fix: avoid parsing ridiculously long strings (#489)

* fix: avoid parsing ridiculously long strings

* fix: debug

* update e2e

* fix: bun install after module update

* fix: install deps

.github/workflows/bun-clickhouse.js.yml

@@ -39,9 +39,10 @@ jobs:
       with:
         bun-version: ${{ matrix.bun-version }}
     - run: rm -rf package-lock.json
-    - run: bun install
     - run: git submodule init
     - run: git submodule update
+    - run: bun install
+    - run: cd test/e2e && bun install
     - name: Workflow Telemetry
       uses: runforesight/[email protected]
       if: github.event_name != 'pull_request'

parser/bnf.js

@@ -70,4 +70,36 @@ for (const [name, rule] of Object.entries(compiler.languages.logql.rules)) {
   }
 }
 
+compiler._ParseScript = compiler.ParseScript
+/**
+ * hack to avoid ridiculously long strings
+ * @param script {string}
+ * @constructor
+ */
+compiler.ParseScript = function (script) {
+  const qLiterals = []
+  const aqLiterals = []
+  const quotedStrings = script.replaceAll(/"([^"\\]|\\.)+"/g, (str) => {
+    qLiterals.push(str)
+    return `"QL_${qLiterals.length - 1}"`
+  })
+  const aQuotedStrings = quotedStrings.replaceAll(/`([^`\\]|\\.)+`/g, (str) => {
+    aqLiterals.push(str)
+    return `\`AL_${aqLiterals.length - 1}\``
+  })
+  const parsedScript = this._ParseScript(aQuotedStrings)
+  if (!parsedScript) {
+    return parsedScript
+  }
+  for (const t of parsedScript.rootToken.Children('QLITERAL')) {
+    t._value = qLiterals[parseInt(t.value.slice(4, t.value.length - 1))]
+    t.tokens = []
+  }
+  for (const t of parsedScript.rootToken.Children('AQLITERAL')) {
+    t._value = aqLiterals[parseInt(t.value.slice(4, t.value.length - 1))]
+    t.tokens = []
+  }
+  return parsedScript
+}
+
 module.exports = compiler