Docx parser: implement PAGEREF fields

These fields, often used in tables of contents, can be a hyperlink.

8 files changed, 41 insertions, 7 deletions

diff --git a/src/Text/Pandoc/Readers/Docx.hs b/src/Text/Pandoc/Readers/Docx.hs
index 462e3c679..5c8f20c18 100644
--- a/src/Text/Pandoc/Readers/Docx.hs
+++ b/src/Text/Pandoc/Readers/Docx.hs
@@ -448,6 +448,7 @@ parPartToInlines' (PlainOMath exps) =
 parPartToInlines' (Field info children) =
   case info of
     HyperlinkField url -> parPartToInlines' $ ExternalHyperLink url children
+    PagerefField fieldAnchor True -> parPartToInlines' $ InternalHyperLink fieldAnchor children
     _ -> smushInlines <$> mapM parPartToInlines' children
 parPartToInlines' NullParPart = return mempty
 
diff --git a/src/Text/Pandoc/Readers/Docx/Fields.hs b/src/Text/Pandoc/Readers/Docx/Fields.hs
index 442bc3466..5f090b6be 100644
--- a/src/Text/Pandoc/Readers/Docx/Fields.hs
+++ b/src/Text/Pandoc/Readers/Docx/Fields.hs
@@ -21,8 +21,11 @@ import Text.Parsec
 import Text.Parsec.Text (Parser)
 
 type URL = T.Text
+type Anchor = T.Text
 
 data FieldInfo = HyperlinkField URL
+                -- The boolean indicates whether the field is a hyperlink.
+               | PagerefField Anchor Bool
                | UnknownField
                deriving (Show)
 
@@ -33,6 +36,8 @@ fieldInfo :: Parser FieldInfo
 fieldInfo =
   try (HyperlinkField <$> hyperlink)
   <|>
+  try ((uncurry PagerefField) <$> pageref) 
+  <|>
   return UnknownField
 
 escapedQuote :: Parser T.Text
@@ -72,3 +77,23 @@ hyperlink = do
               ("\\l", s) : _ -> farg <> "#" <> s
               _              -> farg
   return url
+
+-- See §17.16.5.45
+pagerefSwitch :: Parser (T.Text, T.Text)
+pagerefSwitch = do
+  sw <- string "\\h"
+  spaces
+  farg <- fieldArgument
+  return (T.pack sw, farg)
+
+pageref :: Parser (Anchor, Bool)
+pageref = do
+  many space
+  string "PAGEREF"
+  spaces
+  farg <- fieldArgument
+  switches <- spaces *> many pagerefSwitch
+  let isLink = case switches of
+              ("\\h", _) : _ -> True 
+              _              -> False
+  return (farg, isLink)
diff --git a/test/Tests/Readers/Docx.hs b/test/Tests/Readers/Docx.hs
index af6023836..ea4094c82 100644
--- a/test/Tests/Readers/Docx.hs
+++ b/test/Tests/Readers/Docx.hs
@@ -152,6 +152,10 @@ tests = [ testGroup "document"
             "docx/nested_instrText.docx"
             "docx/nested_instrText.native"
           , testCompare
+            "pageref hyperlinks in <w:instrText> tag"
+            "docx/pageref.docx"
+            "docx/pageref.native"
+          , testCompare
             "inline image"
             "docx/image.docx"
             "docx/image_no_embed.native"
diff --git a/test/docx/0_level_headers.native b/test/docx/0_level_headers.native
index ed589b029..7080063f9 100644
--- a/test/docx/0_level_headers.native
+++ b/test/docx/0_level_headers.native
@@ -39,9 +39,9 @@
  [])
 ,Para [Str "CONTENTS"]
 ,Para [Strong [Str "Section",Space,Str "Page"]]
-,Para [Str "FIGURES",Space,Str "iv"]
-,Para [Str "TABLES",Space,Str "v"]
-,Para [Str "SECTION",Space,Str "1",Space,Str "Introduction",Space,Str "2"]
+,Para [Str "FIGURES",Space,Link ("",[],[]) [Str "iv"] ("#figures","")]
+,Para [Str "TABLES",Space,Link ("",[],[]) [Str "v"] ("#tables","")]
+,Para [Str "SECTION",Space,Str "1",Space,Str "Introduction",Space,Link ("",[],[]) [Str "2"] ("#introduction","")]
 ,Header 1 ("figures",["Heading-0"],[]) [Str "FIGURES"]
 ,Para [Strong [Str "Figure",Space,Str "Page"]]
 ,Para [Strong [Str "No",Space,Str "table",Space,Str "of",Space,Str "figures",Space,Str "entries",Space,Str "found."]]
diff --git a/test/docx/golden/nested_anchors_in_header.docx b/test/docx/golden/nested_anchors_in_header.docx
index 88dd21abd..9d89070d9 100644
--- a/test/docx/golden/nested_anchors_in_header.docx
+++ b/test/docx/golden/nested_anchors_in_header.docx
diff --git a/test/docx/nested_anchors_in_header.native b/test/docx/nested_anchors_in_header.native
index 314b31663..9cc256d28 100644
--- a/test/docx/nested_anchors_in_header.native
+++ b/test/docx/nested_anchors_in_header.native
@@ -1,8 +1,8 @@
 [Header 1 ("\1086\1075\1083\1072\1074\1083\1077\1085\1080\1077",["TOC-Heading"],[]) [Str "\1054\1075\1083\1072\1074\1083\1077\1085\1080\1077"]
-,Para [Link ("",[],[]) [Str "Short",Space,Str "instructions",Space,Str "1"] ("#short-instructions","")]
-,Para [Link ("",[],[]) [Str "Some",Space,Str "instructions",Space,Str "1"] ("#some-instructions","")]
-,Para [Link ("",[],[]) [Str "Remote",Space,Str "folder",Space,Str "or",Space,Str "longlonglonglonglong",Space,Str "file",Space,Str "with",Space,Str "manymanymanymany",Space,Str "letters",Space,Str "inside",Space,Str "opening",Space,Str "2"] ("#remote-folder-or-longlonglonglonglong-file-with-manymanymanymany-letters-inside-opening","")]
-,Para [Link ("",[],[]) [Str "Remote",Space,Str "folder",Space,Str "or",Space,Str "longlonglonglonglong",Space,Str "file",Space,Str "with",Space,Str "manymanymanymany",Space,Str "letters",Space,Str "inside",Space,Str "closing",Space,Str "2"] ("#remote-folder-or-longlonglonglonglong-file-with-manymanymanymany-letters-inside-closing","")]
+,Para [Link ("",[],[]) [Str "Short",Space,Str "instructions",Space,Link ("",[],[]) [Str "1"] ("#short-instructions","")] ("#short-instructions","")]
+,Para [Link ("",[],[]) [Str "Some",Space,Str "instructions",Space,Link ("",[],[]) [Str "1"] ("#some-instructions","")] ("#some-instructions","")]
+,Para [Link ("",[],[]) [Str "Remote",Space,Str "folder",Space,Str "or",Space,Str "longlonglonglonglong",Space,Str "file",Space,Str "with",Space,Str "manymanymanymany",Space,Str "letters",Space,Str "inside",Space,Str "opening",Space,Link ("",[],[]) [Str "2"] ("#remote-folder-or-longlonglonglonglong-file-with-manymanymanymany-letters-inside-opening","")] ("#remote-folder-or-longlonglonglonglong-file-with-manymanymanymany-letters-inside-opening","")]
+,Para [Link ("",[],[]) [Str "Remote",Space,Str "folder",Space,Str "or",Space,Str "longlonglonglonglong",Space,Str "file",Space,Str "with",Space,Str "manymanymanymany",Space,Str "letters",Space,Str "inside",Space,Str "closing",Space,Link ("",[],[]) [Str "2"] ("#remote-folder-or-longlonglonglonglong-file-with-manymanymanymany-letters-inside-closing","")] ("#remote-folder-or-longlonglonglonglong-file-with-manymanymanymany-letters-inside-closing","")]
 ,Header 1 ("short-instructions",[],[]) [Str "Short",Space,Str "instructions"]
 ,Para [Link ("",[],[]) [Str "Open",Space,Str "remote",Space,Str "folder"] ("#remote-folder-or-longlonglonglonglong-file-with-manymanymanymany-letters-inside-opening","")]
 ,Para [Str "Do",Space,Str "staff"]
diff --git a/test/docx/pageref.docx b/test/docx/pageref.docx
new file mode 100644
index 000000000..0a9159cab
--- /dev/null
+++ b/test/docx/pageref.docx
diff --git a/test/docx/pageref.native b/test/docx/pageref.native
new file mode 100644
index 000000000..6c683de67
--- /dev/null
+++ b/test/docx/pageref.native
@@ -0,0 +1,4 @@
+[Para [Str "Title",Space,Link ("",[],[]) [Str "2"] ("#title","")]
+,Para [Str "Title2",Space,Link ("",[],[]) [Str "2"] ("#title2","")]
+,Header 1 ("title", [],[]) [Str "Title"]
+,Header 1 ("title2",[],[]) [Str "Title2"]]