feat(cache): reuse prepared sqlite statements. Also make :memory: cache the default.

Author: IsakT

lib/interceptor/cache.js

@@ -1,13 +1,14 @@
 import assert from 'node:assert'
 import { DecoratorHandler, parseHeaders, parseCacheControl } from '../utils.js'
 import { DatabaseSync } from 'node:sqlite' // --experimental-sqlite
+import * as BJSON from 'buffer-json'
 
 class CacheHandler extends DecoratorHandler {
   #handler
   #store
   #key
   #opts
-  #value = null
+  #value
 
   constructor({ key, handler, store, opts = [] }) {
     super(handler)
@@ -25,13 +26,11 @@ class CacheHandler extends DecoratorHandler {
   }
 
   onHeaders(statusCode, rawHeaders, resume, statusMessage, headers = parseHeaders(rawHeaders)) {
-    if (statusCode !== 307) {
+    if (statusCode !== 307 || statusCode !== 200) {
       return this.#handler.onHeaders(statusCode, rawHeaders, resume, statusMessage, headers)
     }
 
-    // TODO (fix): Support vary header.
     const cacheControl = parseCacheControl(headers['cache-control'])
-
     const contentLength = headers['content-length'] ? Number(headers['content-length']) : Infinity
     const maxEntrySize = this.#store.maxEntrySize ?? Infinity
 
@@ -66,7 +65,7 @@ class CacheHandler extends DecoratorHandler {
             (rawHeaders?.reduce((xs, x) => xs + x.length, 0) ?? 0) +
             (statusMessage?.length ?? 0) +
             64,
-          expires: Date.now() + ttl, // in ms!
+          expires: Date.now() + ttl,
         }
       }
     }
@@ -89,20 +88,14 @@ class CacheHandler extends DecoratorHandler {
 
   onComplete(rawTrailers) {
     if (this.#value) {
+      const reqHeaders = this.#opts
       const resHeaders = parseHeaders(this.#value.data.rawHeaders)
 
       // Early return if Vary = *, uncacheable.
       if (resHeaders.vary === '*') {
         return this.#handler.onComplete(rawTrailers)
       }
 
-      const reqHeaders = this.#opts
-
-      // If Range header present, assume that the response varies based on Range.
-      if (reqHeaders.headers?.range) {
-        resHeaders.vary += ', Range'
-      }
-
       this.#value.data.rawTrailers = rawTrailers
       this.#value.size = this.#value.size
         ? this.#value.size + rawTrailers?.reduce((xs, x) => xs + x.length, 0)
@@ -112,6 +105,7 @@ class CacheHandler extends DecoratorHandler {
 
       this.#store.set(this.#key, this.#value)
     }
+
     return this.#handler.onComplete(rawTrailers)
   }
 }
@@ -120,20 +114,27 @@ function formatVaryData(resHeaders, reqHeaders) {
   return resHeaders.vary
     ?.split(',')
     .map((key) => key.trim().toLowerCase())
-    .map((key) => [key, reqHeaders[key] ?? reqHeaders.headers[key]])
-    .filter(([_key, val]) => val)
+    .map((key) => [key, reqHeaders[key] ?? ''])
+    .filter(([, val]) => val)
 }
 
 export class CacheStore {
-  constructor() {
-    this.database = null
-    this.init()
-  }
+  #database
+
+  #insertquery
+  #getQuery
+  #purgeQuery
+
+  #size = 0
+  #maxSize = 128e9
+
+  constructor(location = ':memory:', opts) {
+    // TODO (fix): Validate args...
 
-  init() {
-    this.database = new DatabaseSync('file:memdb1?mode=memory&cache=shared')
+    this.#maxSize = opts.maxSize ?? this.#maxSize
+    this.#database = new DatabaseSync(location)
 
-    this.database.exec(`
+    this.#database.exec(`
       CREATE TABLE IF NOT EXISTS cacheInterceptor(
         key TEXT,
         data TEXT,
@@ -142,91 +143,51 @@ export class CacheStore {
         expires INTEGER
       ) STRICT
     `)
-  }
-
-  set(key, entry) {
-    if (!this.database) {
-      throw new Error('Database not initialized')
-    }
 
-    entry.data = JSON.stringify(entry.data)
-    entry.vary = JSON.stringify(entry.vary)
-
-    const insert = this.database.prepare(
+    this.#insertquery = this.#database.prepare(
       'INSERT INTO cacheInterceptor (key, data, vary, size, expires) VALUES (?, ?, ?, ?, ?)',
     )
 
-    insert.run(key, entry.data, entry.vary, entry.size, entry.expires)
+    this.#getQuery = this.#database.prepare(
+      'SELECT * FROM cacheInterceptor WHERE key = ? AND expires > ? ',
+    )
+
+    this.#purgeQuery = this.#database.prepare('DELETE FROM cacheInterceptor WHERE expires < ?')
 
-    this.purge()
+    this.#maybePurge()
   }
 
-  get(key) {
-    if (!this.database) {
-      throw new Error('Database not initialized')
-    }
-    this.purge()
-    const query = this.database.prepare(
-      'SELECT * FROM cacheInterceptor WHERE key = ? AND expires > ? ',
-    )
-    const rows = query.all(key, Date.now())
-    rows.map((i) => {
-      i.data = JSON.parse(i.data)
-      i.vary = JSON.parse(i.vary)
-      i.data = {
-        ...i.data,
-        // JSON.parse doesn't convert a Buffer object back to a Buffer object once it has been stringified.
-        body: this.#convertToBuffer(i.data.body),
-        rawHeaders: this.#convertToBuffer(i.data.rawHeaders),
-        rawTrailers: this.#convertToBuffer(i.data.rawTrailers),
-      }
-      return i
-    })
+  set(key, { data, vary, size, expires }) {
+    this.#insertquery.run(key, JSON.stringify(data), BJSON.stringify(vary), size, expires)
 
-    return rows
+    this.#size += size
+    this.#maybePurge()
   }
 
-  purge() {
-    const query = this.database.prepare('DELETE FROM cacheInterceptor WHERE expires < ?')
-    query.run(Date.now())
+  get(key) {
+    return this.#getQuery.all(key, Date.now()).map(({ data, vary, size, expires }) => ({
+      data: BJSON.parse(data),
+      vary: JSON.parse(vary),
+      size: parseInt(size), // TODO (fix): Is parseInt necessary?
+      expores: parseInt(expires), // TODO (fix): Is parseInt necessary?
+    }))
   }
 
-  deleteAll() {
-    const query = this.database.prepare('DELETE FROM cacheInterceptor')
-    query.run()
+  close() {
+    this.#database.close()
   }
 
-  #convertToBuffer(bufferArray) {
-    if (Array.isArray(bufferArray) && bufferArray.length > 0) {
-      return bufferArray.map((ba) => {
-        return typeof ba === 'object' ? Buffer.from(ba.data) : ba
-      })
+  #maybePurge() {
+    if (this.#size == null || this.#size > this.#maxSize) {
+      this.#purgeQuery.run(Date.now())
+      this.#size = this.#database.exec('SELECT SUM(size) FROM cacheInterceptor')[0].values[0][0]
     }
-    return []
   }
 }
 
-/*
-  Sort entries by number of vary headers in descending order, because
-  we need to compare the most complex response to the request first.
-  A cached response with an empty ´vary´ field will otherwise win every time.
-*/
-function sortEntriesByVary(entries) {
-  entries.sort((a, b) => {
-    const lengthA = a.vary ? a.vary.length : 0
-    const lengthB = b.vary ? b.vary.length : 0
-    return lengthB - lengthA
-  })
-}
-
 function findEntryByHeaders(entries, reqHeaders) {
-  sortEntriesByVary(entries)
-
   return entries?.find(
-    (entry) =>
-      entry.vary?.every(([key, val]) => {
-        return reqHeaders?.headers[key] === val
-      }) ?? true,
+    (entry) => entry.vary?.every(([key, val]) => reqHeaders?.headers[key] === val) ?? true,
   )
 }
 
@@ -276,7 +237,7 @@ export default (opts) => (dispatch) => (opts, handler) => {
 
   const key = `${opts.method}:${opts.path}`
 
-  const entries = (store.get(key) ?? opts.method === 'HEAD') ? store.get(`GET:${opts.path}`) : null
+  const entries = store.get(key) ?? (opts.method === 'HEAD' ? store.get(`GET:${opts.path}`) : null)
 
   const entry = findEntryByHeaders(entries, opts)
 

package.json

@@ -9,6 +9,7 @@
     "lib/*"
   ],
   "dependencies": {
+    "buffer-json": "^2.0.0",
     "cache-control-parser": "^2.0.6",
     "cacheable-lookup": "^7.0.0",
     "http-errors": "^2.0.0",

test/cache.js

@@ -133,7 +133,7 @@ test('If no matching entry found, store the response in cache. Else return a mat
     // response not found in cache, response should be added to cache.
     const response = await undici.request(`http://0.0.0.0:${serverPort}`, {
       dispatcher: new undici.Agent().compose(interceptors.cache()),
-      cache: true,
+      cache,
     })
     let str = ''
     for await (const chunk of response.body) {
@@ -154,7 +154,7 @@ test('If no matching entry found, store the response in cache. Else return a mat
         'User-Agent': 'Chrome',
         origin2: 'www.google.com/images',
       },
-      cache: true,
+      cache,
     })
     let str2 = ''
     for await (const chunk of response2.body) {
@@ -200,7 +200,7 @@ test('Responses with header Vary: * should not be cached', async (t) => {
     // But the server returns Vary: *, and thus shouldn't be cached.
     const response = await undici.request(`http://0.0.0.0:${serverPort}`, {
       dispatcher: new undici.Agent().compose(interceptors.cache()),
-      cache: true,
+      cache,
       headers: {
         Accept: 'application/txt',
         'User-Agent': 'Chrome',
@@ -247,7 +247,7 @@ test('Store 307-status-responses that happen to be dependent on the Range header
 
     const request1 = undici.request(`http://0.0.0.0:${serverPort}`, {
       dispatcher: new undici.Agent().compose(interceptors.cache()),
-      cache: true,
+      cache,
       headers: {
         range: 'bytes=0-999',
       },
@@ -269,7 +269,7 @@ test('Store 307-status-responses that happen to be dependent on the Range header
 
     const request2 = undici.request(`http://0.0.0.0:${serverPort}`, {
       dispatcher: new undici.Agent().compose(interceptors.cache()),
-      cache: true,
+      cache,
       headers: {
         range: 'bytes=0-999',
       },