LaTeX writer: Improve escaping of URIs in href, url.

Closes #8992.
jgm · Aug 9, 2023 · cbb33fe · cbb33fe
1 parent cc371a9
commit cbb33fe
Show file tree

Hide file tree

Showing 3 changed files with 28 additions and 17 deletions.
diff --git a/src/Text/Pandoc/Writers/LaTeX/Util.hs b/src/Text/Pandoc/Writers/LaTeX/Util.hs
@@ -29,6 +29,7 @@ import Text.Pandoc.Writers.LaTeX.Types (LW, WriterState(..))
 import Text.Pandoc.Writers.LaTeX.Lang (toBabel)
 import Text.Pandoc.Highlighting (toListingsLanguage)
 import Text.DocLayout
+import Network.URI (escapeURIString)
 import Text.Pandoc.Definition
 import Text.Pandoc.ImageSize (showFl)
 import Control.Monad.State.Strict (gets, modify)
@@ -91,6 +92,13 @@ stringToLaTeX context zs = do
             '\'':_ -> cs <> "\\," <> xs -- add thin space
             _      -> cs <> xs
     in case x of
+         '\\'| isUrl -> emitc '/' -- NB. / works as path sep even on Windows
+         c | isUrl ->
+             if c `elem` ['{', '}', '|', '^', '~', '[', ']', '`']
+                then emits (escapeURIString (const False) [c])
+                else emitc c
+         '{' -> emits "\\{"
+         '}' -> emits "\\}"
          '?' | ligatures ->  -- avoid ?` ligature
            case xs of
              '`':_ -> emits "?{}"
@@ -99,27 +107,24 @@ stringToLaTeX context zs = do
            case xs of
              '`':_ -> emits "!{}"
              _     -> emitc x
-         '{' -> emits "\\{"
-         '}' -> emits "\\}"
          '`' | ctx == CodeString -> emitcseq "\\textasciigrave"
-         '$' | not isUrl -> emits "\\$"
+         '$' -> emits "\\$"
          '%' -> emits "\\%"
-         '&' | not isUrl -> emits "\\&"
-         '_' | not isUrl -> emits "\\_"
+         '&' -> emits "\\&"
+         '_' -> emits "\\_"
          '#' -> emits "\\#"
-         '-' | not isUrl -> case xs of
+         '-' -> case xs of
                      -- prevent adjacent hyphens from forming ligatures
                      ('-':_) -> emits "-\\/"
                      _       -> emitc '-'
-         '~' | not isUrl -> emitcseq "\\textasciitilde"
+         '~' -> emitcseq "\\textasciitilde"
          '^' -> emits "\\^{}"
-         '\\'| isUrl     -> emitc '/' -- NB. / works as path sep even on Windows
-             | otherwise -> emitcseq "\\textbackslash"
-         '|' | not isUrl -> emitcseq "\\textbar"
-         '<' -> emitcseq "\\textless"
-         '>' -> emitcseq "\\textgreater"
-         '[' -> emits "{[}"  -- to avoid interpretation as
-         ']' -> emits "{]}"  -- optional arguments
+         '\\' -> emitcseq "\\textbackslash"
+         '|'  -> emitcseq "\\textbar"
+         '<'  -> emitcseq "\\textless"
+         '>'  -> emitcseq "\\textgreater"
+         '['  -> emits "{[}"  -- to avoid interpretation as
+         ']'  -> emits "{]}"  -- optional arguments
          '\'' -> emitcseq "\\textquotesingle"
          '\160' -> emits "~"
          '\x200B' -> emits "\\hspace{0pt}"  -- zero-width space

diff --git a/test/command/5340.md b/test/command/5340.md
@@ -8,7 +8,7 @@
 ^D
 \url{https://example.com/foo-bar}
 \url{https://example.com/foo--bar}
-\url{https://example.com/foo\%2Dbar}
-\url{https://example.com/foo\%2D\%2Dbar}
-\url{https://example.com/foo\%2D\%2Dbar}
+\url{https://example.com/foo%2Dbar}
+\url{https://example.com/foo%2D%2Dbar}
+\url{https://example.com/foo%2D%2Dbar}
 ```
diff --git a/test/command/8992.md b/test/command/8992.md
@@ -0,0 +1,6 @@
+```
+% pandoc -t latex
+[A theorem](https://en.wikipedia.org/wiki/Rice's_theorem).
+^D
+\href{https://en.wikipedia.org/wiki/Rice's_theorem}{A theorem}.
+```