DocBook reader: More improvements, passes tests.

3 files changed, 32 insertions, 20 deletions

diff --git a/src/Text/Pandoc/Readers/DocBook.hs b/src/Text/Pandoc/Readers/DocBook.hs
index ba2b79049..dcdd2de56 100644
--- a/src/Text/Pandoc/Readers/DocBook.hs
+++ b/src/Text/Pandoc/Readers/DocBook.hs
@@ -89,7 +89,7 @@ List of all DocBook tags, with [x] indicating implemented,
 [ ] collab - Identifies a collaborator
 [ ] collabname - The name of a collaborator
 [ ] colophon - Text at the back of a book describing facts about its production
-[ ] colspec - Specifications for a column in a table
+[x] colspec - Specifications for a column in a table
 [ ] command - The name of an executable program or other software command
 [ ] computeroutput - Data, generally text, displayed or presented by a computer
 [ ] confdates - The dates of a conference for which a document was written
@@ -570,7 +570,8 @@ getImage e = do
                 Just z   -> case filterChild (named "imagedata") z of
                               Nothing -> return mempty
                               Just i -> return $ attrValue "fileref" i
-  caption <- case filterChild (named "caption") e of
+  caption <- case filterChild
+                  (\x -> named "caption" x || named "textobject" x) e of
                Nothing  -> return mempty
                Just z   -> mconcat <$> (mapM parseInline $ elContent z)
   return $ image imageUrl "" caption
@@ -709,15 +710,26 @@ parseBlock (Elem e) =
                                                 Just "right"  -> AlignRight
                                                 Just "center" -> AlignCenter
                                                 _             -> AlignDefault
+                      let toWidth c = case findAttr (unqual "colwidth") c of
+                                                Just w -> read $ filter (\x ->
+                                                     (x >= '0' && x <= '9')
+                                                      || x == '.') w
+                                                Nothing -> 0 :: Double
                       let numrows = maximum $ map length bodyrows
                       let aligns = case colspecs of
                                      []  -> replicate numrows AlignDefault
                                      cs  -> map toAlignment cs
+                      let widths = case colspecs of
+                                     []  -> replicate numrows 0
+                                     cs  -> let ws = map toWidth cs
+                                                tot = sum ws
+                                            in  if all (> 0) ws
+                                                   then map (/ tot) ws
+                                                   else replicate numrows 0
                       let headrows' = if null headrows
                                          then replicate numrows mempty
                                          else headrows
-                      return $ table caption
-                                 (zip aligns (repeat 0))
+                      return $ table caption (zip aligns widths)
                                  headrows' bodyrows
          isEntry x  = named "entry" x || named "td" x || named "th" x
          parseRow = mapM getBlocks . filterChildren isEntry
diff --git a/tests/docbook-reader.docbook b/tests/docbook-reader.docbook
index 1b195602d..eb70dd33c 100644
--- a/tests/docbook-reader.docbook
+++ b/tests/docbook-reader.docbook
@@ -1259,10 +1259,10 @@ or here: &lt;http://example.com/&gt;
       Here's the caption. It may span multiple lines.
     </title>
     <tgroup cols="4">
-      <colspec colwidth="15*" align="center" />
-      <colspec colwidth="13*" align="left" />
-      <colspec colwidth="16*" align="right" />
-      <colspec colwidth="33*" align="left" />
+      <colspec colwidth="2*" align="center" />
+      <colspec colwidth="2*" align="left" />
+      <colspec colwidth="3*" align="right" />
+      <colspec colwidth="3*" align="left" />
       <thead>
         <row>
           <entry>
@@ -1316,10 +1316,10 @@ or here: &lt;http://example.com/&gt;
   </para>
   <informaltable>
     <tgroup cols="4">
-      <colspec colwidth="15*" align="center" />
-      <colspec colwidth="13*" align="left" />
-      <colspec colwidth="16*" align="right" />
-      <colspec colwidth="33*" align="left" />
+      <colspec colwidth="1*" align="center" />
+      <colspec colwidth="2*" align="left" />
+      <colspec colwidth="3*" align="right" />
+      <colspec colwidth="4*" align="left" />
       <thead>
         <row>
           <entry>
@@ -1428,10 +1428,10 @@ or here: &lt;http://example.com/&gt;
   </para>
   <informaltable>
     <tgroup cols="4">
-      <colspec colwidth="15*" align="center" />
-      <colspec colwidth="13*" align="left" />
-      <colspec colwidth="16*" align="right" />
-      <colspec colwidth="33*" align="left" />
+      <colspec colwidth="10*" align="center" />
+      <colspec colwidth="10*" align="left" />
+      <colspec colwidth="10*" align="right" />
+      <colspec colwidth="10*" align="left" />
       <tbody>
         <row>
           <entry>
diff --git a/tests/docbook-reader.native b/tests/docbook-reader.native
index aa459eff7..3d403648b 100644
--- a/tests/docbook-reader.native
+++ b/tests/docbook-reader.native
@@ -261,7 +261,7 @@ Pandoc (Meta {docTitle = [Str "Pandoc",Space,Str "Test",Space,Str "Suite"], docA
 ,Header 1 [Str "Images"]
 ,Para [Str "From",Space,Quoted DoubleQuote [Str "Voyage",Space,Str "dans",Space,Str "la",Space,Str "Lune"],Space,Str "by",Space,Str "Georges",Space,Str "Melies",Space,Str "(1902):"]
 ,Para [Image [Str "lalune"] ("lalune.jpg","")]
-,Para [Str "Here",Space,Str "is",Space,Str "a",Space,Str "movie",Space,Image [Str "movie"] ("movie.jpg",""),Space,Str "icon."]
+,Para [Str "Here",Space,Str "is",Space,Str "a",Space,Str "movie",Space,Image [] ("movie.jpg",""),Space,Str "icon."]
 ,Header 1 [Str "Footnotes"]
 ,Para [Str "Here",Space,Str "is",Space,Str "a",Space,Str "footnote",Space,Str "reference,",Note [Para [Str "Here",Space,Str "is",Space,Str "the",Space,Str "footnote.",Space,Str "It",Space,Str "can",Space,Str "go",Space,Str "anywhere",Space,Str "after",Space,Str "the",Space,Str "footnote",Space,Str "reference.",Space,Str "It",Space,Str "need",Space,Str "not",Space,Str "be",Space,Str "placed",Space,Str "at",Space,Str "the",Space,Str "end",Space,Str "of",Space,Str "the",Space,Str "document."]],Space,Str "and",Space,Str "another.",Note [Para [Str "Here\8217s",Space,Str "the",Space,Str "long",Space,Str "note.",Space,Str "This",Space,Str "one",Space,Str "contains",Space,Str "multiple",Space,Str "blocks."],Para [Str "Subsequent",Space,Str "blocks",Space,Str "are",Space,Str "indented",Space,Str "to",Space,Str "show",Space,Str "that",Space,Str "they",Space,Str "belong",Space,Str "to",Space,Str "the",Space,Str "footnote",Space,Str "(as",Space,Str "with",Space,Str "list",Space,Str "items)."],CodeBlock ("",[],[]) "  { <code> }",Para [Str "If",Space,Str "you",Space,Str "want,",Space,Str "you",Space,Str "can",Space,Str "indent",Space,Str "every",Space,Str "line,",Space,Str "but",Space,Str "you",Space,Str "can",Space,Str "also",Space,Str "be",Space,Str "lazy",Space,Str "and",Space,Str "just",Space,Str "indent",Space,Str "the",Space,Str "first",Space,Str "line",Space,Str "of",Space,Str "each",Space,Str "block."]],Space,Str "This",Space,Str "should",Space,Emph [Str "not"],Space,Str "be",Space,Str "a",Space,Str "footnote",Space,Str "reference,",Space,Str "because",Space,Str "it",Space,Str "contains",Space,Str "a",Space,Str "space.[^my",Space,Str "note]",Space,Str "Here",Space,Str "is",Space,Str "an",Space,Str "inline",Space,Str "note.",Note [Para [Str "This",Space,Str "is",Space,Emph [Str "easier"],Space,Str "to",Space,Str "type.",Space,Str "Inline",Space,Str "notes",Space,Str "may",Space,Str "contain",Space,Link [Str "links"] ("http://google.com",""),Space,Str "and",Space,Code ("",[],[]) "]",Space,Str "verbatim",Space,Str "characters,",Space,Str "as",Space,Str "well",Space,Str "as",Space,Str "[bracketed",Space,Str "text]."]]]
 ,BlockQuote
@@ -325,7 +325,7 @@ Pandoc (Meta {docTitle = [Str "Pandoc",Space,Str "Test",Space,Str "Suite"], docA
   ,[Plain [Str "1"]]
   ,[Plain [Str "1"]]]]
 ,Para [Str "Multiline",Space,Str "table",Space,Str "with",Space,Str "caption:"]
-,Table [Str "Here's",Space,Str "the",Space,Str "caption.",Space,Str "It",Space,Str "may",Space,Str "span",Space,Str "multiple",Space,Str "lines."] [AlignCenter,AlignLeft,AlignRight,AlignLeft] [0.15,0.1375,0.1625,0.3375]
+,Table [Str "Here's",Space,Str "the",Space,Str "caption.",Space,Str "It",Space,Str "may",Space,Str "span",Space,Str "multiple",Space,Str "lines."] [AlignCenter,AlignLeft,AlignRight,AlignLeft] [0.2,0.2,0.3,0.3]
  [[Plain [Str "Centered",Space,Str "Header"]]
  ,[Plain [Str "Left",Space,Str "Aligned"]]
  ,[Plain [Str "Right",Space,Str "Aligned"]]
@@ -339,7 +339,7 @@ Pandoc (Meta {docTitle = [Str "Pandoc",Space,Str "Test",Space,Str "Suite"], docA
   ,[Plain [Str "5.0"]]
   ,[Plain [Str "Here's",Space,Str "another",Space,Str "one.",Space,Str "Note",Space,Str "the",Space,Str "blank",Space,Str "line",Space,Str "between",Space,Str "rows."]]]]
 ,Para [Str "Multiline",Space,Str "table",Space,Str "without",Space,Str "caption:"]
-,Table [] [AlignCenter,AlignLeft,AlignRight,AlignLeft] [0.15,0.1375,0.1625,0.3375]
+,Table [] [AlignCenter,AlignLeft,AlignRight,AlignLeft] [0.1,0.2,0.3,0.4]
  [[Plain [Str "Centered",Space,Str "Header"]]
  ,[Plain [Str "Left",Space,Str "Aligned"]]
  ,[Plain [Str "Right",Space,Str "Aligned"]]
@@ -371,7 +371,7 @@ Pandoc (Meta {docTitle = [Str "Pandoc",Space,Str "Test",Space,Str "Suite"], docA
   ,[Plain [Str "1"]]
   ,[Plain [Str "1"]]]]
 ,Para [Str "Multiline",Space,Str "table",Space,Str "without",Space,Str "column",Space,Str "headers:"]
-,Table [] [AlignCenter,AlignLeft,AlignRight,AlignLeft] [0.15,0.1375,0.1625,0.3375]
+,Table [] [AlignCenter,AlignLeft,AlignRight,AlignLeft] [0.25,0.25,0.25,0.25]
  [[]
  ,[]
  ,[]