Bash: fix comment detection inside a word

A comment in bash is defined as "[...] a word beginning with # causes
that word and all remaining characters on that line to be ignored".

A word is defined as "a sequence of characters considered as a single
unit by the shell";  and there is a set of metacharacters defined as
"a character that, when unquoted, separates words.  One of the
following:  | & ; ( ) < > space tab".

In practice, "foo#bar" is one single word, not "foo" followed by a
comment.  Trickier, "foo\;#bar" is also a single word, but "foo;bar"
are 2 words and a control character.

So, fix the Bash lexer to check whether the character preceding the
hash sign to be either a metacharacter or part of a word.
A maybe better fix would be to understand the Bash conception of a
word, and analyze those, but it would require a large rewrite.

1 files changed, 7 insertions, 1 deletions

diff --git a/lexers/LexBash.cxx b/lexers/LexBash.cxx
index 5f582e401..513a5dff8 100644
--- a/lexers/LexBash.cxx
+++ b/lexers/LexBash.cxx
@@ -108,6 +108,7 @@ static void ColouriseBashDoc(unsigned int startPos, int length, int initStyle,
 	CharacterSet setWordStart(CharacterSet::setAlpha, "_");
 	// note that [+-] are often parts of identifiers in shell scripts
 	CharacterSet setWord(CharacterSet::setAlphaNum, "._+-");
+	CharacterSet setMetaCharacter(CharacterSet::setNone, "|&;()<> \t\r\n");
 	CharacterSet setBashOperator(CharacterSet::setNone, "^&%()-+=|{}[]:;>,*/<?!.~@");
 	CharacterSet setSingleCharOp(CharacterSet::setNone, "rwxoRWXOezsfdlpSbctugkTBMACahGLNn");
 	CharacterSet setParam(CharacterSet::setAlphaNum, "$_");
@@ -627,7 +628,12 @@ static void ColouriseBashDoc(unsigned int startPos, int length, int initStyle,
 			} else if (setWordStart.Contains(sc.ch)) {
 				sc.SetState(SCE_SH_WORD);
 			} else if (sc.ch == '#') {
-				sc.SetState(SCE_SH_COMMENTLINE);
+				if (stylePrev != SCE_SH_WORD && stylePrev != SCE_SH_IDENTIFIER &&
+					(sc.currentPos == 0 || setMetaCharacter.Contains(sc.chPrev))) {
+					sc.SetState(SCE_SH_COMMENTLINE);
+				} else {
+					sc.SetState(SCE_SH_WORD);
+				}
 			} else if (sc.ch == '\"') {
 				sc.SetState(SCE_SH_STRING);
 				QuoteStack.Start(sc.ch, BASH_DELIM_STRING);