNames, so macros wouldn't need to be passed both variants. E.g. you could do map_mac CeruleanCity and then internally it could do db CERULEAN_CITY_MAP_ID as well as dw CeruleanCityNPCData.

I haven't fully thought out the edge cases of how these functions would work, beyond "keep them ASCII-only" (since STRLWR and STRUPR are ASCII-only too, and symbol names can only be ASCII anyway). But the Rust paste! macro (a more powerful version of concat_idents!) can be used for reference: https://docs.rs/paste/latest/paste/#case-conversion

See also guess_map_constant in Polished Map, which tries to derive a CONSTANT_NAME from the ConstantName.blk filename it's given. It handles at least some edge cases, like DarkCaveB3F becoming DARK_CAVE_B3F.

[ISSOtm]

I dunno, I feel like it's a better idea to implement them as macros? And if they prove popular enough, build those into RGBDS. Right?

[Rangi42]

That's definitely one option. I was thinking that a built-in solution would be relatively simple (we have lots of string functions already), as opposed to things like structs or arrays or 2D section packing, which need careful thought for how to implement. (And a macro would have to be, like, strsnake FOO, "TheString" and then use "{FOO}" and then purge FOO, instead of using a string expression directly.) Regardless, I could at least start by implementing these as macros so we can test out algorithm edge cases.

[Rangi42]

Desired behavior for camel-to-snake case:

• STRSNAKE("camelCase") → "camel_Case"
• STRSNAKE("CeruleanCity") → "Cerulean_City"
• STRSNAKE("HTTPRequest") → "HTTP_Request"
• STRSNAKE("DarkCaveB3F") → "Dark_Cave_B3F"
• STRSNAKE("DeptStore4F") → "Dept_Store_4F"
• STRSNAKE("Level4") → "Level_4"
• STRSNAKE("DECToHEX") → "DEC_To_HEX"
• STRSNAKE("ConvertASMToBIN") → "Convert_ASM_To_BIN"
• STRSNAKE("The quick brownFox jumped! Over theLazy Dogs.") → "The quick brown_Fox jumped! Over the_Lazy Dogs."

And for snake-to-camel case:

• STRCAMEL("snake_case") → "snakeCase"
• STRCAMEL("Cerulean_City") → "ceruleanCity"
• STRCAMEL("CERULEAN_CITY") → "ceruleanCity"
• STRCAMEL("Dark_Cave_B3F") → "darkCaveB3F"
• STRCAMEL("HEX2DEC") → "hex2dec"
• STRCAMEL("HEX_2_DEC") → "hex2Dec"
• STRCAMEL("Convert_ASM_To_BIN") → "convertAsmToBin"
• STRCAMEL("The quick brown_fox jumped! Over the_lazy Dogs.") → "the quick brownFox jumped! Over theLazy Dogs."

[calc84maniac]

What's the difference between "CERULEAN_CITY" and "Convert_ASM_To_BIN" that causes CERULEAN and CITY to become camel-cased but not ASM or BIN? Is it that the original string contains no lowercase characters?

[Rangi42]

No, just a mistake, thank you for noticing. :P "Convert_ASM_To_BIN" does not have any all-caps next to all-caps, so I was able to render it camel-case as "ConvertASMToBIN" without introducing ambiguity. However, "CERULEAN_CITY", if it preserved the caps, would camel-case as "CERULEANCITY", which is clearly wrong.

Two options there for camel-casing adjacent all-caps words:

1. Leave an underscore between them. Not useful, since a large reason for these functions would be to derive a label name from a constant name.
2. Make them TitleCase (and do so for all all-caps words, for consistency). That should be fine.

[ISSOtm]

(we have lots of string functions already)

I don't think that's sufficient reason to add more; look at PHP's list of functions topics for example. I feel like a better set of criteria to balance functionality and code we have to own, is:

• “elementary” functionality that's impossible or really difficult to achieve using macros; or
• functionality that's used commonly enough that the corresponding boilerplate is really annoying to have to make each time

Regardless, I could at least start by implementing these as macros so we can test out algorithm edge cases.

Yes, please.

[Rangi42]

This doesn't exactly match the examples I previously wrote, but it matches what I would actually find useful, which I thought might be inconsistent but turns out to work:

#!/usr/bin/env python3

def isupperordigit(s):
	return s.isupper() or s.isdigit()

def camel2snake(camel):
	snake = ''
	for i in range(len(camel)):
		snake += camel[i]
		if (camel[i].islower() and i+1 < len(camel) and isupperordigit(camel[i+1])) or \
		   (isupperordigit(camel[i]) and i+2 < len(camel) and isupperordigit(camel[i+1]) and camel[i+2].islower()):
			snake += '_'
	return snake

def snake2camel(snake):
	camel = ''
	prev_letter = True
	for char in snake:
		if char.isalpha():
			if not prev_letter:
				camel += char.upper()
			elif camel:
				camel += char.lower()
			else:
				camel += char
			prev_letter = True
		else:
			if char != '_':
				camel += char
			prev_letter = False
	return camel

def test(fn, data):
	print(f'{fn.__name__}:')
	for k in data:
		v = fn(k)
		if v == data[k]:
			print(f'PASS: "{k}" => "{v}"')
		else:
			print(f'FAIL: "{k}" => "{v}" not "{data[k]}"')

test(camel2snake, {
	'trivial': 'trivial',
	'camelCase': 'camel_Case',
	'CeruleanCity': 'Cerulean_City',
	'HTTPRequest': 'HTTP_Request',
	'DarkCaveB3F': 'Dark_Cave_B3F',
	'DeptStore4F': 'Dept_Store_4F',
	'Level4': 'Level_4',
	'DECToHEX': 'DEC_To_HEX',
	'ConvertASMToBIN': 'Convert_ASM_To_BIN',
	'The quick brownFox jumped! Over theLazy Dogs.': 'The quick brown_Fox jumped! Over the_Lazy Dogs.',
})

test(snake2camel, {
	'trivial': 'trivial',
	'snake_case': 'snakeCase',
	'Cerulean_City': 'CeruleanCity',
	'CERULEAN_CITY': 'CeruleanCity',
	'Dark_Cave_B3F': 'DarkCaveB3F',
	'HEX2DEC': 'Hex2Dec',
	'HEX_2_DEC': 'Hex2Dec',
	'Convert_ASM_To_BIN': 'ConvertAsmToBin',
	'The quick brown_fox jumped! Over the_lazy Dogs.': 'The Quick BrownFox Jumped! Over TheLazy Dogs.',
})

[Rangi42]

This is too ambiguous to be a built-in feature. The exact case-conversion rules depend on your application, and might need special-casing for particular substrings (e.g. treating "B3F" or acronyms like "HTTP" as all-caps words, or whether snake2camel("lower_case_UPPER_CASE") should be "lowerCaseUpperCase" or "lowerCaseUPPER_CASE" or what).