JavaScript: extract variables and constants in destructuring assignment

Close universal-ctags#1112. Signed-off-by: Masatake YAMATO <[email protected]>
masatake · Aug 26, 2022 · ec3b205 · ec3b205
1 parent dfffd3d
commit ec3b205
Show file tree

Hide file tree

Showing 6 changed files with 275 additions and 2 deletions.
diff --git a/Units/parser-javascript.r/js-destructural-binding.d/args.ctags b/Units/parser-javascript.r/js-destructural-binding.d/args.ctags
@@ -0,0 +1 @@
+--sort=no
diff --git a/Units/parser-javascript.r/js-destructural-binding.d/expected.tags b/Units/parser-javascript.r/js-destructural-binding.d/expected.tags
@@ -0,0 +1,59 @@
+x	input.js	/^const [x] = [1];$/;"	C
+y	input.js	/^const [y, z] = [1, 2, 3, 4, 5];$/;"	C
+z	input.js	/^const [y, z] = [1, 2, 3, 4, 5];$/;"	C
+a	input.js	/^let [a=5, b=7] = [1];$/;"	v
+b	input.js	/^let [a=5, b=7] = [1];$/;"	v
+c	input.js	/^let [c, , d] = [1, 2, 3];$/;"	v
+d	input.js	/^let [c, , d] = [1, 2, 3];$/;"	v
+e	input.js	/^let [e, f = 0, , g] = [1, 2, 3, 4];$/;"	v
+f	input.js	/^let [e, f = 0, , g] = [1, 2, 3, 4];$/;"	v
+g	input.js	/^let [e, f = 0, , g] = [1, 2, 3, 4];$/;"	v
+h	input.js	/^const [h, i, ...[j, k]] = [1, 2, 3, 4];$/;"	C
+i	input.js	/^const [h, i, ...[j, k]] = [1, 2, 3, 4];$/;"	C
+j	input.js	/^const [h, i, ...[j, k]] = [1, 2, 3, 4];$/;"	C
+k	input.js	/^const [h, i, ...[j, k]] = [1, 2, 3, 4];$/;"	C
+l	input.js	/^const [l, m, ...[n, o, ...[p, q]]] = [1, 2, 3, 4, 5, 6];$/;"	C
+m	input.js	/^const [l, m, ...[n, o, ...[p, q]]] = [1, 2, 3, 4, 5, 6];$/;"	C
+n	input.js	/^const [l, m, ...[n, o, ...[p, q]]] = [1, 2, 3, 4, 5, 6];$/;"	C
+o	input.js	/^const [l, m, ...[n, o, ...[p, q]]] = [1, 2, 3, 4, 5, 6];$/;"	C
+p	input.js	/^const [l, m, ...[n, o, ...[p, q]]] = [1, 2, 3, 4, 5, 6];$/;"	C
+q	input.js	/^const [l, m, ...[n, o, ...[p, q]]] = [1, 2, 3, 4, 5, 6];$/;"	C
+A	input.js	/^const [A, B, ...{ C, D }] = []$/;"	C
+B	input.js	/^const [A, B, ...{ C, D }] = []$/;"	C
+C	input.js	/^const [A, B, ...{ C, D }] = []$/;"	C
+D	input.js	/^const [A, B, ...{ C, D }] = []$/;"	C
+E	input.js	/^    E: 42,$/;"	p	variable:user
+F	input.js	/^    F: true$/;"	p	variable:user
+user	input.js	/^const user = {$/;"	v
+E	input.js	/^const {E, F} = user;$/;"	C
+F	input.js	/^const {E, F} = user;$/;"	C
+G	input.js	/^const {E: G, F: H} = user;$/;"	C
+H	input.js	/^const {E: G, F: H} = user;$/;"	C
+I	input.js	/^const {I = 10, J = 5} = {I: 3};$/;"	C
+J	input.js	/^const {I = 10, J = 5} = {I: 3};$/;"	C
+I	input.js	/^const {I = 10, J = 5} = {I: 3};$/;"	p	variable:anonymousObject785a93f40105
+anonymousObject785a93f40105	input.js	/^const {I = 10, J = 5} = {I: 3};$/;"	v
+K	input.js	/^let {a: K = 10, b: L = 5} = {a: 3};$/;"	v
+L	input.js	/^let {a: K = 10, b: L = 5} = {a: 3};$/;"	v
+a	input.js	/^let {a: K = 10, b: L = 5} = {a: 3};$/;"	p	variable:anonymousObject785a93f40205
+anonymousObject785a93f40205	input.js	/^let {a: K = 10, b: L = 5} = {a: 3};$/;"	v
+M	input.js	/^let {M, N, ...O} = {M: 10, N: 20, c: 30, d: 40}$/;"	v
+N	input.js	/^let {M, N, ...O} = {M: 10, N: 20, c: 30, d: 40}$/;"	v
+O	input.js	/^let {M, N, ...O} = {M: 10, N: 20, c: 30, d: 40}$/;"	v
+M	input.js	/^let {M, N, ...O} = {M: 10, N: 20, c: 30, d: 40}$/;"	p	variable:anonymousObject785a93f40305
+N	input.js	/^let {M, N, ...O} = {M: 10, N: 20, c: 30, d: 40}$/;"	p	variable:anonymousObject785a93f40305
+c	input.js	/^let {M, N, ...O} = {M: 10, N: 20, c: 30, d: 40}$/;"	p	variable:anonymousObject785a93f40305
+d	input.js	/^let {M, N, ...O} = {M: 10, N: 20, c: 30, d: 40}$/;"	p	variable:anonymousObject785a93f40305
+anonymousObject785a93f40305	input.js	/^let {M, N, ...O} = {M: 10, N: 20, c: 30, d: 40}$/;"	v
+title	input.js	/^  title: 'Scratchpad',$/;"	p	variable:metadata
+anonymousObject785a93f40405	input.js	/^    {$/;"	v	variable:metadata
+locale	input.js	/^      locale: 'de',$/;"	p	variable:metadata.anonymousObject785a93f40405
+localization_tags	input.js	/^      localization_tags: [],$/;"	p	variable:metadata.anonymousObject785a93f40405
+last_edit	input.js	/^      last_edit: '2014-04-14T08:43:37',$/;"	p	variable:metadata.anonymousObject785a93f40405
+url	input.js	/^      url: '\/de\/docs\/Tools\/Scratchpad',$/;"	p	variable:metadata.anonymousObject785a93f40405
+title	input.js	/^      title: 'JavaScript-Umgebung'$/;"	p	variable:metadata.anonymousObject785a93f40405
+translations	input.js	/^  translations: [$/;"	p	variable:metadata
+url	input.js	/^  url: '\/en-US\/docs\/Tools\/Scratchpad'$/;"	p	variable:metadata
+metadata	input.js	/^const metadata = {$/;"	v
+englishTitle	input.js	/^  title: englishTitle, \/\/ rename$/;"	v
+localeTitle	input.js	/^       title: localeTitle, \/\/ rename$/;"	v
diff --git a/Units/parser-javascript.r/js-destructural-binding.d/input.js b/Units/parser-javascript.r/js-destructural-binding.d/input.js
@@ -0,0 +1,56 @@
+// Derrived from https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/Destructuring_assignment
+const [x] = [1];
+const [y, z] = [1, 2, 3, 4, 5];
+let [a=5, b=7] = [1];
+
+let [c, , d] = [1, 2, 3];
+let [e, f = 0, , g] = [1, 2, 3, 4];
+
+let [,,] = [1, 2, 3];
+let [,]  = [1, 2, 3];
+let []   = [1, 2, 3];
+
+const [h, i, ...[j, k]] = [1, 2, 3, 4];
+const [l, m, ...[n, o, ...[p, q]]] = [1, 2, 3, 4, 5, 6];
+
+const [A, B, ...{ C, D }] = []
+
+const user = {
+    E: 42,
+    F: true
+};
+
+const {E, F} = user;
+
+const {E: G, F: H} = user;
+
+
+const {I = 10, J = 5} = {I: 3};
+
+let {a: K = 10, b: L = 5} = {a: 3};
+
+let {M, N, ...O} = {M: 10, N: 20, c: 30, d: 40}
+
+
+const metadata = {
+  title: 'Scratchpad',
+  translations: [
+    {
+      locale: 'de',
+      localization_tags: [],
+      last_edit: '2014-04-14T08:43:37',
+      url: '/de/docs/Tools/Scratchpad',
+      title: 'JavaScript-Umgebung'
+    }
+  ],
+  url: '/en-US/docs/Tools/Scratchpad'
+};
+
+let {
+  title: englishTitle, // rename
+  translations: [
+    {
+       title: localeTitle, // rename
+    },
+  ],
+} = metadata;
diff --git a/Units/parser-javascript.r/js-destructural-binding.d/validator b/Units/parser-javascript.r/js-destructural-binding.d/validator
@@ -0,0 +1 @@
+node
diff --git a/Units/parser-javascript.r/js-empty-class-name.d/expected.tags b/Units/parser-javascript.r/js-empty-class-name.d/expected.tags
@@ -1,2 +1,3 @@
+prop	input.js	/^var {prop} = { prop: "value" };$/;"	v
 prop	input.js	/^var {prop} = { prop: "value" };$/;"	p	variable:anonymousObject4ca5b60a0105
 anonymousObject4ca5b60a0105	input.js	/^var {prop} = { prop: "value" };$/;"	v
diff --git a/parsers/jscript.c b/parsers/jscript.c
@@ -2637,7 +2637,8 @@ static bool parseStatementRHS (tokenInfo *const name, tokenInfo *const token, st
 
 		if (state->indexForName == CORK_NIL)
 		{
-			state->indexForName = makeJsTag (name, state->isConst ? JSTAG_CONSTANT : JSTAG_VARIABLE, NULL, NULL);
+			if (!vStringIsEmpty (name->string))
+				state->indexForName = makeJsTag (name, state->isConst ? JSTAG_CONSTANT : JSTAG_VARIABLE, NULL, NULL);
 			if (isType (token, TOKEN_IDENTIFIER))
 				canbe_arrowfun = true;
 		}
@@ -2706,6 +2707,143 @@ static bool parseStatementRHS (tokenInfo *const name, tokenInfo *const token, st
 	return true;
 }
 
+static bool parseObjectDestructuring (tokenInfo *const token, bool is_const);
+static bool parseArrayDestructuring (tokenInfo *const token, bool is_const)
+{
+	int nest_level = 1;
+	bool in_left_side = true;
+	bool found = false;
+
+	while (nest_level > 0 && ! isType (token, TOKEN_EOF))
+	{
+		readToken (token);
+		if (isType (token, TOKEN_OPEN_SQUARE))
+		{
+			in_left_side = true;
+			nest_level++;
+		}
+		else if (isType (token, TOKEN_CLOSE_SQUARE))
+		{
+			in_left_side = false;
+			nest_level--;
+		}
+		else if (isType (token, TOKEN_OPEN_CURLY))
+		{
+			in_left_side = false;
+			if (parseObjectDestructuring (token, is_const))
+				found = true;
+		}
+		else if (isType (token, TOKEN_COMMA)
+			 || isType (token, TOKEN_DOTS))
+			in_left_side = true;
+		else if (in_left_side && isType (token, TOKEN_IDENTIFIER))
+		{
+			in_left_side = false;
+			makeJsTag (token,
+				   is_const ? JSTAG_CONSTANT : JSTAG_VARIABLE,
+				   NULL, NULL);
+			found = true;
+		}
+		else if (isType (token, TOKEN_EQUAL_SIGN))
+		{
+			in_left_side = false;
+			/* TODO: SKIP */
+		}
+		else
+			in_left_side = false;
+	}
+
+	return found;
+}
+
+static bool parseObjectDestructuring (tokenInfo *const token, bool is_const)
+{
+	tokenInfo *const name = newToken ();
+	bool found = false;
+
+	/*
+	 * let { k0: v0, k1: v1 = 0, v3 };
+	 *     |	  |  ||	  |  |	  |
+	 *     ^...|..|^...|..|....^.....: start
+	 *     ....^..|....^..|..........: colon
+	 *     .......^.......^..........: emitted (made a tag for an id after colon)
+	 */
+	enum objDestructuringState {
+		OBJ_DESTRUCTURING_START,
+		OBJ_DESTRUCTURING_COLON,
+		OBJ_DESTRUCTURING_EMITTED,
+	} state = OBJ_DESTRUCTURING_START;
+
+	while (! isType (token, TOKEN_EOF))
+	{
+		readToken (token);
+		if (isType (token, TOKEN_OPEN_CURLY))
+		{
+			if (parseObjectDestructuring (token, is_const))
+				found = true;
+			if (state == OBJ_DESTRUCTURING_COLON)
+				state = OBJ_DESTRUCTURING_EMITTED;
+		}
+		else if (isType (token, TOKEN_CLOSE_CURLY))
+		{
+			if (!vStringIsEmpty(name->string))
+			{
+				makeJsTag (name,
+					   is_const ? JSTAG_CONSTANT : JSTAG_VARIABLE,
+					   NULL, NULL);
+				found = true;
+			}
+			break;
+		}
+		else if (isType (token, TOKEN_OPEN_SQUARE))
+		{
+			if (parseArrayDestructuring (token, is_const))
+				found = true;
+			if (state == OBJ_DESTRUCTURING_COLON)
+				state = OBJ_DESTRUCTURING_EMITTED;
+		}
+		else if (isType (token, TOKEN_IDENTIFIER))
+		{
+			if (state == OBJ_DESTRUCTURING_COLON)
+			{
+				makeJsTag (token,
+					   is_const ? JSTAG_CONSTANT : JSTAG_VARIABLE,
+					   NULL, NULL);
+				found = true;
+				state = OBJ_DESTRUCTURING_EMITTED;
+			}
+			else if (state == OBJ_DESTRUCTURING_START
+				 && vStringIsEmpty(name->string))
+				copyToken(name, token, true);
+		}
+		else if (isType (token, TOKEN_COMMA))
+		{
+			if (!vStringIsEmpty(name->string))
+			{
+				makeJsTag (name,
+					   is_const ? JSTAG_CONSTANT : JSTAG_VARIABLE,
+					   NULL, NULL);
+				found = true;
+				vStringClear (name->string);
+			}
+			state = OBJ_DESTRUCTURING_START;
+		}
+		else if (isType (token, TOKEN_COLON))
+		{
+			vStringClear (name->string);
+			state = OBJ_DESTRUCTURING_COLON;
+		}
+		else
+		{
+			if (state == OBJ_DESTRUCTURING_COLON)
+				state = OBJ_DESTRUCTURING_EMITTED;
+		}
+	}
+
+	deleteToken (name);
+	return found;
+}
+
 static bool parseStatement (tokenInfo *const token, bool is_inside_class)
 {
 	TRACE_ENTER_TEXT("is_inside_class: %s", is_inside_class? "yes": "no");
@@ -2773,6 +2911,22 @@ static bool parseStatement (tokenInfo *const token, bool is_inside_class)
 			state.isGlobal = true;
 		}
 		readToken(token);
+
+		if (state.isGlobal)
+		{
+			bool found = false;
+			if (isType (token, TOKEN_OPEN_CURLY))
+				found = parseObjectDestructuring (token, state.isConst);
+			else if (isType (token, TOKEN_OPEN_SQUARE))
+				found = parseArrayDestructuring (token, state.isConst);
+
+			if (found)
+			{
+				/* Adjust the context to the code for non-destructing. */
+				found_lhs = true;
+				readToken(token);
+			}
+		}
 	}
 
 nextVar:
@@ -2860,7 +3014,8 @@ static bool parseStatement (tokenInfo *const token, bool is_inside_class)
 			 * Handles this syntax:
 			 *     var g_var2;
 			 */
-			state.indexForName = makeJsTag (name, state.isConst ? JSTAG_CONSTANT : JSTAG_VARIABLE, NULL, NULL);
+			if (!vStringIsEmpty (name->string))
+				state.indexForName = makeJsTag (name, state.isConst ? JSTAG_CONSTANT : JSTAG_VARIABLE, NULL, NULL);
 		}
 		/*
 		 * Statement has ended.