Feature Request #3039490 Improved syntactic highlighting for Haskell

'as' is a keyword but in all other contexts it is a normal identifier.
The highligthting was failing if the operators are not separated by spaces: 10+20
There was highlighting for data types, classes and instances but it didn't
work correctly. This was removed.

1 files changed, 171 insertions, 81 deletions

diff --git a/lexers/LexHaskell.cxx b/lexers/LexHaskell.cxx
index 2f34c66aa..37d85d0cf 100644
--- a/lexers/LexHaskell.cxx
+++ b/lexers/LexHaskell.cxx
@@ -7,6 +7,7 @@
  *
  *    Written by Tobias Engvall - tumm at dtek dot chalmers dot se
  *
+ *    Several bug fixes by Krasimir Angelov - kr.angelov at gmail.com
  *
  *    TODO:
  *    * Implement a folder :)
@@ -48,11 +49,13 @@ using namespace Scintilla;
 
 #endif
 
-// Max level of nested comments
-#define SCE_HA_COMMENTMAX SCE_HA_COMMENTBLOCK3
-
-
-enum kwType { kwOther, kwClass, kwData, kwInstance, kwImport, kwModule, kwType};
+#define HA_MODE_DEFAULT     0
+#define HA_MODE_IMPORT1     1
+#define HA_MODE_IMPORT2     2
+#define HA_MODE_IMPORT3     3
+#define HA_MODE_MODULE      4
+#define HA_MODE_FFI         5
+#define HA_MODE_TYPE        6
 
 static inline bool IsNewline(const int ch) {
    return (ch == '\n' || ch == '\r');
@@ -76,146 +79,234 @@ static void ColorizeHaskellDoc(unsigned int startPos, int length, int initStyle,
                                WordList *keywordlists[], Accessor &styler) {
 
    WordList &keywords = *keywordlists[0];
-
-   int kwLast = kwOther;
+   WordList &ffi      = *keywordlists[1];
 
    StyleContext sc(startPos, length, initStyle, styler);
 
-   for (; sc.More(); sc.Forward()) {
+   int lineCurrent = styler.GetLine(startPos);
+   int state = lineCurrent ? styler.GetLineState(lineCurrent-1)
+                           : HA_MODE_DEFAULT;
+   int mode  = state & 0xF;
+   int xmode = state >> 4;
 
+   while (sc.More()) {
       // Check for state end
+
          // Operator
       if (sc.state == SCE_HA_OPERATOR) {
-         kwLast = kwOther;
-         sc.SetState(SCE_HA_DEFAULT);
+         if (isascii(sc.ch) && isoperator(static_cast<char>(sc.ch))) {
+            sc.Forward();
+         } else {
+            styler.ColourTo(sc.currentPos - 1, sc.state);
+            sc.ChangeState(SCE_HA_DEFAULT);
+         }
       }
          // String
       else if (sc.state == SCE_HA_STRING) {
          if (sc.ch == '\"') {
-            sc.ForwardSetState(SCE_HA_DEFAULT);
+			sc.Forward();
+            styler.ColourTo(sc.currentPos-1, sc.state);
+            sc.ChangeState(SCE_HA_DEFAULT);
          } else if (sc.ch == '\\') {
-            sc.Forward();
-         }
+            sc.Forward(2);
+         } else if (sc.atLineEnd) {
+			styler.ColourTo(sc.currentPos-1, sc.state);
+			sc.ChangeState(SCE_HA_DEFAULT);
+		 } else {
+			sc.Forward();
+		 }
       }
          // Char
       else if (sc.state == SCE_HA_CHARACTER) {
          if (sc.ch == '\'') {
-            sc.ForwardSetState(SCE_HA_DEFAULT);
+			sc.Forward();
+            styler.ColourTo(sc.currentPos-1, sc.state);
+            sc.ChangeState(SCE_HA_DEFAULT);
          } else if (sc.ch == '\\') {
-            sc.Forward();
-         }
+            sc.Forward(2);
+         } else if (sc.atLineEnd) {
+			styler.ColourTo(sc.currentPos-1, sc.state);
+			sc.ChangeState(SCE_HA_DEFAULT);
+		 } else {
+			sc.Forward();
+		 }
       }
          // Number
       else if (sc.state == SCE_HA_NUMBER) {
-         if (!IsADigit(sc.ch)) {
-            sc.SetState(SCE_HA_DEFAULT);
-         }
-      }
-         // Types, constructors, etc.
-      else if (sc.state == SCE_HA_CAPITAL) {
-         if (!IsAWordChar(sc.ch) || sc.ch == '.') {
-            sc.SetState(SCE_HA_DEFAULT);
+         if (IsADigit(sc.ch, xmode)) {
+            sc.Forward();
+         } else if ((xmode == 10) &&
+                    (sc.ch == 'e' || sc.ch == 'E') &&
+                    (IsADigit(sc.chNext) || sc.chNext == '+' || sc.chNext == '-')) {
+			sc.Forward();
+			if (sc.ch == '+' || sc.ch == '-')
+				sc.Forward();
+         } else {
+            styler.ColourTo(sc.currentPos - 1, sc.state);
+            sc.ChangeState(SCE_HA_DEFAULT);
          }
       }
          // Identifier
       else if (sc.state == SCE_HA_IDENTIFIER) {
-         if (!IsAWordChar(sc.ch)) {
+         if (IsAWordChar(sc.ch)) {
+            sc.Forward();
+         } else {
             char s[100];
             sc.GetCurrent(s, sizeof(s));
-            int style = SCE_HA_IDENTIFIER;
-            if ((kwLast == kwImport) || (strcmp(s,"qualified") == 0) || (strcmp(s,"as") == 0)) {
-               style = SCE_HA_IMPORT;
-            } else if (keywords.InList(s)) {
+            int style = sc.state;
+            int new_mode = 0;
+            if (keywords.InList(s)) {
                style = SCE_HA_KEYWORD;
-            } else if (kwLast == kwData) {
-               style = SCE_HA_DATA;
-            } else if (kwLast == kwClass) {
-               style = SCE_HA_CLASS;
-            } else if (kwLast == kwModule) {
-               style = SCE_HA_MODULE;
             } else if (isupper(s[0])) {
-               style = SCE_HA_CAPITAL;
-            }
-            sc.ChangeState(style);
-            sc.SetState(SCE_HA_DEFAULT);
-            if (style == SCE_HA_KEYWORD) {
-               if (0 == strcmp(s, "class"))
-                  kwLast = kwClass;
-               else if (0 == strcmp(s, "data"))
-                  kwLast = kwData;
-               else if (0 == strcmp(s, "instance"))
-                  kwLast = kwInstance;
-               else if (0 == strcmp(s, "import"))
-                  kwLast = kwImport;
-               else if (0 == strcmp(s, "module"))
-                  kwLast = kwModule;
+               if (mode >= HA_MODE_IMPORT1 && mode <= HA_MODE_IMPORT3) {
+                  style    = SCE_HA_MODULE;
+                  new_mode = HA_MODE_IMPORT2;
+               } else if (mode == HA_MODE_MODULE)
+                  style = SCE_HA_MODULE;
                else
-                  kwLast = kwOther;
-            } else if (style == SCE_HA_CLASS || style == SCE_HA_IMPORT ||
-                       style == SCE_HA_MODULE || style == SCE_HA_CAPITAL ||
-                       style == SCE_HA_DATA || style == SCE_HA_INSTANCE) {
-               kwLast = kwOther;
+                  style = SCE_HA_CAPITAL;
+            } else if (mode == HA_MODE_IMPORT1 &&
+                       strcmp(s,"qualified") == 0) {
+                style    = SCE_HA_KEYWORD;
+                new_mode = HA_MODE_IMPORT1;
+            } else if (mode == HA_MODE_IMPORT2) {
+                if (strcmp(s,"as") == 0) {
+                   style    = SCE_HA_KEYWORD;
+                   new_mode = HA_MODE_IMPORT3;
+               } else if (strcmp(s,"hiding") == 0) {
+                   style     = SCE_HA_KEYWORD;
+               }
+            } else if (mode == HA_MODE_FFI) {
+			   if (ffi.InList(s)) {
+                  style = SCE_HA_KEYWORD;
+                  new_mode = HA_MODE_FFI;
+               }
             }
+            else if (mode == HA_MODE_TYPE) {
+               if (strcmp(s,"family") == 0)
+                  style    = SCE_HA_KEYWORD;
+			}
+            styler.ColourTo(sc.currentPos - 1, style);
+            if (strcmp(s,"import") == 0 && mode != HA_MODE_FFI)
+               new_mode = HA_MODE_IMPORT1;
+            else if (strcmp(s,"module") == 0)
+               new_mode = HA_MODE_MODULE;
+            else if (strcmp(s,"foreign") == 0)
+               new_mode = HA_MODE_FFI;
+            else if (strcmp(s,"type") == 0)
+               new_mode = HA_MODE_TYPE;
+            sc.ChangeState(SCE_HA_DEFAULT);
+            mode = new_mode;
          }
       }
+
          // Comments
             // Oneliner
       else if (sc.state == SCE_HA_COMMENTLINE) {
-         if (IsNewline(sc.ch))
-            sc.SetState(SCE_HA_DEFAULT);
+         if (sc.atLineEnd) {
+            styler.ColourTo(sc.currentPos - 1, sc.state);
+            sc.ChangeState(SCE_HA_DEFAULT);
+         } else {
+            sc.Forward();
+         }
       }
             // Nested
-      else if (sc.state >= SCE_HA_COMMENTBLOCK) {
+      else if (sc.state == SCE_HA_COMMENTBLOCK) {
          if (sc.Match("{-")) {
-            if (sc.state < SCE_HA_COMMENTMAX)
-               sc.SetState(sc.state + 1);
+            sc.Forward(2);
+            xmode++;
          }
          else if (sc.Match("-}")) {
+            sc.Forward(2);
+            xmode--;
+            if (xmode == 0) {
+               styler.ColourTo(sc.currentPos - 1, sc.state);
+               sc.ChangeState(SCE_HA_DEFAULT);
+            }
+         } else {
+            if (sc.atLineEnd) {
+				// Remember the line state for future incremental lexing
+				styler.SetLineState(lineCurrent, (xmode << 4) | mode);
+				lineCurrent++;
+			}
             sc.Forward();
-            if (sc.state == SCE_HA_COMMENTBLOCK)
-               sc.ForwardSetState(SCE_HA_DEFAULT);
-            else
-               sc.ForwardSetState(sc.state - 1);
          }
       }
       // New state?
       if (sc.state == SCE_HA_DEFAULT) {
          // Digit
-         if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
-            sc.SetState(SCE_HA_NUMBER);
-            if (sc.ch == '0' && (sc.chNext == 'X' || sc.chNext == 'x')) { // Match anything starting with "0x" or "0X", too
-               sc.Forward(1);
-            }
+         if (IsADigit(sc.ch) ||
+             (sc.ch == '.' && IsADigit(sc.chNext)) ||
+             (sc.ch == '-' && IsADigit(sc.chNext))) {
+            styler.ColourTo(sc.currentPos - 1, sc.state);
+            sc.ChangeState(SCE_HA_NUMBER);
+            if (sc.ch == '0' && (sc.chNext == 'X' || sc.chNext == 'x')) {
+				// Match anything starting with "0x" or "0X", too
+				sc.Forward(2);
+				xmode = 16;
+            } else if (sc.ch == '0' && (sc.chNext == 'O' || sc.chNext == 'o')) {
+				// Match anything starting with "0x" or "0X", too
+				sc.Forward(2);
+				xmode = 8;
+            } else {
+				sc.Forward();
+				xmode = 10;
+			}
+            mode = HA_MODE_DEFAULT;
          }
          // Comment line
          else if (sc.Match("--")) {
-            sc.SetState(SCE_HA_COMMENTLINE);
+            styler.ColourTo(sc.currentPos - 1, sc.state);
+            sc.Forward(2);
+            sc.ChangeState(SCE_HA_COMMENTLINE);
          // Comment block
          }
          else if (sc.Match("{-")) {
-            sc.SetState(SCE_HA_COMMENTBLOCK);
+            styler.ColourTo(sc.currentPos - 1, sc.state);
+            sc.Forward(2);
+            sc.ChangeState(SCE_HA_COMMENTBLOCK);
+            xmode = 1;
          }
          // String
          else if (sc.Match('\"')) {
-            sc.SetState(SCE_HA_STRING);
+            styler.ColourTo(sc.currentPos - 1, sc.state);
+            sc.Forward();
+            sc.ChangeState(SCE_HA_STRING);
          }
          // Character
          else if (sc.Match('\'')) {
-            sc.SetState(SCE_HA_CHARACTER);
-         }
-         // Stringstart
-         else if (sc.Match('\"')) {
-            sc.SetState(SCE_HA_STRING);
+            styler.ColourTo(sc.currentPos - 1, sc.state);
+            sc.Forward();
+            sc.ChangeState(SCE_HA_CHARACTER);
          }
+         else if (sc.ch == '(' || sc.ch == ')' ||
+                  sc.ch == '{' || sc.ch == '}' ||
+                  sc.ch == '[' || sc.ch == ']') {
+			styler.ColourTo(sc.currentPos - 1, sc.state);
+			sc.Forward();
+			styler.ColourTo(sc.currentPos - 1, SCE_HA_OPERATOR);
+			mode = HA_MODE_DEFAULT;
+		 }
          // Operator
          else if (isascii(sc.ch) && isoperator(static_cast<char>(sc.ch))) {
-            sc.SetState(SCE_HA_OPERATOR);
+            styler.ColourTo(sc.currentPos - 1, sc.state);
+            sc.Forward();
+            sc.ChangeState(SCE_HA_OPERATOR);
+            mode = HA_MODE_DEFAULT;
          }
          // Keyword
          else if (IsAWordStart(sc.ch)) {
-               sc.SetState(SCE_HA_IDENTIFIER);
+            styler.ColourTo(sc.currentPos - 1, sc.state);
+            sc.Forward();
+            sc.ChangeState(SCE_HA_IDENTIFIER);
+         } else {
+            if (sc.atLineEnd) {
+				// Remember the line state for future incremental lexing
+				styler.SetLineState(lineCurrent, (xmode << 4) | mode);
+				lineCurrent++;
+			}
+            sc.Forward();
          }
-
       }
    }
    sc.Complete();
@@ -275,4 +366,3 @@ void EXT_LEXER_DECL GetLexerName(unsigned int Index, char *name, int buflength)
 #endif
 
 LexerModule lmHaskell(SCLEX_HASKELL, ColorizeHaskellDoc, "haskell");
-