By extracting the following data from the games:

• Game TAGs (Player names, ELOs, site, event, ...)
• If the game doesn't begin with the standard starting position, the Ocr attempts to detect the starting position based on the associated image, and if it is successful in detecting it, that starting position is assigned to the game
• Game variants (using parentheses, brackets and braces to determine the subvariants)

The games in the books must be in algebraic notation or figurine algebraic notation. You can choose from several languages for the notation, and if you don't see a language, you can manually specify the letters to identify each type of piece.

Yes, in the application's main base directory, there is a subfolder _scripts.
There you can find sample scripts to prompt the command interface of the application(pdf2pgn)

Yes, version v1.26 is planned to extract games in figurine algebraic notation.
You will have to train the piece identifier first, but it is a relatively quick process.
Keep in mind that the quality of the extracted games in this notation may not be very good This is mainly due to:

• the poor text quality of the PDF (sometimes there are spaces in the middle of the words, or some characters are switched with others)
• And because when brackets or square brackets aren't used to define the subvariants in the games, it's more difficult to extract the correct game tree

It is done through a complex configuration based on regular expressions.
A basic set up has been created based on some formats of TAGs that are repeated throughout the books, which can be used if they match the format of the TAGs in your books.
If no suitable format exists to export the TAGS from your books, the new format can always be added to the set up (if you contact me, I am more than happy to add this new format to the basic set up for later versions of the application)
You can also attempt to set up the new format yourself (You can read how to do so in the user manual )

The Ocr works based on the type of board.
The type of board is obtained from some parameters of the image:

• size of the squares in the image
• color of the white squares
• color of the black squares

To get the type of board, the board grid is first needed (the corners of the squares that make up the board)
Once taken, the type of board is obtained, based on the previous parameters.
For each type of board, a list of normalized images exists (with a concrete size, which always stays the same at 10x10) for each type of piece and square color.
This pattern list is obtained by training the OCR with pairs (FEN chain, image).
For each board square to be recognized, the code of the piece whose image returns the root mean square error (RMSE) message is chosen, provided that it is within the threshold.
If successful, the FEN chain of the position is generated
If unsuccessful, we can ask the user to "train" the OCR by opening a window with a board where the user can identify the unidentified pieces

The OCR trains with pairs of (FEN chain, Image).
These pairs can appear from:

• self-training that the application does when the games are exported from a pdf (with the known positions of the images detected, from the exported variant up to the point of the images in games that have started since the initial position
• when the FEN of an image requested by the user cannot be detected, but the grid of the board could, the application offers the user the option to train the OCR by setting the position on the board

The application hosts a register of configured UIC engines.
To set up a new one, you need to:

• download it (For example, in this URL you can find some of them. )
• unzip it onto your hard drive
• Open the set up window of the application
• Select the "engines" tab
• click the "new engine" button (+)
• Assign the engine binary (open the JFileChooser in a new window (button:(...)), and select an .exe file of the place where you have unzipped the engine)
• From there, you can edit the engine set up

Yes, to do so:

• open the analysis window
• create a new analysis panel (button (+))
• choose an engine configuration

Yes, to do so, you only have to:

• open the game analysis window (in the Game Menu)
• set the analysis details in the window that appears
• after the analysis, the selected game is copied, with the main variant already analyzed

Yes, you can play against an engine.

• Start a new game(Game Menu -> New Game).
• Make yourself a player.
• Choose a customized engine as an opponent

Suggestion: Some engines can be set up so that they can run at or below their maximum ELO level.
It's recommended to use the application's set up option of the engines, in the case that the engine plays "too" well

Yes, you can open it in the application itself, or in this link.

ChessPdfBrowser has native language version in:

• Spanish
• English
• Catalan

And translation created from a machine translator in:

• French
• German
• Portuguese
• Italian
• Greek
• Ukrainian
• Russian
• Japanese
• Chinese
• Hindi
• Arabic