Fix/re2 wasm heap leak

.github/workflows/docker.yml

@@ -3,6 +3,8 @@ name: Docker Build
 on:
   push:
     branches: [ main ]
+  workflow_dispatch:
+
 
 jobs:
   docker:
@@ -32,4 +34,4 @@ jobs:
           platforms: linux/amd64,linux/arm64
           tags: |
             ghcr.io/${{ env.REPO_LC }}:cibuild
-            ghcr.io/${{ env.REPO_LC }}:cibuild-${{ github.sha }}
+            ghcr.io/${{ env.REPO_LC }}:cibuild-${{ github.sha }}

library/regex-utilities.js

@@ -1,13 +1,27 @@
-const { RE2 } = require('re2-wasm');
+const { RE2 } = require('re2-wasm');
 
 class RegExUtilities {
 
+  constructor() {
+    this._cache = new Map();
+  }
+
   compile(pattern, flags) {
-    // RE2 requires the unicode flag; add it if not already present
+    // re2-wasm has a fixed 16 MB WASM heap with no real free(): every
+    // new RE2(...) permanently consumes a few KB. Cache by (pattern, flags)
+    // so the same regex compiles at most once.
+    // TODO: replace re2-wasm with native re2 to eliminate the underlying leak.
     const re2Flags = flags && flags.includes('u') ? flags : (flags || '') + 'u';
-    return new RE2(pattern, re2Flags);
+    const key = pattern + '|' + re2Flags;
+    let compiled = this._cache.get(key);
+    if (!compiled) {
+      compiled = new RE2(pattern, re2Flags);
+      this._cache.set(key, compiled);
+    }
+    return compiled;
   }
 
 }
 
 module.exports = new RegExUtilities();
+

test-scripts/repro-re2-wasm-leak.js

@@ -0,0 +1,33 @@
+// Manual reproducer for the re2-wasm WASM heap leak.
+// See library/regex-utilities.js for the workaround.
+//
+//   node test-scripts/repro-re2-wasm-leak.js            # same-pattern stress
+//   node test-scripts/repro-re2-wasm-leak.js --unique   # unique-pattern stress
+//
+// Without the cache, same-pattern OOMs at ~2965 iterations.
+// With the cache, same-pattern runs indefinitely.
+// Unique-pattern still OOMs (each pattern is a real compile and the underlying
+// re2-wasm heap leak still applies). A proper fix is to replace re2-wasm with
+// the native `re2` package.
+
+const re = require('../library/regex-utilities');
+
+const mode = process.argv.includes('--unique') ? 'unique' : 'same';
+const basePattern =
+  'CYTO|HL7\\.CYTOGEN|HL7\\.GENETICS|^PATH(\\..*)?|^MOLPATH(\\..*)?|NR STATS|H&P\\.HX\\.LAB|CHALSKIN|LABORDERS';
+
+console.log(`mode: ${mode}-pattern`);
+
+let i = 0;
+try {
+  for (;;) {
+    const pattern = mode === 'unique' ? basePattern + `|UNIQUE${i}` : basePattern;
+    re.compile(pattern);
+    i++;
+    if (i % 100 === 0) console.log(`iter ${i}`);
+  }
+} catch (e) {
+  console.error(`OOMed at iteration ${i}: ${e.message}`);
+  process.exit(1);
+}
+