123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120 |
- (function (Prism) {
- Prism.languages.scheme = {
- // this supports "normal" single-line comments:
- // ; comment
- // and (potentially nested) multiline comments:
- // #| comment #| nested |# still comment |#
- // (only 1 level of nesting is supported)
- 'comment': /;.*|#;\s*(?:\((?:[^()]|\([^()]*\))*\)|\[(?:[^\[\]]|\[[^\[\]]*\])*\])|#\|(?:[^#|]|#(?!\|)|\|(?!#)|#\|(?:[^#|]|#(?!\|)|\|(?!#))*\|#)*\|#/,
- 'string': {
- pattern: /"(?:[^"\\]|\\.)*"/,
- greedy: true
- },
- 'symbol': {
- pattern: /'[^()\[\]#'\s]+/,
- greedy: true
- },
- 'char': {
- pattern: /#\\(?:[ux][a-fA-F\d]+\b|[-a-zA-Z]+\b|[\uD800-\uDBFF][\uDC00-\uDFFF]|\S)/,
- greedy: true
- },
- 'lambda-parameter': [
- // https://www.cs.cmu.edu/Groups/AI/html/r4rs/r4rs_6.html#SEC30
- {
- pattern: /((?:^|[^'`#])[(\[]lambda\s+)(?:[^|()\[\]'\s]+|\|(?:[^\\|]|\\.)*\|)/,
- lookbehind: true
- },
- {
- pattern: /((?:^|[^'`#])[(\[]lambda\s+[(\[])[^()\[\]']+/,
- lookbehind: true
- }
- ],
- 'keyword': {
- pattern: /((?:^|[^'`#])[(\[])(?:begin|case(?:-lambda)?|cond(?:-expand)?|define(?:-library|-macro|-record-type|-syntax|-values)?|defmacro|delay(?:-force)?|do|else|except|export|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: true
- },
- 'builtin': {
- // all functions of the base library of R7RS plus some of built-ins of R5Rs
- 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: true
- },
- 'operator': {
- pattern: /((?:^|[^'`#])[(\[])(?:[-+*%/]|[<>]=?|=>?)(?=[()\[\]\s]|$)/,
- lookbehind: true
- },
- 'number': {
- // The number pattern from [the R7RS spec](https://small.r7rs.org/attachment/r7rs.pdf).
- //
- // <number> := <num 2>|<num 8>|<num 10>|<num 16>
- // <num R> := <prefix R><complex R>
- // <complex R> := <real R>(?:@<real R>|<imaginary R>)?|<imaginary R>
- // <imaginary R> := [+-](?:<ureal R>|(?:inf|nan)\.0)?i
- // <real R> := [+-]?<ureal R>|[+-](?:inf|nan)\.0
- // <ureal R> := <uint R>(?:\/<uint R>)?
- // | <decimal R>
- //
- // <decimal 10> := (?:\d+(?:\.\d*)?|\.\d+)(?:e[+-]?\d+)?
- // <uint R> := <digit R>+
- // <prefix R> := <radix R>(?:#[ei])?|(?:#[ei])?<radix R>
- // <radix 2> := #b
- // <radix 8> := #o
- // <radix 10> := (?:#d)?
- // <radix 16> := #x
- // <digit 2> := [01]
- // <digit 8> := [0-7]
- // <digit 10> := \d
- // <digit 16> := [0-9a-f]
- //
- // The problem with this grammar is that the resulting regex is way to complex, so we simplify by grouping all
- // non-decimal bases together. This results in a decimal (dec) and combined binary, octal, and hexadecimal (box)
- // pattern:
- pattern: RegExp(SortedBNF({
- '<ureal dec>': /\d+(?:\/\d+)|(?:\d+(?:\.\d*)?|\.\d+)(?:[esfdl][+-]?\d+)?/.source,
- '<real dec>': /[+-]?<ureal dec>|[+-](?:inf|nan)\.0/.source,
- '<imaginary dec>': /[+-](?:<ureal dec>|(?:inf|nan)\.0)?i/.source,
- '<complex dec>': /<real dec>(?:@<real dec>|<imaginary dec>)?|<imaginary dec>/.source,
- '<num dec>': /(?:#d(?:#[ei])?|#[ei](?:#d)?)?<complex dec>/.source,
- '<ureal box>': /[0-9a-f]+(?:\/[0-9a-f]+)?/.source,
- '<real box>': /[+-]?<ureal box>|[+-](?:inf|nan)\.0/.source,
- '<imaginary box>': /[+-](?:<ureal box>|(?:inf|nan)\.0)?i/.source,
- '<complex box>': /<real box>(?:@<real box>|<imaginary box>)?|<imaginary box>/.source,
- '<num box>': /#[box](?:#[ei])?|(?:#[ei])?#[box]<complex box>/.source,
- '<number>': /(^|[()\[\]\s])(?:<num dec>|<num box>)(?=[()\[\]\s]|$)/.source,
- }), 'i'),
- lookbehind: true
- },
- 'boolean': {
- pattern: /(^|[()\[\]\s])#(?:[ft]|false|true)(?=[()\[\]\s]|$)/,
- lookbehind: true
- },
- 'function': {
- pattern: /((?:^|[^'`#])[(\[])(?:[^|()\[\]'\s]+|\|(?:[^\\|]|\\.)*\|)(?=[()\[\]\s]|$)/,
- lookbehind: true
- },
- 'identifier': {
- pattern: /(^|[()\[\]\s])\|(?:[^\\|]|\\.)*\|(?=[()\[\]\s]|$)/,
- lookbehind: true,
- greedy: true
- },
- 'punctuation': /[()\[\]']/
- };
- /**
- * Given a topologically sorted BNF grammar, this will return the RegExp source of last rule of the grammar.
- *
- * @param {Record<string, string>} grammar
- * @returns {string}
- */
- function SortedBNF(grammar) {
- for (var key in grammar) {
- grammar[key] = grammar[key].replace(/<[\w\s]+>/g, function (key) {
- return '(?:' + grammar[key].trim() + ')';
- });
- }
- // return the last item
- return grammar[key];
- }
- }(Prism));
|