Merge pull request #25 from HTMLParseErrorWG/review-fixes2

inikulin · web-flow · commit 9ec9f264702f · 2017-06-08T20:23:46.000+03:00
Review fixes2
diff --git a/tokenizer/README.md b/tokenizer/README.md
@@ -14,13 +14,16 @@ Basic Structure
         "output": [expected_output_tokens],
         "initialStates": [initial_states],
         "lastStartTag": last_start_tag,
-        "ignoreErrorOrder": ignore_error_order
+        "errors": [parse_errors]
         }
     ]}
 
 Multiple tests per file are allowed simply by adding more objects to the
 "tests" list.
 
+Each parse error is an object that contains error `code` and one-based
+error location indices: `line` and `col`.
+
 `description`, `input` and `output` are always present. The other values
 are optional.
 
diff --git a/tokenizer/test4.test b/tokenizer/test4.test
@@ -355,7 +355,6 @@
 {"description":"U+0000 in lookahead region after non-matching character",
 "input":"<!doc>\u0000",
 "output":[["Comment", "doc"], ["Character", "\u0000"]],
-"ignoreErrorOrder":true,
 "errors":[
     { "code": "incorrectly-opened-comment", "line": 1, "col": 3 },
     { "code": "unexpected-null-character", "line": 1, "col": 7 }
@@ -364,15 +363,13 @@
 {"description":"U+0000 in lookahead region",
 "input":"<!doc\u0000",
 "output":[["Comment", "doc\uFFFD"]],
-"ignoreErrorOrder":true,
 "errors":[
     { "code": "incorrectly-opened-comment", "line": 1, "col": 3 }
 ]},
 
 {"description":"U+0080 in lookahead region",
 "input":"<!doc\u0080",
 "output":[["Comment", "doc\u0080"]],
-"ignoreErrorOrder":true,
 "errors":[
     { "code": "incorrectly-opened-comment", "line": 1, "col": 3 },
     { "code": "control-character-in-input-stream", "line": 1, "col": 6 }
@@ -381,7 +378,6 @@
 {"description":"U+FDD1 in lookahead region",
 "input":"<!doc\uFDD1",
 "output":[["Comment", "doc\uFDD1"]],
-"ignoreErrorOrder":true,
 "errors":[
     { "code": "incorrectly-opened-comment", "line": 1, "col": 3 },
     { "code": "noncharacter-in-input-stream", "line": 1, "col": 6 }
@@ -390,7 +386,6 @@
 {"description":"U+1FFFF in lookahead region",
 "input":"<!doc\uD83F\uDFFF",
 "output":[["Comment", "doc\uD83F\uDFFF"]],
-"ignoreErrorOrder":true,
 "errors":[
     { "code": "incorrectly-opened-comment", "line": 1, "col": 3 },
     { "code": "noncharacter-in-input-stream", "line": 1, "col": 6 }
diff --git a/tokenizer/unicodeCharsProblematic.test b/tokenizer/unicodeCharsProblematic.test
@@ -3,7 +3,6 @@
 "doubleEscaped":true,
 "input": "\\uDFFF",
 "output":[["Character", "\\uDFFF"]],
-"ignoreErrorOrder":true,
 "errors":[
     { "code": "surrogate-in-input-stream", "line": 1, "col": 1 }
 ]},
@@ -12,7 +11,6 @@
 "doubleEscaped":true,
 "input": "\\uD800",
 "output":[["Character", "\\uD800"]],
-"ignoreErrorOrder":true,
 "errors":[
     { "code": "surrogate-in-input-stream", "line": 1, "col": 1 }
 ]},
@@ -21,7 +19,6 @@
 "doubleEscaped":true,
 "input": "a\\uDFFF",
 "output":[["Character", "a"], ["Character", "\\uDFFF"]],
-"ignoreErrorOrder":true,
 "errors":[
     { "code": "surrogate-in-input-stream", "line": 1, "col": 2 }
 ]},
@@ -30,15 +27,13 @@
 "doubleEscaped":true,
 "input": "\\uD800a",
 "output":[["Character", "\\uD800a"]],
-"ignoreErrorOrder":true,
 "errors":[
     { "code": "surrogate-in-input-stream", "line": 1, "col": 1 }
 ]},
 
 {"description":"CR followed by U+0000",
 "input":"\r\u0000",
 "output":[["Character", "\n"], ["Character", "\u0000"]],
-"ignoreErrorOrder":true,
 "errors":[
     { "code": "unexpected-null-character", "line": 2, "col": 1 }
 ]}
diff --git a/tokenizer/xmlViolation.test b/tokenizer/xmlViolation.test
@@ -2,12 +2,10 @@
 
 {"description":"Non-XML character",
 "input":"a\uFFFFb",
-"ignoreErrorOrder":true,
 "output":[["Character","a\uFFFDb"]]},
 
 {"description":"Non-XML space",
 "input":"a\u000Cb",
-"ignoreErrorOrder":true,
 "output":[["Character","a b"]]},
 
 {"description":"Double hyphen in comment",
diff --git a/tree-construction/doctype01.dat b/tree-construction/doctype01.dat
@@ -378,7 +378,7 @@
 |   <body>
 
 #data
-<!DOCTYPE root-element [SYSTEM OR PUBLIC FPI] "uri" [
+<!DOCTYPE root-element [SYSTEM OR PUBLIC FPI] "uri" [ 
 <!-- internal declarations -->
 ]>
 #errors
diff --git a/tree-construction/html5test-com.dat b/tree-construction/html5test-com.dat
@@ -12,7 +12,7 @@
 #data
 <div foo<bar=''>
 #errors
-(1,9): unexpected-character-in-attribute-name
+(1,9): invalid-character-in-attribute-name
 (1,16): expected-doctype-but-got-start-tag
 (1,16): expected-closing-tag-but-got-eof
 #new-errors
@@ -28,7 +28,7 @@
 <div foo=`bar`>
 #errors
 (1,10): equals-in-unquoted-attribute-value
-(1,14): missing-whitespace-between-attributes
+(1,14): unexpected-character-in-unquoted-attribute-value
 (1,15): expected-doctype-but-got-start-tag
 (1,15): expected-closing-tag-but-got-eof
 #new-errors
@@ -44,7 +44,7 @@
 #data
 <div \"foo=''>
 #errors
-(1,7): unexpected-character-in-attribute-name
+(1,7): invalid-character-in-attribute-name
 (1,14): expected-doctype-but-got-start-tag
 (1,14): expected-closing-tag-but-got-eof
 #new-errors
diff --git a/tree-construction/tests2.dat b/tree-construction/tests2.dat
@@ -234,7 +234,7 @@
 #data
 <!DOCTYPE html></b test<b &=&amp>X
 #errors
-(1,24): unexpected-character-in-attribute-name
+(1,24): invalid-character-in-attribute-name
 (1,32): named-entity-without-semicolon
 (1,33): attributes-in-end-tag
 (1,33): unexpected-end-tag-before-html
diff --git a/tree-construction/tests26.dat b/tree-construction/tests26.dat
@@ -265,7 +265,7 @@
 
 #errors
 (1,3): expected-doctype-but-got-start-tag
-(1,11): unexpected-character-in-attribute-name
+(1,11): invalid-character-in-attribute-name
 (1,12): unexpected-character-after-solidus-in-tag
 (1,21): unexpected-end-tag
 (2,0): expected-closing-tag-but-got-eof
diff --git a/tree-construction/webkit01.dat b/tree-construction/webkit01.dat
@@ -622,7 +622,7 @@ console.log("FOO<span>BAR</span>BAZ");
 #data
 <img <="" FAIL>
 #errors
-(1,6): unexpected-character-in-attribute-name
+(1,6): invalid-character-in-attribute-name
 (1,15): expected-doctype-but-got-start-tag
 #new-errors
 (1:6) unexpected-character-in-attribute-name
diff --git a/tree-construction/webkit02.dat b/tree-construction/webkit02.dat
@@ -59,7 +59,7 @@
 <html><body><img src="" border="0" alt="><div>A</div></body></html>
 #errors
 (1,6): expected-doctype-but-got-start-tag
-(1,67): eof-in-tag-double-quote
+(1,67): eof-in-attribute-value-double-quote
 #new-errors
 (1:68) eof-in-tag
 #document
