diff --git a/CHANGELOG.md b/CHANGELOG.md --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -1,5 +1,15 @@ # Revision history for asciidoc-hs +## 0.1.0.2 -- 2026-03-17 + + * Allow fenced constructions to end with end-of-input (#9). + + * Fix issue in parsing bracketed arguments (#8). + + * Fix bug in table parsing with rowspans (#5). + + * Improved parsing of line and block comments (#4). + ## 0.1.0.1 -- 2026-02-01 * Fix character escaping issue (#3). Unconstrained forms of diff --git a/asciidoc.cabal b/asciidoc.cabal --- a/asciidoc.cabal +++ b/asciidoc.cabal @@ -1,6 +1,6 @@ cabal-version: 3.4 name: asciidoc -version: 0.1.0.1 +version: 0.1.0.2 synopsis: AsciiDoc parser. description: A parser for AsciiDoc syntax. license: BSD-3-Clause @@ -13,6 +13,7 @@ extra-doc-files: CHANGELOG.md extra-source-files: test/asciidoctor/**/*.test test/regression/*.test + test/regression/issue_9_include.adoc test/feature/**/*.test test/feature/include/**/*.adoc diff --git a/src/AsciiDoc/Parse.hs b/src/AsciiDoc/Parse.hs --- a/src/AsciiDoc/Parse.hs +++ b/src/AsciiDoc/Parse.hs @@ -12,7 +12,7 @@ import Prelude hiding (takeWhile) import Text.HTML.TagSoup.Entity (lookupNamedEntity) -import Data.Maybe (isNothing, listToMaybe, fromMaybe) +import Data.Maybe (isNothing, listToMaybe, fromMaybe, catMaybes) import Data.Bifunctor (first) import Data.Either (lefts, rights) import qualified Data.Map as M @@ -254,7 +254,7 @@ bs <- (case M.lookup "hardbreaks-option" attr' of Just "" -> withHardBreaks _ -> id) $ - withBlockContext (SectionContext (minSectionLevel - 1)) (many pBlock) + withBlockContext (SectionContext (minSectionLevel - 1)) pBlocks skipWhile isSpace endOfInput attr <- gets docAttrs @@ -371,8 +371,8 @@ skipBlankLines = do contexts <- asks blockContexts case contexts of - ListContext{} : _ -> void $ many $ vchar '+' *> pBlankLine - _ -> void $ many pBlankLine + ListContext{} : _ -> skipMany $ vchar '+' *> pBlankLine + _ -> skipMany pBlankLine pBlankLine :: P () pBlankLine = takeWhile (\c -> c == ' ' || c == '\t') *> (pLineComment <|> endOfLine) @@ -397,7 +397,7 @@ skipBlankLines skipMany pDocAttribute skipBlankLines - pure bs + pure $ catMaybes bs parseAsciidoc :: Text -> P Document parseAsciidoc = parseWith pDocument . (<> "\n") . T.strip @@ -415,14 +415,13 @@ parseInlines :: Text -> P [Inline] parseInlines = parseWith pInlines . T.strip -pBlock :: P Block +pBlock :: P (Maybe Block) pBlock = do contexts <- asks blockContexts skipBlankLines skipMany pDocAttribute skipBlankLines (mbtitle, attr) <- pTitlesAndAttributes - skipMany (pCommentBlock attr) case contexts of ListContext{} : _ -> skipWhile (== ' ') _ -> pure () @@ -432,7 +431,8 @@ | Just opts <- M.lookup "options" kvs -> "hardbreaks" `T.isInfixOf` opts _ -> False - (if hardbreaks then withHardBreaks else id) $ + (Nothing <$ pCommentBlock attr) <|> fmap Just + ((if hardbreaks then withHardBreaks else id) $ pBlockMacro mbtitle attr <|> pDiscreteHeading mbtitle attr <|> pExampleBlock mbtitle attr @@ -454,7 +454,7 @@ , pDefinitionList , pIndentedLiteral , pPara - ] + ]) pIndentedLiteral :: P BlockType @@ -556,7 +556,7 @@ guard (lev > sectionLevel && lev >= 0 && lev <= 5) vchar ' ' title <- pLine >>= parseInlines - contents <- withBlockContext (SectionContext lev) $ many pBlock + contents <- withBlockContext (SectionContext lev) pBlocks -- note: we use sectionLevel, not lev, so in improperly nested content, e.g., -- == foo -- ==== bar @@ -629,7 +629,7 @@ let newContext = case contexts of ListContext ':' n : _ -> ListContext ':' (n + 1) _ -> ListContext ':' 1 - defn <- withBlockContext newContext (many pBlock) + defn <- withBlockContext newContext pBlocks void $ many pBlankLine pure (term, defn) @@ -700,14 +700,14 @@ pListItem = do mbCheckboxState <- optional pCheckbox skipWhile (==' ') - bs <- many pBlock + bs <- pBlocks pure $ ListItem mbCheckboxState bs pDelimitedLiteralBlock :: Char -> Int -> P [T.Text] pDelimitedLiteralBlock c minimumNumber = do len <- length <$> some (vchar c) <* pBlankLine guard $ len >= minimumNumber - let endFence = count len (vchar c) *> pBlankLine + let endFence = count len (vchar c) *> (pBlankLine <|> endOfInput) manyTill pLine endFence pDelimitedBlock :: Char -> Int -> P [Block] @@ -716,7 +716,7 @@ guard $ len >= minimumNumber let endFence = count len (vchar c) *> pBlankLine withBlockContext (DelimitedContext c len) $ - manyTill pBlock endFence + catMaybes <$> manyTill pBlock endFence pPassBlock :: Maybe BlockTitle -> Attr -> P Block pPassBlock mbtitle attr = do @@ -905,6 +905,7 @@ pNormalLine :: P Text pNormalLine = do + notFollowedBy (string "////" *> pBlankLine) -- block comment t <- pLine fp <- asks filePath guard $ not $ T.all (\c -> c == ' ' || c == '\t') t @@ -970,16 +971,16 @@ -- if there are rowspans from rows above, we need to skip some: let mbspecs' = case mbspecs of Nothing -> Nothing - Just specs' -> Just [s | (s,0) <- zip specs' rowspans] + Just specs' -> Just [s | (s,n) <- zip specs' rowspans, n <= 0] row@(TableRow cells) <- pTableRow tableOpts mbspecs' let numcols = sum (map cellColspan cells) let specs = fromMaybe (replicate numcols defaultColumnSpec) mbspecs -- now, update rowspans in light of new row let updateRowspans [] rs = rs updateRowspans (c:cs) rs = - map (+ (cellRowspan c - 1)) (take (cellColspan c) rs) + map (+ (cellRowspan c)) (take (cellColspan c) rs) ++ updateRowspans cs (drop (cellColspan c) rs) - let rowspans' = updateRowspans cells rowspans + let rowspans' = updateRowspans cells (map (\x -> x - 1) rowspans) (\(rows, colspecs') -> (row:rows, case rows of [] -> specs _ -> colspecs')) @@ -1289,17 +1290,18 @@ pure $ T.pack result pInlines' :: [Char] -> P [Inline] -pInlines' cs = - (do il' <- pInline cs - let il = case il' of - Inline (Attr ps kvs) (Span ils) - | Nothing <- M.lookup "role" kvs - -> Inline (Attr ps kvs) (Highlight ils) - _ -> il' - addStr . (il:) <$> pInlines' []) - <|> (do c <- anyChar - pInlines' (c:cs)) - <|> (addStr [] <$ endOfInput) +pInlines' cs = do + (pLineComment *> pInlines' cs) + <|> (do il' <- pInline cs + let il = case il' of + Inline (Attr ps kvs) (Span ils) + | Nothing <- M.lookup "role" kvs + -> Inline (Attr ps kvs) (Highlight ils) + _ -> il' + addStr . (il:) <$> pInlines' []) + <|> (do c <- anyChar + pInlines' (c:cs)) + <|> (addStr [] <$ endOfInput) where addStr = case cs of [] -> id @@ -1348,7 +1350,6 @@ (d:_) -> isSpace d || isPunctuation d || d == '+' [] -> True let inMatched = pInMatched maybeUnconstrained - skipMany pLineComment (do attr <- pFormattedTextAttributes <|> pure mempty c <- peekChar' case c of @@ -1391,7 +1392,7 @@ <$> (string "+++" *> manyTill anyChar (string "+++")) pLineComment :: P () -pLineComment = string "//" *> skipWhile (== ' ') *> void pLine +pLineComment = string "//" *> satisfy (\c -> c == ' ' || c == '\t') *> void pLine pCrossReference :: P Inline pCrossReference = do @@ -1600,7 +1601,7 @@ (T.pack <$> some ((vchar '\\' *> char ']') <|> (vchar '+' *> vchar '+' *> char ']' <* vchar '+' <* vchar '+') <|> - satisfy (\c -> c /= ']' && not (isEndOfLine c)) <|> + satisfy (\c -> c /= ']' && c /= '[' && not (isEndOfLine c)) <|> (' ' <$ (vchar '\\' <* endOfLine))) <|> ((\x -> "[" <> x <> "]") <$> pBracketedText))) <* vchar ']' diff --git a/test/regression/issue_4a.test b/test/regression/issue_4a.test new file mode 100644 --- /dev/null +++ b/test/regression/issue_4a.test @@ -0,0 +1,23 @@ +<<< +Asciidoc text followed by a line comment. +// LINE COMMENT +>>> +Document + { docMeta = + Meta + { docTitle = [] + , docTitleAttributes = Nothing + , docAuthors = [] + , docRevision = Nothing + , docAttributes = fromList [ ( "sectids" , "" ) ] + } + , docBlocks = + [ Block mempty Nothing PageBreak + , Block + mempty + Nothing + (Paragraph + [ Inline mempty (Str "Asciidoc text followed by a line comment.\n") + ]) + ] + } diff --git a/test/regression/issue_4b.test b/test/regression/issue_4b.test new file mode 100644 --- /dev/null +++ b/test/regression/issue_4b.test @@ -0,0 +1,26 @@ +<<< +Asciidoc text followed by a blank line and a line comment. + +// LINE COMMENT +>>> +Document + { docMeta = + Meta + { docTitle = [] + , docTitleAttributes = Nothing + , docAuthors = [] + , docRevision = Nothing + , docAttributes = fromList [ ( "sectids" , "" ) ] + } + , docBlocks = + [ Block mempty Nothing PageBreak + , Block + mempty + Nothing + (Paragraph + [ Inline + mempty + (Str "Asciidoc text followed by a blank line and a line comment.") + ]) + ] + } diff --git a/test/regression/issue_4c.test b/test/regression/issue_4c.test new file mode 100644 --- /dev/null +++ b/test/regression/issue_4c.test @@ -0,0 +1,26 @@ +<<< +Asciidoc text followed by a line comment +// LINE COMMENT +followed by text. +>>> +Document + { docMeta = + Meta + { docTitle = [] + , docTitleAttributes = Nothing + , docAuthors = [] + , docRevision = Nothing + , docAttributes = fromList [ ( "sectids" , "" ) ] + } + , docBlocks = + [ Block mempty Nothing PageBreak + , Block + mempty + Nothing + (Paragraph + [ Inline + mempty + (Str "Asciidoc text followed by a line comment\nfollowed by text.") + ]) + ] + } diff --git a/test/regression/issue_4d.test b/test/regression/issue_4d.test new file mode 100644 --- /dev/null +++ b/test/regression/issue_4d.test @@ -0,0 +1,29 @@ +<<< +Asciidoc text followed by a line comment. +// LINE COMMENT + +Followed by a new paragraph. +>>> +Document + { docMeta = + Meta + { docTitle = [] + , docTitleAttributes = Nothing + , docAuthors = [] + , docRevision = Nothing + , docAttributes = fromList [ ( "sectids" , "" ) ] + } + , docBlocks = + [ Block mempty Nothing PageBreak + , Block + mempty + Nothing + (Paragraph + [ Inline mempty (Str "Asciidoc text followed by a line comment.\n") + ]) + , Block + mempty + Nothing + (Paragraph [ Inline mempty (Str "Followed by a new paragraph.") ]) + ] + } diff --git a/test/regression/issue_4e.test b/test/regression/issue_4e.test new file mode 100644 --- /dev/null +++ b/test/regression/issue_4e.test @@ -0,0 +1,30 @@ +<<< +Asciidoc text followed by a line comment. + +// LINE COMMENT + +Followed by a new paragraph. +>>> +Document + { docMeta = + Meta + { docTitle = [] + , docTitleAttributes = Nothing + , docAuthors = [] + , docRevision = Nothing + , docAttributes = fromList [ ( "sectids" , "" ) ] + } + , docBlocks = + [ Block mempty Nothing PageBreak + , Block + mempty + Nothing + (Paragraph + [ Inline mempty (Str "Asciidoc text followed by a line comment.") + ]) + , Block + mempty + Nothing + (Paragraph [ Inline mempty (Str "Followed by a new paragraph.") ]) + ] + } diff --git a/test/regression/issue_4f.test b/test/regression/issue_4f.test new file mode 100644 --- /dev/null +++ b/test/regression/issue_4f.test @@ -0,0 +1,30 @@ +<<< +Asciidoc text followed by a block comment +//// +BLOCK COMMENT +//// +followed by text. +>>> +Document + { docMeta = + Meta + { docTitle = [] + , docTitleAttributes = Nothing + , docAuthors = [] + , docRevision = Nothing + , docAttributes = fromList [ ( "sectids" , "" ) ] + } + , docBlocks = + [ Block mempty Nothing PageBreak + , Block + mempty + Nothing + (Paragraph + [ Inline mempty (Str "Asciidoc text followed by a block comment") + ]) + , Block + mempty + Nothing + (Paragraph [ Inline mempty (Str "followed by text.") ]) + ] + } diff --git a/test/regression/issue_4g.test b/test/regression/issue_4g.test new file mode 100644 --- /dev/null +++ b/test/regression/issue_4g.test @@ -0,0 +1,32 @@ +<<< +Asciidoc text followed by a block comment. + +//// +BLOCK COMMENT +//// + +Followed by a new paragraph. +>>> +Document + { docMeta = + Meta + { docTitle = [] + , docTitleAttributes = Nothing + , docAuthors = [] + , docRevision = Nothing + , docAttributes = fromList [ ( "sectids" , "" ) ] + } + , docBlocks = + [ Block mempty Nothing PageBreak + , Block + mempty + Nothing + (Paragraph + [ Inline mempty (Str "Asciidoc text followed by a block comment.") + ]) + , Block + mempty + Nothing + (Paragraph [ Inline mempty (Str "Followed by a new paragraph.") ]) + ] + } diff --git a/test/regression/issue_4h.test b/test/regression/issue_4h.test new file mode 100644 --- /dev/null +++ b/test/regression/issue_4h.test @@ -0,0 +1,36 @@ +<<< +Asciidoc text followed by a multi-line block comment. + +//// +BLOCK COMMENT LINE 1 +LINE 2 +LINE 3 +//// + +Followed by a new paragraph. +>>> +Document + { docMeta = + Meta + { docTitle = [] + , docTitleAttributes = Nothing + , docAuthors = [] + , docRevision = Nothing + , docAttributes = fromList [ ( "sectids" , "" ) ] + } + , docBlocks = + [ Block mempty Nothing PageBreak + , Block + mempty + Nothing + (Paragraph + [ Inline + mempty + (Str "Asciidoc text followed by a multi-line block comment.") + ]) + , Block + mempty + Nothing + (Paragraph [ Inline mempty (Str "Followed by a new paragraph.") ]) + ] + } diff --git a/test/regression/issue_5.test b/test/regression/issue_5.test new file mode 100644 --- /dev/null +++ b/test/regression/issue_5.test @@ -0,0 +1,116 @@ +<<< +|=== +|col1|col2 +|11 .2+|12|13 +|21 .2+|22|23 +|=== +>>> +Document + { docMeta = + Meta + { docTitle = [] + , docTitleAttributes = Nothing + , docAuthors = [] + , docRevision = Nothing + , docAttributes = fromList [ ( "sectids" , "" ) ] + } + , docBlocks = + [ Block mempty Nothing PageBreak + , Block + mempty + Nothing + (Table + [ ColumnSpec + { colHorizAlign = Nothing + , colVertAlign = Nothing + , colWidth = Nothing + , colStyle = Nothing + } + , ColumnSpec + { colHorizAlign = Nothing + , colVertAlign = Nothing + , colWidth = Nothing + , colStyle = Nothing + } + ] + (Just + [ TableRow + [ TableCell + { cellContent = + [ Block mempty Nothing (Paragraph [ Inline mempty (Str "col1") ]) ] + , cellHorizAlign = Nothing + , cellVertAlign = Nothing + , cellColspan = 1 + , cellRowspan = 1 + } + , TableCell + { cellContent = + [ Block mempty Nothing (Paragraph [ Inline mempty (Str "col2") ]) ] + , cellHorizAlign = Nothing + , cellVertAlign = Nothing + , cellColspan = 1 + , cellRowspan = 1 + } + ] + ]) + [ TableRow + [ TableCell + { cellContent = + [ Block mempty Nothing (Paragraph [ Inline mempty (Str "11") ]) ] + , cellHorizAlign = Nothing + , cellVertAlign = Nothing + , cellColspan = 1 + , cellRowspan = 1 + } + , TableCell + { cellContent = + [ Block mempty Nothing (Paragraph [ Inline mempty (Str "12") ]) ] + , cellHorizAlign = Nothing + , cellVertAlign = Nothing + , cellColspan = 1 + , cellRowspan = 2 + } + ] + , TableRow + [ TableCell + { cellContent = + [ Block mempty Nothing (Paragraph [ Inline mempty (Str "13") ]) ] + , cellHorizAlign = Nothing + , cellVertAlign = Nothing + , cellColspan = 1 + , cellRowspan = 1 + } + ] + , TableRow + [ TableCell + { cellContent = + [ Block mempty Nothing (Paragraph [ Inline mempty (Str "21") ]) ] + , cellHorizAlign = Nothing + , cellVertAlign = Nothing + , cellColspan = 1 + , cellRowspan = 1 + } + , TableCell + { cellContent = + [ Block mempty Nothing (Paragraph [ Inline mempty (Str "22") ]) ] + , cellHorizAlign = Nothing + , cellVertAlign = Nothing + , cellColspan = 1 + , cellRowspan = 2 + } + ] + ] + (Just + [ TableRow + [ TableCell + { cellContent = + [ Block mempty Nothing (Paragraph [ Inline mempty (Str "23") ]) ] + , cellHorizAlign = Nothing + , cellVertAlign = Nothing + , cellColspan = 1 + , cellRowspan = 1 + } + ] + ])) + ] + } diff --git a/test/regression/issue_8.test b/test/regression/issue_8.test new file mode 100644 --- /dev/null +++ b/test/regression/issue_8.test @@ -0,0 +1,40 @@ +A sentence.footnote:[See https://pandoc.org/MANUAL.html[manual] and other https://pandoc.org/demos.html[demos].] +>>> +Document + { docMeta = + Meta + { docTitle = [] + , docTitleAttributes = Nothing + , docAuthors = [] + , docRevision = Nothing + , docAttributes = fromList [ ( "sectids" , "" ) ] + } + , docBlocks = + [ Block + mempty + Nothing + (Paragraph + [ Inline mempty (Str "A sentence.") + , Inline + mempty + (Footnote + Nothing + [ Inline mempty (Str "See ") + , Inline + mempty + (Link + URLLink + (Target "https://pandoc.org/MANUAL.html") + [ Inline mempty (Str "manual") ]) + , Inline mempty (Str " and other ") + , Inline + mempty + (Link + URLLink + (Target "https://pandoc.org/demos.html") + [ Inline mempty (Str "demos") ]) + , Inline mempty (Str ".") + ]) + ]) + ] + } diff --git a/test/regression/issue_9.test b/test/regression/issue_9.test new file mode 100644 --- /dev/null +++ b/test/regression/issue_9.test @@ -0,0 +1,27 @@ +include::issue_9_include.adoc[] +>>> +Document + { docMeta = + Meta + { docTitle = [] + , docTitleAttributes = Nothing + , docAuthors = [] + , docRevision = Nothing + , docAttributes = fromList [ ( "sectids" , "" ) ] + } + , docBlocks = + [ Block + mempty + Nothing + (Include + "test/regression/issue_9_include.adoc" + (Just + [ Block + Attr + ( [ "source" , "cpp" ] , fromList [] ) + Nothing + (LiteralBlock + "#include \"MyFile.h\"\n\nMyClass::MyClass(LibParent *parent,\n LibCollectionPtr inputCollection)\n : LibClass(parent) {\n \n myMemberFn(inputCollection, this->getLibItem());\n \n}\n") + ])) + ] + } diff --git a/test/regression/issue_9_include.adoc b/test/regression/issue_9_include.adoc new file mode 100644 --- /dev/null +++ b/test/regression/issue_9_include.adoc @@ -0,0 +1,12 @@ +[source, cpp] +.... +#include "MyFile.h" + +MyClass::MyClass(LibParent *parent, + LibCollectionPtr inputCollection) + : LibClass(parent) { + + myMemberFn(inputCollection, this->getLibItem()); + +} +....