lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mark <javam...@gmail.com>
Subject Re: indexing java byte code in classes / jars
Date Fri, 08 May 2015 21:09:17 GMT
To answer why bytecode - because mostly the use case I have is looking to
index as much detail from jars/classes.

extract class names,
method names
signatures
packages / imports

I am considering using ASM in order to generate an analysis view of the
class

The sort of usecases I have would be method / signature searches.

For example;

1) show any classes with a method named parse*

2) show any classes with a method named parse that passes in a type *json*

...etc

In the past I have written something to reverse out javadocs from just java
bytecode, using solr would move this idea considerably much more powerful.

Thanks for the suggestions so far







On 8 May 2015 at 21:19, Erik Hatcher <erik.hatcher@gmail.com> wrote:

> Oh, and sorry, I omitted a couple of details:
>
> # creating the “java” core/collection
> bin/solr create -c java
>
> # I ran this from my Solr source code checkout, so that
> SolrLogFormatter.class just happened to be handy
>
>         Erik
>
>
>
>
> > On May 8, 2015, at 4:11 PM, Erik Hatcher <erik.hatcher@gmail.com> wrote:
> >
> > What kinds of searches do you want to run?  Are you trying to extract
> class names, method names, and such and make those searchable?   If that’s
> the case, you need some kind of “parser” to reverse engineer that
> information from .class and .jar files before feeding it to Solr, which
> would happen before analysis.   Java itself comes with a javap command that
> can do this; whether this is the “best” way to go for your scenario I don’t
> know, but here’s an interesting example pasted below (using Solr 5.x).
> >
> > —
> > Erik Hatcher, Senior Solutions Architect
> > http://www.lucidworks.com
> >
> >
> > javap
> build/solr-core/classes/java/org/apache/solr/SolrLogFormatter.class >
> test.txt
> > bin/post -c java test.txt
> >
> > now search for "coreInfoMap"
> http://localhost:8983/solr/java/browse?q=coreInfoMap
> >
> > I tried to be cleverer and use the stdin option of bin/post, like this:
> > javap
> build/solr-core/classes/java/org/apache/solr/SolrLogFormatter.class |
> bin/post -c java -url http://localhost:8983/solr/java/update/extract
> -type text/plain -params "literal.id=SolrLogFormatter" -out yes -d
> > but something isn’t working right with the stdin detection like that (it
> does work to `cat test.txt | bin/post…` though, hmmm)
> >
> > test.txt looks like this, `cat test.txt`:
> > Compiled from "SolrLogFormatter.java"
> > public class org.apache.solr.SolrLogFormatter extends
> java.util.logging.Formatter {
> >  long startTime;
> >  long lastTime;
> >  java.util.Map<org.apache.solr.SolrLogFormatter$Method,
> java.lang.String> methodAlias;
> >  public boolean shorterFormat;
> >  java.util.Map<org.apache.solr.core.SolrCore,
> org.apache.solr.SolrLogFormatter$CoreInfo> coreInfoMap;
> >  public java.util.Map<java.lang.String, java.lang.String> classAliases;
> >  static java.lang.ThreadLocal<java.lang.String> threadLocal;
> >  public org.apache.solr.SolrLogFormatter();
> >  public void setShorterFormat();
> >  public java.lang.String format(java.util.logging.LogRecord);
> >  public void appendThread(java.lang.StringBuilder,
> java.util.logging.LogRecord);
> >  public java.lang.String _format(java.util.logging.LogRecord);
> >  public java.lang.String getHead(java.util.logging.Handler);
> >  public java.lang.String getTail(java.util.logging.Handler);
> >  public java.lang.String formatMessage(java.util.logging.LogRecord);
> >  public static void main(java.lang.String[]) throws java.lang.Exception;
> >  public static void go() throws java.lang.Exception;
> >  static {};
> > }
> >
> >> On May 8, 2015, at 3:31 PM, Mark <javamark@gmail.com> wrote:
> >>
> >> I looking to use Solr search over the byte code in Classes and Jars.
> >>
> >> Does anyone know or have experience of Analyzers, Tokenizers, and Token
> >> Filters for such a task?
> >>
> >> Regards
> >>
> >> Mark
> >
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message