PXML

From Lingoport Wiki
Jump to: navigation, search

Introduction

Resource files for software localization come in different standard format, such as JSON, .resx, .properties and so on. To simplify showing context in a TMS or in a CAT tool like Trados Studio, best is to send strings to be translated as PXML file, i.e. Preview XML. This is a Lingoport format.

Main advantages:

  • The URL for the context on InContext server is easily parsed
  • Files like JSON without support for comments are not well suited to pass attributes such as context URLs
  • The parsing on the TMS or CAT tool is for one file format, not for all the different resource file types, so one parser/filter configuration
  • Other attributes coming from Command Center can also be associated

Configuration

The prerequisites for sending context are the following:

  • an InContext server configured
  • a TMS configuration pointing to the InContext server
  • a Localyzer project configured with that TMS

In TMS configuration, check the "Convert to PXML" checkbox.

Any file sent to the TMS will be converted to a PXML file format.

PXML File Naming Convention

Repository files converted to PXML can be of many standard resource file types, such as:

  • JSON
  • resx
  • properties
  • YML
  • .strings
  • xml (many schemas here)
  • etc.

Also, some of the files to be translated may have the same name, but be placed in different directory. A typical case is 'en.json'.

For those reasons, in order to know which converted PXML corresponds to which repository files, the naming convention looks like:

  • lrmjson_LID349823_message.pxml

where the prefix indicates the source file format (here json), an ID of the path in the repo (here 349823), and the file name root itself (here message).

PXML File Format

The following sample PXML shows the different elements: 


<?xml version="1.0" encoding="utf-8"?>
<pxml>
<!-- System prompt is a concatenation of the translation connection system prompt and the project prompt in Localyzer -->
<system_prompt><![CDATA[As a software application translator, translate an application user interface text from English to French. Special considerations for UI: Brevity is crucial, Clarity over creativity, Consistency with UI patterns, Action-oriented language. Return only the translated text-no explanation. This is for a hardware store.]]></system_prompt>
<resources>
  <string translate="1" segmentID="1" minLength="" maxLength="">
    <lrm_incontext>https://olivier.lingoport.io/incontext-server/lookup?key=183163_254&token=Vozsmt6xmKxw5qVQ3c8w921Hj9NEKX70</lrm_incontext>
    <llm_prompt></llm_prompt>
    <SID>paint.coat</SID>
    <value><![CDATA[Coat]]></value>
  </string>
  <string translate="1" segmentID="3" minLength="" maxLength="">
    <lrm_incontext>https://olivier.lingoport.io/incontext-server/lookup?key=183165_254&token=Vozsmt6xmKxw5qVQ3c8w921Hj9NEKX70</lrm_incontext>
    <llm_prompt></llm_prompt>
    <SID>machine.washer</SID>
    <value><![CDATA[Washer]]></value>
  </string>
</resources>
</pxml>

The PXML elements:

  • <pxml>: The root element for this schema
  • <system_prompt>: If any prompt was provided at the translation configuration and or the project level, those are concatenated as the system prompt
  • <resources>: Many <string/> resources are under this level.
  • <string>: the string element contains the items to translated with a number of associated attributes
    • SID: the unique segment identifier for that string, in effect the key for that string value.
    • translate: Upcoming, for now always 1 meaning the value is to be translated. 0 means do not translate this string.
    • segmentID: a placeholder, for now a counter starting at 1 in the PXML file and incremented for each <string> element
    • minLength: Upcoming - a value for the minimum length of the string, for now an empty string. This optional attribute would be set in String Manager
    • maxLength: Upcoming - a value for the maximum length of the string, for now an empty string. This optional attribute would be set in String Manager
    • lrm_incontext: the URL to the context associated with that string
    • llm_prompt: Upcoming - a prompt to help translate the value. See: Prompting
    • value: the string to be translated.

Note: Texts use CDATA so special characters added by users will not be misconstrued.

Context URL

The context URL looks like the following: 

  https://poc.lingoport.io/incontext-server/lookup?key=1573_105&amp;token=8JZGug2O004Q6q9Z00ruO5g04Lr8w7SD
  • https://poc.lingoport.io/incontext-server/ is the URL of the InContext server. This server must be accessible to translators
  • lookup: the method to retrieve the context
  • key: the context identifier
  • token: only those with the right token can view the context


Note:

In the XML-based PXML file, the ampersand (&) character is escaped as &amp; to comply with XML standards. This ensures the file remains valid and can be parsed by TMS and CAT tools.

However, when accessing the context URL directly in a web browser, you must replace &amp; with a regular ampersand (&). For example:

XML format (escaped for validity):

Browser format (actual usable link):

Retrieved from "http://wiki.lingoport.com/index.php?title=PXML&oldid=99735"