mirror of
https://gerrit.wikimedia.org/r/mediawiki/extensions/DiscussionTools
synced 2024-11-12 09:58:17 +00:00
14fb013515
PHP was counting UTF-8 bytes, JS was counting UTF-16 bytes. Both should have been counting codepoints (although it doesn't really matter as long as they both count the same things). I noticed the issue after adding some tests using the Cyrillic script, when one case had different results in PHP and JS: Id25b537fecd789640c209ff7f30e777455a3aece. Change-Id: Ic31240678f71ba48e6ec202126bf490cea12bb66
995 lines
30 KiB
JavaScript
995 lines
30 KiB
JavaScript
'use strict';
|
|
/* global $:off */
|
|
|
|
/**
|
|
* @external ThreadItem
|
|
*/
|
|
|
|
var
|
|
utils = require( './utils.js' ),
|
|
codePointLength = require( 'mediawiki.String' ).codePointLength,
|
|
CommentItem = require( './CommentItem.js' ),
|
|
HeadingItem = require( './HeadingItem.js' ),
|
|
// Hooks::getLocalData()
|
|
data = require( './parser/data.json' ),
|
|
moment = require( './lib/moment-timezone/moment-timezone-with-data-1970-2030.js' );
|
|
|
|
/**
|
|
* Utilities for detecting and parsing components of discussion pages: signatures, timestamps,
|
|
* comments and threads.
|
|
*
|
|
* @class mw.dt.Parser
|
|
*/
|
|
|
|
function Parser( rootNode ) {
|
|
this.rootNode = rootNode;
|
|
this.threadItems = null;
|
|
this.commentItems = null;
|
|
this.commentsById = null;
|
|
this.threads = null;
|
|
}
|
|
|
|
OO.initClass( Parser );
|
|
|
|
/**
|
|
* Get text of localisation messages in content language.
|
|
*
|
|
* @private
|
|
* @param {string[]} messages
|
|
* @return {string[]}
|
|
*/
|
|
function getMessages( messages ) {
|
|
return messages.map( function ( code ) {
|
|
return data.contLangMessages[ code ];
|
|
} );
|
|
}
|
|
|
|
/**
|
|
* Get a regexp that matches timestamps generated using the given date format.
|
|
*
|
|
* This only supports format characters that are used by the default date format in any of
|
|
* MediaWiki's languages, namely: D, d, F, G, H, i, j, l, M, n, Y, xg, xkY (and escape characters),
|
|
* and only dates when MediaWiki existed, let's say 2000 onwards (Thai dates before 1941 are
|
|
* complicated).
|
|
*
|
|
* @private
|
|
* @param {string} format Date format, as used by MediaWiki
|
|
* @param {string} digitsRegexp Regular expression matching a single localised digit, e.g. `[0-9]`
|
|
* @param {Object} tzAbbrs Map of localised timezone abbreviations to IANA abbreviations
|
|
* for the local timezone, e.g. `{EDT: "EDT", EST: "EST"}`
|
|
* @return {string} Regular expression
|
|
*/
|
|
Parser.prototype.getTimestampRegexp = function ( format, digitsRegexp, tzAbbrs ) {
|
|
var s, p, num, code, endQuote, tzRegexp, regexp;
|
|
|
|
function regexpGroup( r ) {
|
|
return '(' + r + ')';
|
|
}
|
|
|
|
function regexpAlternateGroup( array ) {
|
|
return '(' + array.map( mw.util.escapeRegExp ).join( '|' ) + ')';
|
|
}
|
|
|
|
s = '';
|
|
// Adapted from Language::sprintfDate()
|
|
for ( p = 0; p < format.length; p++ ) {
|
|
num = false;
|
|
code = format[ p ];
|
|
if ( code === 'x' && p < format.length - 1 ) {
|
|
code += format[ ++p ];
|
|
}
|
|
if ( code === 'xk' && p < format.length - 1 ) {
|
|
code += format[ ++p ];
|
|
}
|
|
|
|
switch ( code ) {
|
|
case 'xx':
|
|
s += 'x';
|
|
break;
|
|
case 'xg':
|
|
s += regexpAlternateGroup( getMessages( [
|
|
'january-gen', 'february-gen', 'march-gen', 'april-gen', 'may-gen', 'june-gen',
|
|
'july-gen', 'august-gen', 'september-gen', 'october-gen', 'november-gen',
|
|
'december-gen'
|
|
] ) );
|
|
break;
|
|
case 'd':
|
|
num = '2';
|
|
break;
|
|
case 'D':
|
|
s += regexpAlternateGroup( getMessages( [
|
|
'sun', 'mon', 'tue', 'wed', 'thu', 'fri', 'sat'
|
|
] ) );
|
|
break;
|
|
case 'j':
|
|
num = '1,2';
|
|
break;
|
|
case 'l':
|
|
s += regexpAlternateGroup( getMessages( [
|
|
'sunday', 'monday', 'tuesday', 'wednesday', 'thursday',
|
|
'friday', 'saturday'
|
|
] ) );
|
|
break;
|
|
case 'F':
|
|
s += regexpAlternateGroup( getMessages( [
|
|
'january', 'february', 'march', 'april', 'may_long', 'june',
|
|
'july', 'august', 'september', 'october', 'november',
|
|
'december'
|
|
] ) );
|
|
break;
|
|
case 'M':
|
|
s += regexpAlternateGroup( getMessages( [
|
|
'jan', 'feb', 'mar', 'apr', 'may', 'jun', 'jul', 'aug',
|
|
'sep', 'oct', 'nov', 'dec'
|
|
] ) );
|
|
break;
|
|
case 'n':
|
|
num = '1,2';
|
|
break;
|
|
case 'Y':
|
|
num = '4';
|
|
break;
|
|
case 'xkY':
|
|
num = '4';
|
|
break;
|
|
case 'G':
|
|
num = '1,2';
|
|
break;
|
|
case 'H':
|
|
num = '2';
|
|
break;
|
|
case 'i':
|
|
num = '2';
|
|
break;
|
|
case '\\':
|
|
// Backslash escaping
|
|
if ( p < format.length - 1 ) {
|
|
s += mw.util.escapeRegExp( format[ ++p ] );
|
|
} else {
|
|
s += mw.util.escapeRegExp( '\\' );
|
|
}
|
|
break;
|
|
case '"':
|
|
// Quoted literal
|
|
if ( p < format.length - 1 ) {
|
|
endQuote = format.indexOf( '"', p + 1 );
|
|
if ( endQuote === -1 ) {
|
|
// No terminating quote, assume literal "
|
|
s += '"';
|
|
} else {
|
|
s += mw.util.escapeRegExp( format.substr( p + 1, endQuote - p - 1 ) );
|
|
p = endQuote;
|
|
}
|
|
} else {
|
|
// Quote at end of string, assume literal "
|
|
s += '"';
|
|
}
|
|
break;
|
|
default:
|
|
s += mw.util.escapeRegExp( format[ p ] );
|
|
}
|
|
if ( num !== false ) {
|
|
s += regexpGroup( digitsRegexp + '{' + num + '}' );
|
|
}
|
|
}
|
|
|
|
tzRegexp = regexpAlternateGroup( Object.keys( tzAbbrs ) );
|
|
// Hard-coded parentheses and space like in Parser::pstPass2
|
|
// Ignore some invisible Unicode characters that often sneak into copy-pasted timestamps (T245784)
|
|
regexp = s + '[\\u200E\\u200F]? [\\u200E\\u200F]?\\(' + tzRegexp + '\\)';
|
|
|
|
return regexp;
|
|
};
|
|
|
|
/**
|
|
* Get a function that parses timestamps generated using the given date format, based on the result
|
|
* of matching the regexp returned by #getTimestampRegexp.
|
|
*
|
|
* @private
|
|
* @param {string} format Date format, as used by MediaWiki
|
|
* @param {array|null} digits Localised digits from 0 to 9, e.g. `[ '0', '1', ..., '9' ]`
|
|
* @param {string} localTimezone Local timezone IANA name, e.g. `America/New_York`
|
|
* @param {Object} tzAbbrs Map of localised timezone abbreviations to IANA abbreviations
|
|
* for the local timezone, e.g. `{EDT: "EDT", EST: "EST"}`
|
|
* @return {TimestampParser} Timestamp parser function
|
|
*/
|
|
Parser.prototype.getTimestampParser = function ( format, digits, localTimezone, tzAbbrs ) {
|
|
var p, code, endQuote, matchingGroups = [];
|
|
for ( p = 0; p < format.length; p++ ) {
|
|
code = format[ p ];
|
|
if ( code === 'x' && p < format.length - 1 ) {
|
|
code += format[ ++p ];
|
|
}
|
|
if ( code === 'xk' && p < format.length - 1 ) {
|
|
code += format[ ++p ];
|
|
}
|
|
|
|
switch ( code ) {
|
|
case 'xx':
|
|
break;
|
|
case 'xg':
|
|
case 'd':
|
|
case 'j':
|
|
case 'D':
|
|
case 'l':
|
|
case 'F':
|
|
case 'M':
|
|
case 'n':
|
|
case 'Y':
|
|
case 'xkY':
|
|
case 'G':
|
|
case 'H':
|
|
case 'i':
|
|
matchingGroups.push( code );
|
|
break;
|
|
case '\\':
|
|
// Backslash escaping
|
|
if ( p < format.length - 1 ) {
|
|
++p;
|
|
}
|
|
break;
|
|
case '"':
|
|
// Quoted literal
|
|
if ( p < format.length - 1 ) {
|
|
endQuote = format.indexOf( '"', p + 1 );
|
|
if ( endQuote !== -1 ) {
|
|
p = endQuote;
|
|
}
|
|
}
|
|
break;
|
|
default:
|
|
break;
|
|
}
|
|
}
|
|
|
|
function untransformDigits( text ) {
|
|
if ( !digits ) {
|
|
return text;
|
|
}
|
|
return text.replace(
|
|
new RegExp( '[' + digits.join( '' ) + ']', 'g' ),
|
|
function ( m ) {
|
|
return digits.indexOf( m );
|
|
}
|
|
);
|
|
}
|
|
|
|
/**
|
|
* @typedef {function(Array):moment} TimestampParser
|
|
*/
|
|
|
|
/**
|
|
* Timestamp parser
|
|
*
|
|
* @param {Array} match RegExp match data
|
|
* @return {moment} Moment date object
|
|
*/
|
|
return function timestampParser( match ) {
|
|
var
|
|
year = 0,
|
|
monthIdx = 0,
|
|
day = 0,
|
|
hour = 0,
|
|
minute = 0,
|
|
tzAbbr,
|
|
i, code2, text,
|
|
date;
|
|
for ( i = 0; i < matchingGroups.length; i++ ) {
|
|
code2 = matchingGroups[ i ];
|
|
text = match[ i + 1 ];
|
|
|
|
switch ( code2 ) {
|
|
case 'xg':
|
|
monthIdx = getMessages( [
|
|
'january-gen', 'february-gen', 'march-gen', 'april-gen', 'may-gen', 'june-gen',
|
|
'july-gen', 'august-gen', 'september-gen', 'october-gen', 'november-gen',
|
|
'december-gen'
|
|
] ).indexOf( text );
|
|
break;
|
|
case 'd':
|
|
case 'j':
|
|
day = Number( untransformDigits( text ) );
|
|
break;
|
|
case 'D':
|
|
case 'l':
|
|
// Day of the week - unused
|
|
break;
|
|
case 'F':
|
|
monthIdx = getMessages( [
|
|
'january', 'february', 'march', 'april', 'may_long', 'june',
|
|
'july', 'august', 'september', 'october', 'november',
|
|
'december'
|
|
] ).indexOf( text );
|
|
break;
|
|
case 'M':
|
|
monthIdx = getMessages( [
|
|
'jan', 'feb', 'mar', 'apr', 'may', 'jun', 'jul', 'aug',
|
|
'sep', 'oct', 'nov', 'dec'
|
|
] ).indexOf( text );
|
|
break;
|
|
case 'n':
|
|
monthIdx = Number( untransformDigits( text ) ) - 1;
|
|
break;
|
|
case 'Y':
|
|
year = Number( untransformDigits( text ) );
|
|
break;
|
|
case 'xkY':
|
|
// Thai year
|
|
year = Number( untransformDigits( text ) ) - 543;
|
|
break;
|
|
case 'G':
|
|
case 'H':
|
|
hour = Number( untransformDigits( text ) );
|
|
break;
|
|
case 'i':
|
|
minute = Number( untransformDigits( text ) );
|
|
break;
|
|
default:
|
|
throw new Error( 'Not implemented' );
|
|
}
|
|
}
|
|
// The last matching group is the timezone abbreviation
|
|
tzAbbr = tzAbbrs[ match[ match.length - 1 ] ];
|
|
|
|
// Most of the time, the timezone abbreviation is not necessary to parse the date, since we
|
|
// can assume all times are in the wiki's local timezone.
|
|
date = moment.tz( [ year, monthIdx, day, hour, minute ], localTimezone );
|
|
|
|
// But during the "fall back" at the end of DST, some times will happen twice. Per the docs,
|
|
// "Moment Timezone handles this by always using the earlier instance of a duplicated hour."
|
|
// https://momentjs.com/timezone/docs/#/using-timezones/parsing-ambiguous-inputs/
|
|
|
|
// Since the timezone abbreviation disambiguates the DST/non-DST times, we can detect when
|
|
// that behavior was incorrect...
|
|
if ( date.zoneAbbr() !== tzAbbr ) {
|
|
// ...and force the correct parsing. I can't find proper documentation for this feature,
|
|
// but this pull request explains it: https://github.com/moment/moment-timezone/pull/101
|
|
moment.tz.moveAmbiguousForward = true;
|
|
date = moment.tz( [ year, monthIdx, day, hour, minute ], localTimezone );
|
|
moment.tz.moveAmbiguousForward = false;
|
|
if ( date.zoneAbbr() !== tzAbbr ) {
|
|
// This should not be possible for "genuine" timestamps generated by MediaWiki.
|
|
// But bots and humans get it wrong when marking up unsigned comments…
|
|
// https://pl.wikipedia.org/w/index.php?title=Wikipedia:Kawiarenka/Artykuły&diff=prev&oldid=54772606
|
|
date.discussionToolsWarning = 'Timestamp has timezone abbreviation for the wrong time';
|
|
} else {
|
|
date.discussionToolsWarning = 'Ambiguous time at DST switchover was parsed';
|
|
}
|
|
}
|
|
|
|
return date;
|
|
};
|
|
};
|
|
|
|
/**
|
|
* Get a regexp that matches timestamps in the local date format.
|
|
*
|
|
* This calls #getTimestampRegexp with predefined data for the current wiki.
|
|
*
|
|
* @private
|
|
* @return {string} Regular expression
|
|
*/
|
|
Parser.prototype.getLocalTimestampRegexp = function () {
|
|
return this.getTimestampRegexp(
|
|
data.dateFormat,
|
|
data.digits ? '[' + data.digits.join( '' ) + ']' : '\\d',
|
|
data.timezones
|
|
);
|
|
};
|
|
|
|
/**
|
|
* Get a function that parses timestamps in the local date format, based on the result
|
|
* of matching the regexp returned by #getLocalTimestampRegexp.
|
|
*
|
|
* This calls #getTimestampParser with predefined data for the current wiki.
|
|
*
|
|
* @private
|
|
* @return {TimestampParser} Timestamp parser function
|
|
*/
|
|
Parser.prototype.getLocalTimestampParser = function () {
|
|
return this.getTimestampParser(
|
|
data.dateFormat,
|
|
data.digits,
|
|
data.localTimezone,
|
|
data.timezones
|
|
);
|
|
};
|
|
|
|
/**
|
|
* Callback for document.createTreeWalker that will skip over nodes where we don't want to detect
|
|
* comments (or section headings).
|
|
*
|
|
* @param {Node} node
|
|
* @return {number} Appropriate NodeFilter constant
|
|
*/
|
|
function acceptOnlyNodesAllowingComments( node ) {
|
|
// The table of contents has a heading that gets erroneously detected as a section
|
|
if ( node.id === 'toc' ) {
|
|
return NodeFilter.FILTER_REJECT;
|
|
}
|
|
return NodeFilter.FILTER_ACCEPT;
|
|
}
|
|
|
|
/**
|
|
* Find a timestamps in a given text node
|
|
*
|
|
* @private
|
|
* @param {Text} node Text node
|
|
* @param {string} timestampRegex Timestamp regex
|
|
* @return {Array} Regexp match data, which specifies the location of the match,
|
|
* and which can be parsed using #getLocalTimestampParser
|
|
*/
|
|
Parser.prototype.findTimestamp = function ( node, timestampRegex ) {
|
|
var matchData,
|
|
nodeText = '',
|
|
offset = 0;
|
|
while ( node ) {
|
|
nodeText = node.nodeValue + nodeText;
|
|
|
|
// In Parsoid HTML, entities are represented as a 'mw:Entity' node, rather than normal HTML
|
|
// entities. On Arabic Wikipedia, the "UTC" timezone name contains some non-breaking spaces,
|
|
// which apparently are often turned into entities by buggy editing tools. To handle
|
|
// this, we must piece together the text, so that our regexp can match those timestamps.
|
|
if (
|
|
node.previousSibling &&
|
|
node.previousSibling.nodeType === Node.ELEMENT_NODE &&
|
|
node.previousSibling.getAttribute( 'typeof' ) === 'mw:Entity'
|
|
) {
|
|
nodeText = node.previousSibling.firstChild.nodeValue + nodeText;
|
|
offset += node.previousSibling.firstChild.nodeValue.length;
|
|
|
|
// If the entity is followed by more text, do this again
|
|
if (
|
|
node.previousSibling.previousSibling &&
|
|
node.previousSibling.previousSibling.nodeType === Node.TEXT_NODE
|
|
) {
|
|
offset += node.previousSibling.previousSibling.nodeValue.length;
|
|
node = node.previousSibling.previousSibling;
|
|
} else {
|
|
node = null;
|
|
}
|
|
} else {
|
|
node = null;
|
|
}
|
|
}
|
|
|
|
// Technically, there could be multiple matches in a single text node. However, the ultimate
|
|
// point of this is to find the signatures which precede the timestamps, and any later
|
|
// timestamps in the text node can't be directly preceded by a signature (as we require them to
|
|
// have links), so we only concern ourselves with the first match.
|
|
matchData = nodeText.match( timestampRegex );
|
|
if ( matchData ) {
|
|
matchData.offset = offset;
|
|
return matchData;
|
|
}
|
|
return null;
|
|
};
|
|
|
|
/**
|
|
* Get a MediaWiki page title from a URL.
|
|
*
|
|
* @private
|
|
* @param {string} url
|
|
* @return {mw.Title|null} Page title, or null if this isn't a link to a page
|
|
*/
|
|
function getTitleFromUrl( url ) {
|
|
var articlePathRegexp, match;
|
|
|
|
try {
|
|
url = new mw.Uri( url );
|
|
} catch ( err ) {
|
|
// T106244: URL encoded values using fallback 8-bit encoding (invalid UTF-8) cause mediawiki.Uri to crash
|
|
return null;
|
|
}
|
|
articlePathRegexp = new RegExp(
|
|
mw.util.escapeRegExp( mw.config.get( 'wgArticlePath' ) )
|
|
.replace( mw.util.escapeRegExp( '$1' ), '(.*)' )
|
|
);
|
|
|
|
if ( ( match = url.path.match( articlePathRegexp ) ) ) {
|
|
return mw.Title.newFromText( decodeURIComponent( match[ 1 ] ) );
|
|
}
|
|
if ( url.query.title ) {
|
|
return mw.Title.newFromText( url.query.title );
|
|
}
|
|
return null;
|
|
}
|
|
|
|
/**
|
|
* Find a user signature preceding a timestamp.
|
|
*
|
|
* The signature includes the timestamp node.
|
|
*
|
|
* A signature must contain at least one link to the user's userpage, discussion page or
|
|
* contributions (and may contain other links). The link may be nested in other elements.
|
|
*
|
|
* @private
|
|
* @param {Text} timestampNode Text node
|
|
* @param {Node} [until] Node to stop searching at
|
|
* @return {Array} Result, a tuple contaning:
|
|
* - {Node[]} Sibling nodes comprising the signature, in reverse order (with
|
|
* `timestampNode` or its parent node as the first element)
|
|
* - {string|null} Username, null for unsigned comments
|
|
*/
|
|
Parser.prototype.findSignature = function ( timestampNode, until ) {
|
|
var node, sigNodes, sigUsername, length, lastLinkNode, links, nodes;
|
|
|
|
// Support timestamps being linked to the diff introducing the comment:
|
|
// if the timestamp node is the only child of a link node, use the link node instead
|
|
if (
|
|
!timestampNode.previousSibling && !timestampNode.nextSibling &&
|
|
timestampNode.parentNode.nodeName.toLowerCase() === 'a'
|
|
) {
|
|
timestampNode = timestampNode.parentNode;
|
|
}
|
|
|
|
node = timestampNode;
|
|
sigNodes = [ node ];
|
|
sigUsername = null;
|
|
length = 0;
|
|
lastLinkNode = timestampNode;
|
|
|
|
while ( ( node = node.previousSibling ) && length < data.signatureScanLimit && node !== until ) {
|
|
sigNodes.push( node );
|
|
length += node.textContent ? codePointLength( node.textContent ) : 0;
|
|
if ( node.nodeType !== Node.ELEMENT_NODE ) {
|
|
continue;
|
|
}
|
|
links = [];
|
|
if ( node.tagName.toLowerCase() === 'a' ) {
|
|
links.push( node );
|
|
} else {
|
|
// Handle links nested in formatting elements.
|
|
// Helpful accidental feature: users whose signature is not detected in full (due to
|
|
// text formatting) can just wrap it in a <span> to fix that.
|
|
// "Ten Pound Hammer • (What did I screw up now?)"
|
|
// "« Saper // dyskusja »"
|
|
nodes = node.getElementsByTagName( 'a' );
|
|
links.push.apply( links, nodes );
|
|
}
|
|
if ( !links.length ) {
|
|
continue;
|
|
}
|
|
// Use .some() rather than .every() to permit vanity links
|
|
// "TonyTheTiger (T / C / WP:FOUR / WP:CHICAGO / WP:WAWARD)"
|
|
// eslint-disable-next-line no-loop-func
|
|
if ( links.some( function ( link ) {
|
|
var username, title;
|
|
title = getTitleFromUrl( link.href );
|
|
if ( !title ) {
|
|
return false;
|
|
}
|
|
if (
|
|
title.getNamespaceId() === mw.config.get( 'wgNamespaceIds' ).user ||
|
|
title.getNamespaceId() === mw.config.get( 'wgNamespaceIds' ).user_talk
|
|
) {
|
|
username = title.getMainText();
|
|
if ( username.indexOf( '/' ) !== -1 ) {
|
|
return false;
|
|
}
|
|
} else if (
|
|
title.getNamespaceId() === mw.config.get( 'wgNamespaceIds' ).special &&
|
|
title.getMainText().split( '/' )[ 0 ] === data.specialContributionsName
|
|
) {
|
|
username = title.getMainText().split( '/' )[ 1 ];
|
|
// Normalize the username: users may link to their contributions with an unnormalized name
|
|
username = mw.Title.makeTitle( mw.config.get( 'wgNamespaceIds' ).user, username ).getMainText();
|
|
}
|
|
if ( !username ) {
|
|
return false;
|
|
}
|
|
if ( mw.util.isIPv6Address( username ) ) {
|
|
// Bot-generated links "Preceding unsigned comment added by" have non-standard case
|
|
username = username.toUpperCase();
|
|
}
|
|
|
|
// Accept the first link to the user namespace, then only accept links to that user
|
|
if ( !sigUsername ) {
|
|
sigUsername = username;
|
|
}
|
|
return username === sigUsername;
|
|
} ) ) {
|
|
lastLinkNode = node;
|
|
}
|
|
// Keep looking if a node with links wasn't a link to a user page
|
|
// "Doc James (talk · contribs · email)"
|
|
}
|
|
// Pop excess text nodes
|
|
while ( sigNodes[ sigNodes.length - 1 ] !== lastLinkNode ) {
|
|
sigNodes.pop();
|
|
}
|
|
return [ sigNodes, sigUsername ];
|
|
};
|
|
|
|
/**
|
|
* Return the next leaf node in the tree order that is not an empty or whitespace-only text node.
|
|
*
|
|
* In other words, this returns a Text node with content other than whitespace, or an Element node
|
|
* with no children, that follows the given node in the HTML source.
|
|
*
|
|
* @private
|
|
* @param {Node} node Node to start searching at. If it isn't a leaf node, its children are ignored.
|
|
* @return {Node}
|
|
*/
|
|
Parser.prototype.nextInterestingLeafNode = function ( node ) {
|
|
var treeWalker,
|
|
rootNode = this.rootNode;
|
|
|
|
treeWalker = rootNode.ownerDocument.createTreeWalker(
|
|
rootNode,
|
|
// eslint-disable-next-line no-bitwise
|
|
NodeFilter.SHOW_ELEMENT | NodeFilter.SHOW_TEXT,
|
|
function ( n ) {
|
|
// Ignore this node and its descendants
|
|
// (unless it's the root node, this is a special case for "fakeHeading" handling)
|
|
if ( node !== rootNode && ( n === node || n.parentNode === node ) ) {
|
|
return NodeFilter.FILTER_REJECT;
|
|
}
|
|
if (
|
|
( n.nodeType === Node.TEXT_NODE && utils.htmlTrim( n.textContent ) !== '' ) ||
|
|
( n.nodeType === Node.CDATA_SECTION_NODE && utils.htmlTrim( n.textContent ) !== '' ) ||
|
|
( n.nodeType === Node.ELEMENT_NODE && !n.firstChild )
|
|
) {
|
|
return NodeFilter.FILTER_ACCEPT;
|
|
}
|
|
return NodeFilter.FILTER_SKIP;
|
|
},
|
|
false
|
|
);
|
|
treeWalker.currentNode = node;
|
|
treeWalker.nextNode();
|
|
if ( !treeWalker.currentNode ) {
|
|
throw new Error( 'nextInterestingLeafNode not found' );
|
|
}
|
|
return treeWalker.currentNode;
|
|
};
|
|
|
|
/**
|
|
* Get all discussion comments (and headings) within a DOM subtree.
|
|
*
|
|
* This returns a flat list, use #getThreads to get a tree structure starting at section headings.
|
|
*
|
|
* For example, for a MediaWiki discussion like this (we're dealing with HTML DOM here, the wikitext
|
|
* syntax is just for illustration):
|
|
*
|
|
* == A ==
|
|
* B. ~~~~
|
|
* : C.
|
|
* : C. ~~~~
|
|
* :: D. ~~~~
|
|
* ::: E. ~~~~
|
|
* ::: F. ~~~~
|
|
* : G. ~~~~
|
|
* H. ~~~~
|
|
* : I. ~~~~
|
|
*
|
|
* This function would return a structure like:
|
|
*
|
|
* [
|
|
* HeadingItem( { level: 0, range: (h2: A) } ),
|
|
* CommentItem( { level: 1, range: (p: B) } ),
|
|
* CommentItem( { level: 2, range: (li: C, li: C) } ),
|
|
* CommentItem( { level: 3, range: (li: D) } ),
|
|
* CommentItem( { level: 4, range: (li: E) } ),
|
|
* CommentItem( { level: 4, range: (li: F) } ),
|
|
* CommentItem( { level: 2, range: (li: G) } ),
|
|
* CommentItem( { level: 1, range: (p: H) } ),
|
|
* CommentItem( { level: 2, range: (li: I) } )
|
|
* ]
|
|
*
|
|
* @param {HTMLElement} rootNode
|
|
* @return {ThreadItem[]} Thread items
|
|
*/
|
|
Parser.prototype.getThreadItems = function () {
|
|
if ( !this.threadItems ) {
|
|
this.buildThreads();
|
|
}
|
|
return this.threadItems;
|
|
};
|
|
|
|
/**
|
|
* Same as getFlatThreadItems, but only returns the CommentItems
|
|
*
|
|
* @return {CommentItem[]} Comment items
|
|
*/
|
|
Parser.prototype.getCommentItems = function () {
|
|
if ( !this.commentItems ) {
|
|
this.buildThreads();
|
|
}
|
|
return this.commentItems;
|
|
};
|
|
|
|
/**
|
|
* Find a CommentItem by its ID
|
|
*
|
|
* @param {string} id Comment ID
|
|
* @return {CommentItem|null} Comment item, null if not found
|
|
*/
|
|
Parser.prototype.findCommentById = function ( id ) {
|
|
if ( !this.commentsById ) {
|
|
this.buildThreads();
|
|
}
|
|
return this.commentsById[ id ] || null;
|
|
};
|
|
|
|
Parser.prototype.buildThreadItems = function () {
|
|
var
|
|
dfParser = this.getLocalTimestampParser(),
|
|
timestampRegex = this.getLocalTimestampRegexp(),
|
|
commentItems = [],
|
|
threadItems = [],
|
|
treeWalker,
|
|
node, range, fakeHeading, curComment,
|
|
foundSignature, firstSigNode, lastSigNode, sigRange, author, startNode, endNode, length,
|
|
match, lastSigNodeOffset, startLevel, endLevel, level, dateTime, warnings;
|
|
|
|
treeWalker = this.rootNode.ownerDocument.createTreeWalker(
|
|
this.rootNode,
|
|
// eslint-disable-next-line no-bitwise
|
|
NodeFilter.SHOW_ELEMENT | NodeFilter.SHOW_TEXT,
|
|
acceptOnlyNodesAllowingComments,
|
|
false
|
|
);
|
|
|
|
// Placeholder heading in case there are comments in the 0th section
|
|
range = {
|
|
startContainer: this.rootNode,
|
|
startOffset: 0,
|
|
endContainer: this.rootNode,
|
|
endOffset: 0
|
|
};
|
|
fakeHeading = new HeadingItem( range, true );
|
|
fakeHeading.rootNode = this.rootNode;
|
|
|
|
curComment = fakeHeading;
|
|
|
|
while ( ( node = treeWalker.nextNode() ) ) {
|
|
if ( node.tagName && node.tagName.match( /^h[1-6]$/i ) ) {
|
|
range = {
|
|
startContainer: node,
|
|
startOffset: 0,
|
|
endContainer: node,
|
|
endOffset: node.childNodes.length
|
|
};
|
|
curComment = new HeadingItem( range );
|
|
curComment.rootNode = this.rootNode;
|
|
threadItems.push( curComment );
|
|
} else if ( node.nodeType === Node.TEXT_NODE && ( match = this.findTimestamp( node, timestampRegex ) ) ) {
|
|
warnings = [];
|
|
foundSignature = this.findSignature( node, lastSigNode );
|
|
author = foundSignature[ 1 ];
|
|
firstSigNode = foundSignature[ 0 ][ foundSignature[ 0 ].length - 1 ];
|
|
lastSigNode = foundSignature[ 0 ][ 0 ];
|
|
|
|
if ( !author ) {
|
|
// Ignore timestamps for which we couldn't find a signature. It's probably not a real
|
|
// comment, but just a false match due to a copypasted timestamp.
|
|
continue;
|
|
}
|
|
|
|
lastSigNodeOffset = lastSigNode === node ?
|
|
match.index + match[ 0 ].length - match.offset :
|
|
utils.childIndexOf( lastSigNode ) + 1;
|
|
sigRange = {
|
|
startContainer: firstSigNode.parentNode,
|
|
startOffset: utils.childIndexOf( firstSigNode ),
|
|
endContainer: lastSigNode === node ? node : lastSigNode.parentNode,
|
|
endOffset: lastSigNodeOffset
|
|
};
|
|
|
|
// Everything from the last comment up to here is the next comment
|
|
startNode = this.nextInterestingLeafNode( curComment.range.endContainer );
|
|
endNode = lastSigNode;
|
|
// Skip to the end of the "paragraph". This only looks at tag names and can be fooled by CSS, but
|
|
// avoiding that would be more difficult and slower.
|
|
while ( endNode.nextSibling && !( endNode.nextSibling instanceof HTMLElement && ve.isBlockElement( endNode.nextSibling ) ) ) {
|
|
endNode = endNode.nextSibling;
|
|
}
|
|
|
|
if ( endNode === lastSigNode ) {
|
|
range = {
|
|
startContainer: startNode.parentNode,
|
|
startOffset: utils.childIndexOf( startNode ),
|
|
endContainer: sigRange.endContainer,
|
|
endOffset: sigRange.endOffset
|
|
};
|
|
} else {
|
|
length = endNode.nodeType === Node.TEXT_NODE ?
|
|
endNode.textContent.replace( /[\t\n\f\r ]+$/, '' ).length :
|
|
endNode.childNodes.length;
|
|
range = {
|
|
startContainer: startNode.parentNode,
|
|
startOffset: utils.childIndexOf( startNode ),
|
|
endContainer: endNode,
|
|
endOffset: length
|
|
};
|
|
}
|
|
|
|
startLevel = utils.getIndentLevel( startNode, this.rootNode ) + 1;
|
|
endLevel = utils.getIndentLevel( node, this.rootNode ) + 1;
|
|
if ( startLevel !== endLevel ) {
|
|
warnings.push( 'Comment starts and ends with different indentation' );
|
|
}
|
|
// Should this use the indent level of `startNode` or `node`?
|
|
level = Math.min( startLevel, endLevel );
|
|
|
|
dateTime = dfParser( match );
|
|
if ( dateTime.discussionToolsWarning ) {
|
|
warnings.push( dateTime.discussionToolsWarning );
|
|
}
|
|
|
|
// Avoid generating multiple comments when there is more than one signature on a single "line".
|
|
// Often this is done when someone edits their comment later and wants to add a note about that.
|
|
// (Or when another person corrects a typo, or strikes out a comment, etc.) Multiple comments
|
|
// within one paragraph/list-item result in a confusing double "Reply" button, and we also have
|
|
// no way to indicate which one you're replying to (this might matter in the future for
|
|
// notifications or something).
|
|
if (
|
|
curComment instanceof CommentItem &&
|
|
( utils.closestElement( node, [ 'li', 'dd', 'p' ] ) || node.parentNode ) ===
|
|
( utils.closestElement( curComment.range.endContainer, [ 'li', 'dd', 'p' ] ) || curComment.range.endContainer.parentNode )
|
|
) {
|
|
// Merge this with the previous comment. Use that comment's author and timestamp.
|
|
curComment.signatureRanges.push( sigRange );
|
|
|
|
if (
|
|
curComment.range.endContainer === range.endContainer &&
|
|
curComment.range.endOffset <= range.endOffset
|
|
) {
|
|
// We've already skipped over this signature, and the `range` and `level` are messed up,
|
|
// because that causes `startNode` to be after `endNode`
|
|
continue;
|
|
}
|
|
|
|
curComment.range.endContainer = range.endContainer;
|
|
curComment.range.endOffset = range.endOffset;
|
|
curComment.level = Math.min( level, curComment.level );
|
|
|
|
continue;
|
|
}
|
|
|
|
curComment = new CommentItem(
|
|
level,
|
|
range,
|
|
[ sigRange ],
|
|
dateTime,
|
|
author
|
|
);
|
|
curComment.rootNode = this.rootNode;
|
|
if ( warnings.length ) {
|
|
curComment.warnings = warnings;
|
|
}
|
|
commentItems.push( curComment );
|
|
threadItems.push( curComment );
|
|
}
|
|
}
|
|
|
|
// Insert the fake placeholder heading if there are any comments in the 0th section
|
|
// (before the first real heading)
|
|
if ( threadItems.length && !( threadItems[ 0 ] instanceof HeadingItem ) ) {
|
|
threadItems.unshift( fakeHeading );
|
|
}
|
|
|
|
this.commentItems = commentItems;
|
|
this.threadItems = threadItems;
|
|
};
|
|
|
|
/**
|
|
* Group discussion comments into threads and associate replies to original messages.
|
|
*
|
|
* Each thread must begin with a heading. Original messages in the thread are treated as replies to
|
|
* its heading. Other replies are associated based on the order and indentation level.
|
|
*
|
|
* Note that the objects in `comments` are extended in-place with the additional data.
|
|
*
|
|
* For example, for a MediaWiki discussion like this (we're dealing with HTML DOM here, the wikitext
|
|
* syntax is just for illustration):
|
|
*
|
|
* == A ==
|
|
* B. ~~~~
|
|
* : C.
|
|
* : C. ~~~~
|
|
* :: D. ~~~~
|
|
* ::: E. ~~~~
|
|
* ::: F. ~~~~
|
|
* : G. ~~~~
|
|
* H. ~~~~
|
|
* : I. ~~~~
|
|
*
|
|
* This function would return a structure like:
|
|
*
|
|
* [
|
|
* HeadingItem( { level: 0, range: (h2: A), replies: [
|
|
* CommentItem( { level: 1, range: (p: B), replies: [
|
|
* CommentItem( { level: 2, range: (li: C, li: C), replies: [
|
|
* CommentItem( { level: 3, range: (li: D), replies: [
|
|
* CommentItem( { level: 4, range: (li: E), replies: [] },
|
|
* CommentItem( { level: 4, range: (li: F), replies: [] },
|
|
* ] },
|
|
* ] },
|
|
* CommentItem( { level: 2, range: (li: G), replies: [] },
|
|
* ] },
|
|
* CommentItem( { level: 1, range: (p: H), replies: [
|
|
* CommentItem( { level: 2, range: (li: I), replies: [] },
|
|
* ] },
|
|
* ] } )
|
|
* ]
|
|
*
|
|
* @return {HeadingItem[]} Tree structure of comments, top-level items are the headings.
|
|
*/
|
|
Parser.prototype.getThreads = function () {
|
|
if ( !this.threads ) {
|
|
this.buildThreads();
|
|
}
|
|
return this.threads;
|
|
};
|
|
|
|
Parser.prototype.buildThreads = function () {
|
|
var
|
|
threads = [],
|
|
replies = [],
|
|
i, threadItem, id, number;
|
|
|
|
if ( !this.threadItems ) {
|
|
this.buildThreadItems();
|
|
}
|
|
|
|
this.commentsById = {};
|
|
for ( i = 0; i < this.threadItems.length; i++ ) {
|
|
threadItem = this.threadItems[ i ];
|
|
|
|
if ( threadItem instanceof HeadingItem ) {
|
|
// We don't need ids for section headings right now, but we might in the future
|
|
// e.g. if we allow replying directly to sections (adding top-level comments)
|
|
id = null;
|
|
} else {
|
|
// username+timestamp
|
|
id = [
|
|
threadItem.author || '',
|
|
threadItem.timestamp.toISOString()
|
|
].join( '|' );
|
|
|
|
// If there would be multiple comments with the same ID (i.e. the user left multiple comments
|
|
// in one edit, or within a minute), append sequential numbers
|
|
number = 0;
|
|
while ( this.commentsById[ id + '|' + number ] ) {
|
|
number++;
|
|
}
|
|
id = id + '|' + number;
|
|
this.commentsById[ id ] = threadItem;
|
|
// This modifies the original objects in `comments`!
|
|
threadItem.id = id;
|
|
}
|
|
|
|
if ( replies.length < threadItem.level ) {
|
|
// Someone skipped an indentation level (or several). Pretend that the previous reply
|
|
// covers multiple indentation levels, so that following comments get connected to it.
|
|
threadItem.warnings = threadItem.warnings || [];
|
|
threadItem.warnings.push( 'Comment skips indentation level' );
|
|
while ( replies.length < threadItem.level ) {
|
|
replies[ replies.length ] = replies[ replies.length - 1 ];
|
|
}
|
|
}
|
|
|
|
if ( threadItem instanceof HeadingItem ) {
|
|
// New root (thread)
|
|
threads.push( threadItem );
|
|
} else if ( replies[ threadItem.level - 1 ] ) {
|
|
// Add as a reply to the closest less-nested comment
|
|
threadItem.parent = replies[ threadItem.level - 1 ];
|
|
threadItem.parent.replies.push( threadItem );
|
|
} else {
|
|
threadItem.warnings = threadItem.warnings || [];
|
|
threadItem.warnings.push( 'Comment could not be connected to a thread' );
|
|
}
|
|
|
|
replies[ threadItem.level ] = threadItem;
|
|
// Cut off more deeply nested replies
|
|
replies.length = threadItem.level + 1;
|
|
}
|
|
|
|
this.threads = threads;
|
|
};
|
|
|
|
module.exports = Parser;
|