chess-web/chessParser
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

more cleanPgn() cleanup: trim lines, del initial blanks, one space only...

Browse Source
This commit is contained in:
Bruno Raoult
2023-06-18 21:01:59 +02:00
parent ce0923d0e9
commit 8187168ca2
2 changed files with 19 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
PgnGameParser.php
View File

@@ -4,6 +4,7 @@
class PgnGameParser{ class PgnGameParser{
private $pgnGame; private $pgnGame;
private $moveBuilder;
private $defaultFen = 'rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1'; private $defaultFen = 'rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1';
private $gameData = array(); private $gameData = array();

32
PgnParser.php
View File

@@ -75,9 +75,21 @@ class PgnParser
$c = str_replace("0-0-0", "O-O-O", $c); $c = str_replace("0-0-0", "O-O-O", $c);
$c = str_replace("0-0", "O-O", $c); $c = str_replace("0-0", "O-O", $c);
/* basic trimming:
* - remove initial blanks
* - keep only one space
* - trim lines
* - max 2 consecutive '\n'
*/
$c = preg_replace("/^\s+([^\s])/", "$1", $c);
$c = preg_replace("/ +/", " ", $c);
$c = preg_replace("/\s*\n\s*/", "\n", $c);
$c = preg_replace("/\n{3,}/s", "\n\n", $c);
/* replace '[' between '{' and '}' with '//--SB--//' /* replace '[' between '{' and '}' with '//--SB--//'
*/ */
$c = preg_replace('/(?:\G(?!\A)|\{)[^}[]*\K\[/', "//--SB--//", $c); $c = preg_replace('/(?:\G(?!\A)|\{)[^\}\[]*\K\[/', "//--SB--//", $c);
$c = preg_replace('/(?:\G(?!\A)|\{)[^\}\]]*\K\]/', "//--SE--//", $c);
/* set one '\n' between tags. /* set one '\n' between tags.
* This is possible because brackets within comments are protected above * This is possible because brackets within comments are protected above
@@ -85,25 +97,15 @@ class PgnParser
$c = preg_replace('/"\]\s*\[/s', "\"]\n[", $c); $c = preg_replace('/"\]\s*\[/s', "\"]\n[", $c);
/* set '\n\n' between tags and movetext section /* set '\n\n' between tags and movetext section
* set '\n\n' between non tag line and tag line (between two games)
* This is possible because brackets within comments are protected above * This is possible because brackets within comments are protected above
*/ */
$c = preg_replace('/"\]\s*([\.0-9]|{)/s', "\"]\n\n$1", $c); $c = preg_replace('/"\]\s*([\.0-9]|{)/s', "\"]\n\n$1", $c);
$c = preg_replace("/([^\]])\n+\[/", "$1\n\n[", $c);
/* remove space before '['
* This is possible because brackets within comments are protected above
*/
$c = str_replace(" +[", "[", $c);
/* set '\n\n' between non tag line and tag line (between two games)
* This is possible because brackets within comments are protected above
*/
$c = preg_replace("/([^\]])(\n+)\[/si", "$1\n\n[", $c);
/* revert brackets within movetext comments */ /* revert brackets within movetext comments */
$c = str_replace("//--SB--//", "[", $c); $c = str_replace("//--SB--//", "[", $c);
$c = str_replace("//--SE--//", "]", $c);
/* max 2 consecutive '\n */
$c = preg_replace("/\n{3,}/s", "\n\n", $c);
return $c; return $c;
} }
@@ -124,6 +126,8 @@ class PgnParser
$content = "\n\n" . $pgn; $content = "\n\n" . $pgn;
$games = preg_split("/\n\n\[/s", $content, -1, PREG_SPLIT_DELIM_CAPTURE); $games = preg_split("/\n\n\[/s", $content, -1, PREG_SPLIT_DELIM_CAPTURE);
//file_put_contents("/tmp/parsed.pgn", $content);
for ($i = 1, $count = count($games); $i < $count; $i++) { for ($i = 1, $count = count($games); $i < $count; $i++) {
$gameContent = trim("[" . $games[$i]); $gameContent = trim("[" . $games[$i]);
if (strlen($gameContent) > 10) { if (strlen($gameContent) > 10) {