htmlparser: add parse error handling

1 files changed, 70 insertions, 0 deletions

diff --git a/src/html/parseerror.nim b/src/html/parseerror.nim
new file mode 100644
index 00000000..d99b2fed
--- /dev/null
+++ b/src/html/parseerror.nim
@@ -0,0 +1,70 @@
+type ParseError* = enum
+  #TODO write a description for all error codes
+  ABRUPT_CLOSING_OF_EMPTY_COMMENT
+  ABRUPT_DOCTYPE_PUBLIC_IDENTIFIER
+  ABRUPT_DOCTYPE_SYSTEM_IDENTIFIER
+  ABSENCE_OF_DIGITS_IN_NUMERIC_CHARACTER_REFERENCE
+  CDATA_IN_HTML_CONTENT
+  CHARACTER_REFERENCE_OUTSIDE_UNICODE_RANGE
+  CONTROL_CHARACTER_IN_INPUT_STREAM
+  CONTROL_CHARACTER_REFERENCE
+  END_TAG_WITH_ATTRIBUTES
+  DUPLICATE_ATTRIBUTE
+  END_TAG_WITH_TRAILING_SOLIDUS
+  EOF_BEFORE_TAG_NAME
+  EOF_IN_CDATA
+  EOF_IN_COMMENT
+  EOF_IN_DOCTYPE
+  EOF_IN_SCRIPT_HTML_COMMENT_LIKE_TEXT
+  EOF_IN_TAG
+  INCORRECTLY_CLOSED_COMMENT
+  INCORRECTLY_OPENED_COMMENT
+  INVALID_CHARACTER_SEQUENCE_AFTER_DOCTYPE_NAME
+  INVALID_FIRST_CHARACTER_OF_TAG_NAME
+  MISSING_ATTRIBUTE_VALUE
+  MISSING_DOCTYPE_NAME
+  MISSING_DOCTYPE_PUBLIC_IDENTIFIER
+  MISSING_DOCTYPE_SYSTEM_IDENTIFIER
+  MISSING_END_TAG_NAME
+  MISSING_QUOTE_BEFORE_DOCTYPE_PUBLIC_IDENTIFIER
+  MISSING_QUOTE_BEFORE_DOCTYPE_SYSTEM_IDENTIFIER
+  MISSING_SEMICOLON_AFTER_CHARACTER_REFERENCE
+  MISSING_WHITESPACE_AFTER_DOCTYPE_PUBLIC_KEYWORD
+  MISSING_WHITESPACE_AFTER_DOCTYPE_SYSTEM_KEYWORD
+  MISSING_WHITESPACE_BEFORE_DOCTYPE_NAME
+  MISSING_WHITESPACE_BETWEEN_ATTRIBUTES
+  MISSING_WHITESPACE_BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS
+  NESTED_COMMENT
+  NONCHARACTER_CHARACTER_REFERENCE
+  NONCHARACTER_IN_INPUT_STREAM
+  NON_VOID_HTML_ELEMENT_START_TAG_WITH_TRAILING_SOLIDUS
+  NULL_CHARACTER_REFERENCE
+  SURROGATE_CHARACTER_REFERENCE
+  SURROGATE_IN_INPUT_STREAM
+  UNEXPECTED_CHARACTER_AFTER_DOCTYPE_SYSTEM_IDENTIFIER
+  UNEXPECTED_CHARACTER_IN_ATTRIBUTE_NAME
+  UNEXPECTED_CHARACTER_IN_UNQUOTED_ATTRIBUTE_VALUE
+  UNEXPECTED_EQUALS_SIGN_BEFORE_ATTRIBUTE_NAME
+  UNEXPECTED_NULL_CHARACTER
+  UNEXPECTED_QUESTION_MARK_INSTEAD_OF_TAG_NAME
+  UNEXPECTED_SOLIDUS_IN_TAG
+  UNKNOWN_NAMED_CHARACTER_REFERENCE
+  LAST_SPECIFIED_ERROR # never returned
+  # From here on, error code names have not been specified by the standard.
+  MISMATCHED_TAGS = "Mismatched start and end tags"
+  INVALID_DOCTYPE = "Unrecognized document type"
+  UNEXPECTED_DOCTYPE = "Unexpected document type"
+  UNEXPECTED_INITIAL_TOKEN = "Unexpected token in initial state"
+  UNEXPECTED_START_TAG = "Unexpected start tag"
+  UNEXPECTED_END_TAG = "Unexpected end tag"
+  ELEMENT_NOT_IN_OPEN_ELEMENTS = "Element has not been added to open elements"
+  ELEMENT_NOT_IN_SCOPE = "Element not in appropriate scope"
+  ELEMENT_NOT_CURRENT_NODE = "Element is not current node"
+  #TODO merge with UNEXPECTED_NULL_CHARACTER?
+  UNEXPECTED_NULL = "Unexpected null character"
+  NESTED_TAGS = "Non-nestable nested tags"
+  UNEXPECTED_SPECIAL_ELEMENT = "Unexpected special element on open elements"
+  UNEXPECTED_EOF = "Unexpected end of file"
+  INVALID_TEXT_PARENT = "Invalid parent element for text node"
+  NON_SPACE_TABLE_TEXT = "Non-space table text"
+  UNEXPECTED_AFTER_BODY_TOKEN = "Unexpected token after body"