Merge pull request #4019 from masatake/haskell--handle-newline-just-a…

…fter-id

Haskell: skip multi-line type signature

Units/parser-haskell.r/multi-line-type-signature.d/README

@@ -0,0 +1 @@
+Taken from #4013 submitted by Robin Palotai (@robinp).

Units/parser-haskell.r/multi-line-type-signature.d/args.ctags

@@ -0,0 +1 @@
+--sort=no

Units/parser-haskell.r/multi-line-type-signature.d/expected.tags

@@ -0,0 +1,2 @@
+Foo	input.hs	/^module Foo () where$/;"	m
+thing	input.hs	/^thing x = pure x$/;"	f

Units/parser-haskell.r/multi-line-type-signature.d/input.hs

@@ -0,0 +1,9 @@
+module Foo () where
+
+
+thing
+    :: App m
+    => Int
+    -> m Int
+thing x = pure x
+

parsers/haskell.c

@@ -27,6 +27,7 @@
 #include "routines.h"
 
 
+#define HASKELL_STRMAXLEN 1000
 /*
 *   DATA DEFINITIONS
 */
@@ -64,7 +65,7 @@ static int get_line(char *buf)
 	do {
 		c = getcFromInputFile();
 		buf[i++] = c;
-	} while (c != EOF && c != '\n' && i < 1000);
+	} while (c != EOF && c != '\n' && i < HASKELL_STRMAXLEN);
 	buf[i] = '\0';
 	return i;
 }
@@ -114,7 +115,7 @@ static int get_token(char *token, int n)
 {
 	int c = getcFromInputFile();
 	int i = n;
-	while (c != EOF && isident(c) && i < 1000) {
+	while (c != EOF && isident(c) && i < HASKELL_STRMAXLEN) {
 		token[i] = c;
 		i++;
 		c = getcFromInputFile();
@@ -136,7 +137,7 @@ static int inside_datatype(vString *name)
 {
 	enum Find_State st = Find_Eq;
 	int c;
-	char token[1001];
+	char token[HASKELL_STRMAXLEN + 1];
 
 	while (1) {
 		if (st == Find_Eq)
@@ -221,7 +222,7 @@ static int inside_datatype(vString *name)
 static void findHaskellTags (int is_literate)
 {
 	vString *name = vStringNew ();
-	char token[1001], arg[1001];
+	char token[HASKELL_STRMAXLEN + 1], arg[HASKELL_STRMAXLEN + 1];
 	int c;
 	int in_tex_lit_code = 0;
 	c = get_next_char();
@@ -310,6 +311,32 @@ static void findHaskellTags (int is_literate)
 				 strcmp(token, "import") == 0)
 			;
 		else {
+			/*
+			 * "::" can be after [\n][ \t]+
+			 * ------------------------------
+			 * thing
+			 *   :: App m
+			 *   => Int
+			 *   -> m Int
+			 * ------------------------------
+			 * Skip them to find ':'.
+			 */
+			if (arg[0] == '\n' && arg[1] == '\0') {
+				bool indented = false;
+				while (1)
+				{
+					if ((c = getcFromInputFile()) == EOF)
+						return;
+					if (!isspace(c))
+						break;
+					indented = true;
+				}
+				if (indented)
+					arg[0] = c;
+				else
+					ungetcToInputFile(c);
+			}
+
 			if (arg[0] != ':')
 				add_tag(token, K_FUNCTION, name);
 		}