/** * @import { * Construct, * State, * TokenizeContext, * Tokenizer * } from 'micromark-util-types' */ import { asciiAlphanumeric, asciiAlpha, asciiAtext, asciiControl } from 'micromark-util-character'; /** @type {Construct} */ export const autolink = { name: 'autolink', tokenize: tokenizeAutolink }; /** * @this {TokenizeContext} * Context. * @type {Tokenizer} */ function tokenizeAutolink(effects, ok, nok) { let size = 0; return start; /** * Start of an autolink. * * ```markdown * > | ab * ^ * > | ab * ^ * ``` * * @type {State} */ function start(code) { effects.enter("autolink"); effects.enter("autolinkMarker"); effects.consume(code); effects.exit("autolinkMarker"); effects.enter("autolinkProtocol"); return open; } /** * After `<`, at protocol or atext. * * ```markdown * > | ab * ^ * > | ab * ^ * ``` * * @type {State} */ function open(code) { if (asciiAlpha(code)) { effects.consume(code); return schemeOrEmailAtext; } if (code === 64) { return nok(code); } return emailAtext(code); } /** * At second byte of protocol or atext. * * ```markdown * > | ab * ^ * > | ab * ^ * ``` * * @type {State} */ function schemeOrEmailAtext(code) { // ASCII alphanumeric and `+`, `-`, and `.`. if (code === 43 || code === 45 || code === 46 || asciiAlphanumeric(code)) { // Count the previous alphabetical from `open` too. size = 1; return schemeInsideOrEmailAtext(code); } return emailAtext(code); } /** * In ambiguous protocol or atext. * * ```markdown * > | ab * ^ * > | ab * ^ * ``` * * @type {State} */ function schemeInsideOrEmailAtext(code) { if (code === 58) { effects.consume(code); size = 0; return urlInside; } // ASCII alphanumeric and `+`, `-`, and `.`. if ((code === 43 || code === 45 || code === 46 || asciiAlphanumeric(code)) && size++ < 32) { effects.consume(code); return schemeInsideOrEmailAtext; } size = 0; return emailAtext(code); } /** * After protocol, in URL. * * ```markdown * > | ab * ^ * ``` * * @type {State} */ function urlInside(code) { if (code === 62) { effects.exit("autolinkProtocol"); effects.enter("autolinkMarker"); effects.consume(code); effects.exit("autolinkMarker"); effects.exit("autolink"); return ok; } // ASCII control, space, or `<`. if (code === null || code === 32 || code === 60 || asciiControl(code)) { return nok(code); } effects.consume(code); return urlInside; } /** * In email atext. * * ```markdown * > | ab * ^ * ``` * * @type {State} */ function emailAtext(code) { if (code === 64) { effects.consume(code); return emailAtSignOrDot; } if (asciiAtext(code)) { effects.consume(code); return emailAtext; } return nok(code); } /** * In label, after at-sign or dot. * * ```markdown * > | ab * ^ ^ * ``` * * @type {State} */ function emailAtSignOrDot(code) { return asciiAlphanumeric(code) ? emailLabel(code) : nok(code); } /** * In label, where `.` and `>` are allowed. * * ```markdown * > | ab * ^ * ``` * * @type {State} */ function emailLabel(code) { if (code === 46) { effects.consume(code); size = 0; return emailAtSignOrDot; } if (code === 62) { // Exit, then change the token type. effects.exit("autolinkProtocol").type = "autolinkEmail"; effects.enter("autolinkMarker"); effects.consume(code); effects.exit("autolinkMarker"); effects.exit("autolink"); return ok; } return emailValue(code); } /** * In label, where `.` and `>` are *not* allowed. * * Though, this is also used in `emailLabel` to parse other values. * * ```markdown * > | ab * ^ * ``` * * @type {State} */ function emailValue(code) { // ASCII alphanumeric or `-`. if ((code === 45 || asciiAlphanumeric(code)) && size++ < 63) { const next = code === 45 ? emailValue : emailLabel; effects.consume(code); return next; } return nok(code); } }