Bug 976556 - Parse translation results back into the TranslationDocument and TranslationItem data structures. r=florian

This commit is contained in:
Felipe Gomes 2014-05-19 17:31:23 -03:00
Родитель 61232aa1de
Коммит de0ef149d6
2 изменённых файлов: 151 добавлений и 2 удалений

Просмотреть файл

@ -122,8 +122,47 @@ this.BingTranslation.prototype = {
} }
}, },
_parseChunkResult() { /**
// note: this function is implemented in the patch from bug 976556 * This function parses the result returned by Bing's Http.svc API,
* which is a XML file that contains a number of elements. To our
* particular interest, the only part of the response that matters
* are the <TranslatedText> nodes, which contains the resulting
* items that were sent to be translated.
*
* @param request The request sent to the server.
* @returns boolean True if parsing of this chunk was successful.
*/
_parseChunkResult: function(bingRequest) {
let domParser = Cc["@mozilla.org/xmlextras/domparser;1"]
.createInstance(Ci.nsIDOMParser);
let results;
try {
let doc = domParser.parseFromString(bingRequest.networkRequest
.response.body, "text/xml");
results = doc.querySelectorAll("TranslatedText");
} catch (e) {
return false;
}
let len = results.length;
if (len != bingRequest.translationData.length) {
// This should never happen, but if the service returns a different number
// of items (from the number of items submitted), we can't use this chunk
// because all items would be paired incorrectly.
return false;
}
let error = false;
for (let i = 0; i < len; i++) {
try {
bingRequest.translationData[i][0].parseResult(
results[i].firstChild.nodeValue
);
} catch (e) { error = true; }
}
return !error;
}, },
/** /**

Просмотреть файл

@ -167,6 +167,38 @@ this.TranslationDocument.prototype = {
* This class represents an item for translation. It's basically our * This class represents an item for translation. It's basically our
* wrapper class around a node returned by getTranslationNode, with * wrapper class around a node returned by getTranslationNode, with
* more data and structural information on it. * more data and structural information on it.
*
* At the end of the translation process, besides the properties below,
* a TranslationItem will contain two other properties: one called "original"
* and one called "translation". They are twin objects, one which reflect
* the structure of that node in its original state, and the other in its
* translated state.
*
* The "original" array is generated in the generateTextForItem function,
* and the "translation" array is generated when the translation results
* are parsed.
*
* They are both arrays, which contain a mix of strings and references to
* child TranslationItems. The references in both arrays point to the * same *
* TranslationItem object, but they might appear in different orders between the
* "original" and "translation" arrays.
*
* An example:
*
* English: <div id="n1">Welcome to <b id="n2">Mozilla's</b> website</div>
* Portuguese: <div id="n1">Bem vindo a pagina <b id="n2">da Mozilla</b></div>
*
* TranslationItem n1 = {
* id: 1,
* original: ["Welcome to", ptr to n2, "website"]
* translation: ["Bem vindo a pagina", ptr to n2]
* }
*
* TranslationItem n2 = {
* id: 2,
* original: ["Mozilla's"],
* translation: ["da Mozilla"]
* }
*/ */
function TranslationItem(node, id, isRoot) { function TranslationItem(node, id, isRoot) {
this.nodeRef = node; this.nodeRef = node;
@ -185,5 +217,83 @@ TranslationItem.prototype = {
: ''; : '';
return "[object TranslationItem: <" + this.nodeRef.localName + ">" return "[object TranslationItem: <" + this.nodeRef.localName + ">"
+ rootType + "]"; + rootType + "]";
},
/**
* This function will parse the result of the translation of one translation
* item. If this item was a simple root, all we sent was a plain-text version
* of it, so the result is also straightforward text.
*
* For non-simple roots, we sent a simplified HTML representation of that
* node, and we'll first parse that into an HTML doc and then call the
* parseResultNode helper function to parse it.
*
* While parsing, the result is stored in the "translation" field of the
* TranslationItem, which will be used to display the final translation when
* all items are finished. It remains stored too to allow back-and-forth
* switching between the "Show Original" and "Show Translation" functions.
*
* @param result A string with the textual result received from the server,
* which can be plain-text or a serialized HTML doc.
*/
parseResult: function(result) {
if (this.isSimpleRoot) {
this.translation = [result];
return;
}
let domParser = Cc["@mozilla.org/xmlextras/domparser;1"]
.createInstance(Ci.nsIDOMParser);
let doc = domParser.parseFromString(result, "text/html");
parseResultNode(this, doc.body.firstChild);
},
/**
* This function finds a child TranslationItem
* with the given id.
* @param id The id to look for, in the format "n#"
* @returns A TranslationItem with the given id, or null if
* it was not found.
*/
getChildById: function(id) {
let foundChild = null;
for (let child of item.children) {
if (("n" + child.id) == id) {
foundChild = child;
break;
}
}
return foundChild;
}
};
/**
* Helper function to parse a HTML doc result.
* How it works:
*
* An example result string is:
*
* <div id="n1">Hello <b id="n2">World</b> of Mozilla.</div>
*
* For an element node, we look at its id and find the corresponding
* TranslationItem that was associated with this node, and then we
* walk down it repeating the process.
*
* For text nodes we simply add it as a string.
*/
function parseResultNode(item, node) {
item.translation = [];
for (let child of node.childNodes) {
if (child.nodeType == TEXT_NODE) {
item.translation.push(child.nodeValue);
} else {
let translationItemChild = item.getChildById(child.id);
if (translationItemChild) {
item.translation.push(translationItemChild);
parseResultNode(translationItemChild, child);
}
}
} }
} }