From c3eec896ab9ea0c88795972fecb7699fffa4e1ab Mon Sep 17 00:00:00 2001
From: John MacFarlane <jgm@berkeley.edu>
Date: Tue, 1 Oct 2024 10:15:50 -0700
Subject: [PATCH] Text.Pandoc.Class: export extractURIData.

[API change]
---
 src/Text/Pandoc/Class/PandocMonad.hs | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/src/Text/Pandoc/Class/PandocMonad.hs b/src/Text/Pandoc/Class/PandocMonad.hs
index fdec5616d..51d017c06 100644
--- a/src/Text/Pandoc/Class/PandocMonad.hs
+++ b/src/Text/Pandoc/Class/PandocMonad.hs
@@ -40,6 +40,7 @@ module Text.Pandoc.Class.PandocMonad
   , setUserDataDir
   , getUserDataDir
   , fetchItem
+  , extractURIData
   , getInputFiles
   , setInputFiles
   , getOutputFile
@@ -78,6 +79,8 @@ import Text.Pandoc.MediaBag (MediaBag, lookupMedia, MediaItem(..))
 import Text.Pandoc.Shared (safeRead, makeCanonical, tshow)
 import Text.Pandoc.URI (uriPathToPath)
 import Text.Pandoc.Walk (walkM)
+import qualified Text.Pandoc.UTF8 as UTF8
+import Data.ByteString.Base64 (decodeLenient)
 import Text.Parsec (ParsecT, getPosition, sourceLine, sourceName)
 import qualified Data.ByteString as B
 import qualified Data.ByteString.Lazy as BL
@@ -372,6 +375,16 @@ downloadOrRead s = do
          convertSlash '\\' = '/'
          convertSlash x    = x
 
+-- Extract data from a data URI's path component.
+extractURIData :: String -> (B.ByteString, Maybe MimeType)
+extractURIData upath =
+  case break (== ';') (filter (/= ' ') mimespec) of
+     (mime', ";base64") -> (decodeLenient contents, Just (T.pack mime'))
+     (mime', _) -> (contents, Just (T.pack mime'))
+  where
+    (mimespec, rest) = break (== ',') $ unEscapeString upath
+    contents = UTF8.fromString $ drop 1 rest
+
 -- | Checks if the file path is relative to a parent directory.
 isRelativeToParentDir :: FilePath -> Bool
 isRelativeToParentDir fname =