- 0.5.3

- **BUGFIX** introduced in 0.5.0 when certain shells don't accept `[foo=bar, …]` parameters which should correctly be `[foo, bar]` Thanks to [@egze](https://github.com/egze) for submitting a patch. - Refactored `PathAgent` that holds configuration state for readability and more fashionable and extensible error messages. Extensible towards new generators. - Updated README to be more elaborative on how to install `wkhtmltopdf` and `chrome-headless-render-pdf`
robertpeacock22 · Apr 30, 2019 · 645d165 · 645d165
2 parents d6f4453 + 7d586ed
commit 645d165
Show file tree

Hide file tree

Showing 6 changed files with 118 additions and 43 deletions.
diff --git a/CHANGES.md b/CHANGES.md
@@ -1,4 +1,23 @@
 # Changes
+
+  - 0.5.3
+    - **BUGFIX** introduced in 0.5.0 when certain shells don't accept
+      `["foo=bar", …]` parameters which should correctly be `["foo", "bar"]`
+      Thanks to [@egze](https://github.com/egze) for submitting a patch.
+    - Refactored `PathAgent` that holds configuration state for readability and
+      more fashionable and extensible error messages. Extensible towards new
+      generators.
+    - Updated README to be more elaborative on how to install `wkhtmltopdf` and
+      `chrome-headless-render-pdf`
+  - 0.5.2
+    - **BUGFIX** introduced in 0.5.0 when global options to wkhtmltopdf weren't
+      accepted any more due to wrong shell parameter order. Thanks to
+      [manukall](https://github.com/manukall) for reporting.
+  - 0.5.1
+    - allow chrome to be executed as root via default config option
+      `disable_chrome_sandbox` – this is required for an easy usage within a
+      docker container as in
+      [elixir-pdf-server](https://github.com/gutschilla/elixir-pdf-server)
   - 0.5.0
     - **Got rid of Porcelain** dependency as it interferes with many builds using
       plain `System.cmd/3`. Please note, that as of the documentation

diff --git a/README.md b/README.md
@@ -1,7 +1,9 @@
 # elixir-pdf-generator
 
 A wrapper for both wkhtmltopdf and chrome-headless plus PDFTK (adds in
-encryption) for use in Elixir projects. If available, it will use xvfb-run (x
+encryption) for use in Elixir projects. 
+
+If available, it will use xvfb-run (x
 virtual frame buffer) to use wkhtmltopdf on systems that have no X installed,
 e.g. a server.
 
@@ -33,53 +35,84 @@ e.g. a server.
     - Refactored `PathAgent` that holds configuration state for readability and
       more fashionable and extensible error messages. Extensible towards new
       generators.
+    - Updated README to be more elaborative on how to install `wkhtmltopdf` and
+      `chrome-headless-render-pdf`
 
 For a proper changelog, see [CHANGES](CHANGES.md)
 
-# System prerequisites (either wkhtmltopdf or nodejs and maybe pdftk)
+# System prerequisites 
+
+It's either 
+
+* wkhtmltopdf or 
+
+* nodejs and possibly chrome/chromium
 
 ## chrome-headless
 
-1. Run `npm install`. This requires [nodejs](https://nodejs.org), of course.
-   This will install a recent chromium and chromedriver to run Chrome in
-   headless mode and use this browser and its API to print PDFs.
+This will allow you to make more use of Javascript and advanced CSS as it's just
+your Chrome/Chromium browser rendering your web page as HTML and printing it as
+PDF. Rendering _tend_ to be a bit faster than with wkhtmltopdf. The price tag is
+that PDFs printed with chrome/chromium are usually considerably bigger than
+those generated with wkhtmltopdf.
+
+1. Run `npm -g install chrome-headless-render-pdf puppeteer`. 
+
+   This requires [nodejs](https://nodejs.org), of course. This will install a
+   recent chromium and chromedriver to run Chrome in headless mode and use this
+   browser and its API to print PDFs globally on your machine.
 
+   If you prefer a project-local install, just use `npm install` This will
+   install dependencies under `./node_modules`. Be aware that those won't be
+   packaged in your distribution (I will add support for this later).
+
    On some machines, this doesn't install Chromium and fails. Here's how to get
    this running on Ubtunu 18:
 
    ```
-   
+   DEBIAN_FRONTEND=noninteractive PUPPETEER_SKIP_CHROMIUM_DOWNLOAD=TRUE \
+     apt-get install -y chromium-chromedriver \
+     && npm -g install chrome-headless-render-pdf puppeteer
    ```
+
+## wkhtmltopdf
 
 2. Download wkhtmltopdf and place it in your $PATH. Current binaries can be
    found here: http://wkhtmltopdf.org/downloads.html
 
-   D
+   For the impatient:
 
-   * _(optional)_ To use wkhtmltopdf on systems without an X window server
-     installed, please install `xvfb-run` from your repository (on
-     Debian/Ubuntu: `sudo apt-get install xvfb`).
+   ```
+   apt-get -y install xfonts-base xfonts-75dpi \
+    && wget https://downloads.wkhtmltopdf.org/0.12/0.12.5/wkhtmltox_0.12.5-1.bionic_amd64.deb \
+    && dpkg -i wkhtmltox_0.12.5-1.bionic_amd64.deb
+   ```
 
-  * On current (2018) Macintosh computers `/usr/X11/bin/xvfb` should be
-    available and is reported to do the same thing. _warning:** This is untested.
-    PLS report to me if you ran this successfully on a Mac.
+3. _optional:_ Install `xvfb` (shouldn't be required with the binary mentioned above):
 
-**AND MAYBE**
+   To use other wkhtmltopdf executables comiled with an unpatched Qt on systems
+   without an X window server installed, please install `xvfb-run` from your
+   repository (on Debian/Ubuntu: `sudo apt-get install xvfb`).
+
+   I haven't heard any feedback of people using this feature since a while since
+   the wkhtmltopdf projects ships ready-made binaries. I will deprecate this
+   starting in `0.6.0` since, well, YAGNI.
 
-3. _(optional)_ Install pdftk via your package manager or homebrew. The project
-   page also contains a Windows installer
+4. _optional:_ Install `pdftk` via your package manager or homebrew. The project
+   page also contains a Windows installer. On Debian/Ubuntu just type:
+   `apt-get -y install pdftk`
 
 # Usage
 
 Add this to your dependencies in your mix.exs:
 
     def application do
         [applications: [
-            :logger, 
+            :logger,
             :pdf_generator # <-- add this
         ]]
     end
-    
+
     defp deps do
         [
             # ... whatever else
@@ -95,7 +128,7 @@ $ iex -S mix
 html = "<html><body><p>Hi there!</p></body></html>"
 # be aware, this may take a while...
 {:ok, filename}    = PdfGenerator.generate(html, page_size: "A5")
-{:ok, pdf_content} = File.read(filename) 
+{:ok, pdf_content} = File.read(filename)
 
 # or, if you prefer methods that raise on error:
 filename = PdfGenerator.generate!(html, generator: :chrome)
@@ -125,19 +158,19 @@ pdf_binary = PdfGenerator.generate_binary! "<html>..."
 
 This module will automatically try to finde both `wkhtmltopdf` and `pdftk` in
 your path. But you may override or explicitly set their paths in your
-`config/config.exs`. 
+`config/config.exs`.
 
 ```
 config :pdf_generator,
     wkhtml_path:    "/usr/bin/wkhtmltopdf",   # <-- this program actually does the heavy lifting
     pdftk_path:     "/usr/bin/pdftk"          # <-- only needed for PDF encryption
 ```
- 
+
 or, if you prefer shrome-headless
 
 ```
 config :pdf_generator,
-    use_chrome: true             # <-- will be default by 0.6.0 
+    use_chrome: true             # <-- will be default by 0.6.0
     pdftk_path: "/usr/bin/pdftk" # <-- only needed for PDF encryption
 ```
 
@@ -150,7 +183,7 @@ an X Window server, but your server (or Mac) does not have one installed,
 you may find the `command_prefix` handy:
 
 ```
-PdfGenerator.generate "<html..", command_prefix: "xvfb-run" 
+PdfGenerator.generate "<html..", command_prefix: "xvfb-run"
 ```
 
 This can also be configured globally in your `config/config.exs`:
@@ -174,20 +207,20 @@ config :pdf_generator,
 
 - `filename` - filename for the output pdf file (without .pdf extension, defaults to a random string)
 
-- `page_size`:        
-  * defaults to `A4`, see `wkhtmltopdf` for more options 
+- `page_size`:
+  * defaults to `A4`, see `wkhtmltopdf` for more options
   * A4 will be translated to `page-height 11` and `page-width 8.5` when
     chrome-headless is used
 
 - `open_password`:    requires `pdftk`, set password to encrypt PDFs with
 
 - `edit_password`:    requires `pdftk`, set password for edit permissions on PDF
 
-- `shell_params`:     pass custom parameters to `wkhtmltopdf`. **CAUTION: BEWARE OF SHELL INJECTIONS!** 
+- `shell_params`:     pass custom parameters to `wkhtmltopdf`. **CAUTION: BEWARE OF SHELL INJECTIONS!**
 
 - `command_prefix`:   prefix `wkhtmltopdf` with some command or a command with options
                       (e.g. `xvfb-run -a`, `sudo` ..)
-                      
+
 - `delete_temporary`: immediately remove temp files after generation
 
 ## Heroku Setup
@@ -202,7 +235,7 @@ https://github.com/gjaldon/phoenix-static-buildpack
 ```
 
 __note:__ The list also includes Elixir and Phoenix buildpacks to show you that they
-must be placed after `pdftk` and `wkhtmltopdf`. It won't work if you load the 
+must be placed after `pdftk` and `wkhtmltopdf`. It won't work if you load the
 Elixir and Phoenix buildpacks first.
 
 # Documentation

diff --git a/lib/pdf_generator.ex b/lib/pdf_generator.ex
@@ -137,14 +137,14 @@ defmodule PdfGenerator do
          :ok                         <- maybe_write_html(content, html_file),
          {executable, arguments}     <- make_command(generator, options, content, {html_file, pdf_file}),
          {:cmd, {stderr, exit_code}} <- {:cmd, System.cmd(executable, arguments, stderr_to_stdout: true)},       # unfortuantely wkhtmltopdf returns 0 on errors as well :-/
-         {:result_ok, true, stderr}  <- {:result_ok, result_ok(generator, stderr, exit_code), stderr},           # so we inspect stderr instead
+         {:result_ok, true, _err}    <- {:result_ok, result_ok(generator, stderr, exit_code), stderr},           # so we inspect stderr instead
          {:rm, :ok}                  <- {:rm, maybe_delete_temp(delete_temp, html_file)},
          {:ok, encrypted_pdf}        <- maybe_encrypt_pdf(pdf_file, open_password, edit_password) do
       {:ok, encrypted_pdf}
     else
-      {:error, reason}        -> {:error, reason}
-      {:result_ok, _, stderr} -> {:error, {:generator_failed, stderr}}
-      reason                  -> {:error, reason}
+      {:error, reason}     -> {:error, reason}
+      {:result_ok, _, err} -> {:error, {:generator_failed, err}}
+      reason               -> {:error, reason}
     end
   end
 
@@ -174,16 +174,17 @@ defmodule PdfGenerator do
 
   @spec make_command(generator, opts, content, {html_path, pdf_path}) :: {path, list()}
   def make_command(:chrome, options, content, {html_path, pdf_path}) do
-    executable_on_path = System.find_executable("chrome-headless-render-pdf")
-    nodejs_on_path     = System.find_executable("nodejs")
+    chrome_executable  = PdfGenerator.PathAgent.get.chrome_path
+    node_executable    = PdfGenerator.PathAgent.get.node_path
     disable_sandbox    = Application.get_env(:pdf_generator, :disable_chrome_sandbox) || options[:no_sandbox]
+    # FIXME: this won't work in releases
     js_file = Application.app_dir(:pdf_generator) <> "/../../../../node_modules/chrome-headless-render-pdf/dist/cli/chrome-headless-render-pdf.js"
 
     {executable, executable_args} =
-      if options[:prefer_system_executable] && is_binary(executable_on_path) do
-        {executable_on_path, []}
+      if options[:prefer_system_executable] && is_binary(chrome_executable) do
+        {chrome_executable, []}
       else
-        {nodejs_on_path, [js_file]}
+        {node_executable, [js_file]}
       end
 
     {width, height} = make_dimensions(options)

diff --git a/lib/pdf_generator_path_agent.ex b/lib/pdf_generator_path_agent.ex
@@ -2,13 +2,18 @@ defmodule PdfGenerator.PathAgent do
   require Logger
   defstruct [
     wkhtml_path: nil,
-    pdftk_path: nil,
-    chrome: nil,
+    pdftk_path:  nil,
+    chrome_path: nil,
+    node_path:   nil,
   ]
 
   @moduledoc """
   Will check for system executables at startup time and store paths. If
   configured as such, will raise an error when no usable executable was found.
+
+  The idea is to perform this kind of auto-configuration at startup and then
+  store the values in this agent. The same could probably be achieved with
+  Application.put_env/3. Comments on this welcome. Until then, this just owrks.
   """
 
   @name __MODULE__
@@ -24,6 +29,7 @@ defmodule PdfGenerator.PathAgent do
         wkhtml_path: System.find_executable("wkhtmltopdf"),
         pdftk_path:  System.find_executable("pdftk"),
         chrome_path: System.find_executable("chrome-headless-render-pdf"),
+        node_path:   System.find_executable("nodejs") || System.find_executable("node"),
       ]
       ++ paths_from_options
       |> Enum.dedup()
@@ -43,6 +49,15 @@ defmodule PdfGenerator.PathAgent do
     Agent.get( @name, fn( data ) -> data end )
   end
 
+  @doc """
+  Checks options for generator binaries (chrome-headless-render-pdf and
+  wkhtmltopdf) exists on path. Raises an error if corresponding options are set
+  to true:
+
+  - :raise_on_missing_wkhtmltopdf_binary
+  - :raise_on_missing_chrome_binary
+  - :raise_on_missing_binaries -> raises on eiter binary missing
+  """
   def raise_or_continue(options) do
     wkhtml_exists = File.exists?(options[:wkhtml_path])
     chrome_exists = File.exists?(options[:chrome_path])
@@ -54,11 +69,13 @@ defmodule PdfGenerator.PathAgent do
     maybe_raise(:wkhtml, raise_on_wkhtml_missing, wkhtml_exists)
     maybe_raise(:chrome, raise_on_chrome_missing, chrome_exists)
     maybe_raise(:any,    raise_on_any_missing, wkhtml_exists or chrome_exists)
+
+    options
   end
 
-  defp maybe_raise(generator, config_says_raise = true,  wkhtml_exists = false), do: generator |> missing_message() |> raise()
-  defp maybe_raise(generator, config_says_raise = false, wkhtml_exists = false), do: generator |> missing_message() |> Logger.warn()
-  defp maybe_raise(generator, config_says_raise = _,     wkhtml_exists = _    ), do: :noop
+  defp maybe_raise(generator,  _config_says_raise = true,  _executable_exists = false), do: generator |> missing_message() |> raise()
+  defp maybe_raise(generator,  _config_says_raise = false, _executable_exists = false), do: generator |> missing_message() |> Logger.warn()
+  defp maybe_raise(_generator, _config_says_raise = _,     _executable_exists = _    ), do: :noop
 
   defp missing_message(:wkhtml), do: "wkhtmltopdf executable was not found on your system"
   defp missing_message(:chrome), do: "chrome-headless-render-pdf executable was not found on your system"

diff --git a/mix.exs b/mix.exs
@@ -5,7 +5,7 @@ defmodule PdfGenerator.Mixfile do
     [
       app: :pdf_generator,
       name: "PDF Generator",
-      version: "0.5.2",
+      version: "0.5.3",
       elixir: ">= 1.1.0",
       deps: deps(),
       description: description(),

diff --git a/test/pdf_generator_test.exs b/test/pdf_generator_test.exs
@@ -49,6 +49,11 @@ defmodule PdfGeneratorTest do
     assert result |> File.read! |> String.slice(0, 6) == "%PDF-1"
   end
 
+  test "chrome's no-sandbox option doesn't crash" do
+    {_status, result} = PdfGenerator.generate({:url, "http://google.com"}, generator: :chrome, disable_chrome_sandbox: true)
+    assert result |> File.read! |> String.slice(0, 6) == "%PDF-1"
+  end
+
   test "generate! returns a filename" do
     @html
     |> PdfGenerator.generate!