-// todo remove refs and lens, js, ls
-// run test suite!
+// todo remove unused variables
+// todo remove debug log, or make a way to access it
// Copyright 2015 Jason Woofenden
// This file implements an HTML5 parser
//
// See README.md for how to run this file in the browser or in node.js.
//
-// This file exports a single useful function: parse_tml, and some constants
+// This file exports a single useful function: parse, and some constants
// (see the bottom of this file for those.)
//
// Call it like this:
//
// peach_parser.parse("<p><b>hi</p>", {fragment: "body"})
//
-// return value is an array of Nodes, see "class Node" below.
+// return value is an array of Nodes, A Node contains:
+// type: one of: "tag", "text", "comment", "doctype"
+// text: contents for text/comment nodes
+// attrs: object of attributes, eg {href: "#main"}
+// children: array of Nodes
+// namespace: one of: "html", "mathml", "svg"
+// parent: another Node or null
// This code is a work in progress, eg try search this file for "fixfull",
// "TODO" and "FIXME"
// 2: c "next", "after", "lower", "below"
// 1: b
// 0: a "end of the list", "current node", "bottommost", "last"
+(function () {
+
+var NS_HTML, NS_MATHML, NS_SVG, QUIRKS_LIMITED, QUIRKS_NO, QUIRKS_YES, TYPE_AAA_BOOKMARK, TYPE_AFE_MARKER, TYPE_COMMENT, TYPE_DOCTYPE, TYPE_END_TAG, TYPE_EOF, TYPE_START_TAG, TYPE_TAG, TYPE_TEXT, _decode_named_char_ref, adjust_foreign_attributes, adjust_mathml_attributes, adjust_svg_attributes, adp_els, alnum, context, debug_log, debug_log_each, debug_log_reset, decode_named_char_ref, decode_named_char_ref_cache, decode_named_char_ref_el, digits, el_is_special, el_is_special_not_adp, end_tag_implied, exports, foreign_attr_fixes, formatting_elements, foster_parenting_targets, from_code_point, g_debug_log, h_tags, hex_chars, is_html_integration, is_input_hidden_tok, is_lc_alpha, is_mathml_text_integration_point, is_space, is_space_tok, is_uc_alpha, lc_alpha, legacy_char_refs, mathml_elements, mathml_text_integration, new_aaa_bookmark, new_afe_marker, new_character_token, new_comment_token, new_doctype_token, new_element, new_end_tag, new_eof_token, new_open_tag, new_text_node, parse_html, prev_node_id, quirks_yes_pi_prefixes, space_chars, special_elements, svg_attribute_fixes, svg_elements, svg_name_fixes, tag_name_chars, uc_alpha, unicode_fixes, whitespace_chars
if ((typeof module) !== 'undefined' && (module.exports != null)) {
context = 'module'
- exports = module.exports
} else {
context = 'browser'
window.peach_parser = {}
- exports = window.peach_parser
}
from_code_point = function (x) {
this.name = args.name != null ? args.name : '' // tag name
this.text = args.text != null ? args.text : '' // contents for text/comment nodes
this.attrs = args.attrs != null ? args.attrs : {}
- this.attrs_a = args.attr_k != null ? args.attr_k : [] // attrs in progress, TYPE_START_TAG only
this.children = args.children != null ? args.children : []
this.namespace = args.namespace != null ? args.namespace : NS_HTML
this.parent = args.parent != null ? args.parent : null
+ // private:
+ this.attrs_a = args.attr_k != null ? args.attr_k : [] // attrs in progress, TYPE_START_TAG only
this.token = args.token != null ? args.token : null
this.flags = args.flags != null ? args.flags : {}
if (args.id != null) {
}
parse_html = function (args_html, args) {
- var adjusted_current_node, adjusted_insertion_location, adoption_agency, afe, afe_push, afe_push_marker, button_scopers, clear_afe_to_marker, clear_stack_to_table_body_context, clear_stack_to_table_context, clear_stack_to_table_row_context, clear_to_table_body_stopers, clear_to_table_row_stopers, clear_to_table_stopers, close_p_element, close_p_if_in_button_scope, close_the_cell, context_element, cur, doc, eat_next_token_if_newline, el_is_in_scope, flag_foster_parenting, flag_fragment_parsing, flag_frameset_ok, flag_parsing, flag_scripting, form_element_pointer, fragment_root, generate_implied_end_tags, has_color_face_or_size, head_element_pointer, in_body_any_other_end_tag, in_foreign_content, in_foreign_content_end_script, in_foreign_content_other_start, ins_mode, ins_mode_after_after_body, ins_mode_after_after_frameset, ins_mode_after_body, ins_mode_after_frameset, ins_mode_after_head, ins_mode_after_head_else, ins_mode_before_head, ins_mode_before_html, ins_mode_in_body, ins_mode_in_caption, ins_mode_in_cell, ins_mode_in_column_group, ins_mode_in_frameset, ins_mode_in_head, ins_mode_in_head_else, ins_mode_in_head_noscript, ins_mode_in_head_noscript_else, ins_mode_in_row, ins_mode_in_select, ins_mode_in_select_in_table, ins_mode_in_table, ins_mode_in_table_body, ins_mode_in_table_else, ins_mode_in_table_text, ins_mode_in_template, ins_mode_initial, ins_mode_text, insert_character, insert_comment, insert_foreign_element, insert_html_element, is_appropriate_end_tag, is_in_button_scope, is_in_li_scope, is_in_scope, is_in_scope_x, is_in_scope_x_y, is_in_select_scope, is_in_table_scope, is_quirks_limited_doctype, is_quirks_yes_doctype, li_scopers, open_els, original_ins_mode, parse_character_reference, parse_error, parse_generic_raw_text, parse_generic_rcdata_text, parse_init, parse_main_loop, pending_table_character_tokens, process_token, reconstruct_afe, ref, reset_ins_mode, standard_scopers, stop_parsing, table_scopers, template_ins_modes, template_tag_is_open, temporary_buffer, tok_cur_tag, tok_state, tok_state_after_attribute_name, tok_state_after_attribute_value_quoted, tok_state_after_doctype_name, tok_state_after_doctype_public_identifier, tok_state_after_doctype_public_keyword, tok_state_after_doctype_system_identifier, tok_state_after_doctype_system_keyword, tok_state_attribute_name, tok_state_attribute_value_double_quoted, tok_state_attribute_value_single_quoted, tok_state_attribute_value_unquoted, tok_state_before_attribute_name, tok_state_before_attribute_value, tok_state_before_doctype_name, tok_state_before_doctype_public_identifier, tok_state_before_doctype_system_identifier, tok_state_between_doctype_public_and_system_identifiers, tok_state_bogus_comment, tok_state_bogus_doctype, tok_state_cdata_section, tok_state_comment, tok_state_comment_end, tok_state_comment_end_bang, tok_state_comment_end_dash, tok_state_comment_start, tok_state_comment_start_dash, tok_state_data, tok_state_doctype, tok_state_doctype_name, tok_state_doctype_public_identifier_double_quoted, tok_state_doctype_public_identifier_single_quoted, tok_state_doctype_system_identifier_double_quoted, tok_state_doctype_system_identifier_single_quoted, tok_state_end_tag_open, tok_state_markup_declaration_open, tok_state_plaintext, tok_state_rawtext, tok_state_rawtext_end_tag_name, tok_state_rawtext_end_tag_open, tok_state_rawtext_less_than_sign, tok_state_rcdata, tok_state_rcdata_end_tag_name, tok_state_rcdata_end_tag_open, tok_state_rcdata_less_than_sign, tok_state_script_data, tok_state_script_data_double_escape_end, tok_state_script_data_double_escape_start, tok_state_script_data_double_escaped, tok_state_script_data_double_escaped_dash, tok_state_script_data_double_escaped_dash_dash, tok_state_script_data_double_escaped_less_than_sign, tok_state_script_data_end_tag_name, tok_state_script_data_end_tag_open, tok_state_script_data_escape_start, tok_state_script_data_escape_start_dash, tok_state_script_data_escaped, tok_state_script_data_escaped_dash, tok_state_script_data_escaped_dash_dash, tok_state_script_data_escaped_end_tag_name, tok_state_script_data_escaped_end_tag_open, tok_state_script_data_escaped_less_than_sign, tok_state_script_data_less_than_sign, tok_state_self_closing_start_tag, tok_state_tag_name, tok_state_tag_open, token_to_element, txt
+ var adjusted_current_node, adjusted_insertion_location, adoption_agency, afe, afe_push, afe_push_marker, button_scopers, clear_afe_to_marker, clear_stack_to_table_body_context, clear_stack_to_table_context, clear_stack_to_table_row_context, clear_to_table_body_stopers, clear_to_table_row_stopers, clear_to_table_stopers, close_p_element, close_p_if_in_button_scope, close_the_cell, context_element, cur, doc, eat_next_token_if_newline, el_is_in_scope, flag_foster_parenting, flag_fragment_parsing, flag_frameset_ok, flag_parsing, flag_scripting, form_element_pointer, fragment_root, generate_implied_end_tags, has_color_face_or_size, head_element_pointer, in_body_any_other_end_tag, in_foreign_content, in_foreign_content_end_script, in_foreign_content_other_start, ins_mode, ins_mode_after_after_body, ins_mode_after_after_frameset, ins_mode_after_body, ins_mode_after_frameset, ins_mode_after_head, ins_mode_after_head_else, ins_mode_before_head, ins_mode_before_html, ins_mode_in_body, ins_mode_in_caption, ins_mode_in_cell, ins_mode_in_column_group, ins_mode_in_frameset, ins_mode_in_head, ins_mode_in_head_else, ins_mode_in_head_noscript, ins_mode_in_head_noscript_else, ins_mode_in_row, ins_mode_in_select, ins_mode_in_select_in_table, ins_mode_in_table, ins_mode_in_table_body, ins_mode_in_table_else, ins_mode_in_table_text, ins_mode_in_template, ins_mode_initial, ins_mode_text, insert_character, insert_comment, insert_foreign_element, insert_html_element, is_appropriate_end_tag, is_in_button_scope, is_in_li_scope, is_in_scope, is_in_scope_x, is_in_scope_x_y, is_in_select_scope, is_in_table_scope, is_quirks_limited_doctype, is_quirks_yes_doctype, li_scopers, open_els, original_ins_mode, parse_character_reference, parse_error, parse_generic_raw_text, parse_generic_rcdata_text, parse_init, parse_main_loop, pending_table_character_tokens, process_token, reconstruct_afe, reset_ins_mode, standard_scopers, stop_parsing, table_scopers, template_ins_modes, template_tag_is_open, temporary_buffer, tok_cur_tag, tok_state, tok_state_after_attribute_name, tok_state_after_attribute_value_quoted, tok_state_after_doctype_name, tok_state_after_doctype_public_identifier, tok_state_after_doctype_public_keyword, tok_state_after_doctype_system_identifier, tok_state_after_doctype_system_keyword, tok_state_attribute_name, tok_state_attribute_value_double_quoted, tok_state_attribute_value_single_quoted, tok_state_attribute_value_unquoted, tok_state_before_attribute_name, tok_state_before_attribute_value, tok_state_before_doctype_name, tok_state_before_doctype_public_identifier, tok_state_before_doctype_system_identifier, tok_state_between_doctype_public_and_system_identifiers, tok_state_bogus_comment, tok_state_bogus_doctype, tok_state_cdata_section, tok_state_comment, tok_state_comment_end, tok_state_comment_end_bang, tok_state_comment_end_dash, tok_state_comment_start, tok_state_comment_start_dash, tok_state_data, tok_state_doctype, tok_state_doctype_name, tok_state_doctype_public_identifier_double_quoted, tok_state_doctype_public_identifier_single_quoted, tok_state_doctype_system_identifier_double_quoted, tok_state_doctype_system_identifier_single_quoted, tok_state_end_tag_open, tok_state_markup_declaration_open, tok_state_plaintext, tok_state_rawtext, tok_state_rawtext_end_tag_name, tok_state_rawtext_end_tag_open, tok_state_rawtext_less_than_sign, tok_state_rcdata, tok_state_rcdata_end_tag_name, tok_state_rcdata_end_tag_open, tok_state_rcdata_less_than_sign, tok_state_script_data, tok_state_script_data_double_escape_end, tok_state_script_data_double_escape_start, tok_state_script_data_double_escaped, tok_state_script_data_double_escaped_dash, tok_state_script_data_double_escaped_dash_dash, tok_state_script_data_double_escaped_less_than_sign, tok_state_script_data_end_tag_name, tok_state_script_data_end_tag_open, tok_state_script_data_escape_start, tok_state_script_data_escape_start_dash, tok_state_script_data_escaped, tok_state_script_data_escaped_dash, tok_state_script_data_escaped_dash_dash, tok_state_script_data_escaped_end_tag_name, tok_state_script_data_escaped_end_tag_open, tok_state_script_data_escaped_less_than_sign, tok_state_script_data_less_than_sign, tok_state_self_closing_start_tag, tok_state_tag_name, tok_state_tag_open, token_to_element, txt
if (args == null) {
args = {}
}
// http://www.w3.org/TR/html5/syntax.html#push-onto-the-list-of-active-formatting-elements
// "Noah's Ark clause" but with three
afe_push = function (new_el) {
- var attrs_match, el, i, j, k, len, matches, ref, ref1, v
+ var attrs_match, el, i, j, k, matches, v
matches = 0
for (i = 0; i < afe.length; ++i) {
el = afe[i]
// http://www.w3.org/TR/html5/syntax.html#misnested-tags:-b-p-/b-/p
// http://www.w3.org/TR/html5/syntax.html#unclosed-formatting-elements
adoption_agency = function (subject) {
- var aa, ab, ac, ad, ae, af, bookmark, c, ca, dest, el, fb, fb_of_open_els, fe, fe_of_afe, fe_of_open_els, i, in_afe, in_open_els, inner, j, l, last_node, len, len1, len10, len11, len12, len13, len14, len15, len16, len17, len2, len3, len4, len5, len6, len7, len8, len9, m, n, new_node, node, node_above, node_in_afe, node_next, o, outer, q, r, ref, ref1, s, t, u, w, y, z
+ var aa, ab, ac, ad, ae, af, bookmark, c, ca, dest, el, fb, fb_of_open_els, fe, fe_of_afe, fe_of_open_els, i, in_afe, in_open_els, inner, j, l, last_node, len, len1, len10, len11, len12, len13, len14, len15, len16, len17, len2, len3, len4, len5, len6, len7, len8, len9, m, n, new_node, node, node_above, node_in_afe, node_next, o, outer, q, r, s, t, u, w, y, z
// this block implements tha W3C spec
// # 1. If the current node is an HTML element whose tag name is subject,
// # then run these substeps:
// the list of active formatting elements, then remove node from
// the list of active formatting elements.
node_in_afe = false
- for (i = 0; i < afe.length; ++i) {
- t = afe[i]
- if (t === node) {
- if (inner > 3) {
- afe.splice(i, 1)
- } else {
- node_in_afe = true
- }
- break
+ if ((i = afe.indexOf(node)) !== -1) {
+ if (inner > 3) {
+ afe.splice(i, 1)
+ } else {
+ node_in_afe = true
}
}
// 6. If node is not in the list of active formatting elements,
// then remove node from the stack of open elements and then go
// back to the step labeled inner loop.
if (!node_in_afe) {
- for (i = 0; i < open_els.length; ++i) {
- t = open_els[i]
- if (t === node) {
- node_above = open_els[i + 1]
- open_els.splice(i, 1)
- break
- }
+ if ((i = open_els.indexOf(node)) !== -1) {
+ node_above = open_els[i + 1]
+ open_els.splice(i, 1)
}
continue
}
// elements with an entry for the new element, and let node be
// the new element.
new_node = token_to_element(node.token, NS_HTML, ca)
- for (i = 0; i < afe.length; ++i) {
- t = afe[i]
- if (t === node) {
- afe[i] = new_node
- break
- }
+ if ((i = afe.indexOf(node)) !== -1) {
+ afe[i] = new_node
}
- for (i = 0; i < open_els.length; ++i) {
- t = open_els[i]
- if (t === node) {
- node_above = open_els[i + 1]
- open_els[i] = new_node
- break
- }
+ if ((i = open_els.indexOf(node)) !== -1) {
+ node_above = open_els[i + 1]
+ open_els[i] = new_node
}
node = new_node
// 8. If last node is furthest block, then move the
// aforementioned bookmark to be immediately after the new node
// in the list of active formatting elements.
if (last_node === fb) {
- for (i = 0; i < afe.length; ++i) {
- t = afe[i]
- if (t === bookmark) {
- afe.splice(i, 1)
- break
- }
+ if ((i = afe.indexOf(bookmark)) !== -1) {
+ afe.splice(i, 1)
}
- for (i = 0; i < afe.length; ++i) {
- t = afe[i]
- if (t === node) {
- // "after" means lower
- afe.splice(i, 0, bookmark) // "after as <-
- break
- }
+ if ((i = afe.indexOf(node)) !== -1) {
+ // "after" means lower
+ afe.splice(i, 0, bookmark) // "after as <-
}
}
// 9. Insert last node into node, first removing it from its
// previous parent node if any.
if (last_node.parent != null) {
- for (i = 0; i < last_node.parent.children.length; ++i) {
- c = last_node.parent.children[i]
- if (c === last_node) {
- last_node.parent.children.splice(i, 1)
- break
- }
+ if ((i = last_node.parent.children.indexOf(last_node)) !== -1) {
+ last_node.parent.children.splice(i, 1)
}
}
node.children.push(last_node)
// * last_node is fb
// * last_node is still in the tree (not a duplicate)
if (last_node.parent != null) {
- for (i = 0; i < last_node.parent.children.length; ++i) {
- c = last_node.parent.children[i]
- if (c === last_node) {
- last_node.parent.children.splice(i, 1)
- break
- }
+ if ((i = last_node.parent.children.indexOf(last_node)) !== -1) {
+ last_node.parent.children.splice(i, 1)
}
}
// can't use standard insert token thing, because it's already in
// elements, and insert the new element into the list of active
// formatting elements at the position of the aforementioned
// bookmark.
- for (i = 0; i < afe.length; ++i) {
- t = afe[i]
- if (t === fe) {
- afe.splice(i, 1)
- break
- }
+ if ((i = afe.indexOf(fe)) !== -1) {
+ afe.splice(i, 1)
}
- for (i = 0; i < afe.length; ++i) {
- t = afe[i]
- if (t === bookmark) {
- afe[i] = new_element
- break
- }
+ if ((i = afe.indexOf(bookmark)) !== -1) {
+ afe[i] = new_element
}
// 19. Remove formatting element from the stack of open elements,
// and insert the new element into the stack of open elements
// immediately below the position of furthest block in that stack.
- for (i = 0; i < open_els.length; ++i) {
- t = open_els[i]
- if (t === fe) {
- open_els.splice(i, 1)
- break
- }
+ if ((i = open_els.indexOf(fe)) !== -1) {
+ open_els.splice(i, 1)
}
- for (i = 0; i < open_els.length; ++i) {
- t = open_els[i]
- if (t === fb) {
- open_els.splice(i, 0, new_element)
- break
- }
+ if ((i = open_els.indexOf(fb)) !== -1) {
+ open_els.splice(i, 0, new_element)
}
// 20. Jump back to the step labeled outer loop.
}
// http://www.w3.org/TR/html5/syntax.html#creating-and-inserting-nodes
// http://www.w3.org/TR/html5/syntax.html#appropriate-place-for-inserting-a-node
adjusted_insertion_location = function (override_target) {
- var c, el, i, j, l, last_table, last_table_i, last_template, last_template_i, len, len1, len2, m, previous_element, ref, target, target_i
+ var c, el, i, j, l, last_table, last_table_i, last_template, last_template_i, len, len1, len2, m, previous_element, target, target_i
// 1. If there was an override target specified, then let target be the
// override target.
if (override_target != null) {
position = adjusted_insertion_location()
}
position[0].children.splice(position[1], 0, t)
+ t.parent = position[0]
return
}
el.flag('parser-inserted', true)
// fixfull frament case
ail[0].children.splice(ail[1], 0, el)
+ el.parent = ail[0]
open_els.unshift(el)
tok_state = tok_state_script_data
original_ins_mode = ins_mode // make sure orig... is defined
}
}
ins_mode_in_body = function (t) {
- var a, aa, ab, ac, el, found, h_in_scope, i, input_el, j, l, len, len1, len10, len11, len12, len13, len14, len2, len3, len4, len5, len6, len7, len8, len9, m, n, node, o, ok_tags, prompt, q, r, ref, ref1, ref2, ref3, ref4, root_attrs, s, second, second_i, u, w, y, z
+ var a, aa, ab, ac, el, found, h_in_scope, i, input_el, j, l, len, len1, len10, len11, len12, len13, len14, len2, len3, len4, len5, len6, len7, len8, len9, m, n, node, o, ok_tags, prompt, q, r, root_attrs, s, second, second_i, u, w, y, z
if (t.type === TYPE_TEXT && t.text === "\u0000") {
parse_error()
return
return doc.children
}
-exports.parse = parse_html
-exports.Node = Node
-exports.debug_log_reset = debug_log_reset
-exports.debug_log_each = debug_log_each
-exports.TYPE_TAG = TYPE_TAG
-exports.TYPE_TEXT = TYPE_TEXT
-exports.TYPE_COMMENT = TYPE_COMMENT
-exports.TYPE_DOCTYPE = TYPE_DOCTYPE
-exports.NS_HTML = NS_HTML
-exports.NS_MATHML = NS_MATHML
-exports.NS_SVG = NS_SVG
-exports.QUIRKS_NO = QUIRKS_NO
-exports.QUIRKS_LIMITED = QUIRKS_LIMITED
-exports.QUIRKS_YES = QUIRKS_YES
+var this_module = {
+ parse: parse_html,
+ Node: Node,
+}
+
+if (context === 'module') {
+ module.exports = this_module
+} else {
+ window.peach_parser = this_module
+}
+
+}).call(this)