Parse Markdown links at beginning of line within a paragraph (#1470)

* Catch Markdown links at beginning of line within paragraph

Per Issue #774, Markdown links were being parsed as ordinary text when
they occurred at the beginning of a line other than the first line of
the paragraph. This occurred because the parser was not interpreting a
left square bracket as a special character that could delimit special
markup. A space character was considered a special character, so, if a
space occurred at the beginning of the new line, then the parser would
interpret the space by itself and then continue parsing, thereby
catching the Markdown link. '\n' was not treated as a special character,
so the parser did not catch a Markdown link that may have followed.

Note that this will allow for Markdown links that are not surrounded by
spaces. For example, the following text includes a Markdown link that
will be parsed:

  Hello, world[label](url)

This is consistent with how the parser handles other types of markup.

* Remove obsolete documentation hint

Commit 6b9aeafddf20efc65d3725c16e3fc43a20aac343 should eliminate the
need for the workaround suggested in the documentation.

2 files changed, 29 insertions, 19 deletions

diff --git a/haddock-library/src/Documentation/Haddock/Parser.hs b/haddock-library/src/Documentation/Haddock/Parser.hs
index fa15c275..c6d7e59b 100644
--- a/haddock-library/src/Documentation/Haddock/Parser.hs
+++ b/haddock-library/src/Documentation/Haddock/Parser.hs
@@ -173,7 +173,7 @@ encodedChar = "&#" *> c <* ";"
 -- Once we have checked for any of these and tried to parse the
 -- relevant markup, we can assume they are used as regular text.
 specialChar :: [Char]
-specialChar = "_/<@\"&'`# "
+specialChar = "_/<@\"&'`#[ "
 
 -- | Plain, regular parser for text. Called as one of the last parsers
 -- to ensure that we have already given a chance to more meaningful parsers
diff --git a/haddock-library/test/Documentation/Haddock/ParserSpec.hs b/haddock-library/test/Documentation/Haddock/ParserSpec.hs
index 5fa73ecd..9cf7c537 100644
--- a/haddock-library/test/Documentation/Haddock/ParserSpec.hs
+++ b/haddock-library/test/Documentation/Haddock/ParserSpec.hs
@@ -576,27 +576,37 @@ spec = do
         it "turns it into a code block" $ do
           "@foo@" `shouldParseTo` DocCodeBlock "foo"
 
-      context "when a paragraph starts with a markdown link" $ do
-        it "correctly parses it as a text paragraph (not a definition list)" $ do
-          "[label](url)" `shouldParseTo`
-            DocParagraph (hyperlink "url" "label")
+      context "when a paragraph contains a markdown link" $ do
+        it "correctly parses the link" $ do
+          "Blah [label](url)" `shouldParseTo`
+            DocParagraph ("Blah " <> hyperlink "url" "label")
 
-        it "can be followed by an other paragraph" $ do
-          "[label](url)\n\nfoobar" `shouldParseTo`
-            DocParagraph (hyperlink "url" "label") <> DocParagraph "foobar"
-
-        context "when paragraph contains additional text" $ do
-          it "accepts more text after the link" $ do
-            "[label](url) foo bar baz" `shouldParseTo`
-              DocParagraph (hyperlink "url" "label" <> " foo bar baz")
-
-          it "accepts a newline right after the markdown link" $ do
-            "[label](url)\nfoo bar baz" `shouldParseTo`
-              DocParagraph (hyperlink "url" "label" <> " foo bar baz")
+        context "when the paragraph starts with the markdown link" $ do
+          it "correctly parses it as a text paragraph (not a definition list)" $ do
+            "[label](url)" `shouldParseTo`
+              DocParagraph (hyperlink "url" "label")
 
           it "can be followed by an other paragraph" $ do
-            "[label](url)foo\n\nbar" `shouldParseTo`
-              DocParagraph (hyperlink "url" "label" <> "foo") <> DocParagraph "bar"
+            "[label](url)\n\nfoobar" `shouldParseTo`
+              DocParagraph (hyperlink "url" "label") <> DocParagraph "foobar"
+
+          context "when paragraph contains additional text" $ do
+            it "accepts more text after the link" $ do
+              "[label](url) foo bar baz" `shouldParseTo`
+                DocParagraph (hyperlink "url" "label" <> " foo bar baz")
+
+            it "accepts a newline right after the markdown link" $ do
+              "[label](url)\nfoo bar baz" `shouldParseTo`
+                DocParagraph (hyperlink "url" "label" <> " foo bar baz")
+
+            it "can be followed by an other paragraph" $ do
+              "[label](url)foo\n\nbar" `shouldParseTo`
+                DocParagraph (hyperlink "url" "label" <> "foo") <> DocParagraph "bar"
+
+        context "when the link starts on a new line not at the beginning of the paragraph" $ do
+          it "correctly parses the link" $ do
+            "Bla\n[label](url)" `shouldParseTo`
+              DocParagraph ("Bla\n" <> hyperlink "url" "label")
 
     context "when parsing birdtracks" $ do
       it "parses them as a code block" $ do