perf: improved tokenizer

Faster and simpler tokenization. Refs: #133
scriptcoded · Oct 4, 2023 · f282666 · f282666
1 parent bf2e2c2
commit f282666
Show file tree

Hide file tree

Showing 2 changed files with 47 additions and 85 deletions.
diff --git a/lib/index.d.ts b/lib/index.d.ts
@@ -21,4 +21,5 @@ declare module 'sql-highlight' {
 
   export function getSegments(sqlString: string): Array<Segment>;
   export function highlight(sqlString: string, options?: HighlightOptions): string;
+  export const DEFAULT_OPTIONS: HighlightOptions;
 }
diff --git a/lib/index.js b/lib/index.js
@@ -18,106 +18,66 @@ const DEFAULT_OPTIONS = {
   }
 }
 
-const SPLIT_CHARS = '[^a-zA-Z_]'
-
 const DEFAULT_KEYWORD = 'default'
 
 const highlighters = [
-  {
-    name: 'keyword',
-    group: 1,
-    regex: new RegExp(`(^|${SPLIT_CHARS})(${keywords.join('|')})(?=${SPLIT_CHARS}|$)`, 'gi')
-  },
-  {
-    name: 'special',
-    regex: /(=|!=|%|\/|\*|-|,|;|:|\+|<|>)/g
-  },
-  {
-    name: 'function',
-    regex: /(\w+?)\(/g,
-    trimEnd: 1
-  },
-  {
-    name: 'number',
-    regex: /(\b\d+(?:\.\d+)?)/g
-  },
-  {
-    name: 'string',
-    regex: /(['](?:\\'|.)*?[']|["](?:\\"|.)*?["]|[`](?:\\`|.)*?[`])/g
-  },
-  {
-    name: 'bracket',
-    regex: /([()])/g
-  }
-]
+  /\b(?<number>\d+(?:\.\d+)?)\b/,
 
-function getSegments (sqlString) {
-  const matches = []
-
-  for (const hl of highlighters) {
-    let match
-
-    // This is probably the one time when an assignment inside a condition makes sense
-    // eslint-disable-next-line no-cond-assign
-    while (match = hl.regex.exec(sqlString)) {
-      let text = match[0]
-      let boringLength = 0
-
-      // If a specific group is requested, use that group instead, and make sure
-      // we offset the index by the length of the preceding groups
-      if (hl.group) {
-        text = match[hl.group + 1]
-        for (let i = 1; i <= hl.group; i++) {
-          boringLength += match[i].length
-        }
-      }
+  // Note: Repeating string escapes like 'sql''server' will also work as they are just repeating strings
+  /(?<string>'(?:[^'\\]|\\.)*'|"(?:[^"\\]|\\.)*"|`(?:[^`\\]|\\.)*`)/,
 
-      const trimmedText = hl.trimEnd
-        ? text.substring(0, text.length - hl.trimEnd)
-        : text
-      matches.push({
-        name: hl.name,
-        start: match.index + boringLength,
-        length: trimmedText.length
-      })
-    }
-  }
+  /\b(?<function>\w+)(?=\s*\()/,
+
+  /(?<bracket>[()])/,
 
-  const sortedMatches = matches.slice().sort((a, b) => a.start - b.start)
+  /(?<special>!=|[=%*/\-+,;:<>])/
+]
+
+function getRegexString (regex) {
+  const str = regex.toString()
+  return str.replace(/^\/|\/\w*$/g, '')
+}
 
-  // filter/exclude nested matches (matches within the last match)
+// Regex of the shape /(.*?)|((?<token1>...)|(?<token2>...)|...|$)/y
+const tokenizer = new RegExp(
+  '(.*?)(' +
+  '\\b(?<keyword>' + keywords.join('|') + ')\\b|' +
+  highlighters.map(getRegexString).join('|') +
+  '|$)', // $ needed to to match "default" till the end of string
+  'isy'
+)
+
+function getSegments (sqlString) {
   const segments = []
-  let upperBound = 0
-  for (let i = 0; i < sortedMatches.length; i++) {
-    if (sortedMatches[i].start < upperBound) { continue }
+  let match
+
+  // Reset the starting position
+  tokenizer.lastIndex = 0
 
-    // If no match, add a default segment
-    if (sortedMatches[i].start > upperBound) {
+  // This is probably the one time when an assignment inside a condition makes sense
+  // eslint-disable-next-line no-cond-assign
+  while (match = tokenizer.exec(sqlString)) {
+    if (match[1]) {
       segments.push({
         name: DEFAULT_KEYWORD,
-        content: sqlString.substring(upperBound, sortedMatches[i].start)
+        content: match[1]
       })
     }
 
-    segments.push({
-      name: sortedMatches[i].name,
-      content: sqlString.substring(
-        sortedMatches[i].start,
-        sortedMatches[i].start + sortedMatches[i].length
-      )
-    })
-    upperBound = sortedMatches[i].start + sortedMatches[i].length
-  }
+    if (match[2]) {
+      const name = Object.keys(match.groups).find(key => match.groups[key])
+      segments.push({
+        name,
+        content: match.groups[name]
+      })
+    }
 
-  if (upperBound < sqlString.length - 1) {
-    segments.push({
-      name: DEFAULT_KEYWORD,
-      content: sqlString.substring(
-        upperBound,
-        upperBound + sqlString.length + 1
-      )
-    })
+    // Stop at the end of string
+    if (match.index + match[0].length >= sqlString.length) {
+      break
+    }
   }
+
   return segments
 }
 
@@ -140,5 +100,6 @@ function highlight (sqlString, options) {
 
 module.exports = {
   getSegments,
-  highlight
+  highlight,
+  DEFAULT_OPTIONS
 }