Add section identifiers support for FB2 writer (#5315)

Closes #5229.
2026-08-28 17:20:47 +08:00 · 2019-02-22 10:43:43 +03:00
parent 69d433d37a
commit c75b558cbc
4 changed files with 62 additions and 69 deletions
@@ -52,13 +52,15 @@ import Network.HTTP (urlEncode)
 import Text.XML.Light
 import qualified Text.XML.Light as X
 import qualified Text.XML.Light.Cursor as XC
+import qualified Text.XML.Light.Input as XI

 import Text.Pandoc.Class (PandocMonad, report)
 import qualified Text.Pandoc.Class as P
 import Text.Pandoc.Definition
 import Text.Pandoc.Logging
 import Text.Pandoc.Options (HTMLMathMethod (..), WriterOptions (..), def)
-import Text.Pandoc.Shared (capitalize, isHeaderBlock, isURI, orderedListMarkers)
+import Text.Pandoc.Shared (capitalize, isURI, orderedListMarkers, hierarchicalize)
+import qualified Text.Pandoc.Shared as Shared (Element(Blk, Sec))

 -- | Data to be written at the end of the document:
 -- (foot)notes, URLs, references, images.
@@ -177,20 +179,28 @@ docdate meta' = do
 -- representation.
 renderSections :: PandocMonad m => Int -> [Block] -> FBM m [Content]
 renderSections level blocks = do
-    let secs = splitSections level blocks
-    mapM (renderSection level) secs
+    let elements = hierarchicalize blocks
+    let isSection Shared.Sec{} = True
+        isSection _ = False
+    let (initialBlocks, secs) = break isSection elements
+    let elements' = if null initialBlocks
+        then secs
+        else Shared.Sec 1 [] nullAttr mempty initialBlocks : secs
+    cMapM (renderSection level) elements'

-renderSection :: PandocMonad m => Int -> ([Inline], [Block]) -> FBM m Content
-renderSection level (ttl, body) = do
-    title <- if null ttl
+
+
+renderSection :: PandocMonad m =>  Int -> Shared.Element -> FBM m [Content]
+renderSection _   (Shared.Blk block) = blockToXml block
+renderSection lvl (Shared.Sec _ _num (id',_,_) title elements) = do
+  content <- cMapM (renderSection (lvl + 1)) elements
+  title' <- if null title
            then return []
-            else list . el "title" <$> formatTitle ttl
-    content <- if hasSubsections body
-               then renderSections (level + 1) body
-               else cMapM blockToXml body
-    return $ el "section" (title ++ content)
-  where
-    hasSubsections = any isHeaderBlock
+            else list . el "title" <$> formatTitle title
+  let sectionContent = if null id'
+      then el "section" (title' ++ content)
+      else el "section" ([uattr "id" id'], (title' ++ content))
+  return $ [sectionContent]

 -- | Only <p> and <empty-line> are allowed within <title> in FB2.
 formatTitle :: PandocMonad m => [Inline] -> FBM m [Content]
@@ -206,24 +216,6 @@ isLineBreak :: Inline -> Bool
 isLineBreak LineBreak = True
 isLineBreak _         = False

-- | Divide the stream of block elements into sections: [(title, blocks)].
-splitSections :: Int -> [Block] -> [([Inline], [Block])]
-splitSections level blocks = reverse $ revSplit (reverse blocks)
-  where
-  revSplit [] = []
-  revSplit rblocks =
-    let (lastsec, before) = break sameLevel rblocks
-        (header, prevblocks) =
-            case before of
-              (Header n _ title:prevblocks') ->
-                  if n == level
-                     then (title, prevblocks')
-                     else ([], before)
-              _ -> ([], before)
-    in (header, reverse lastsec) : revSplit prevblocks
-  sameLevel (Header n _ _) = n == level
-  sameLevel _              = False
-
 -- | Make another FictionBook body with footnotes.
 renderFootnotes :: PandocMonad m => FBM m [Content]
 renderFootnotes = do
@@ -325,9 +317,10 @@ blockToXml (Para [Image atr alt (src,'f':'i':'g':':':tit)]) =
 blockToXml (Para ss) = (list . el "p") <$> cMapM toXml ss
 blockToXml (CodeBlock _ s) = return . spaceBeforeAfter .
                             map (el "p" . el "code") . lines $ s
-blockToXml b@(RawBlock _ _) = do
-  report $ BlockNotRendered b
-  return []
+blockToXml (RawBlock f str) = do
+  if f == Format "fb2"
+    then return $ XI.parseXML str
+    else return []
 blockToXml (Div _ bs) = cMapM blockToXml bs
 blockToXml (BlockQuote bs) = (list . el "cite") <$> cMapM blockToXml bs
 blockToXml (LineBlock lns) =
@@ -12,15 +12,15 @@
 <title>
 <p />
 </title>
-<section>
+<section id="top-level-title">
 <title>
 <p>Top-level title</p>
 </title>
-<section>
+<section id="section">
 <title>
 <p>Section</p>
 </title>
-<section>
+<section id="subsection">
 <title>
 <p>Subsection</p>
 </title>
@@ -1,3 +1,3 @@
 <?xml version="1.0" encoding="UTF-8"?>
-<FictionBook xmlns="http://www.gribuser.ru/xml/fictionbook/2.0" xmlns:l="http://www.w3.org/1999/xlink"><description><title-info><genre>unrecognised</genre></title-info><document-info><program-used>pandoc</program-used></document-info></description><body><title><p /></title><section><title><p>Simple title</p></title><p>This example tests FictionBook titles.</p></section><section><title><p><emphasis>Emphasized</emphasis> <strong>Strong</strong> Title</p></title></section></body></FictionBook>
+<FictionBook xmlns="http://www.gribuser.ru/xml/fictionbook/2.0" xmlns:l="http://www.w3.org/1999/xlink"><description><title-info><genre>unrecognised</genre></title-info><document-info><program-used>pandoc</program-used></document-info></description><body><title><p /></title><section id="simple-title"><title><p>Simple title</p></title><p>This example tests FictionBook titles.</p></section><section id="emphasized-strong-title"><title><p><emphasis>Emphasized</emphasis> <strong>Strong</strong> Title</p></title></section></body></FictionBook>

@@ -26,25 +26,25 @@
 John Gruber’s markdown test suite.</p>
 <empty-line />
 </section>
-<section>
+<section id="headers">
 <title>
 <p>Headers</p>
 </title>
-<section>
+<section id="level-2-with-an-embedded-link">
 <title>
 <p>Level 2 with an <a l:href="/url">embedded link</a>
 </p>
 </title>
-<section>
+<section id="level-3-with-emphasis">
 <title>
 <p>Level 3 with <emphasis>emphasis</emphasis>
 </p>
 </title>
-<section>
+<section id="level-4">
 <title>
 <p>Level 4</p>
 </title>
-<section>
+<section id="level-5">
 <title>
 <p>Level 5</p>
 </title>
@@ -53,23 +53,23 @@ John Gruber’s markdown test suite.</p>
 </section>
 </section>
 </section>
-<section>
+<section id="level-1">
 <title>
 <p>Level 1</p>
 </title>
-<section>
+<section id="level-2-with-emphasis">
 <title>
 <p>Level 2 with <emphasis>emphasis</emphasis>
 </p>
 </title>
-<section>
+<section id="level-3">
 <title>
 <p>Level 3</p>
 </title>
 <p>with no blank line</p>
 </section>
 </section>
-<section>
+<section id="level-2">
 <title>
 <p>Level 2</p>
 </title>
@@ -77,7 +77,7 @@ John Gruber’s markdown test suite.</p>
 <empty-line />
 </section>
 </section>
-<section>
+<section id="paragraphs">
 <title>
 <p>Paragraphs</p>
 </title>
@@ -93,7 +93,7 @@ list item.</p>
 here.</p>
 <empty-line />
 </section>
-<section>
+<section id="block-quotes">
 <title>
 <p>Block Quotes</p>
 </title>
@@ -131,7 +131,7 @@ It is pretty short.</p>
 <p>And a following paragraph.</p>
 <empty-line />
 </section>
-<section>
+<section id="code-blocks">
 <title>
 <p>Code Blocks</p>
 </title>
@@ -176,11 +176,11 @@ It is pretty short.</p>
 <empty-line />
 <empty-line />
 </section>
-<section>
+<section id="lists">
 <title>
 <p>Lists</p>
 </title>
-<section>
+<section id="unordered">
 <title>
 <p>Unordered</p>
 </title>
@@ -218,7 +218,7 @@ It is pretty short.</p>
 <p>• Minus 3</p>
 <empty-line />
 </section>
-<section>
+<section id="ordered">
 <title>
 <p>Ordered</p>
 </title>
@@ -255,7 +255,7 @@ back.</p>
 <p>3. Item 3.</p>
 <empty-line />
 </section>
-<section>
+<section id="nested">
 <title>
 <p>Nested</p>
 </title>
@@ -280,7 +280,7 @@ back.</p>
 <p>3. Third</p>
 <empty-line />
 </section>
-<section>
+<section id="tabs-and-spaces">
 <title>
 <p>Tabs and spaces</p>
 </title>
@@ -297,7 +297,7 @@ indented with tabs</p>
 indented with spaces</p>
 <empty-line />
 </section>
-<section>
+<section id="fancy-list-markers">
 <title>
 <p>Fancy list markers</p>
 </title>
@@ -326,7 +326,7 @@ starting with 4</p>
 <empty-line />
 </section>
 </section>
-<section>
+<section id="definition-lists">
 <title>
 <p>Definition Lists</p>
 </title>
@@ -440,7 +440,7 @@ crisp, pleasant to taste</p>
 <p>1. sublist</p>
 <p>2. sublist</p>
 </section>
-<section>
+<section id="html-blocks">
 <title>
 <p>HTML Blocks</p>
 </title>
@@ -484,7 +484,7 @@ crisp, pleasant to taste</p>
 <p>Hr’s:</p>
 <empty-line />
 </section>
-<section>
+<section id="inline-markup">
 <title>
 <p>Inline Markup</p>
 </title>
@@ -521,7 +521,7 @@ crisp, pleasant to taste</p>
 because of the unescaped spaces: a^b c^d, a~b c~d.</p>
 <empty-line />
 </section>
-<section>
+<section id="smart-quotes-ellipses-dashes">
 <title>
 <p>Smart quotes, ellipses, dashes</p>
 </title>
@@ -537,7 +537,7 @@ So is ‘pine.’</p>
 <p>Ellipses…and…and….</p>
 <empty-line />
 </section>
-<section>
+<section id="latex">
 <title>
 <p>LaTeX</p>
 </title>
@@ -564,7 +564,7 @@ So is ‘pine.’</p>
 <p>Here’s a LaTeX table:</p>
 <empty-line />
 </section>
-<section>
+<section id="special-characters">
 <title>
 <p>Special Characters</p>
 </title>
@@ -597,11 +597,11 @@ So is ‘pine.’</p>
 <p>Minus: -</p>
 <empty-line />
 </section>
-<section>
+<section id="links">
 <title>
 <p>Links</p>
 </title>
-<section>
+<section id="explicit">
 <title>
 <p>Explicit</p>
 </title>
@@ -627,7 +627,7 @@ So is ‘pine.’</p>
 <p>
 <a l:href="">Empty</a>.</p>
 </section>
-<section>
+<section id="reference">
 <title>
 <p>Reference</p>
 </title>
@@ -647,7 +647,7 @@ So is ‘pine.’</p>
 <p>Foo <a l:href="/url/">bar</a>.</p>
 <p>Foo <a l:href="/url/">biz</a>.</p>
 </section>
-<section>
+<section id="with-ampersands">
 <title>
 <p>With ampersands</p>
 </title>
@@ -656,7 +656,7 @@ So is ‘pine.’</p>
 <p>Here’s an <a l:href="/script?foo=1&amp;bar=2">inline link</a>.</p>
 <p>Here’s an <a l:href="/script?foo=1&amp;bar=2">inline link in pointy braces</a>.</p>
 </section>
-<section>
+<section id="autolinks">
 <title>
 <p>Autolinks</p>
 </title>
@@ -682,7 +682,7 @@ So is ‘pine.’</p>
 <empty-line />
 </section>
 </section>
-<section>
+<section id="images">
 <title>
 <p>Images</p>
 </title>
@@ -691,7 +691,7 @@ So is ‘pine.’</p>
 <p>Here is a movie <image l:href="#image2" l:type="inlineImageType" alt="movie" /> icon.</p>
 <empty-line />
 </section>
-<section>
+<section id="footnotes">
 <title>
 <p>Footnotes</p>
 </title>