com.redhat.rhn.common.filediff

Class Trace

java.lang.Object

com.redhat.rhn.common.filediff.Trace

public class Trace
extends java.lang.Object

A single diff trace through two files. Represents one trace of a linked list of traces.

Constructor Summary

Constructors

Constructor and Description
Trace(int oldSize, int newSize)

Method Summary

Modifier and Type	Method and Description
int	bestPossible()
java.util.List<Hunk>	createHunks(java.lang.String[] oldFile, java.lang.String[] newFile) Since the diff was performed backwards, "popping" the resulting edits from the backwards tree gives them in forward order.
int	getMatches()
boolean	isDone()
Trace	next()
void	setNext(Trace nextIn)
boolean	step(java.lang.String[] oldFile, java.lang.String[] newFile) Step once in this trace.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

Trace

public Trace(int oldSize,
             int newSize)

Parameters:

oldSize - The size of the old file.

newSize - The size of the new file.

Method Detail

next

public Trace next()

Returns:

The next trace in the linked list.

setNext

public void setNext(Trace nextIn)

Parameters:

nextIn - The next trace in the linked list.

getMatches

public int getMatches()

Returns:

The number of matched lines in this trace.

isDone

public boolean isDone()

Returns:

whether this trace has terminated.

bestPossible

public int bestPossible()

Returns:

The best possible number of matched lines for this trace.

step

public boolean step(java.lang.String[] oldFile,
                    java.lang.String[] newFile)

Step once in this trace. The power of this algorithm is the fact that different traces are explored in parallel. This method is recursive while it keeps finding matching lines. This is because finding matching lines makes this trace much more possibly optimal. This behaviour is what Myers called the "furthest reaching D-path"

This step method steps backward through the files. This is for the simple reason that when creating hunks, we want to create them in forward order, but we have to visit the edits backwards from how we diffed them. Two backwards make a forward, two negatives make a positive, and two wrongs make a right.

Parameters:

oldFile - The old(first, from) file

newFile - The new(second, to) file

Returns:

whether this step forked. (needed for incrementation by the step controller)

createHunks

public java.util.List<Hunk> createHunks(java.lang.String[] oldFile,
                                        java.lang.String[] newFile)

Since the diff was performed backwards, "popping" the resulting edits from the backwards tree gives them in forward order.

Parameters:

oldFile - The old(first, from) file

newFile - The new(second, to) file

Returns:

A list of hunks representing the edit to make oldFile into newFile.