/*
* [Randomiser.java]
*
* Summary: Select a "random" quotation from a collection of quotations.
*
* Copyright: (c) 2009-2017 Roedy Green, Canadian Mind Products, http://mindprod.com
*
* Licence: This software may be copied and used freely for any purpose but military.
* http://mindprod.com/contact/nonmil.html
*
* Requires: JDK 1.8+
*
* Created with: JetBrains IntelliJ IDEA IDE http://www.jetbrains.com/idea/
*
* Version History:
* 1.8 2009-02-06 include go package in ZIP bundle.
* will be inserted, and where on the page it will be inserted.
* this is not a Macro. It is a collection of methods to help select random quotations,
* random advertisement,
* PSAs, that change periodically. The selections are not random. They produce the same results every
* time
* the code is run. This allows choices to remain stable for weeks at a time. There is nothing
* specific to
* ads or quotations in this class.
*/
package com.mindprod.htmlmacros.support;
import com.mindprod.common18.FNV1a64Digester;
import java.io.File;
import static java.lang.System.*;
/**
* Select a "random" quotation from a collection of quotations.
*
* Selection is based on the time and the digest of the name of the file where the quotation is being inserted and
* where on the page it is being inserted.
*
* It is not really random. It is completely reproducible.
*
* This is not a Macro. It is a collection of methods to help select random quotations, random advertisement,
* PSAs, that change periodically. The selections are not random. They produce the same results every time
* the code is run. This allows choices to remain stable for weeks at a time. There is nothing specific to
* ads or quotations in this class.
*
* @author Roedy Green, Canadian Mind Products
* @version 1.8 2009-02-06 include go package in ZIP bundle.
* @since 2009
*/
class Randomiser
{
// todo: convert to enum
// declarations
private static final boolean DEBUGGING = false;
/**
* which invocation on the page is this, 0 is first 1 is second etc.
*/
private static int sequenceOnPage = 0;
/**
* previous fileBeingProcessed
*/
private static File prevFileBeingProcessed = null;
// /declarations
// methods
/**
* Generate a 32 bit random digest number that roughly speaking depends on letters in file name,
* where we are on page, current time, changeIntervall. So it will stay stable for a while.
*
* @param changeIntervalInMillis how often to change the ad or quotation in millis.
* @param fileBeingProcessed the file currently being processed.
*
* @return random 32-bit number that changes when the file name changes, and changes over time,
* once every changeIntervalInMillis
* but usually remains stable.
*/
public static int getHashForFilename( final long changeIntervalInMillis, final File fileBeingProcessed )
{
// Why is this so complicated?
// 1. we want to change psa ads, quotations, non-mil etc on a
// regular basis, some pages more frequently.
// 2. we don't want all pages changing at once triggering big uploads, so we stagger them.
// 3. if the quotations change, even if the hash do does not change, the chosen quote
// will change, triggering mass updates. So we choke this back by only updating
// a random selection for each run using set resetfooterpercentage.
if ( fileBeingProcessed.equals( prevFileBeingProcessed ) )
{
sequenceOnPage++;
}
else
{
sequenceOnPage = 0;
prevFileBeingProcessed = fileBeingProcessed;
}
// Used both for ads and quotations. Don't insert any code there specific to either.
// We want to want to ensure each ad rotates at a given interval, but not all ads should
// change on the same day. We want to make sure quotation at top and bottom of the page
// change at the same time, but don't select the same quotation.
// Use SHA-1 instead of CRC32 to get better scrambling.
// We are not hashing the content of any page, just the file name.
// throw in hashCode so all randoms won't change at once.
// seed will change every changeIntervalInMillis. The value itself has no physical meaning.
// if the changeInterval is greater than a day, we arrange things so that the change occurs at midnight
// UTC, and
// roughly 1/n of them change on any given day, where n is the change interval in days.
// if the changeInterval is greater than an hour, we arrange things so that the change occurs on the hour
// UTC,
// roughly 1/n of them change on any given hour, where n in the change interval in hours.
// we adjust the file timestamp by a stable random amount so all hashes will not change at once
// for Quotations, this will be called twice, once to select the flock and once to select
// the the quotation within the flock.
final int fileHash = fileBeingProcessed.hashCode();
final long accelerateChangeByMillis = ( fileHash & Integer.MAX_VALUE )
% changeIntervalInMillis;
// seed will change only once every changeIntervalMillis, but not all quotes/ads will change at once
// because they have different accelerateChangeByMillis.
// seed is like a timer that increments every changeInterval
final long now = System.currentTimeMillis();
final int seed = ( int ) ( ( now + accelerateChangeByMillis ) /
changeIntervalInMillis );
// hash 4-bytes of the seed in big-endian order.
FNV1a64Digester digester = new FNV1a64Digester();
digester.updateInt( seed );
digester.updateInt( fileHash );
// hash in index on page.
// only low bits will be occupied.
digester.update( ( sequenceOnPage * 41 ) & 0xff );
// extract scrambled digest
final long digest = digester.getValue();
// collapse 64-bit hash down to 32
final int hash = ( int ) ( ( digest >>> 32 ) ^ digest );
if ( DEBUGGING )
{
out.println( "\nhash:" + hash
+ " fileHash:" + fileHash
+ " interval:" + changeIntervalInMillis
+ " seed:" + seed
+ " file:" + fileBeingProcessed.getAbsolutePath()
+ " seq:" + sequenceOnPage );
}
return hash;
}
// /method
/**
* given a 32-bit hash/digest, returns an a selection number.
*
* @param hash 32 bits, possibly signed.
* @param possibilities how many possible choices there are
*
* @return a repeatable number in the range 0..possibilities.-1, If possibilities is 0, returns -1.
*/
@SuppressWarnings( { "WeakerAccess" } )
public static int getRandomSelectorIndexForHash( final int hash, final int possibilities )
{
if ( possibilities == 0 )
{
return -1;
}
// strip sign bit and take modulus
return ( hash & Integer.MAX_VALUE ) % possibilities;
}
// /method
/**
* given a 32-bit hash/digest, returns an a selection number.
*
* @param hash 32 bits, possibly signed.
* @param weights weight for choices 0..n-1. Not necessarily normalised to 100.
*
* @return a repeatable number in the range 0..n-1 based on hash.
*/
@SuppressWarnings( { "WeakerAccess" } )
public static int getWeightedSelectorIndexForHash( final int hash, final int... weights )
{
// we need to normalise the weights
int sum1 = 0;
for ( int weight : weights )
{
sum1 += weight;
}
// generate a randomised number 0..sum1-1
final int cutoff = ( hash & Integer.MAX_VALUE ) % sum1;
// if weights were 5, 10, 5, then sum is 20 and assignments
// 0..4 -> 0
// 5..14 -> 1
// 15..19 -> 2
// code works even if some weights are zero.
int sum2 = 0;
for ( int choice = 0; choice < weights.length; choice++ )
{
sum2 += weights[ choice ];
if ( sum2 > cutoff )
{
return choice;
}
}
assert false : "program should never get here";
return weights.length - 1;
}
// /method
// /methods
}