grover 1.1.8 → 1.1.10

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 7bfcdc3d84dade94866bbb29eca6c1e91b9459a5f020ebd175341f740dfdd585
-  data.tar.gz: 539a10195c08de66249983fccdf3f5314f1adec83b400d6058f87d127e731f86
+  metadata.gz: 89de769dbb4ef96309c640d664ac90d9771968c02f596b6e6370d2f77b10cd1e
+  data.tar.gz: 5df2d5e82e1155757d859c95e01b0ed24dd49fdae6d4d8cb5c23276ef774603d
 SHA512:
-  metadata.gz: 6322c7cd633f776d58bb2fcad105d067a3df4782dcbc2fb15aff4fe60dcc2e8e1a58075e4972bb9f65362075eebe973e51f8aa385bd651d89ad41078e00fd4ca
-  data.tar.gz: e9431f9a613f4b40f5d836616ad2e8afd6470a606a8daf80ad5553e248a461035228c62e290f0778e742aa2f2a9b663fe23f0590600cdda592cd8a26339ba391
+  metadata.gz: 9a6b7be08fb0d719cd464033d1d9ebf63343f8cf7126ea4dfead19eb2b022d9bf124898b58b6842690714f7d7d05fc01e847340a9ed47f9966b6c8eddfb768fe
+  data.tar.gz: 90acf98c742ad66686108735828b0a378776c07375562df8c4db6f483a9f1669c7a0409381764caddb9a788c6ddbe9cfc7802edfa05961bedeb4c7896da4558d

data/lib/grover/configuration.rb

@@ -7,7 +7,7 @@ class Grover
   class Configuration
     attr_accessor :options, :meta_tag_prefix, :ignore_path, :ignore_request,
                   :root_url, :use_pdf_middleware, :use_png_middleware,
-                  :use_jpeg_middleware, :node_env_vars
+                  :use_jpeg_middleware, :node_env_vars, :allow_file_uris
 
     def initialize
       @options = {}
@@ -19,6 +19,7 @@ class Grover
       @use_png_middleware = false
       @use_jpeg_middleware = false
       @node_env_vars = {}
+      @allow_file_uris = false
     end
   end
 end

data/lib/grover/errors.rb

@@ -15,4 +15,5 @@ class Grover
       const_set name, Class.new(Error)
     end
   end
+  UnsafeConfigurationError = Class.new(Error)
 end

data/lib/grover/js/processor.cjs

@@ -26,7 +26,7 @@ const fs = require('fs');
 const os = require('os');
 const path = require('path');
 
-const _processPage = (async (convertAction, urlOrHtml, options) => {
+const _processPage = (async (convertAction, uriOrHtml, options) => {
   let browser, page, errors = [], tmpDir, wsConnection = false;
 
   try {
@@ -173,10 +173,12 @@ const _processPage = (async (convertAction, urlOrHtml, options) => {
     }
 
     const waitUntil = options.waitUntil; delete options.waitUntil;
-    if (urlOrHtml.match(/^http/i)) {
+    const allowFileUri = options.allowFileUri; delete options.allowFileUri;
+    const uriRegex = allowFileUri ? /^(https?|file):\/\//i : /^https?:\/\//i;
+    if (uriOrHtml.match(uriRegex)) {
       // Request is for a URL, so request it
       requestOptions.waitUntil = waitUntil || 'networkidle2';
-      await page.goto(urlOrHtml, requestOptions);
+      await page.goto(uriOrHtml, requestOptions);
     } else {
       // Request is some HTML content. Use request interception to assign the body
       requestOptions.waitUntil = waitUntil || 'networkidle0';
@@ -188,7 +190,7 @@ const _processPage = (async (convertAction, urlOrHtml, options) => {
           request.continue();
         else {
           htmlIntercepted = true
-          request.respond({ body: urlOrHtml === '' ? ' ' : urlOrHtml });
+          request.respond({ body: uriOrHtml === '' ? ' ' : uriOrHtml });
         }
       });
       const displayUrl = options.displayUrl; delete options.displayUrl;
@@ -280,7 +282,7 @@ const _processPage = (async (convertAction, urlOrHtml, options) => {
 
     // If we're running puppeteer in headless mode, return the converted PDF
     if (debug === undefined || (typeof debug === 'object' && (debug.headless === undefined || debug.headless))) {
-      return await page[convertAction](options);
+      return Buffer.from(await page[convertAction](options));
     }
   } finally {
     if (browser) {

data/lib/grover/middleware.rb

@@ -24,11 +24,14 @@ class Grover
       dup._call(env)
     end
 
-    def _call(env)
+    def _call(env) # rubocop:disable Metrics/MethodLength
       @request = Rack::Request.new(env)
       identify_request_type
 
-      configure_env_for_grover_request(env) if grover_request?
+      if grover_request?
+        check_file_uri_configuration
+        configure_env_for_grover_request(env)
+      end
       status, headers, response = @app.call(env)
       response = update_response response, headers if grover_request? && html_content?(headers)
 
@@ -45,6 +48,14 @@ class Grover
 
     attr_reader :pdf_request, :png_request, :jpeg_request
 
+    def check_file_uri_configuration
+      return unless Grover.configuration.allow_file_uris
+
+      # The combination of middleware and allowing file URLs is exceptionally
+      # unsafe as it can lead to data exfiltration from the host system.
+      raise UnsafeConfigurationError, 'using `allow_file_uris` configuration with middleware is exceptionally unsafe'
+    end
+
     def identify_request_type
       @pdf_request = Grover.configuration.use_pdf_middleware && path_matches?(PDF_REGEX)
       @png_request = Grover.configuration.use_png_middleware && path_matches?(PNG_REGEX)

data/lib/grover/options_builder.rb

@@ -8,12 +8,12 @@ class Grover
   # Build options from Grover.configuration, meta_options, and passed-in options
   #
   class OptionsBuilder < Hash
-    def initialize(options, url)
+    def initialize(options, uri)
       super()
-      @url = url
+      @uri = uri
       combined = grover_configuration
       Utils.deep_merge! combined, Utils.deep_stringify_keys(options)
-      Utils.deep_merge! combined, meta_options unless url_source?
+      Utils.deep_merge! combined, meta_options unless uri_source?
 
       update OptionsFixer.new(combined).run
     end
@@ -41,11 +41,11 @@ class Grover
     end
 
     def meta_tags
-      Nokogiri::HTML(@url).xpath('//meta')
+      Nokogiri::HTML(@uri).xpath('//meta')
     end
 
-    def url_source?
-      @url.match(/\Ahttp/i)
+    def uri_source?
+      @uri.match?(%r{\A(https?|file)://}i)
     end
   end
 end

data/lib/grover/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 class Grover
-  VERSION = '1.1.8'
+  VERSION = '1.1.10'
 end

data/lib/grover.rb

@@ -28,40 +28,40 @@ class Grover
   attr_reader :front_cover_path, :back_cover_path
 
   #
-  # @param [String] url URL of the page to convert
+  # @param [String] uri URI of the page to convert
   # @param [Hash] options Optional parameters to pass to PDF processor
   #   see https://github.com/puppeteer/puppeteer/blob/main/docs/api/puppeteer.pdfoptions.md
   #   and https://github.com/puppeteer/puppeteer/blob/main/docs/api/puppeteer.screenshotoptions.md
   #
-  def initialize(url, **options)
-    @url = url.to_s
-    @options = OptionsBuilder.new(options, @url)
+  def initialize(uri, **options)
+    @uri = uri.to_s
+    @options = OptionsBuilder.new(options, @uri)
     @root_path = @options.delete 'root_path'
     @front_cover_path = @options.delete 'front_cover_path'
     @back_cover_path = @options.delete 'back_cover_path'
   end
 
   #
-  # Request URL with provided options and create PDF
+  # Request URI with provided options and create PDF
   #
   # @param [String] path Optional path to write the PDF to
   # @return [String] The resulting PDF data
   #
   def to_pdf(path = nil)
-    processor.convert :pdf, @url, normalized_options(path: path)
+    processor.convert :pdf, @uri, normalized_options(path: path)
   end
 
   #
-  # Request URL with provided options and render HTML
+  # Request URI with provided options and render HTML
   #
   # @return [String] The resulting HTML string
   #
   def to_html
-    processor.convert :content, @url, normalized_options(path: nil)
+    processor.convert :content, @uri, normalized_options(path: nil)
   end
 
   #
-  # Request URL with provided options and create screenshot
+  # Request URI with provided options and create screenshot
   #
   # @param [String] path Optional path to write the screenshot to
   # @param [String] format Optional format of the screenshot
@@ -70,11 +70,11 @@ class Grover
   def screenshot(path: nil, format: nil)
     options = normalized_options(path: path)
     options['type'] = format if %w[png jpeg].include? format
-    processor.convert :screenshot, @url, options
+    processor.convert :screenshot, @uri, options
   end
 
   #
-  # Request URL with provided options and create PNG
+  # Request URI with provided options and create PNG
   #
   # @param [String] path Optional path to write the screenshot to
   # @return [String] The resulting PNG data
@@ -84,7 +84,7 @@ class Grover
   end
 
   #
-  # Request URL with provided options and create JPEG
+  # Request URI with provided options and create JPEG
   #
   # @param [String] path Optional path to write the screenshot to
   # @return [String] The resulting JPEG data
@@ -116,10 +116,10 @@ class Grover
   #
   def inspect
     format(
-      '#<%<class_name>s:0x%<object_id>p @url="%<url>s">',
+      '#<%<class_name>s:0x%<object_id>p @uri="%<uri>s">',
       class_name: self.class.name,
       object_id: object_id,
-      url: @url
+      uri: @uri
     )
   end
 
@@ -147,6 +147,7 @@ class Grover
   def normalized_options(path:)
     normalized_options = Utils.normalize_object @options, excluding: ['extraHTTPHeaders']
     normalized_options['path'] = path if path.is_a? ::String
+    normalized_options['allowFileUri'] = Grover.configuration.allow_file_uris == true
     normalized_options
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: grover
 version: !ruby/object:Gem::Version
-  version: 1.1.8
+  version: 1.1.10
 platform: ruby
 authors:
 - Andrew Bromwich
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-06-23 00:00:00.000000000 Z
+date: 2024-09-01 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: combine_pdf