[forge-issues] [JBoss JIRA] (ROASTER-75) Encoding of source[?] code cannot be easily configured.

[George Gastaldi (JIRA)]

    [ https://issues.jboss.org/browse/ROASTER-75?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13112693#comment-13112693 ] 

George Gastaldi commented on ROASTER-75:
----------------------------------------

I have reduced the parseUnit methods in {{JavaParser}} interface to having a {{org.jboss.forge.roaster.spi.JavaParser.parseUnit(String)}} method. 
We wouldn't be able to reuse the same method for binary parsing, as it wouldn't be possible to detect if the {{InputStream}} is a source or a binary file. 

As this SPI is only used internally, we believe it's safe to change it without impact on consumers.

> Encoding of source[?] code cannot be easily configured.
> -------------------------------------------------------
>
>                 Key: ROASTER-75
>                 URL: https://issues.jboss.org/browse/ROASTER-75
>             Project: Roaster
>          Issue Type: Enhancement
>          Components: API, JDT
>    Affects Versions: 2.14.0.Final, 2.15.0.Final
>         Environment: Java 8 64bit, Eclipse Mars, Windows 8.1 Pro 64bit
>            Reporter: Shoji Kuzukami
>            Priority: Minor
>             Fix For: 2.16.1.Final
>
>
> The parser encoding for souce code can only be configured by the 'file.encoding' system parameter, as we can see in JavaParserImpl, "String encoding = System.getProperty("file.encoding", "ISO8859_1");".
> Even though the Roaster API can receive a 'String' source as "public static JavaType<?> parse(final String data)" in the 'Roaster.java', the 'String' of source can only contain "ISO8859_1" charsets if we don't change the "file.encoding".
> The parser SPI in 'JavaParser.java'  is defined as "JavaUnit parseUnit(final InputStream data);" instead of "JavaUnit parseUnit(Reader source);". I think it is adequate considering that the Roaster API will accept some binary sources (such as bytecode) in near future. 
> I recommend that the SPI "JavaUnit parseUnit(final InputStream data);"  is expanded into "JavaUnit parseUnit(final InputStream data, Charset encodingIfText );" and some encoding refactors are done in the 'Roaster.java' without any change of semantics in the Roaster APIs except for the 'String' souce APIs as "public static JavaType<?> parse(final String data)". 
> The 'Charset' argument is used only if the 'data' of InputStream must be interpreted as 'String' ( as a case of source code ).
> If this idea is accepted by this project design, I'll refactor the code and do a pull-requeust.



--
This message was sent by Atlassian JIRA
(v6.4.11#64026)