resolves asciidoctor#1507 introduce an in-memory cache (Node.js)

Author: ggrossetie

packages/core/lib/asciidoctor/js/asciidoctor_ext/node.rb

@@ -2,3 +2,4 @@
 require 'asciidoctor/js/asciidoctor_ext/node/open_uri'
 require 'asciidoctor/js/asciidoctor_ext/node/stylesheet'
 require 'asciidoctor/js/asciidoctor_ext/node/template'
+require 'asciidoctor/js/asciidoctor_ext/node/helpers'

packages/core/lib/asciidoctor/js/asciidoctor_ext/node/helpers.rb

@@ -0,0 +1,35 @@
+module Asciidoctor
+# Internal: Except where noted, a module that contains internal helper functions.
+module Helpers
+  module_function
+
+  # preserve the original require_library method
+  alias original_require_library require_library
+
+  # Public: Require the specified library using Kernel#require.
+  #
+  # Attempts to load the library specified in the first argument using the
+  # Kernel#require. Rescues the LoadError if the library is not available and
+  # passes a message to Kernel#raise if on_failure is :abort or Kernel#warn if
+  # on_failure is :warn to communicate to the user that processing is being
+  # aborted or functionality is disabled, respectively. If a gem_name is
+  # specified, the message communicates that a required gem is not available.
+  #
+  # name       - the String name of the library to require.
+  # gem_name   - a Boolean that indicates whether this library is provided by a RubyGem,
+  #              or the String name of the RubyGem if it differs from the library name
+  #              (default: true)
+  # on_failure - a Symbol that indicates how to handle a load failure (:abort, :warn, :ignore) (default: :abort)
+  #
+  # Returns The [Boolean] return value of Kernel#require if the library can be loaded.
+  # Otherwise, if on_failure is :abort, Kernel#raise is called with an appropriate message.
+  # Otherwise, if on_failure is :warn, Kernel#warn is called with an appropriate message and nil returned.
+  # Otherwise, nil is returned.
+  def require_library name, gem_name = true, on_failure = :abort
+    if name == 'open-uri/cached'
+      `return Opal.Asciidoctor.Cache.enable()`
+    end
+    original_require_library name, gem_name, on_failure
+  end
+end
+end

packages/core/lib/asciidoctor/js/opal_ext/uri.rb

@@ -1,6 +1,8 @@
 module URI
   def self.parse str
-    str.extend URI
+    # REMIND: Cannot create property '$$meta' on string in strict mode!
+    #str.extend URI
+    str
   end
 
   def path

packages/core/lib/open-uri/cached.rb

@@ -0,0 +1,156 @@
+# copied from https://github.com/tigris/open-uri-cached/blob/master/lib/open-uri/cached.rb
+module OpenURI
+  class << self
+    # preserve the original open_uri method
+    alias original_open_uri open_uri
+    def cache_open_uri(uri, *rest, &block)
+      response = Cache.get(uri.to_s) ||
+                 Cache.set(uri.to_s, original_open_uri(uri, *rest))
+
+      if block_given?
+        begin
+          yield response
+        ensure
+          response.close
+        end
+      else
+        response
+      end
+    end
+    # replace the existing open_uri method
+    alias open_uri cache_open_uri
+  end
+
+  class Cache
+    class << self
+
+      %x{
+        // largely inspired by https://github.com/isaacs/node-lru-cache/blob/master/index.js
+        let cache = new Map()
+        let max = 16000000 // bytes
+        let length = 0
+        let lruList = []
+
+        class Entry {
+          constructor (key, value, length) {
+            this.key = key
+            this.value = value
+            this.length = length
+          }
+        }
+
+        const trim = () => {
+          while (length > max) {
+            pop()
+          }
+        }
+
+        const reset = () => {
+          cache = new Map()
+          length = 0
+          lruList = []
+        }
+
+        const pop = () => {
+          const leastRecentEntry = lruList.pop()
+          if (leastRecentEntry) {
+            length -= leastRecentEntry.length
+            cache.delete(leastRecentEntry.key)
+          }
+        }
+
+        const del = (entry) => {
+          if (entry) {
+            length -= entry.length
+            cache.delete(entry.key)
+            const entryIndex = lruList.indexOf(entry)
+            if (entryIndex > -1) {
+              lruList.splice(entryIndex, 1)
+            }
+          }
+        }
+      }
+
+      ##
+      # Retrieve file content and meta data from cache
+      # @param [String] key
+      # @return [StringIO]
+      def get(key)
+        %x{
+          const cacheKey = crypto.createHash('sha256').update(key).digest('hex')
+          if (cache.has(cacheKey)) {
+            const entry = cache.get(cacheKey)
+            const io = Opal.$$$('::', 'StringIO').$new()
+            io['$<<'](entry.value)
+            io.$rewind()
+            return io
+          }
+        }
+
+        nil
+      end
+
+      # Cache file content
+      # @param [String] key
+      #   URL of content to be cached
+      # @param [StringIO] value
+      #   value to be cached, typically StringIO returned from `original_open_uri`
+      # @return [StringIO]
+      #   Returns value
+      def set(key, value)
+        %x{
+          const cacheKey = crypto.createHash('sha256').update(key).digest('hex')
+          const contents = value.string
+          const len = contents.length
+          if (cache.has(cacheKey)) {
+            if (len > max) {
+              // oversized object, dispose the current entry.
+              del(cache.get(cacheKey))
+              return value
+            }
+            // update current entry
+            const entry = cache.get(cacheKey)
+            // remove existing entry in the LRU list (unless the entry is already the head).
+            const listIndex = lruList.indexOf(entry)
+            if (listIndex > 0) {
+              lruList.splice(listIndex, 1)
+              lruList.unshift(entry)
+            }
+            entry.value = value
+            length += len - entry.length
+            entry.length = len
+            trim()
+            return value
+          }
+
+          const entry = new Entry(cacheKey, value, len)
+          // oversized objects fall out of cache automatically.
+          if (entry.length > max) {
+            return value
+          }
+
+          length += entry.length
+          lruList.unshift(entry)
+          cache.set(cacheKey, entry)
+          trim()
+          return value
+        }
+      end
+
+      def max=(maxLength)
+        %x{
+          if (typeof maxLength !== 'number' || maxLength < 0) {
+            throw new TypeError('max must be a non-negative number')
+          }
+
+          max = maxLength || Infinity
+          trim()
+        }
+      end
+
+      def reset
+        `reset()`
+      end
+    end
+  end
+end

packages/core/package.json

@@ -20,7 +20,7 @@
   "scripts": {
     "test:graalvm": "node tasks/graalvm.cjs",
     "test:node": "mocha spec/*/*.spec.cjs && npm run test:node:esm",
-    "test:node:esm": "mocha --experimental-json-modules spec/node/asciidoctor.spec.js",
+    "test:node:esm": "mocha spec/node/asciidoctor.spec.js",
     "test:browser": "node spec/browser/run.cjs",
     "test:types": "rm -f types/tests.js && eslint types --ext .ts && tsc --build types/tsconfig.json && node --input-type=commonjs types/tests.js",
     "test": "node tasks/test/unsupported-features.cjs && npm run test:node && npm run test:browser && npm run test:types",

packages/core/spec/fixtures/images/cc-heart.svg

@@ -9,14 +9,12 @@ import dirtyChai from 'dirty-chai'
 import dot from 'dot'
 import nunjucks from 'nunjucks'
 import Opal from 'asciidoctor-opal-runtime' // for testing purpose only
-
 import semVer from '../share/semver.cjs'
 import MockServer from '../share/mock-server.cjs'
 import shareSpec from '../share/asciidoctor-spec.cjs'
 import includeHttpsSpec from '../share/asciidoctor-include-https-spec.cjs'
 import { fileExists, isWin, removeFile, resolveFixture, truncateFile } from './helpers.js'
 import Asciidoctor from '../../build/asciidoctor-node.js'
-import packageJson from '../../package.json'
 
 import fooBarPostProcessor from '../share/extensions/foo-bar-postprocessor.cjs'
 import loveTreeProcessor from '../share/extensions/love-tree-processor.cjs'
@@ -30,6 +28,10 @@ import chartBlockMacro from '../share/extensions/chart-block.cjs'
 import smileyInlineMacro from '../share/extensions/smiley-inline-macro.cjs'
 import loremBlockMacro from '../share/extensions/lorem-block-macro.cjs'
 
+import { createRequire } from 'module'
+const require = createRequire(import.meta.url)
+const packageJson = require('../../package.json')
+
 const expect = chai.expect
 chai.use(dirtyChai)
 
@@ -2186,6 +2188,109 @@ content
 content`, options)
       expect(html).to.contain('0. Chapter A')
     })
+
+    describe('Cache', () => {
+      it('should cache remote SVG when allow-uri-read, cache-uri, and inline option are set', async () => {
+        try {
+          const input = `
+
+image::${testOptions.remoteBaseUri}/cc-zero.svg[opts=inline]
+
+image::${testOptions.remoteBaseUri}/cc-zero.svg[opts=inline]
+
+image::${testOptions.remoteBaseUri}/cc-zero.svg[opts=inline]
+`
+          await mockServer.resetRequests()
+          asciidoctor.convert(input, { safe: 'safe', attributes: { 'allow-uri-read': '', 'cache-uri': '' } })
+          const requestsReceived = await mockServer.getRequests()
+          expect(requestsReceived.length).to.equal(1)
+        } finally {
+          asciidoctor.Cache.disable()
+        }
+      })
+
+      it('should not cache remote SVG when cache-uri is absent (undefined)', async () => {
+        const input = `
+
+image::${testOptions.remoteBaseUri}/cc-zero.svg[opts=inline]
+
+image::${testOptions.remoteBaseUri}/cc-zero.svg[opts=inline]
+
+image::${testOptions.remoteBaseUri}/cc-zero.svg[opts=inline]
+`
+        await mockServer.resetRequests()
+        asciidoctor.convert(input, { safe: 'safe', attributes: { 'allow-uri-read': '' } })
+        const requestsReceived = await mockServer.getRequests()
+        expect(requestsReceived.length).to.equal(3)
+      })
+
+      it('should cache remote include when cache-uri is set', async () => {
+        try {
+          const input = `
+
+include::${testOptions.remoteBaseUri}/include-lines.adoc[lines=1..2]
+
+include::${testOptions.remoteBaseUri}/include-lines.adoc[lines=3..4]
+`
+          await mockServer.resetRequests()
+          asciidoctor.convert(input, { safe: 'safe', attributes: { 'allow-uri-read': true, 'cache-uri': '' } })
+          const requestsReceived = await mockServer.getRequests()
+          expect(requestsReceived.length).to.equal(1)
+        } finally {
+          asciidoctor.Cache.disable()
+        }
+      })
+
+      it('should not cache file if the size exceed the max cache', async () => {
+        try {
+          asciidoctor.Cache.setMax(1)
+          const input = `
+
+include::${testOptions.remoteBaseUri}/include-lines.adoc[lines=1..2]
+
+include::${testOptions.remoteBaseUri}/include-lines.adoc[lines=3..4]
+`
+          await mockServer.resetRequests()
+          asciidoctor.convert(input, { safe: 'safe', attributes: { 'allow-uri-read': true, 'cache-uri': '' } })
+          const requestsReceived = await mockServer.getRequests()
+          expect(requestsReceived.length).to.equal(2)
+        } finally {
+          asciidoctor.Cache.disable()
+          asciidoctor.Cache.reset()
+          asciidoctor.Cache.setMax(Opal.Asciidoctor.Cache.DEFAULT_MAX)
+        }
+      })
+
+      it('should not exceed max cache size', async () => {
+        try {
+          // cc-zero.svg exact size/length
+          const contentLength = fs.readFileSync(path.join(__dirname, '..', 'fixtures', 'images', 'cc-zero.svg'), 'utf-8').length
+          asciidoctor.Cache.setMax(contentLength)
+          const input = `
+
+image::${testOptions.remoteBaseUri}/cc-zero.svg[opts=inline]
+
+image::${testOptions.remoteBaseUri}/cc-zero.svg[opts=inline]
+
+// will remove cc-zero.svg from the cache!
+image::${testOptions.remoteBaseUri}/cc-heart.svg[opts=inline]
+
+image::${testOptions.remoteBaseUri}/cc-heart.svg[opts=inline]
+
+image::${testOptions.remoteBaseUri}/cc-zero.svg[opts=inline]
+`
+          await mockServer.resetRequests()
+          asciidoctor.convert(input, { safe: 'safe', attributes: { 'allow-uri-read': true, 'cache-uri': '' } })
+          const requestsReceived = await mockServer.getRequests()
+          expect(requestsReceived.length).to.equal(3)
+          expect(requestsReceived.map((request) => request.pathname)).to.have.members(['/cc-zero.svg', '/cc-heart.svg', '/cc-heart.svg'])
+        } finally {
+          asciidoctor.Cache.disable()
+          asciidoctor.Cache.reset()
+          asciidoctor.Cache.setMax(Opal.Asciidoctor.Cache.DEFAULT_MAX)
+        }
+      })
+    })
   })
 
   describe('Docinfo files', () => {