EAS - for Stefan

Using some old code, consider this as a first version, the hardest part (truncate the PGN) is already done.

Next -

1. Skip moves available in a book, analyzing 1.e4 c5 2.Nf3 etc makes no sense.
2. Stop analyzing after a blunder.

Analyzing multiple games is the next obstacle but in programming nothing is impossible.

Admin wrote:

Using some old code, consider this as a first version, the hardest part (truncate the PGN) is already done.

Next -

1. Skip moves available in a book, analyzing 1.e4 c5 2.Nf3 etc makes no sense.
2. Stop analyzing after a blunder.

Analyzing multiple games is the next obstacle but in programming nothing is impossible.

Anything to test?

Busy with net training, no free PC, but haven't forgotten, maybe within a couple of days.

https://github.com/egh-s/somu

Changes -

. Stops analysis after a blunder is found.
. Added polyglot book support to speed-up the analysis.

In book.ini you select your favorite polyglot book.

Included (and default) is ProDeo.bin which is a SF12 analyzed book I made a couple of years ago, average depth 25+ Since book moves have a score this practical means the book can detect a blunder.

Win10 doesn't allow it to work, somu.exe is quarantined and even after restoring it, clicking "start" doesn't do anything other than opening an empty DOS window. Pressing F5 does nothing, a white blinking cursor is all I get.

Oh well, another false virus alert. The start.bat shortcut needs modification with the folder name you installed somu. But clicking somu.exe should work. That is if W10 accepts it. Else I will do a different compile.

It doesn't like it. It works under Win7, but SF then stops working after a while. Anyway, Win10 is what I aim to use. Win7 is an old machine I have lying around which rarely sees any action.

I have tested it once more with the example pgn blunder-check.pgn the game you posted. It runs flawlessly with the default settings. Then exits with the output file blunder.pgn truncated as you suggested.

[Event "LSS/IECG"]
[Site ""]
[Date "2006.11.04"]
[Round "1"]
[White "Jean, Michel"]
[Black "Ong, Allan"]
[Result "0-1"]
[BlackElo ""]
[WhiteElo ""]
[PlyCount "52"]

1. e4 d5 2. exd5 Qxd5 3. Nc3 Qd6 4. d4 Bf5 5. Nf3 Nc6 6. Bb5 O-O-O 7. Be3 Nb4 8. Ba4 Nf6 9. Qd2 Nbd5 10. Nxd5 Nxd5 11. Bb3 Nxe3 12. fxe3 e6 13. O-O-O Qb6 14. h3 Bb4 15. c3 Be7 16. Ne5 Rhf8 17. Rhe1 Bg5 18. Bc2 Bxc2 19. Qxc2 Qb5 20. Qxh7 g6 21. Nxf7 Rd7 22. Qxg6 Rdxf7 23. Qxe6+ Kb8 24. c4 Qb4

----

Can you 1) check and 2) post the pgn that gives trouble?

I put a new compiled somu.exe on github, use that one.

The new binary is to Win10's liking, so I could run it. These are my findings:

- If it doesn't find a blunder, it doesn't save the entire game in the output and stops processing the rest of the PGN.
- When it finds a blunder, it stops, but it doesn't proceed with the following game, it just saves the first truncated game.
- Upon exiting, it automatically opens blunder.pgn with whichever GUI you have registered in Windows to handle those files.
- I haven't been able to use depth. When selecting level, whatever value I enter is interpreted as ms.

Edit: I just used a PGN with the first four games of twic1497.

1. To analyze with depth use the (L)evel command and type a number <100, that will be the depth.

2. To analyze more games is the last hurdle.

Upon exiting, it automatically opens blunder.pgn

Skip the option ?

If it doesn't find a blunder, it doesn't save the entire game in the output

Okay, will do.

I tried 11, and the output was showing that to be 11ms.

By skipping the option, you mean to not associate the pgn extension with a program?

Ozymandias wrote:

I tried 11, and the output was showing that to be 11ms.

Ah, a glitch, but it is depth=11, see the file temporarily file new.epd in the epd folder.

Ozymandias wrote:

By skipping the option, you mean to not associate the pgn extension with a program?

Here pgn extensions are associated with a text editor for easy checking things I program. But I can imagine that if pgn is associated with a GUI this is annoying. I could try 'notepad', it's standard windows.

Not surprised, I thought there was too much time difference between "11ms" and "20ms"

I changed it to Notepad++, which is orders of magnitude faster than the regular version.

Admin wrote:

To analyze more games is the last hurdle

Big one?

Ozymandias wrote:

Admin wrote:

To analyze more games is the last hurdle

Big one?

Nasty one.

Made a start today.

One day

Code:

//      Count PGN
//      =========
        fp=fopen(pgnnaam,"r");
pgn:    if (feof(fp)) { fclose(fp); goto split; }
        fgets(s,1990,fp);
        if (strstr(s,"[Event ")) dba++;
        goto pgn;

//      Split PGN into single games
//      ===========================
split:  strcpy(w,pgnnaam); l=strlen(w); memmove(w,w+4,l);
        sprintf(s,"copy %s %s >NULL",pgnnaam,w); system(s);
        sprintf(s,"gameSplit %s %d >xyz.xyz",w,dba);
        system(s);

Don't remember much of what I learned on BASIC, 40 years ago, but I seem to see some sort of "if-then-go to" routine. Am I wrong?

You are not wrong.

Admin wrote:

Ozymandias wrote:

Admin wrote:

To analyze more games is the last hurdle

Big one?

Nasty one.

Made a start today.

One day

Was that one day or day one?

Oh well, another weekend of cheap energy that couldn't be put to good use, with some testing. From 8:00 AM, to 19:00, the price was between 0.02418 €/kWh and 0.04167

Oh, solly. "One day" with smiley is ambiguous, it was meant to read as some day. And the latter is true, but I have little time at the moment.

Oh similes, you evasive of meaning.

Question, your goal with the tool, how many games do you intend to use in one run?

At first I got the impression you so now and then wanted to analyze some of the games played at the chess club, hence the one game setup and exit.

As I have it now a pgn collection will be split into single games first and then analyze of the games one by one. IOW, I don't it will be wise to use a PGN of 200,000 games and overload the folder with 200,000 separate pgn files.

Hence I will limit the maximum to 10,000.

The total is 15 million games, but the PGN can easily be split into any number of segments easily, with Norm's tools.

15M is serious business