From 4f8c536de0dec4bf72485b43d6d0edd68fefb0bb Mon Sep 17 00:00:00 2001
From: John MacFarlane <fiddlosopher@gmail.com>
Date: Tue, 8 May 2012 23:25:34 -0700
Subject: [PATCH] DocBook reader: More improvements, more tests pass.
---
src/Text/Pandoc/Readers/DocBook.hs | 24 ++++++++++++----
tests/docbook-reader.docbook | 13 +++------
tests/docbook-reader.native | 44 ++++++++++++++----------------
3 files changed, 44 insertions(+), 37 deletions(-)
diff --git a/src/Text/Pandoc/Readers/DocBook.hs b/src/Text/Pandoc/Readers/DocBook.hs
index 665d89a6a..ba2b79049 100644
--- a/src/Text/Pandoc/Readers/DocBook.hs
+++ b/src/Text/Pandoc/Readers/DocBook.hs
@@ -1,5 +1,5 @@
module Text.Pandoc.Readers.DocBook ( readDocBook ) where
-import Data.Char (toUpper)
+import Data.Char (toUpper, isDigit)
import Text.Pandoc.Parsing (ParserState(..))
import Text.Pandoc.Definition
import Text.Pandoc.Builder
@@ -10,7 +10,7 @@ import Data.Monoid
import Data.Char (isSpace)
import Control.Monad.State
import Control.Applicative ((<$>))
-import Data.List (intersperse, transpose)
+import Data.List (intersperse)
{-
@@ -619,7 +619,20 @@ parseBlock (Elem e) =
"answer" -> addToStart (strong (str "A:") <> str " ") <$> getBlocks e
"abstract" -> blockQuote <$> getBlocks e
"itemizedlist" -> bulletList <$> listitems
- "orderedlist" -> orderedList <$> listitems -- TODO list attributes
+ "orderedlist" -> do
+ let listStyle = case attrValue "numeration" e of
+ "arabic" -> Decimal
+ "loweralpha" -> LowerAlpha
+ "upperalpha" -> UpperAlpha
+ "lowerroman" -> LowerRoman
+ "upperroman" -> UpperRoman
+ _ -> Decimal
+ let start = case attrValue "override" <$>
+ filterElement (named "listitem") e of
+ Just x@(_:_) | all isDigit x -> read x
+ _ -> 1
+ orderedListWith (start,listStyle,DefaultDelim)
+ <$> listitems -- TODO list attributes
"variablelist" -> definitionList <$> deflistitems
"mediaobject" -> para <$> (getImage e)
"caption" -> return mempty
@@ -653,8 +666,8 @@ parseBlock (Elem e) =
parseVarListEntry e' = do
let terms = filterChildren (named "term") e'
let items = filterChildren (named "listitem") e'
- terms' <- mapM ((trimInlines . mconcat <$>) . mapM parseInline . elContent) terms
- items' <- mapM ((mconcat <$>) . mapM parseBlock . elContent) items
+ terms' <- mapM getInlines terms
+ items' <- mapM getBlocks items
return (mconcat $ intersperse (str "; ") terms', items')
getTitle = case filterChild (named "title") e of
Just t -> do
@@ -768,6 +781,7 @@ parseInline (Elem e) =
"foreignphrase" -> emph <$> innerInlines
"emphasis" -> case attrValue "role" e of
"strong" -> strong <$> innerInlines
+ "strikethrough" -> strikeout <$> innerInlines
_ -> emph <$> innerInlines
"footnote" -> (note . mconcat) <$> (mapM parseBlock $ elContent e)
_ -> innerInlines
diff --git a/tests/docbook-reader.docbook b/tests/docbook-reader.docbook
index c66cdf3ec..6aca2c087 100644
--- a/tests/docbook-reader.docbook
+++ b/tests/docbook-reader.docbook
@@ -602,6 +602,8 @@ These should not be escaped: \$ \\ \> \[ \{
<para>
orange fruit
</para>
+ </listitem>
+ <listitem>
<para>
bank
</para>
@@ -620,6 +622,8 @@ These should not be escaped: \$ \\ \> \[ \{
<para>
red fruit
</para>
+ </listitem>
+ <listitem>
<para>
computer
</para>
@@ -708,19 +712,10 @@ These should not be escaped: \$ \\ \> \[ \{
<para>
<quote>A</quote>, <quote>B</quote>, and <quote>C</quote> are letters.
</para>
- <para>
- <quote>Oak,</quote> <quote>elm,</quote> and <quote>beech</quote> are names
- of trees. So is <quote>pine.</quote>
- </para>
<para>
<quote>He said, <quote>I want to go.</quote></quote> Were you alive in the
70’s?
</para>
- <para>
- Here is some quoted <quote><literal>code</literal></quote> and a
- <quote><ulink url="http://example.com/?foo=1&bar=2">quoted
- link</ulink></quote>.
- </para>
<para>
Some dashes: one—two — three—four — five.
</para>
diff --git a/tests/docbook-reader.native b/tests/docbook-reader.native
index b91eed8ab..93095afef 100644
--- a/tests/docbook-reader.native
+++ b/tests/docbook-reader.native
@@ -105,14 +105,14 @@ Pandoc (Meta {docTitle = [Str "Pandoc",Space,Str "Test",Space,Str "Suite"], docA
[[Para [Str "this",Space,Str "is",Space,Str "an",Space,Str "example",Space,Str "list",Space,Str "item",Space,Str "indented",Space,Str "with",Space,Str "tabs"]]
,[Para [Str "this",Space,Str "is",Space,Str "an",Space,Str "example",Space,Str "list",Space,Str "item",Space,Str "indented",Space,Str "with",Space,Str "spaces"]]]]]
,Header 2 [Str "Fancy",Space,Str "list",Space,Str "markers"]
-,OrderedList (2,Decimal,TwoParens)
+,OrderedList (2,Decimal,DefaultDelim)
[[Para [Str "begins",Space,Str "with",Space,Str "2"]]
,[Para [Str "and",Space,Str "now",Space,Str "3"]
,Para [Str "with",Space,Str "a",Space,Str "continuation"]
,OrderedList (4,LowerRoman,DefaultDelim)
[[Para [Str "sublist",Space,Str "with",Space,Str "roman",Space,Str "numerals,",Space,Str "starting",Space,Str "with",Space,Str "4"]]
,[Para [Str "more",Space,Str "items"]
- ,OrderedList (1,UpperAlpha,TwoParens)
+ ,OrderedList (1,UpperAlpha,DefaultDelim)
[[Para [Str "a",Space,Str "subsublist"]]
,[Para [Str "a",Space,Str "subsublist"]]]]]]]
,Para [Str "Nesting:"]
@@ -120,15 +120,15 @@ Pandoc (Meta {docTitle = [Str "Pandoc",Space,Str "Test",Space,Str "Suite"], docA
[[Para [Str "Upper",Space,Str "Alpha"]
,OrderedList (1,UpperRoman,DefaultDelim)
[[Para [Str "Upper",Space,Str "Roman."]
- ,OrderedList (6,Decimal,TwoParens)
+ ,OrderedList (6,Decimal,DefaultDelim)
[[Para [Str "Decimal",Space,Str "start",Space,Str "with",Space,Str "6"]
- ,OrderedList (3,LowerAlpha,OneParen)
+ ,OrderedList (3,LowerAlpha,DefaultDelim)
[[Para [Str "Lower",Space,Str "alpha",Space,Str "with",Space,Str "paren"]]]]]]]]]
,Para [Str "Autonumbering:"]
-,OrderedList (1,DefaultStyle,DefaultDelim)
+,OrderedList (1,Decimal,DefaultDelim)
[[Para [Str "Autonumber."]]
,[Para [Str "More."]
- ,OrderedList (1,DefaultStyle,DefaultDelim)
+ ,OrderedList (1,Decimal,DefaultDelim)
[[Para [Str "Nested."]]]]]
,Para [Str "Should",Space,Str "not",Space,Str "be",Space,Str "a",Space,Str "list",Space,Str "item:"]
,Para [Str "M.A.\160\&2007"]
@@ -184,10 +184,8 @@ Pandoc (Meta {docTitle = [Str "Pandoc",Space,Str "Test",Space,Str "Suite"], docA
,Para [Str "These",Space,Str "should",Space,Str "not",Space,Str "be",Space,Str "superscripts",Space,Str "or",Space,Str "subscripts,",Space,Str "because",Space,Str "of",Space,Str "the",Space,Str "unescaped",Space,Str "spaces:",Space,Str "a^b",Space,Str "c^d,",Space,Str "a~b",Space,Str "c~d."]
,Header 1 [Str "Smart",Space,Str "quotes,",Space,Str "ellipses,",Space,Str "dashes"]
,Para [Quoted DoubleQuote [Str "Hello,"],Space,Str "said",Space,Str "the",Space,Str "spider.",Space,Quoted DoubleQuote [Quoted SingleQuote [Str "Shelob"],Space,Str "is",Space,Str "my",Space,Str "name."]]
-,Para [Quoted SingleQuote [Str "A"],Str ",",Space,Quoted SingleQuote [Str "B"],Str ",",Space,Str "and",Space,Quoted SingleQuote [Str "C"],Space,Str "are",Space,Str "letters."]
-,Para [Quoted SingleQuote [Str "Oak,"],Space,Quoted SingleQuote [Str "elm,"],Space,Str "and",Space,Quoted SingleQuote [Str "beech"],Space,Str "are",Space,Str "names",Space,Str "of",Space,Str "trees.",Space,Str "So",Space,Str "is",Space,Quoted SingleQuote [Str "pine."]]
-,Para [Quoted SingleQuote [Str "He",Space,Str "said,",Space,Quoted DoubleQuote [Str "I",Space,Str "want",Space,Str "to",Space,Str "go."]],Space,Str "Were",Space,Str "you",Space,Str "alive",Space,Str "in",Space,Str "the",Space,Str "70\8217s?"]
-,Para [Str "Here",Space,Str "is",Space,Str "some",Space,Str "quoted",Space,Quoted SingleQuote [Code ("",[],[]) "code"],Space,Str "and",Space,Str "a",Space,Quoted DoubleQuote [Link [Str "quoted",Space,Str "link"] ("http://example.com/?foo=1&bar=2","")],Str "."]
+,Para [Quoted DoubleQuote [Str "A"],Str ",",Space,Quoted DoubleQuote [Str "B"],Str ",",Space,Str "and",Space,Quoted DoubleQuote [Str "C"],Space,Str "are",Space,Str "letters."]
+,Para [Quoted DoubleQuote [Str "He",Space,Str "said,",Space,Quoted SingleQuote [Str "I",Space,Str "want",Space,Str "to",Space,Str "go."]],Space,Str "Were",Space,Str "you",Space,Str "alive",Space,Str "in",Space,Str "the",Space,Str "70\8217s?"]
,Para [Str "Some",Space,Str "dashes:",Space,Str "one\8212two",Space,Str "\8212",Space,Str "three\8212four",Space,Str "\8212",Space,Str "five."]
,Para [Str "Dashes",Space,Str "between",Space,Str "numbers:",Space,Str "5\8211\&7,",Space,Str "255\8211\&66,",Space,Str "1987\8211\&1999."]
,Para [Str "Ellipses\8230and\8230and\8230."]
@@ -223,11 +221,11 @@ Pandoc (Meta {docTitle = [Str "Pandoc",Space,Str "Test",Space,Str "Suite"], docA
,Header 1 [Str "Links"]
,Header 2 [Str "Explicit"]
,Para [Str "Just",Space,Str "a",Space,Link [Str "URL"] ("/url/",""),Str "."]
-,Para [Link [Str "URL",Space,Str "and",Space,Str "title"] ("/url/","title"),Str "."]
-,Para [Link [Str "URL",Space,Str "and",Space,Str "title"] ("/url/","title preceded by two spaces"),Str "."]
-,Para [Link [Str "URL",Space,Str "and",Space,Str "title"] ("/url/","title preceded by a tab"),Str "."]
-,Para [Link [Str "URL",Space,Str "and",Space,Str "title"] ("/url/","title with \"quotes\" in it")]
-,Para [Link [Str "URL",Space,Str "and",Space,Str "title"] ("/url/","title with single quotes")]
+,Para [Link [Str "URL",Space,Str "and",Space,Str "title"] ("/url/",""),Str "."]
+,Para [Link [Str "URL",Space,Str "and",Space,Str "title"] ("/url/",""),Str "."]
+,Para [Link [Str "URL",Space,Str "and",Space,Str "title"] ("/url/",""),Str "."]
+,Para [Link [Str "URL",Space,Str "and",Space,Str "title"] ("/url/","")]
+,Para [Link [Str "URL",Space,Str "and",Space,Str "title"] ("/url/","")]
,Para [Link [Str "with_underscore"] ("/url/with_underscore","")]
,Para [Link [Str "Email",Space,Str "link"] ("mailto:nobody@nowhere.net","")]
,Para [Link [Str "Empty"] ("",""),Str "."]
@@ -242,27 +240,27 @@ Pandoc (Meta {docTitle = [Str "Pandoc",Space,Str "Test",Space,Str "Suite"], docA
,Para [Str "Indented",Space,Link [Str "thrice"] ("/url",""),Str "."]
,Para [Str "This",Space,Str "should",Space,Str "[not][]",Space,Str "be",Space,Str "a",Space,Str "link."]
,CodeBlock ("",[],[]) "[not]: /url"
-,Para [Str "Foo",Space,Link [Str "bar"] ("/url/","Title with \"quotes\" inside"),Str "."]
-,Para [Str "Foo",Space,Link [Str "biz"] ("/url/","Title with \"quote\" inside"),Str "."]
+,Para [Str "Foo",Space,Link [Str "bar"] ("/url/",""),Str "."]
+,Para [Str "Foo",Space,Link [Str "biz"] ("/url/",""),Str "."]
,Header 2 [Str "With",Space,Str "ampersands"]
,Para [Str "Here\8217s",Space,Str "a",Space,Link [Str "link",Space,Str "with",Space,Str "an",Space,Str "ampersand",Space,Str "in",Space,Str "the",Space,Str "URL"] ("http://example.com/?foo=1&bar=2",""),Str "."]
-,Para [Str "Here\8217s",Space,Str "a",Space,Str "link",Space,Str "with",Space,Str "an",Space,Str "amersand",Space,Str "in",Space,Str "the",Space,Str "link",Space,Str "text:",Space,Link [Str "AT&T"] ("http://att.com/","AT&T"),Str "."]
+,Para [Str "Here\8217s",Space,Str "a",Space,Str "link",Space,Str "with",Space,Str "an",Space,Str "amersand",Space,Str "in",Space,Str "the",Space,Str "link",Space,Str "text:",Space,Link [Str "AT&T"] ("http://att.com/",""),Str "."]
,Para [Str "Here\8217s",Space,Str "an",Space,Link [Str "inline",Space,Str "link"] ("/script?foo=1&bar=2",""),Str "."]
,Para [Str "Here\8217s",Space,Str "an",Space,Link [Str "inline",Space,Str "link",Space,Str "in",Space,Str "pointy",Space,Str "braces"] ("/script?foo=1&bar=2",""),Str "."]
,Header 2 [Str "Autolinks"]
-,Para [Str "With",Space,Str "an",Space,Str "ampersand:",Space,Link [Code ("",["url"],[]) "http://example.com/?foo=1&bar=2"] ("http://example.com/?foo=1&bar=2","")]
+,Para [Str "With",Space,Str "an",Space,Str "ampersand:",Space,Link [Code ("",[],[]) "http://example.com/?foo=1&bar=2"] ("http://example.com/?foo=1&bar=2","")]
,BulletList
[[Para [Str "In",Space,Str "a",Space,Str "list?"]]
- ,[Para [Link [Code ("",["url"],[]) "http://example.com/"] ("http://example.com/","")]]
+ ,[Para [Link [Code ("",[],[]) "http://example.com/"] ("http://example.com/","")]]
,[Para [Str "It",Space,Str "should."]]]
-,Para [Str "An",Space,Str "e-mail",Space,Str "address:",Space,Link [Code ("",["url"],[]) "nobody@nowhere.net"] ("mailto:nobody@nowhere.net","")]
+,Para [Str "An",Space,Str "e-mail",Space,Str "address:",Space,Link [Code ("",[],[]) "nobody@nowhere.net"] ("mailto:nobody@nowhere.net","")]
,BlockQuote
- [Para [Str "Blockquoted:",Space,Link [Code ("",["url"],[]) "http://example.com/"] ("http://example.com/","")]]
+ [Para [Str "Blockquoted:",Space,Link [Code ("",[],[]) "http://example.com/"] ("http://example.com/","")]]
,Para [Str "Auto-links",Space,Str "should",Space,Str "not",Space,Str "occur",Space,Str "here:",Space,Code ("",[],[]) "<http://example.com/>"]
,CodeBlock ("",[],[]) "or here: <http://example.com/>"
,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","Voyage dans la Lune")]
+,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."]
,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]."]]]