[mac] org to pdf image absolute path not work

[peng051410]

• pandoc veriosn: pandoc 2.7.2
• os: mac 10.14.3 (18D109)
• org version: 9.1.9

when i run command
pandoc --self-contained --pdf-engine=xelatex --template=/Users/username/soft/plugin/latex/pm-template.latex -f org -t latex apollo-config-swim.org -o test.pdf

I got warn info:
[WARNING] Could not fetch resource 'file:///Users/username/github/blog/images/apollo-home-screenshot.png': PandocHttpError "file:///Users/username/github/blog/images/apollo-home-screenshot.png" (InvalidUrlException "file:///Users/username/github/blog/images/apollo-home-screenshot.png" "Invalid scheme")

org file image content:
[[file:/Users/username/github/blog/images/apollo-home-screenshot.png]]
[[/Users/username/github/blog/images/apollo-home-screenshot.png]]

these all not work!

What should i do ?

[mb21]

You don't want to use --self-contained when outputting to PDF. Also make sure the image file is actually there.

For future questions, please use the pandoc-discuss mailing list instead of this issue tracker. Thanks!

[tarleb]

I can reproduce the problem for this syntax:

[[/Users/username/github/blog/images/apollo-home-screenshot.png]]

while this works as expected for me

[[file:/Users/username/github/blog/images/apollo-home-screenshot.png]]

Reopening, both variants should work.

[tarleb]

We currently convert a link [[file:/path/to/file.jpg]] to a link with source file:///path/to/file.jpg. We then run into problems in later conversion stages.

This is not unique to org-mode. Specifying an image with schema file: leads to problems with other input formats as well:

$ echo '![my image](file:///path/to/existing/image.jpg)' | pandoc --output image-output.pdf
[WARNING] Could not fetch resource …

[mb21]

But only file:///path is a valid URL, right?

I guess this is the place to look: https://github.com/jgm/pandoc/blob/master/src/Text/Pandoc/Class.hs#L577

[jgm]

Mauro Bieg <notifications@github.com> writes:

But only `file:///path` is a valid URL, right?

It's a valid URL, but you can't make an HTTP request to it. I think what's happening is that pandoc is trying to make an HTTP request since it sees it's a URL.

[mb21]

Ah, both file:///path and file:/path seem to be valid, according to wikipedia.

[tarleb]

Ah, both file:///path and file:/path seem to be valid, according to wikipedia.

Interesting, I didn't know. The org reader currently translates `file:/path` into `file:///path`. It is what org does when exporting to HTML, and the behavior is described (or hinted at) in this list here: https://orgmode.org/manual/External-Links.html#External-Links.

[peng051410]

You don't want to use --self-contained when outputting to PDF. Also make sure the image file is actually there.

For future questions, please use the pandoc-discuss mailing list instead of this issue tracker. Thanks!

I remove the --self-contained option, the problem is still.

[mb21]

With echo '![](file:///foo/bar.jpg)' | pandoc --self-contained, I get

InvalidUrlException "file:///foo/bar.jpg" "Invalid scheme"

I wanted to play around with some function in Pandoc.Class, but unfortunately stack ghci seems to be broken for me currently:

<no location info>: error: [-Wmissing-home-modules, -Werror=missing-home-modules]
    Modules are not listed in command line but needed for compilation: 
        Paths_pandoc

[jgm]

InvalidUrlException comes from Network.HTTP.Client, used by openURL.

Check out this code from Text.Pandoc.Class:

downloadOrRead :: PandocMonad m
               => String
               -> m (B.ByteString, Maybe MimeType)
downloadOrRead s = do
  sourceURL <- getsCommonState stSourceURL
  case (sourceURL >>= parseURIReference' .
                       ensureEscaped, ensureEscaped s) of
    (Just u, s') -> -- try fetching from relative path at source
       case parseURIReference' s' of
            Just u' -> openURL $ show $ u' `nonStrictRelativeTo` u
            Nothing -> openURL s' -- will throw error

If the source parses as a URI, then we use openURL, but of course this doesn't make much sense when it's a file:/ URI. So we should check the scheme.

[mb21]

@peng051410 for me the following works:

echo '[[file:/Users/myuser/myfile.jpg]]' | pandoc -f org -o foo.pdf

[peng051410]

It works for me, Thanks!
The content is [[file:/Users/username/github/blog/images/apollo-overall-architecture.png][file:/Users/username/github/blog/images/apollo-overall-architecture.png]]

and [[file:/Users/username/github/blog/images/apollo-overall-architecture.png][/Users/username/github/blog/images/apollo-overall-architecture.png]] not works

[mb21]

Reopening for the file:/// case...