Added readerTrace to ReaderOptions, --trace command line opt.

This is to debug backtracking-related parsing bugs.
So far it is only implemented for markdown, but it would
be good to extend it to latex and html readers.

3 files changed, 22 insertions, 1 deletions

diff --git a/pandoc.hs b/pandoc.hs
index e49b3b9cf..677101746 100644
--- a/pandoc.hs
+++ b/pandoc.hs
@@ -162,6 +162,7 @@ data Opt = Opt
     , optAscii             :: Bool       -- ^ Use ascii characters only in html
     , optTeXLigatures      :: Bool       -- ^ Use TeX ligatures for quotes/dashes
     , optDefaultImageExtension :: String -- ^ Default image extension
+    , optTrace             :: Bool       -- ^ Print debug information
     }
 
 -- | Defaults for command-line options.
@@ -217,6 +218,7 @@ defaultOpts = Opt
     , optAscii                 = False
     , optTeXLigatures          = True
     , optDefaultImageExtension = ""
+    , optTrace                 = False
     }
 
 -- | A list of functions, each transforming the options data structure
@@ -758,6 +760,11 @@ options =
                   (\opt -> return opt { optHTMLMathMethod = GladTeX }))
                  "" -- "Use gladtex for HTML math"
 
+    , Option "" ["trace"]
+                 (NoArg
+                  (\opt -> return opt { optTrace = True }))
+                 "" -- "Turn on diagnostic tracing in readers."
+
     , Option "" ["dump-args"]
                  (NoArg
                   (\opt -> return opt { optDumpArgs = True }))
@@ -952,6 +959,7 @@ main = do
               , optAscii                 = ascii
               , optTeXLigatures          = texLigatures
               , optDefaultImageExtension = defaultImageExtension
+              , optTrace                 = trace
              } = opts
 
   when dumpArgs $
@@ -1074,6 +1082,7 @@ main = do
                       , readerIndentedCodeClasses = codeBlockClasses
                       , readerApplyMacros = not laTeXOutput
                       , readerDefaultImageExtension = defaultImageExtension
+                      , readerTrace = trace
                       }
 
   let writerOptions = def { writerStandalone       = standalone',
diff --git a/src/Text/Pandoc/Options.hs b/src/Text/Pandoc/Options.hs
index 5f65abdde..38220f542 100644
--- a/src/Text/Pandoc/Options.hs
+++ b/src/Text/Pandoc/Options.hs
@@ -210,6 +210,7 @@ data ReaderOptions = ReaderOptions{
        , readerIndentedCodeClasses :: [String] -- ^ Default classes for
                                        -- indented code blocks
        , readerDefaultImageExtension :: String -- ^ Default extension for images
+       , readerTrace           :: Bool -- ^ Print debugging info
 } deriving (Show, Read)
 
 instance Default ReaderOptions
@@ -225,6 +226,7 @@ instance Default ReaderOptions
                , readerApplyMacros           = True
                , readerIndentedCodeClasses   = []
                , readerDefaultImageExtension = ""
+               , readerTrace                 = False
                }
 
 --
diff --git a/src/Text/Pandoc/Readers/Markdown.hs b/src/Text/Pandoc/Readers/Markdown.hs
index dd6fa733a..c73c8f610 100644
--- a/src/Text/Pandoc/Readers/Markdown.hs
+++ b/src/Text/Pandoc/Readers/Markdown.hs
@@ -60,6 +60,8 @@ import System.FilePath (takeExtension, addExtension)
 import Text.HTML.TagSoup
 import Text.HTML.TagSoup.Match (tagOpen)
 import qualified Data.Set as Set
+import Text.Printf (printf)
+import Debug.Trace (trace)
 
 type MarkdownParser = Parser [Char] ParserState
 
@@ -440,7 +442,10 @@ parseBlocks :: MarkdownParser (F Blocks)
 parseBlocks = mconcat <$> manyTill block eof
 
 block :: MarkdownParser (F Blocks)
-block = choice [ mempty <$ blanklines
+block = do
+  tr <- getOption readerTrace
+  pos <- getPosition
+  res <- choice [ mempty <$ blanklines
                , codeBlockFenced
                , yamlMetaBlock
                , guardEnabled Ext_latex_macros *> (macro >>= return . return)
@@ -465,6 +470,11 @@ block = choice [ mempty <$ blanklines
                , para
                , plain
                ] <?> "block"
+  when tr $ do
+    st <- getState
+    trace (printf "line %d: %s" (sourceLine pos)
+           (take 60 $ show $ B.toList $ runF res st)) (return ())
+  return res
 
 --
 -- header blocks