Commit 0e61a7e1ac2c8de9548c57c166aeb3161369778d

Michael Schmidt 2021-01-03T18:18:28

Scheme: Added support for high Unicode characters (#2693)

diff --git a/components/prism-scheme.js b/components/prism-scheme.js
index eea2ae8..77a9255 100644
--- a/components/prism-scheme.js
+++ b/components/prism-scheme.js
@@ -15,7 +15,7 @@
 			greedy: true
 		},
 		'character': {
-			pattern: /#\\(?:[ux][a-fA-F\d]+\b|[-a-zA-Z]+\b|\S)/,
+			pattern: /#\\(?:[ux][a-fA-F\d]+\b|[-a-zA-Z]+\b|[\uD800-\uDBFF][\uDC00-\uDFFF]|\S)/,
 			greedy: true,
 			alias: 'string'
 		},
diff --git a/components/prism-scheme.min.js b/components/prism-scheme.min.js
index 0e5467f..4724e88 100644
--- a/components/prism-scheme.min.js
+++ b/components/prism-scheme.min.js
@@ -1 +1 @@
-Prism.languages.scheme={comment:/;.*|#;\s*\((?:[^()]|\([^()]*\))*\)|#\|(?:[^#|]|#(?!\|)|\|(?!#)|#\|(?:[^#|]|#(?!\|)|\|(?!#))*\|#)*\|#/,string:{pattern:/"(?:[^"\\]|\\.)*"/,greedy:!0},symbol:{pattern:/'[^()#'\s]+/,greedy:!0},character:{pattern:/#\\(?:[ux][a-fA-F\d]+\b|[-a-zA-Z]+\b|\S)/,greedy:!0,alias:"string"},"lambda-parameter":[{pattern:/((?:^|[^'`#])\(lambda\s+)(?:[^|()'\s]+|\|(?:[^\\|]|\\.)*\|)/,lookbehind:!0},{pattern:/((?:^|[^'`#])\(lambda\s+\()[^()']+/,lookbehind:!0}],keyword:{pattern:/((?:^|[^'`#])\()(?:begin|case(?:-lambda)?|cond(?:-expand)?|define(?:-library|-macro|-record-type|-syntax|-values)?|defmacro|delay(?:-force)?|do|else|export|except|guard|if|import|include(?:-ci|-library-declarations)?|lambda|let(?:rec)?(?:-syntax|-values|\*)?|let\*-values|only|parameterize|prefix|(?:quasi-?)?quote|rename|set!|syntax-(?:case|rules)|unless|unquote(?:-splicing)?|when)(?=[()\s]|$)/,lookbehind:!0},builtin:{pattern:/((?:^|[^'`#])\()(?:abs|and|append|apply|assoc|ass[qv]|binary-port\?|boolean=?\?|bytevector(?:-append|-copy|-copy!|-length|-u8-ref|-u8-set!|\?)?|caar|cadr|call-with-(?:current-continuation|port|values)|call\/cc|car|cdar|cddr|cdr|ceiling|char(?:->integer|-ready\?|\?|<\?|<=\?|=\?|>\?|>=\?)|close-(?:input-port|output-port|port)|complex\?|cons|current-(?:error|input|output)-port|denominator|dynamic-wind|eof-object\??|eq\?|equal\?|eqv\?|error|error-object(?:-irritants|-message|\?)|eval|even\?|exact(?:-integer-sqrt|-integer\?|\?)?|expt|features|file-error\?|floor(?:-quotient|-remainder|\/)?|flush-output-port|for-each|gcd|get-output-(?:bytevector|string)|inexact\??|input-port(?:-open\?|\?)|integer(?:->char|\?)|lcm|length|list(?:->string|->vector|-copy|-ref|-set!|-tail|\?)?|make-(?:bytevector|list|parameter|string|vector)|map|max|member|memq|memv|min|modulo|negative\?|newline|not|null\?|number(?:->string|\?)|numerator|odd\?|open-(?:input|output)-(?:bytevector|string)|or|output-port(?:-open\?|\?)|pair\?|peek-char|peek-u8|port\?|positive\?|procedure\?|quotient|raise|raise-continuable|rational\?|rationalize|read-(?:bytevector|bytevector!|char|error\?|line|string|u8)|real\?|remainder|reverse|round|set-c[ad]r!|square|string(?:->list|->number|->symbol|->utf8|->vector|-append|-copy|-copy!|-fill!|-for-each|-length|-map|-ref|-set!|\?|<\?|<=\?|=\?|>\?|>=\?)?|substring|symbol(?:->string|\?|=\?)|syntax-error|textual-port\?|truncate(?:-quotient|-remainder|\/)?|u8-ready\?|utf8->string|values|vector(?:->list|->string|-append|-copy|-copy!|-fill!|-for-each|-length|-map|-ref|-set!|\?)?|with-exception-handler|write-(?:bytevector|char|string|u8)|zero\?)(?=[()\s]|$)/,lookbehind:!0},operator:{pattern:/((?:^|[^'`#])\()(?:[-+*%/]|[<>]=?|=>?)(?=[()\s]|$)/,lookbehind:!0},number:{pattern:RegExp(function(r){for(var e in r)r[e]=r[e].replace(/<[\w\s]+>/g,function(e){return"(?:"+r[e].trim()+")"});return r[e]}({"<ureal dec>":"\\d+(?:/\\d+)?|(?:\\d+(?:\\.\\d*)?|\\.\\d+)(?:e[+-]?\\d+)?","<real dec>":"[+-]?<ureal dec>|[+-](?:inf|nan)\\.0","<imaginary dec>":"[+-](?:<ureal dec>|(?:inf|nan)\\.0)?i","<complex dec>":"<real dec>(?:@<real dec>|<imaginary dec>)?|<imaginary dec>","<num dec>":"(?:#d(?:#[ei])?|#[ei](?:#d)?)?<complex dec>","<ureal box>":"[0-9a-f]+(?:/[0-9a-f]+)?","<real box>":"[+-]?<ureal box>|[+-](?:inf|nan)\\.0","<imaginary box>":"[+-](?:<ureal box>|(?:inf|nan)\\.0)?i","<complex box>":"<real box>(?:@<real box>|<imaginary box>)?|<imaginary box>","<num box>":"#[box](?:#[ei])?|(?:#[ei])?#[box]<complex box>","<number>":"(^|[\\s()])(?:<num dec>|<num box>)(?=[()\\s]|$)"}),"i"),lookbehind:!0},boolean:{pattern:/(^|[\s()])#(?:[ft]|false|true)(?=[()\s]|$)/,lookbehind:!0},function:{pattern:/((?:^|[^'`#])\()(?:[^|()'\s]+|\|(?:[^\\|]|\\.)*\|)(?=[()\s]|$)/,lookbehind:!0},identifier:{pattern:/(^|[\s()])\|(?:[^\\|]|\\.)*\|(?=[()\s]|$)/,lookbehind:!0,greedy:!0},punctuation:/[()']/};
\ No newline at end of file
+Prism.languages.scheme={comment:/;.*|#;\s*\((?:[^()]|\([^()]*\))*\)|#\|(?:[^#|]|#(?!\|)|\|(?!#)|#\|(?:[^#|]|#(?!\|)|\|(?!#))*\|#)*\|#/,string:{pattern:/"(?:[^"\\]|\\.)*"/,greedy:!0},symbol:{pattern:/'[^()#'\s]+/,greedy:!0},character:{pattern:/#\\(?:[ux][a-fA-F\d]+\b|[-a-zA-Z]+\b|[\uD800-\uDBFF][\uDC00-\uDFFF]|\S)/,greedy:!0,alias:"string"},"lambda-parameter":[{pattern:/((?:^|[^'`#])\(lambda\s+)(?:[^|()'\s]+|\|(?:[^\\|]|\\.)*\|)/,lookbehind:!0},{pattern:/((?:^|[^'`#])\(lambda\s+\()[^()']+/,lookbehind:!0}],keyword:{pattern:/((?:^|[^'`#])\()(?:begin|case(?:-lambda)?|cond(?:-expand)?|define(?:-library|-macro|-record-type|-syntax|-values)?|defmacro|delay(?:-force)?|do|else|export|except|guard|if|import|include(?:-ci|-library-declarations)?|lambda|let(?:rec)?(?:-syntax|-values|\*)?|let\*-values|only|parameterize|prefix|(?:quasi-?)?quote|rename|set!|syntax-(?:case|rules)|unless|unquote(?:-splicing)?|when)(?=[()\s]|$)/,lookbehind:!0},builtin:{pattern:/((?:^|[^'`#])\()(?:abs|and|append|apply|assoc|ass[qv]|binary-port\?|boolean=?\?|bytevector(?:-append|-copy|-copy!|-length|-u8-ref|-u8-set!|\?)?|caar|cadr|call-with-(?:current-continuation|port|values)|call\/cc|car|cdar|cddr|cdr|ceiling|char(?:->integer|-ready\?|\?|<\?|<=\?|=\?|>\?|>=\?)|close-(?:input-port|output-port|port)|complex\?|cons|current-(?:error|input|output)-port|denominator|dynamic-wind|eof-object\??|eq\?|equal\?|eqv\?|error|error-object(?:-irritants|-message|\?)|eval|even\?|exact(?:-integer-sqrt|-integer\?|\?)?|expt|features|file-error\?|floor(?:-quotient|-remainder|\/)?|flush-output-port|for-each|gcd|get-output-(?:bytevector|string)|inexact\??|input-port(?:-open\?|\?)|integer(?:->char|\?)|lcm|length|list(?:->string|->vector|-copy|-ref|-set!|-tail|\?)?|make-(?:bytevector|list|parameter|string|vector)|map|max|member|memq|memv|min|modulo|negative\?|newline|not|null\?|number(?:->string|\?)|numerator|odd\?|open-(?:input|output)-(?:bytevector|string)|or|output-port(?:-open\?|\?)|pair\?|peek-char|peek-u8|port\?|positive\?|procedure\?|quotient|raise|raise-continuable|rational\?|rationalize|read-(?:bytevector|bytevector!|char|error\?|line|string|u8)|real\?|remainder|reverse|round|set-c[ad]r!|square|string(?:->list|->number|->symbol|->utf8|->vector|-append|-copy|-copy!|-fill!|-for-each|-length|-map|-ref|-set!|\?|<\?|<=\?|=\?|>\?|>=\?)?|substring|symbol(?:->string|\?|=\?)|syntax-error|textual-port\?|truncate(?:-quotient|-remainder|\/)?|u8-ready\?|utf8->string|values|vector(?:->list|->string|-append|-copy|-copy!|-fill!|-for-each|-length|-map|-ref|-set!|\?)?|with-exception-handler|write-(?:bytevector|char|string|u8)|zero\?)(?=[()\s]|$)/,lookbehind:!0},operator:{pattern:/((?:^|[^'`#])\()(?:[-+*%/]|[<>]=?|=>?)(?=[()\s]|$)/,lookbehind:!0},number:{pattern:RegExp(function(r){for(var e in r)r[e]=r[e].replace(/<[\w\s]+>/g,function(e){return"(?:"+r[e].trim()+")"});return r[e]}({"<ureal dec>":"\\d+(?:/\\d+)?|(?:\\d+(?:\\.\\d*)?|\\.\\d+)(?:e[+-]?\\d+)?","<real dec>":"[+-]?<ureal dec>|[+-](?:inf|nan)\\.0","<imaginary dec>":"[+-](?:<ureal dec>|(?:inf|nan)\\.0)?i","<complex dec>":"<real dec>(?:@<real dec>|<imaginary dec>)?|<imaginary dec>","<num dec>":"(?:#d(?:#[ei])?|#[ei](?:#d)?)?<complex dec>","<ureal box>":"[0-9a-f]+(?:/[0-9a-f]+)?","<real box>":"[+-]?<ureal box>|[+-](?:inf|nan)\\.0","<imaginary box>":"[+-](?:<ureal box>|(?:inf|nan)\\.0)?i","<complex box>":"<real box>(?:@<real box>|<imaginary box>)?|<imaginary box>","<num box>":"#[box](?:#[ei])?|(?:#[ei])?#[box]<complex box>","<number>":"(^|[\\s()])(?:<num dec>|<num box>)(?=[()\\s]|$)"}),"i"),lookbehind:!0},boolean:{pattern:/(^|[\s()])#(?:[ft]|false|true)(?=[()\s]|$)/,lookbehind:!0},function:{pattern:/((?:^|[^'`#])\()(?:[^|()'\s]+|\|(?:[^\\|]|\\.)*\|)(?=[()\s]|$)/,lookbehind:!0},identifier:{pattern:/(^|[\s()])\|(?:[^\\|]|\\.)*\|(?=[()\s]|$)/,lookbehind:!0,greedy:!0},punctuation:/[()']/};
\ No newline at end of file
diff --git a/tests/languages/scheme/character_feature.test b/tests/languages/scheme/character_feature.test
index f95cb70..b699e18 100644
--- a/tests/languages/scheme/character_feature.test
+++ b/tests/languages/scheme/character_feature.test
@@ -14,6 +14,7 @@
 #\x10FFFF
 #\λ
 #\)
+#\💩
 
 ----------------------------------------------------
 
@@ -43,7 +44,8 @@
 	["character", "#\\u0041"],
 	["character", "#\\x10FFFF"],
 	["character", "#\\λ"],
-	["character", "#\\)"]
+	["character", "#\\)"],
+	["character", "#\\💩"]
 ]
 
 ----------------------------------------------------