fix: print: beancount: encode non-ASCII letters in metadata keys, also [#2576]

Beancount metadata keys must match [a-z][a-z0-9_-]*, so non-ASCII
letters like ä, ö, ü must be encoded there too (using the usual
"c<HEXBYTES>" encoding).

Author: simonmichael

hledger-lib/Hledger/Write/Beancount.hs

@@ -132,24 +132,25 @@ toBeancountMetadataName name =
         Just (c,cs) | T.null cs || not (isBeancountMetadataNameStartChar c) -> T.cons beancountMetadataDummyNameStartChar t
         _ -> t
 
--- | Is this a valid character to start a Beancount metadata name (lowercase letter) ?
+-- | Is this a valid character to start a Beancount metadata name (lowercase ASCII letter) ?
 isBeancountMetadataNameStartChar :: Char -> Bool
-isBeancountMetadataNameStartChar c = isLetter c && islowercase c
+isBeancountMetadataNameStartChar = isAsciiLower
 
 -- | Dummy valid starting character to prepend to a Beancount metadata name if needed.
 beancountMetadataDummyNameStartChar :: Char
 beancountMetadataDummyNameStartChar = 'm'
 
--- | Is this a valid character in the middle of a Beancount metadata name (a lowercase letter, digit, _ or -) ?
+-- | Is this a valid character in the middle of a Beancount metadata name (a lowercase ASCII letter, digit, _ or -) ?
 isBeancountMetadataNameChar :: Char -> Bool
-isBeancountMetadataNameChar c = (isLetter c && islowercase c) || isDigit c || c `elem` ['_', '-']
+isBeancountMetadataNameChar c = isAsciiLower c || isDigit c || c `elem` ['_', '-']
 
 -- | Convert a character to one or more characters valid inside a Beancount metadata name.
--- Letters are lowercased, spaces are converted to dashes, and unsupported characters are encoded as c<HEXBYTES>.
+-- ASCII uppercase letters are lowercased, spaces are converted to dashes, and unsupported
+-- characters (including non-ASCII letters) are encoded as c<HEXBYTES>.
 toBeancountMetadataNameChar :: Char -> Text
 toBeancountMetadataNameChar c
   | isBeancountMetadataNameChar c = T.singleton c
-  | isLetter c = T.singleton $ toLower c
+  | isAsciiUpper c = T.singleton $ toLower c
   | isSpace c = "-"
   | otherwise = T.pack $ printf "c%x" c
 
@@ -312,8 +313,6 @@ charToBeancount c = if isSpace c then "-" else printf "C%x" c
 -- https://hackage.haskell.org/package/base-4.20.0.1/docs/Data-Char.html#v:isUpperCase would be more correct,
 -- but isn't available till base 4.18/ghc 9.6. isUpper is close enough in practice.
 isuppercase = isUpper
--- same story, presumably
-islowercase = isLower
 
 -- | Is this a valid character to start a Beancount account name part (capital letter or digit) ?
 isBeancountAccountStartChar :: Char -> Bool

hledger/test/print/beancount.test

@@ -265,3 +265,24 @@ $ hledger -f- print -O beancount
     Equity:Opening
 
 >=
+
+# ** 13. Non-ASCII letters in metadata names are hex-encoded (#2576).
+# Beancount metadata keys must match [a-z][a-z0-9_-]*, so non-ASCII letters
+# like ä cannot just be lowercased; they must be encoded as c<HEXBYTES>.
+# Non-ASCII characters in values are fine - Beancount accepts UTF-8 in strings.
+<
+2024-01-01 Test ; Empfänger: Schäfer
+    assets:cash    -1 EUR
+    expenses:test   1 EUR
+
+$ hledger -f- print -O beancount
+;option "inferred_tolerance_default" "*:0.005"
+2024-01-01 open Assets:Cash
+2024-01-01 open Expenses:Test
+
+2024-01-01 * "Test"  ; Empfänger: Schäfer
+    empfce4nger: "Schäfer"
+    Assets:Cash            -1 EUR
+    Expenses:Test           1 EUR
+
+>=