Add --rebase-relative-paths option.

- Add manual entry for `--rebase-relative-paths`.
- Add option `--rebase-relative-paths`, which rewrites
  relative image and link paths by prepending the (relative)
  directory of the containing file.
- Enable `rebase-relative-paths` in defaults files.
- Add `readerRebaseRelativePaths` to ReaderOptions record
  [API change].
- Make Markdown reader sensitive to `readerRebaseRelativePaths`.
- Add tests for #3752.

Closes #3752.

MANUAL.txt

@@ -526,6 +526,31 @@ header when requesting a document from a URL:
     where X = NUMBER - 1.* Specify the base level for headings
     (defaults to 1).
 
+`--rebase-relative-paths`
+
+:   Rewrite relative paths for Link and Image elements, depending
+    on the path of the file containing the link or image link.
+    For each link or image, pandoc will compute the directory of
+    the containing file, relative to the working directory, and
+    prepend the resulting path to the link or image path.
+
+    The use of this option is best understood by example.
+    Suppose you have a a subdirectory for each chapter of a
+    book, `chap1`, `chap2`, `chap3`. Each contains a file
+    `text.md` and a number of images used in the chapter.  You
+    would like to have `![image](spider.jpg)` in `chap1/text.md`
+    refer to `chap1/spider.jpg` and `![image](spider.jpg)` in
+    `chap2/text.md` refer to `chap2/spider.jpg`.  To do this,
+    use
+
+        pandoc chap*/*.md --rebase-relative-paths
+
+    Without this option, you would have to use
+    `![image](chap1/spider.jpg)` in `chap1/text.md` and
+    `![image](chap2/spider.jpg)` in `chap2/text.md`.  Links with
+    relative paths will be rewritten in the same way as images.
+    *This option currently only affects Markdown input.*
+
 `--strip-empty-paragraphs`
 
 :   *Deprecated.  Use the `+empty_paragraphs` extension instead.*

pandoc.cabal

@@ -214,6 +214,10 @@ extra-source-files:
                  test/command/C.txt
                  test/command/D.txt
                  test/command/01.csv
+                 test/command/chap1/spider.png
+                 test/command/chap2/spider.png
+                 test/command/chap1/text.md
+                 test/command/chap2/text.md
                  test/command/defaults1.yaml
                  test/command/defaults2.yaml
                  test/command/defaults3.yaml

src/Text/Pandoc/App.hs

@@ -214,6 +214,7 @@ convertWithOpts opts = do
           , readerIndentedCodeClasses = optIndentedCodeClasses opts
           , readerDefaultImageExtension =
              optDefaultImageExtension opts
+          , readerRebaseRelativePaths = optRebaseRelativePaths opts
           , readerTrackChanges = optTrackChanges opts
           , readerAbbreviations = abbrevs
           , readerExtensions = readerExts

src/Text/Pandoc/App/CommandLineOptions.hs

@@ -277,6 +277,11 @@ options =
                    "section|chapter|part")
                  "" -- "Use top-level division type in LaTeX, ConTeXt, DocBook"
 
+    , Option "" ["rebase-relative-paths"]
+                 (NoArg
+                  (\opt -> return opt { optRebaseRelativePaths = True }))
+                  "" -- "Rebase relative paths to directory of containing file"
+
     , Option "" ["extract-media"]
                  (ReqArg
                   (\arg opt ->

src/Text/Pandoc/App/Opt.hs

@@ -144,6 +144,8 @@ data Opt = Opt
     , optAscii                 :: Bool       -- ^ Prefer ascii output
     , optDefaultImageExtension :: Text       -- ^ Default image extension
     , optExtractMedia          :: Maybe FilePath -- ^ Path to extract embedded media
+    , optRebaseRelativePaths   :: Bool -- ^ Rebase relative link/image paths
+                                       -- to directory of containing file
     , optTrackChanges          :: TrackChanges -- ^ Accept or reject MS Word track-changes.
     , optFileScope             :: Bool         -- ^ Parse input files before combining
     , optTitlePrefix           :: Maybe Text     -- ^ Prefix for title
@@ -529,6 +531,8 @@ doOpt (k',v) = do
     "extract-media" ->
       parseYAML v >>= \x ->
              return (\o -> o{ optExtractMedia = unpack <$> x })
+    "rebase-relative-paths" ->
+      parseYAML v >>= \x -> return (\o -> o{ optRebaseRelativePaths = x })
     "track-changes" ->
       parseYAML v >>= \x -> return (\o -> o{ optTrackChanges = x })
     "file-scope" ->
@@ -657,6 +661,7 @@ defaultOpts = Opt
     , optAscii                 = False
     , optDefaultImageExtension = ""
     , optExtractMedia          = Nothing
+    , optRebaseRelativePaths   = False
     , optTrackChanges          = AcceptChanges
     , optFileScope             = False
     , optTitlePrefix           = Nothing

src/Text/Pandoc/Lua/Marshaling/ReaderOptions.hs

@@ -46,6 +46,7 @@ instance Pushable ReaderOptions where
           (indentedCodeClasses   :: [Text.Text])
           (abbreviations         :: Set.Set Text.Text)
           (defaultImageExtension :: Text.Text)
+          (rebaseRelativePaths   :: Bool)
           (trackChanges          :: TrackChanges)
           (stripComments         :: Bool)
           = ro
@@ -57,6 +58,7 @@ instance Pushable ReaderOptions where
     LuaUtil.addField "indented_code_classes" indentedCodeClasses
     LuaUtil.addField "abbreviations" abbreviations
     LuaUtil.addField "default_image_extension" defaultImageExtension
+    LuaUtil.addField "rebase_relative_paths" rebaseRelativePaths
     LuaUtil.addField "track_changes" trackChanges
     LuaUtil.addField "strip_comments" stripComments
 

src/Text/Pandoc/Options.hs

@@ -63,6 +63,8 @@ data ReaderOptions = ReaderOptions{
                                        -- indented code blocks
        , readerAbbreviations         :: Set.Set Text -- ^ Strings to treat as abbreviations
        , readerDefaultImageExtension :: Text -- ^ Default extension for images
+       , readerRebaseRelativePaths   :: Bool -- ^ If True, prepend relative
+            -- directory of containing file to paths in links and images
        , readerTrackChanges          :: TrackChanges -- ^ Track changes setting for docx
        , readerStripComments         :: Bool -- ^ Strip HTML comments instead of parsing as raw HTML
                                              -- (only implemented in commonmark)
@@ -80,6 +82,7 @@ instance Default ReaderOptions
                , readerIndentedCodeClasses   = []
                , readerAbbreviations         = defaultAbbrevs
                , readerDefaultImageExtension = ""
+               , readerRebaseRelativePaths   = False
                , readerTrackChanges          = AcceptChanges
                , readerStripComments         = False
                }

src/Text/Pandoc/Readers/Markdown.hs

@@ -29,7 +29,8 @@ import qualified Data.Set as Set
 import Data.Text (Text)
 import qualified Data.Text as T
 import qualified Data.ByteString.Lazy as BL
-import System.FilePath (addExtension, takeExtension)
+import System.FilePath (addExtension, takeExtension, isAbsolute, takeDirectory,
+                        (</>))
 import Text.HTML.TagSoup hiding (Row)
 import Text.Pandoc.Builder (Blocks, Inlines)
 import qualified Text.Pandoc.Builder as B
@@ -1836,9 +1837,12 @@ regLink :: PandocMonad m
         -> MarkdownParser m (F Inlines)
 regLink constructor lab = try $ do
   (src, tit) <- source
+  rebase <- getOption readerRebaseRelativePaths
+  pos <- getPosition
+  let src' = if rebase then rebasePath pos src else src
   attr <- option nullAttr $
           guardEnabled Ext_link_attributes >> attributes
-  return $ constructor attr src tit <$> lab
+  return $ constructor attr src' tit <$> lab
 
 -- a link like [this][ref] or [this][] or [this]
 referenceLink :: PandocMonad m
@@ -1854,6 +1858,8 @@ referenceLink constructor (lab, raw) = do
                          return (mempty, "")))
       <|>
       try ((guardDisabled Ext_spaced_reference_links <|> spnl) >> reference)
+  rebase <- getOption readerRebaseRelativePaths
+  pos <- getPosition
   when (raw' == "") $ guardEnabled Ext_shortcut_reference_links
   let labIsRef = raw' == "" || raw' == "[]"
   let key = toKey $ if labIsRef then raw else raw'
@@ -1878,7 +1884,9 @@ referenceLink constructor (lab, raw) = do
                    Just ((src, tit), _) -> constructor nullAttr src tit <$> lab
                    Nothing              -> makeFallback
             else makeFallback
-       Just ((src,tit), attr) -> constructor attr src tit <$> lab
+       Just ((src,tit), attr) ->
+          let src' = if rebase then rebasePath pos src else src
+           in constructor attr src' tit <$> lab
 
 dropBrackets :: Text -> Text
 dropBrackets = dropRB . dropLB
@@ -1911,15 +1919,30 @@ autoLink = try $ do
   return $ return $ B.linkWith attr (src <> escapeURI extra) ""
                      (B.str $ orig <> extra)
 
+-- | Rebase a relative path, by adding the (relative) directory
+-- of the containing source position.  Absolute links and URLs
+-- are untouched.
+rebasePath :: SourcePos -> Text -> Text
+rebasePath pos path = do
+  let fp = sourceName pos
+   in if isAbsolute (T.unpack path) || isURI path
+         then path
+         else
+           case takeDirectory fp of
+             ""  -> path
+             "." -> path
+             d   -> T.pack (d </> T.unpack path)
+
 image :: PandocMonad m => MarkdownParser m (F Inlines)
 image = try $ do
   char '!'
   (lab,raw) <- reference
   defaultExt <- getOption readerDefaultImageExtension
-  let constructor attr' src = case takeExtension (T.unpack src) of
-                          "" -> B.imageWith attr' (T.pack $ addExtension (T.unpack src)
-                                                          $ T.unpack defaultExt)
-                          _  -> B.imageWith attr' src
+  let constructor attr' src =
+         case takeExtension (T.unpack src) of
+            "" -> B.imageWith attr' (T.pack $ addExtension (T.unpack src)
+                                            $ T.unpack defaultExt)
+            _  -> B.imageWith attr' src
   regLink constructor lab <|> referenceLink constructor (lab,raw)
 
 note :: PandocMonad m => MarkdownParser m (F Inlines)

test/command/3752.md

@@ -0,0 +1,20 @@
+```
+% pandoc command/chap1/text.md command/chap2/text.md --rebase-relative-paths --verbose -t docx > /dev/null
+^D
+[INFO] Loaded command/chap1/spider.png from ./command/chap1/spider.png
+[INFO] Loaded command/chap1/../../lalune.jpg from ./command/chap1/../../lalune.jpg
+[INFO] Loaded command/chap2/spider.png from ./command/chap2/spider.png
+```
+
+```
+% pandoc command/chap1/text.md command/chap2/text.md --rebase-relative-paths -t html
+^D
+<h1 id="chapter-one">Chapter one</h1>
+<p>A spider: <img src="command/chap1/spider.png" alt="spider" /></p>
+<p>The moon: <img src="command/chap1/../../lalune.jpg" alt="moon" /></p>
+<p>Link to <a href="command/chap1/spider.png">spider picture</a>.</p>
+<p>URL left alone: <a href="https://pandoc.org/MANUAL.html">manual</a>.</p>
+<p>Absolute path left alone: <a href="/foo/bar/baz.png">absolute</a>.</p>
+<h1 id="chapter-two">Chapter two</h1>
+<p>A spider: <img src="command/chap2/spider.png" alt="spider" /></p>
+```

test/command/chap1/text.md

@@ -0,0 +1,11 @@
+# Chapter one
+
+A spider: ![spider](spider.png)
+
+The moon: ![moon](../../lalune.jpg)
+
+Link to [spider picture](spider.png).
+
+URL left alone: [manual](https://pandoc.org/MANUAL.html).
+
+Absolute path left alone: [absolute](/foo/bar/baz.png).

test/command/chap2/text.md

@@ -0,0 +1,3 @@
+# Chapter two
+
+A spider: ![spider](spider.png)