[jira] [Commented] (PDFBOX-4774) Add AWS Lambda support to FontFileFinder

["Ben Manes (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/PDFBOX-4774?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17036562#comment-17036562 ] 

Ben Manes commented on PDFBOX-4774:
-----------------------------------

I realized a workaround, which is very much a hack, but enough to unblock me. In addition to static locations, {{UnixFontDirFinder}} also searches in a dynamic path:
{code:java}
System.getProperty("user.home") + "/.fonts"
{code}
While this goes to a non-writeable generated path, e.g. {{/home/sbx_user1051}}, it can be overwritten because there is no security manager. This lets me set the property because PdfBox initializes, e.g.
{code:java}
System.setProperty("user.home", System.getenv("LAMBDA_TASK_ROOT"));
{code}
Then my lambda distribution adds {{src/main/resources/.fonts}} with the extra fonts. This allows for the nicer rendering in {{fixed_page.jpg}}.

It is surely a hack and not a proper solution, so this feature would be nice for others going forward.

> Add AWS Lambda support to FontFileFinder
> ----------------------------------------
>
>                 Key: PDFBOX-4774
>                 URL: https://issues.apache.org/jira/browse/PDFBOX-4774
>             Project: PDFBox
>          Issue Type: Improvement
>            Reporter: Ben Manes
>            Priority: Major
>         Attachments: fixed_page.jpg, original.pdf, rendered_page.jpg
>
>
> The font directory finder is hard coded based on the operating system and is not directly extensible. Instead, if I understand correctly, the fonts have to be explicitly declared in a {{PDFBox_External_Fonts.properties}} file.
> AWS Lambda includes only minimal fonts in its linux distribution. For some documents this is too limiting, so on our EC2 instances we install {{msttcorefonts}}, {{ttf-aenigma}}, and {{fonts-tuffy}}. These go into {{/usr/share/fonts}} which the {{UnixFontDirFinder}} inspects.
> AWS Lambda will unzip the distribution into {{/var/task}}, will unzip layers into {{/opt}} only allows tasks to otherwise write to {{/tmp}}. The common recommendation for fonts is to include them in the lambda, reference them at {{/var/task/fonts}}, and set {{FONTCONFIG_PATH}} to that path for headless tasks like Chrome's print-to-pdf.
> Since PdfBox does not use font-config this solution does not work. Ideally it would be nice if one could specify a custom {{FontDirFinder}} or add custom directories (such as by a system property if not API). Alternatively, PdfBox could include reasonable default locations if the environmental variable {{LAMBDA_TASK_ROOT}} is set.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@pdfbox.apache.org
For additional commands, e-mail: dev-help@pdfbox.apache.org