[WIP] Sh: [EXPERIMENTAL] extract variables in variable assignments

Tmain/ptag-in-optlib-parser.d/stdout-expected.txt

@@ -8,8 +8,9 @@
 !_TAG_FILE_SORTED	1	/0=unsorted, 1=sorted, 2=foldcase/
 !_TAG_KIND_DESCRIPTION!Sh	a,alias	/aliases/
 !_TAG_KIND_DESCRIPTION!Sh	f,function	/functions/
-!_TAG_KIND_DESCRIPTION!Sh	h,heredoc	/label for here document/
+!_TAG_KIND_DESCRIPTION!Sh	h,heredoc	/labels for here document/
 !_TAG_KIND_DESCRIPTION!Sh	s,script	/script files/
+!_TAG_KIND_DESCRIPTION!Sh	v,variable	/variables assigment (experimental)/
 !_TAG_OUTPUT_EXCMD	mixed	/number, pattern, mixed, or combineV2/
 !_TAG_OUTPUT_FILESEP	slash	/slash or backslash/
 !_TAG_OUTPUT_MODE	u-ctags	/u-ctags or e-ctags/

Units/parser-sh.r/sh-comments.d/expected.tags

@@ -2,3 +2,7 @@ afterpound1	input.sh	/^foo="#"; afterpound1() {}$/;"	f
 afterpound2	input.sh	/^foo="#\\""; afterpound2() {}$/;"	f
 afterpound3	input.sh	/^foo='#'; afterpound3() {}$/;"	f
 afterpound4	input.sh	/^foo='#'''; afterpound4() {}$/;"	f
+foo	input.sh	/^foo="#"; afterpound1() {}$/;"	v
+foo	input.sh	/^foo="#\\""; afterpound2() {}$/;"	v
+foo	input.sh	/^foo='#'''; afterpound4() {}$/;"	v
+foo	input.sh	/^foo='#'; afterpound3() {}$/;"	v

Units/parser-sh.r/sh-quotes.d/expected.tags

@@ -2,3 +2,8 @@ afterpound1	input.sh	/^foo="#"; afterpound1() {}$/;"	f
 afterpound2	input.sh	/^foo="#\\""; afterpound2() {}$/;"	f
 afterpound3	input.sh	/^foo='#'; afterpound3() {}$/;"	f
 afterpound4	input.sh	/^foo='#'''; afterpound4() {}$/;"	f
+foo	input.sh	/^foo="#"; afterpound1() {}$/;"	v
+foo	input.sh	/^foo="#\\""; afterpound2() {}$/;"	v
+foo	input.sh	/^foo="nofunction()"$/;"	v
+foo	input.sh	/^foo='#'''; afterpound4() {}$/;"	v
+foo	input.sh	/^foo='#'; afterpound3() {}$/;"	v

parsers/sh.c

@@ -38,6 +38,7 @@
 	K_FUNCTION,
 	K_SCRIPT,
 	K_HEREDOCLABEL,
+	K_VARIABLE,
 } shKind;
 
 typedef enum {
@@ -86,8 +87,10 @@
 	  .referenceOnly = false, FUNCTION_ROLES_SPEC },			\
 	{ true, 's', "script", "script files",						\
 	  .referenceOnly = true, ATTACH_ROLES (SCRIPT_ROLES) },		\
-	{ true, 'h', "heredoc", "label for here document",			\
-	  .referenceOnly = false, ATTACH_ROLES (HEREDOC_ROLES) }
+	{ true, 'h', "heredoc", "labels for here document",			\
+	  .referenceOnly = false, ATTACH_ROLES (HEREDOC_ROLES) },   \
+	{ true, 'v', "variable", "variables assigment (experimental)" }
+
 
 static kindDefinition ShKinds [] = {
 	SH_KINDS_COMMON(ShScriptRoles, ShHeredocRoles,),
@@ -103,6 +106,7 @@
 	KEYWORD_alias,
 	KEYWORD_source,
 	KEYWORD_autoload,
+	KEYWORD_declare,			/* declare, typeset, local, export, readonly, float(zsh), integer */
 };
 
 const char *dialectMap [] = {
@@ -114,6 +118,13 @@
 	{ "alias",     KEYWORD_alias,    { 1, 1 } },
 	{ "source",    KEYWORD_source,   { 1, 1 } },
 	{ "autoload",  KEYWORD_autoload, { 0, 1 } },
+	{ "declare",   KEYWORD_declare,  { 1, 1 } },
+	{ "typset",    KEYWORD_declare,  { 1, 1 } },
+	{ "local",     KEYWORD_declare,  { 1, 1 } },
+	{ "export",    KEYWORD_declare,  { 1, 1 } },
+	{ "readonly",  KEYWORD_declare,  { 1, 1 } },
+	{ "float",     KEYWORD_declare,  { 0, 1 } },
+	{ "integer",   KEYWORD_declare,  { 0, 1 } },
 };
 
 /*
@@ -352,6 +363,10 @@
 		*role = R_SCRIPT_LOADED;
 		*check_char = isFileChar;
 		break;
+	case KEYWORD_declare:
+		*kind = K_VARIABLE;
+		*check_char = isIdentChar;
+		break;
 	default:
 		AssertNotReached();
 		break;
@@ -509,6 +524,54 @@
 	return vStringLength(token);
 }
 
+static bool doesLineCotinue(const unsigned char *start, const unsigned char *cp)
+{
+	cp--;
+	if (start >= cp)
+		return false;
+
+	if (*cp != '\\')
+		return false;
+
+	while (start < cp)
+	{
+		if (*cp == ';' || *cp == '|' || *cp == '&'
+			|| *cp == '(' || *cp == '{')
+			return false;
+		else if (isspace(*cp))
+			cp--;
+		else
+			return true;
+	}
+	return false;
+}
+
+static bool handleVariableAssignment (vString *input)
+{
+	const char *base = vStringValue (input);
+	const char *cp = base;
+
+	while (*cp != '\0')
+	{
+		if (*cp == '=' || (*cp == '+' && *(cp + 1) == '='))
+		{
+			size_t len = cp - base;
+			if (len > 0)
+			{
+				vStringTruncate (input, len);
+				return true;
+			}
+			break;
+		}
+		else if ( ((cp == base)?
+				   isIdentChar0: isIdentChar) ((unsigned char)*cp) )
+			cp++;
+		else
+			break;
+	}
+	return false;
+}
+
 typedef bool (* checkCharFunc) (int);
 static void findShTagsCommon (size_t (* keyword_handler) (int,
 														  vString *,
@@ -527,6 +590,7 @@
 	struct hereDocParsingState hstate;
 	hdocStateInit (&hstate);
 
+	bool cont_line = false;
 	while ((line = readLineFromInputFile ()) != NULL)
 	{
 		const unsigned char* cp = line;
@@ -552,10 +616,12 @@
 				vStringDelete (hereDocDelimiter);
 				hereDocDelimiter = NULL;
 			}
+			cont_line = false;
 			continue;
 		}
 
 		hdocStateClear (&hstate);
+		bool beginning_of_line = !cont_line;
 		while (*cp != '\0')
 		{
 			subparser *sub = NULL;
@@ -693,6 +759,7 @@
 						cp += d;
 					else if (*cp != '\0')
 						++cp;
+					beginning_of_line = false;
 					continue;
 				}
 
@@ -706,8 +773,9 @@
 			while (isspace (*cp))
 				++cp;
 
-			if ((found_kind != K_SCRIPT)
-			    && *cp == '(')
+			if (found_kind == K_SCRIPT)
+				;				/* Do NOTHING */
+			else if (*cp == '(')
 			{
 				++cp;
 				while (isspace (*cp))
@@ -733,6 +801,10 @@
 					++cp;
 				}
 			}
+			else if (beginning_of_line
+					 && found_kind == K_NOTHING
+					 && handleVariableAssignment (name))
+				found_kind = K_VARIABLE;
 
 			if (found_kind != K_NOTHING)
 			{
@@ -749,7 +821,11 @@
 			else if (!hereDocDelimiter)
 				hdocStateUpdateArgs (&hstate, name);
 			vStringClear (name);
+			beginning_of_line = false;
 		}
+		if (*cp == '#')
+			cont_line = false;
+		cont_line = doesLineCotinue (line, cp);
 	}
 	hdocStateFini (&hstate);
 	vStringDelete (name);