reslop 0.2.4 → 0.2.6

package/lib/highlight/bash.js

@@ -1,217 +1,95 @@
 'use strict';
 
 const core = require('./core.js');
-const { SQUOTE, emit, wordStyle, readUntilNl } = core;
+const { words, emit, wordStyles, scanWhile, lineEnd } = core;
 
-const BASH_QUOTE = `${SQUOTE}\``;
+const SINGLE_QUOTES = `'\``;
 
-const KEYWORDS = [
-  'if',
-  'then',
-  'else',
-  'elif',
-  'fi',
-  'for',
-  'while',
-  'until',
-  'do',
-  'done',
-  'case',
-  'esac',
-  'function',
-  'select',
-  'in',
-  'time',
-  'coproc',
-];
+const KEYWORDS = words(`
+  if then else elif fi for while until do done case esac function select in
+  time coproc
+`);
 
-const BUILTINS = [
-  'echo',
-  'printf',
-  'cd',
-  'pwd',
-  'exit',
-  'export',
-  'source',
-  'alias',
-  'unalias',
-  'local',
-  'declare',
-  'typeset',
-  'readonly',
-  'unset',
-  'shift',
-  'read',
-  'test',
-  'true',
-  'false',
-  'exec',
-  'eval',
-  'set',
-  'trap',
-  'wait',
-  'kill',
-  'type',
-  'command',
-  'builtin',
-  'return',
-  'break',
-  'continue',
-  'pushd',
-  'popd',
-  'dirs',
-  'getopts',
-  'mapfile',
-  'readarray',
-  'let',
-  'umask',
-  'ulimit',
-  'fg',
-  'bg',
-  'jobs',
-  'hash',
-  'help',
-  'enable',
-  'caller',
-  'bind',
-  'compgen',
-  'complete',
-];
+const BUILTINS = words(`
+  echo printf cd pwd exit export source alias unalias local declare typeset
+  readonly unset shift read test true false exec eval set trap wait kill type
+  command builtin return break continue pushd popd dirs getopts mapfile
+  readarray let umask ulimit fg bg jobs hash help enable caller bind compgen
+  complete
+`);
 
-const WORD_STYLE = {
-  ...wordStyle(KEYWORDS, 'control'),
-  ...wordStyle(BUILTINS, 'function'),
-};
+const WORD_STYLE = wordStyles({ control: KEYWORDS, function: BUILTINS });
 
-const readBashVar = (source, i) => {
+const varEnd = (source, i) => {
   if (source[i + 1] === '{') {
-    let j = i + 2;
-    while (j < source.length && source[j] !== '}') j += 1;
-    if (j < source.length) j += 1;
-    return { text: source.slice(i, j), end: j };
-  }
-  if (/[0-9#?$!*@-]/.test(source[i + 1] || '')) {
-    return { text: source.slice(i, i + 2), end: i + 2 };
+    const close = source.indexOf('}', i + 2);
+    return close < 0 ? source.length : close + 1;
   }
-  let j = i + 1;
-  while (j < source.length && /[A-Za-z0-9_]/.test(source[j])) j += 1;
-  if (j === i + 1) return { text: '$', end: i + 1 };
-  return { text: source.slice(i, j), end: j };
+  if (/[0-9#?$!*@-]/.test(source[i + 1] || '')) return i + 2;
+  return scanWhile(source, i + 1, /[A-Za-z0-9_]/);
 };
 
-const takeQuoted = (source, i) => {
-  const q = source[i];
-  let j = i + 1;
+const pushDoubleQuoted = (out, source, i) => {
   const n = source.length;
-  while (j < n && source[j] !== q) j += 1;
-  if (j < n) j += 1;
-  return { text: source.slice(i, j), end: j };
-};
-
-const takeBashDouble = (source, i, out) => {
-  let j = i + 1;
   let chunk = '"';
-  const n = source.length;
+  let j = i + 1;
   while (j < n && source[j] !== '"') {
     if (source[j] === '\\' && j + 1 < n) {
       chunk += source.slice(j, j + 2);
       j += 2;
-      continue;
-    }
-    if (source[j] === '$') {
+    } else if (source[j] === '$') {
       emit(out, 'string', chunk);
       chunk = '';
-      const { text, end } = readBashVar(source, j);
-      emit(out, 'variable', text);
+      const end = varEnd(source, j);
+      emit(out, 'variable', source.slice(j, end));
       j = end;
-      continue;
+    } else {
+      chunk += source[j];
+      j += 1;
     }
-    chunk += source[j];
-    j += 1;
   }
   if (j < n) {
     chunk += '"';
     j += 1;
   }
-  if (chunk) emit(out, 'string', chunk);
+  emit(out, 'string', chunk);
   return j;
 };
 
-const bashWordStyle = (word, nextCh) => {
-  const known = WORD_STYLE[word];
-  if (known) return known;
-  if (nextCh === '=') return 'variable';
-  return 'plain';
-};
-
 const highlight = (source) => {
   const out = [];
   let i = 0;
-  const n = source.length;
-  while (i < n) {
+  const take = (style, end) => {
+    emit(out, style, source.slice(i, end));
+    i = end;
+  };
+  while (i < source.length) {
     const ch = source[i];
     const next = source[i + 1];
-
     if (ch === '#' && (i === 0 || /\s/.test(source[i - 1]))) {
-      const { text, end } = readUntilNl(source, i);
-      emit(out, 'comment', text);
-      i = end;
-      continue;
+      take('comment', lineEnd(source, i));
+    } else if (SINGLE_QUOTES.includes(ch)) {
+      const close = source.indexOf(ch, i + 1);
+      take('string', close < 0 ? source.length : close + 1);
+    } else if (ch === '"') {
+      i = pushDoubleQuoted(out, source, i);
+    } else if (ch === '$') {
+      take('variable', varEnd(source, i));
+    } else if (ch === '-' && /[A-Za-z0-9]/.test(next || '')) {
+      take('attr', scanWhile(source, i + 1, /[A-Za-z0-9_-]/));
+    } else if (/[A-Za-z_]/.test(ch)) {
+      const end = scanWhile(source, i + 1, /[A-Za-z0-9_]/);
+      const word = source.slice(i, end);
+      const assign = source[end] === '=' ? 'variable' : 'plain';
+      take(WORD_STYLE[word] || assign, end);
+    } else if (/[0-9]/.test(ch)) {
+      take('number', scanWhile(source, i + 1, /[0-9]/));
+    } else if ('|&;<>(){}'.includes(ch)) {
+      const doubled = '|&><'.includes(ch) && next === ch;
+      take('operator', i + (doubled ? 2 : 1));
+    } else {
+      take('plain', i + 1);
     }
-
-    if (BASH_QUOTE.includes(ch)) {
-      const { text, end } = takeQuoted(source, i);
-      emit(out, 'string', text);
-      i = end;
-      continue;
-    }
-
-    if (ch === '"') {
-      i = takeBashDouble(source, i, out);
-      continue;
-    }
-
-    if (ch === '$') {
-      const { text, end } = readBashVar(source, i);
-      emit(out, 'variable', text);
-      i = end;
-      continue;
-    }
-
-    if (ch === '-' && /[A-Za-z0-9]/.test(next || '')) {
-      let j = i + 1;
-      while (j < n && /[A-Za-z0-9_-]/.test(source[j])) j += 1;
-      emit(out, 'attr', source.slice(i, j));
-      i = j;
-      continue;
-    }
-
-    if (/[A-Za-z_]/.test(ch)) {
-      let j = i + 1;
-      while (j < n && /[A-Za-z0-9_]/.test(source[j])) j += 1;
-      const word = source.slice(i, j);
-      emit(out, bashWordStyle(word, source[j]), word);
-      i = j;
-      continue;
-    }
-
-    if (/[0-9]/.test(ch)) {
-      let j = i + 1;
-      while (j < n && /[0-9]/.test(source[j])) j += 1;
-      emit(out, 'number', source.slice(i, j));
-      i = j;
-      continue;
-    }
-
-    if ('|&;<>(){}'.includes(ch)) {
-      const j = i + ('|&><'.includes(ch) && next === ch ? 2 : 1);
-      emit(out, 'operator', source.slice(i, j));
-      i = j;
-      continue;
-    }
-
-    emit(out, 'plain', ch);
-    i += 1;
   }
   return out;
 };

package/lib/highlight/core.js

@@ -1,34 +1,17 @@
 'use strict';
 
-const SQUOTE = '\x27';
-const QUOTES = `${SQUOTE}"`;
-const ANY_QUOTE = `${QUOTES}\``;
-
-const OP3 = ['===', '!==', '>>>', '**=', '&&=', '||=', '??='];
-
-const OP2 = [
-  '==',
-  '!=',
-  '<=',
-  '>=',
-  '&&',
-  '||',
-  '??',
-  '=>',
-  '++',
-  '--',
-  '<<',
-  '>>',
-  '**',
-  '+=',
-  '-=',
-  '*=',
-  '/=',
-  '%=',
-  '&=',
-  '|=',
-  '^=',
-];
+const QUOTES = `'"`;
+const ANY_QUOTE = `'"\``;
+const IDENT_START = /[A-Za-z_$]/;
+const IDENT_CHAR = /[A-Za-z0-9_$]/;
+const CONSTANT_RE = /^[A-Z][A-Z0-9_]+$/;
+
+const words = (text) => text.trim().split(/\s+/);
+
+const OP3 = words('=== !== >>> **= &&= ||= ??=');
+const OP2 = words(
+  '== != <= >= && || ?? => ++ -- << >> ** += -= *= /= %= &= |= ^=',
+);
 
 const EXPECT_STYLE = {
   function: 'function',
@@ -36,159 +19,201 @@ const EXPECT_STYLE = {
   type: 'className',
 };
 
+const RADIX_DIGIT = { x: /[0-9a-fA-F_]/, b: /[01_]/, o: /[0-7_]/ };
+
 const emit = (out, style, text) => {
-  if (!text) return;
-  out.push({ text, style: style || 'plain' });
+  if (text) out.push({ text, style });
 };
 
+const highlightLines =
+  (highlightLine, separator = '\n') =>
+  (source) => {
+    const lines = source.split(separator);
+    const out = [];
+    for (let i = 0; i < lines.length; i += 1) {
+      if (i) emit(out, 'plain', '\n');
+      for (const token of highlightLine(lines[i])) out.push(token);
+    }
+    return out;
+  };
+
 const tokensText = (out) => {
-  let s = '';
-  for (const t of out) s += t.text;
-  return s;
+  let text = '';
+  for (const token of out) text += token.text;
+  return text;
 };
 
-const wordStyle = (words, style) => {
+const wordStyles = (groups) => {
   const table = Object.create(null);
-  for (const word of words) table[word] = style;
+  for (const style of Object.keys(groups)) {
+    for (const word of groups[style]) table[word] = style;
+  }
   return table;
 };
 
-const peekIdent = (source, i) => {
-  if (!/[A-Za-z_$]/.test(source[i] || '')) return null;
-  let j = i + 1;
-  while (j < source.length && /[A-Za-z0-9_$]/.test(source[j])) j += 1;
-  return source.slice(i, j);
+const appendNode = (out, node) => {
+  for (const token of node.tokens) out.push(token);
+  return node.end;
 };
 
-const skipWs = (source, i) => {
-  while (i < source.length && /\s/.test(source[i])) i += 1;
+const scanWhile = (source, start, re) => {
+  let i = start;
+  while (i < source.length && re.test(source[i])) i += 1;
   return i;
 };
 
-const readUntilNl = (source, i) => {
-  let j = i;
-  const n = source.length;
-  while (j < n && source[j] !== '\n') j += 1;
-  return { text: source.slice(i, j), end: j };
+const lineEnd = (source, i) => {
+  const end = source.indexOf('\n', i);
+  return end < 0 ? source.length : end;
 };
 
-const readBlockComment = (source, i) => {
-  let j = i + 2;
-  const n = source.length;
-  while (j < n && !(source[j] === '*' && source[j + 1] === '/')) {
-    j += 1;
-  }
-  j = Math.min(j + 2, n);
-  return { text: source.slice(i, j), end: j };
+const blockCommentEnd = (source, i) => {
+  const end = source.indexOf('*/', i + 2);
+  return end < 0 ? source.length : end + 2;
 };
 
-const skipQuoted = (source, j) => {
-  const q = source[j];
-  j += 1;
-  const n = source.length;
-  while (j < n && source[j] !== q) {
+const quotedEnd = (source, start) => {
+  const quote = source[start];
+  let j = start + 1;
+  while (j < source.length && source[j] !== quote) {
     if (source[j] === '\\') j += 1;
     j += 1;
   }
-  j += 1;
-  return j;
-};
-
-const readJsNumber = (source, i) => {
-  const n = source.length;
-  let j = i;
-  if (source.startsWith('0x', i) || source.startsWith('0X', i)) {
-    j += 2;
-    while (j < n && /[0-9a-fA-F_]/.test(source[j])) j += 1;
-  } else if (source.startsWith('0b', i) || source.startsWith('0B', i)) {
-    j += 2;
-    while (j < n && /[01_]/.test(source[j])) j += 1;
-  } else if (source.startsWith('0o', i) || source.startsWith('0O', i)) {
-    j += 2;
-    while (j < n && /[0-7_]/.test(source[j])) j += 1;
-  } else {
-    while (j < n && /[0-9_n.]/.test(source[j])) j += 1;
-    if ('eE'.includes(source[j])) {
-      j += 1;
-      if ('+-'.includes(source[j])) j += 1;
-      while (j < n && /[0-9_]/.test(source[j])) j += 1;
-    }
-  }
-  return { text: source.slice(i, j), end: j };
+  return j + 1;
 };
 
-const readJsOperator = (source, i) => {
-  const two = source.slice(i, i + 2);
-  const three = source.slice(i, i + 3);
+const stringEnd = (source, i) => {
+  const quote = source[i];
   let j = i + 1;
-  if (OP3.includes(three)) j = i + 3;
-  else if (OP2.includes(two)) j = i + 2;
-  return { text: source.slice(i, j), end: j };
-};
-
-const readString = (source, i) => {
-  const q = source[i];
-  let j = i + 1;
-  const n = source.length;
-  while (j < n) {
+  while (j < source.length) {
     if (source[j] === '\\') {
       j += 2;
       continue;
     }
-    if (source[j] === q) {
-      j += 1;
-      break;
-    }
+    if (source[j] === quote) return j + 1;
     if (source[j] === '\n') break;
     j += 1;
   }
-  return { text: source.slice(i, j), end: j };
+  return j;
+};
+
+const numberEnd = (source, i) => {
+  const prefix = source[i] === '0' ? source[i + 1] : undefined;
+  const radix = prefix ? RADIX_DIGIT[prefix.toLowerCase()] : undefined;
+  if (radix) return scanWhile(source, i + 2, radix);
+  let j = scanWhile(source, i, /[0-9_n.]/);
+  if (source[j] !== 'e' && source[j] !== 'E') return j;
+  j += 1;
+  if (source[j] === '+' || source[j] === '-') j += 1;
+  return scanWhile(source, j, /[0-9_]/);
+};
+
+const operatorEnd = (source, i) => {
+  if (OP3.includes(source.slice(i, i + 3))) return i + 3;
+  if (OP2.includes(source.slice(i, i + 2))) return i + 2;
+  return i + 1;
 };
 
-const readTemplateInterp = (source, j) => {
+const interpEnd = (source, start) => {
   let depth = 1;
-  const start = j;
-  const n = source.length;
-  while (j < n && depth > 0) {
-    if (ANY_QUOTE.includes(source[j])) {
-      j = skipQuoted(source, j);
+  let j = start;
+  while (j < source.length) {
+    const ch = source[j];
+    if (ANY_QUOTE.includes(ch)) {
+      j = quotedEnd(source, j);
       continue;
     }
-    if (source[j] === '{') depth += 1;
-    if (source[j] === '}') {
+    if (ch === '{') depth += 1;
+    if (ch === '}') {
       depth -= 1;
       if (depth === 0) break;
     }
     j += 1;
   }
-  return { expr: source.slice(start, j), end: j };
+  return j;
 };
 
-const styleConstant = ({ word }) =>
-  /^[A-Z][A-Z0-9_]+$/.test(word) ? 'constant' : null;
+const pushInterpolation = (out, source, start, inner) => {
+  const end = interpEnd(source, start);
+  for (const token of inner(source.slice(start, end))) out.push(token);
+  if (source[end] !== '}') return end;
+  emit(out, 'interpolation', '}');
+  return end + 1;
+};
 
-const stylePascal = ({ word }) => (/^[A-Z]/.test(word) ? 'className' : null);
+const highlightInterpolated = (source, start, body, kind, inner) => {
+  const { style, quote, raw } = kind;
+  const n = source.length;
+  const out = [];
+  emit(out, style, source.slice(start, body));
+  let j = body;
+  let buf = '';
+  const flush = () => {
+    emit(out, style, buf);
+    buf = '';
+  };
+  while (j < n) {
+    const ch = source[j];
+    if (!raw && ch === '\\') {
+      flush();
+      emit(out, 'escape', source.slice(j, j + 2));
+      j += 2;
+      continue;
+    }
+    if (source.startsWith(quote, j)) {
+      flush();
+      emit(out, style, quote);
+      j += quote.length;
+      break;
+    }
+    if (!raw && ch === '$' && source[j + 1] === '{') {
+      flush();
+      emit(out, 'interpolation', '${');
+      j = pushInterpolation(out, source, j + 2, inner);
+      continue;
+    }
+    const isDollar = !raw && ch === '$' && kind.dollarIdent;
+    if (isDollar && IDENT_START.test(source[j + 1] || '')) {
+      flush();
+      const end = scanWhile(source, j + 1, IDENT_CHAR);
+      emit(out, 'interpolation', source.slice(j, end));
+      j = end;
+      continue;
+    }
+    if (!kind.multiline && ch === '\n') break;
+    buf += ch;
+    j += 1;
+  }
+  flush();
+  return { tokens: out, end: j };
+};
 
-const styleCall = ({ after }) => (after === '(' ? 'function' : null);
+const nameStyle = (word, after) => {
+  if (CONSTANT_RE.test(word)) return 'constant';
+  if (/^[A-Z]/.test(word)) return 'className';
+  if (after === '(') return 'function';
+  return null;
+};
 
 module.exports = {
-  SQUOTE,
   QUOTES,
-  ANY_QUOTE,
+  IDENT_START,
+  IDENT_CHAR,
   EXPECT_STYLE,
+  words,
   emit,
+  highlightLines,
   tokensText,
-  wordStyle,
-  peekIdent,
-  skipWs,
-  readUntilNl,
-  readBlockComment,
-  skipQuoted,
-  readJsNumber,
-  readJsOperator,
-  readString,
-  readTemplateInterp,
-  styleConstant,
-  stylePascal,
-  styleCall,
+  wordStyles,
+  appendNode,
+  scanWhile,
+  lineEnd,
+  blockCommentEnd,
+  quotedEnd,
+  stringEnd,
+  numberEnd,
+  operatorEnd,
+  pushInterpolation,
+  highlightInterpolated,
+  nameStyle,
 };