Skip to content

Commit f0e6f15

Browse files
untitakeruntitaker
committed
Relax <select> parser (WHATWG proposal)
html5lib/html5lib-tests#178 The proposal isn't merged yet, and the error codes are off. The forked html5lib-tests makes it more complicated. I recommend to ditch the fork of html5lib-tests and give up on standardizing errors.
1 parent 593e0f4 commit f0e6f15

File tree

4 files changed

+49
-255
lines changed

4 files changed

+49
-255
lines changed

‎packages/parse5/lib/parser/index.ts

+43-211
Original file line numberDiff line numberDiff line change
@@ -54,8 +54,6 @@ enum InsertionMode {
5454
IN_TABLE_BODY,
5555
IN_ROW,
5656
IN_CELL,
57-
IN_SELECT,
58-
IN_SELECT_IN_TABLE,
5957
IN_TEMPLATE,
6058
AFTER_BODY,
6159
IN_FRAMESET,
@@ -706,10 +704,6 @@ export class Parser<T extends TreeAdapterTypeMap> implements TokenHandler, Stack
706704
this.insertionMode = InsertionMode.IN_FRAMESET;
707705
return;
708706
}
709-
case $.SELECT: {
710-
this._resetInsertionModeForSelect(i);
711-
return;
712-
}
713707
case $.TEMPLATE: {
714708
this.insertionMode = this.tmplInsertionModeStack[0];
715709
return;
@@ -739,24 +733,6 @@ export class Parser<T extends TreeAdapterTypeMap> implements TokenHandler, Stack
739733
this.insertionMode = InsertionMode.IN_BODY;
740734
}
741735

742-
/** @protected */
743-
_resetInsertionModeForSelect(selectIdx: number): void {
744-
if (selectIdx > 0) {
745-
for (let i = selectIdx - 1; i > 0; i--) {
746-
const tn = this.openElements.tagIDs[i];
747-
748-
if (tn === $.TEMPLATE) {
749-
break;
750-
} else if (tn === $.TABLE) {
751-
this.insertionMode = InsertionMode.IN_SELECT_IN_TABLE;
752-
return;
753-
}
754-
}
755-
}
756-
757-
this.insertionMode = InsertionMode.IN_SELECT;
758-
}
759-
760736
//Foster parenting
761737

762738
/** @protected */
@@ -859,9 +835,7 @@ export class Parser<T extends TreeAdapterTypeMap> implements TokenHandler, Stack
859835
characterInBody(this, token);
860836
break;
861837
}
862-
case InsertionMode.TEXT:
863-
case InsertionMode.IN_SELECT:
864-
case InsertionMode.IN_SELECT_IN_TABLE: {
838+
case InsertionMode.TEXT: {
865839
this._insertCharacters(token);
866840
break;
867841
}
@@ -974,8 +948,6 @@ export class Parser<T extends TreeAdapterTypeMap> implements TokenHandler, Stack
974948
case InsertionMode.IN_TABLE_BODY:
975949
case InsertionMode.IN_ROW:
976950
case InsertionMode.IN_CELL:
977-
case InsertionMode.IN_SELECT:
978-
case InsertionMode.IN_SELECT_IN_TABLE:
979951
case InsertionMode.IN_TEMPLATE:
980952
case InsertionMode.IN_FRAMESET:
981953
case InsertionMode.AFTER_FRAMESET: {
@@ -1110,14 +1082,6 @@ export class Parser<T extends TreeAdapterTypeMap> implements TokenHandler, Stack
11101082
startTagInCell(this, token);
11111083
break;
11121084
}
1113-
case InsertionMode.IN_SELECT: {
1114-
startTagInSelect(this, token);
1115-
break;
1116-
}
1117-
case InsertionMode.IN_SELECT_IN_TABLE: {
1118-
startTagInSelectInTable(this, token);
1119-
break;
1120-
}
11211085
case InsertionMode.IN_TEMPLATE: {
11221086
startTagInTemplate(this, token);
11231087
break;
@@ -1220,14 +1184,6 @@ export class Parser<T extends TreeAdapterTypeMap> implements TokenHandler, Stack
12201184
endTagInCell(this, token);
12211185
break;
12221186
}
1223-
case InsertionMode.IN_SELECT: {
1224-
endTagInSelect(this, token);
1225-
break;
1226-
}
1227-
case InsertionMode.IN_SELECT_IN_TABLE: {
1228-
endTagInSelectInTable(this, token);
1229-
break;
1230-
}
12311187
case InsertionMode.IN_TEMPLATE: {
12321188
endTagInTemplate(this, token);
12331189
break;
@@ -1285,9 +1241,7 @@ export class Parser<T extends TreeAdapterTypeMap> implements TokenHandler, Stack
12851241
case InsertionMode.IN_COLUMN_GROUP:
12861242
case InsertionMode.IN_TABLE_BODY:
12871243
case InsertionMode.IN_ROW:
1288-
case InsertionMode.IN_CELL:
1289-
case InsertionMode.IN_SELECT:
1290-
case InsertionMode.IN_SELECT_IN_TABLE: {
1244+
case InsertionMode.IN_CELL: {
12911245
eofInBody(this, token);
12921246
break;
12931247
}
@@ -1340,8 +1294,6 @@ export class Parser<T extends TreeAdapterTypeMap> implements TokenHandler, Stack
13401294
case InsertionMode.AFTER_HEAD:
13411295
case InsertionMode.TEXT:
13421296
case InsertionMode.IN_COLUMN_GROUP:
1343-
case InsertionMode.IN_SELECT:
1344-
case InsertionMode.IN_SELECT_IN_TABLE:
13451297
case InsertionMode.IN_FRAMESET:
13461298
case InsertionMode.AFTER_FRAMESET: {
13471299
this._insertCharacters(token);
@@ -2158,6 +2110,13 @@ function hrStartTagInBody<T extends TreeAdapterTypeMap>(p: Parser<T>, token: Tag
21582110
p._closePElement();
21592111
}
21602112

2113+
if (p.openElements.hasInScope($.SELECT)) {
2114+
p.openElements.generateImpliedEndTagsWithExclusion($.OPTGROUP);
2115+
if (p.openElements.hasInScope($.OPTION)) {
2116+
p._err(token, ERR.characterReferenceOutsideUnicodeRange); // TODO correct error type
2117+
}
2118+
}
2119+
21612120
p._appendElement(token, NS.HTML);
21622121
p.framesetOk = false;
21632122
token.ackSelfClosing = true;
@@ -2202,26 +2161,45 @@ function rawTextStartTagInBody<T extends TreeAdapterTypeMap>(p: Parser<T>, token
22022161
}
22032162

22042163
function selectStartTagInBody<T extends TreeAdapterTypeMap>(p: Parser<T>, token: TagToken): void {
2164+
if (p.openElements.hasInScope($.SELECT)) {
2165+
p._err(token, ERR.characterReferenceOutsideUnicodeRange); // TODO correct error type
2166+
p.openElements.popUntilTagNamePopped($.SELECT);
2167+
}
2168+
22052169
p._reconstructActiveFormattingElements();
22062170
p._insertElement(token, NS.HTML);
22072171
p.framesetOk = false;
2172+
}
2173+
2174+
function optionStartTagInBody<T extends TreeAdapterTypeMap>(p: Parser<T>, token: TagToken): void {
2175+
if (p.openElements.hasInScope($.SELECT)) {
2176+
p.openElements.generateImpliedEndTagsWithExclusion($.OPTGROUP);
2177+
if (p.openElements.hasInScope($.OPTION)) {
2178+
p._err(token, ERR.characterReferenceOutsideUnicodeRange); // TODO correct error type
2179+
}
2180+
} else {
2181+
if (p.openElements.currentTagId === $.OPTION) {
2182+
p.openElements.pop();
2183+
}
2184+
p._reconstructActiveFormattingElements();
2185+
}
22082186

2209-
p.insertionMode =
2210-
p.insertionMode === InsertionMode.IN_TABLE ||
2211-
p.insertionMode === InsertionMode.IN_CAPTION ||
2212-
p.insertionMode === InsertionMode.IN_TABLE_BODY ||
2213-
p.insertionMode === InsertionMode.IN_ROW ||
2214-
p.insertionMode === InsertionMode.IN_CELL
2215-
? InsertionMode.IN_SELECT_IN_TABLE
2216-
: InsertionMode.IN_SELECT;
2187+
p._insertElement(token, NS.HTML);
22172188
}
22182189

22192190
function optgroupStartTagInBody<T extends TreeAdapterTypeMap>(p: Parser<T>, token: TagToken): void {
2220-
if (p.openElements.currentTagId === $.OPTION) {
2221-
p.openElements.pop();
2191+
if (p.openElements.hasInScope($.SELECT)) {
2192+
p.openElements.generateImpliedEndTags();
2193+
if (p.openElements.hasInScope($.OPTION) || p.openElements.hasInScope($.OPTGROUP)) {
2194+
p._err(token, ERR.characterReferenceOutsideUnicodeRange); // TODO correct error type
2195+
}
2196+
} else {
2197+
if (p.openElements.currentTagId == $.OPTION) {
2198+
p.openElements.pop();
2199+
}
2200+
p._reconstructActiveFormattingElements();
22222201
}
22232202

2224-
p._reconstructActiveFormattingElements();
22252203
p._insertElement(token, NS.HTML);
22262204
}
22272205

@@ -2444,7 +2422,10 @@ function startTagInBody<T extends TreeAdapterTypeMap>(p: Parser<T>, token: TagTo
24442422
selectStartTagInBody(p, token);
24452423
break;
24462424
}
2447-
case $.OPTION:
2425+
case $.OPTION: {
2426+
optionStartTagInBody(p, token);
2427+
break;
2428+
}
24482429
case $.OPTGROUP: {
24492430
optgroupStartTagInBody(p, token);
24502431
break;
@@ -3269,155 +3250,6 @@ function endTagInCell<T extends TreeAdapterTypeMap>(p: Parser<T>, token: TagToke
32693250
}
32703251
}
32713252

3272-
// The "in select" insertion mode
3273-
//------------------------------------------------------------------
3274-
function startTagInSelect<T extends TreeAdapterTypeMap>(p: Parser<T>, token: TagToken): void {
3275-
switch (token.tagID) {
3276-
case $.HTML: {
3277-
startTagInBody(p, token);
3278-
break;
3279-
}
3280-
case $.OPTION: {
3281-
if (p.openElements.currentTagId === $.OPTION) {
3282-
p.openElements.pop();
3283-
}
3284-
3285-
p._insertElement(token, NS.HTML);
3286-
break;
3287-
}
3288-
case $.OPTGROUP: {
3289-
if (p.openElements.currentTagId === $.OPTION) {
3290-
p.openElements.pop();
3291-
}
3292-
3293-
if (p.openElements.currentTagId === $.OPTGROUP) {
3294-
p.openElements.pop();
3295-
}
3296-
3297-
p._insertElement(token, NS.HTML);
3298-
break;
3299-
}
3300-
case $.HR: {
3301-
if (p.openElements.currentTagId === $.OPTION) {
3302-
p.openElements.pop();
3303-
}
3304-
3305-
if (p.openElements.currentTagId === $.OPTGROUP) {
3306-
p.openElements.pop();
3307-
}
3308-
3309-
p._appendElement(token, NS.HTML);
3310-
token.ackSelfClosing = true;
3311-
break;
3312-
}
3313-
case $.INPUT:
3314-
case $.KEYGEN:
3315-
case $.TEXTAREA:
3316-
case $.SELECT: {
3317-
if (p.openElements.hasInSelectScope($.SELECT)) {
3318-
p.openElements.popUntilTagNamePopped($.SELECT);
3319-
p._resetInsertionMode();
3320-
3321-
if (token.tagID !== $.SELECT) {
3322-
p._processStartTag(token);
3323-
}
3324-
}
3325-
break;
3326-
}
3327-
case $.SCRIPT:
3328-
case $.TEMPLATE: {
3329-
startTagInHead(p, token);
3330-
break;
3331-
}
3332-
default:
3333-
// Do nothing
3334-
}
3335-
}
3336-
3337-
function endTagInSelect<T extends TreeAdapterTypeMap>(p: Parser<T>, token: TagToken): void {
3338-
switch (token.tagID) {
3339-
case $.OPTGROUP: {
3340-
if (
3341-
p.openElements.stackTop > 0 &&
3342-
p.openElements.currentTagId === $.OPTION &&
3343-
p.openElements.tagIDs[p.openElements.stackTop - 1] === $.OPTGROUP
3344-
) {
3345-
p.openElements.pop();
3346-
}
3347-
3348-
if (p.openElements.currentTagId === $.OPTGROUP) {
3349-
p.openElements.pop();
3350-
}
3351-
break;
3352-
}
3353-
case $.OPTION: {
3354-
if (p.openElements.currentTagId === $.OPTION) {
3355-
p.openElements.pop();
3356-
}
3357-
break;
3358-
}
3359-
case $.SELECT: {
3360-
if (p.openElements.hasInSelectScope($.SELECT)) {
3361-
p.openElements.popUntilTagNamePopped($.SELECT);
3362-
p._resetInsertionMode();
3363-
}
3364-
break;
3365-
}
3366-
case $.TEMPLATE: {
3367-
templateEndTagInHead(p, token);
3368-
break;
3369-
}
3370-
default:
3371-
// Do nothing
3372-
}
3373-
}
3374-
3375-
// The "in select in table" insertion mode
3376-
//------------------------------------------------------------------
3377-
function startTagInSelectInTable<T extends TreeAdapterTypeMap>(p: Parser<T>, token: TagToken): void {
3378-
const tn = token.tagID;
3379-
3380-
if (
3381-
tn === $.CAPTION ||
3382-
tn === $.TABLE ||
3383-
tn === $.TBODY ||
3384-
tn === $.TFOOT ||
3385-
tn === $.THEAD ||
3386-
tn === $.TR ||
3387-
tn === $.TD ||
3388-
tn === $.TH
3389-
) {
3390-
p.openElements.popUntilTagNamePopped($.SELECT);
3391-
p._resetInsertionMode();
3392-
p._processStartTag(token);
3393-
} else {
3394-
startTagInSelect(p, token);
3395-
}
3396-
}
3397-
3398-
function endTagInSelectInTable<T extends TreeAdapterTypeMap>(p: Parser<T>, token: TagToken): void {
3399-
const tn = token.tagID;
3400-
3401-
if (
3402-
tn === $.CAPTION ||
3403-
tn === $.TABLE ||
3404-
tn === $.TBODY ||
3405-
tn === $.TFOOT ||
3406-
tn === $.THEAD ||
3407-
tn === $.TR ||
3408-
tn === $.TD ||
3409-
tn === $.TH
3410-
) {
3411-
if (p.openElements.hasInTableScope(tn)) {
3412-
p.openElements.popUntilTagNamePopped($.SELECT);
3413-
p._resetInsertionMode();
3414-
p.onEndTag(token);
3415-
}
3416-
} else {
3417-
endTagInSelect(p, token);
3418-
}
3419-
}
3420-
34213253
// The "in template" insertion mode
34223254
//------------------------------------------------------------------
34233255
function startTagInTemplate<T extends TreeAdapterTypeMap>(p: Parser<T>, token: TagToken): void {

‎packages/parse5/lib/parser/open-element-stack.test.ts

-17
Original file line numberDiff line numberDiff line change
@@ -409,23 +409,6 @@ generateTestsForEachTreeAdapter('open-element-stack', (treeAdapter) => {
409409
assert.ok(stack.hasTableBodyContextInTableScope());
410410
});
411411

412-
test('Has element in select scope', () => {
413-
const stack = new OpenElementStack(treeAdapter.createDocument(), treeAdapter, stackHandler);
414-
415-
assert.ok(stack.hasInSelectScope($.P));
416-
417-
stack.push(createElement(TN.HTML), $.HTML);
418-
stack.push(createElement(TN.DIV), $.DIV);
419-
assert.ok(!stack.hasInSelectScope($.P));
420-
421-
stack.push(createElement(TN.P), $.P);
422-
stack.push(createElement(TN.OPTION), $.OPTION);
423-
assert.ok(stack.hasInSelectScope($.P));
424-
425-
stack.push(createElement(TN.DIV), $.DIV);
426-
assert.ok(!stack.hasInSelectScope($.P));
427-
});
428-
429412
test('Generate implied end tags', () => {
430413
const stack = new OpenElementStack(treeAdapter.createDocument(), treeAdapter, stackHandler);
431414

0 commit comments

Comments
 (0)