<?php

/**
 * Place this script in the doc/ directory of MediaWiki and run it.
 *
 * This script creats a new file hooks.yaml from hooks.txt to make it machine readable.
 * @author akangupt < akanksha2879@gmail.com >
 */

$hookFile = 'hooks.txt';    # complete path for file hooks.txt

if ( !file_exists( "hooks.txt" ) ) {
    die( "Error: hooks.txt not found\n" );
} else {
    $hooksData = file_get_contents( $hookFile );
}

# convert all the tabs in four spaces because YAML forbids the tabs.
$data = preg_replace( "/\t/", "    ", $hooksData );

$hookYamlFile = 'hooks.yaml';
# create hooks.yaml
$handle = fopen( $hookYamlFile, 'w' ) or die( 'Cannot open file: '.$hookYamlFile.' \n' );

# split all the text in two parts.
# Explanatory text(text before the hooks) and rest of the text
$text = preg_split( "/(?=\'\w+\:*\w*\:*\w*\'\:)/", $data, 2 );
if ( $text === "" ) {
    echo "Warning: No explanatory text found in hooks.txt\n";
}

# comment the explanatory text
$explanatoryText = preg_replace( '/^/m', '# ', $text[0] );
fwrite( $handle, $explanatoryText );
hooksInYaml( $handle, $text[1], $hookFile );

function hooksInYaml( $handle, $string, $hookFile ) {

    # extract each hook from rest of the text
    $hooks = preg_split( "/\r?\n\r?\n/", $string );
    foreach ( $hooks as $hook ) {

        # if extracted hook doesn't start with a hook name 
        # i.e. explanatory part so comment that
        $isText = preg_match( "/^\s*\'(\w+\:*\w*\:*\w*)\'\s*\:/msi", $hook );
        if ( $isText != 1 ) {
            $explanatoryText = preg_replace( '/^/m', '# ', $hook );
            fwrite( $handle, $explanatoryText );
        } else {
            $pattern = "/\s*\'(\w+\:*\w*\:*\w*)\'\s*:\s*(DEPRECATED)*!*(\r?\n)*\s*(.*?)(?:(?=\n\&\\$\w+\s*\:|\n\\$\w+\s*\:|\Z))\s*(?:(.*?)\Z)/msi";
            
            # $spaces - indentation which should be added to each line
            $spaces = "    ";

            if ( preg_match( $pattern, $hook, $matches ) ) {

                # $matches[1] has hook's name.
                $string = $matches[1].":\n";
                if ( $matches[1] === "" ) {
                    die( "Error: Couldn't recognize the hook\n" );
                }

                # $matches[2] has information if hook is deprecated or not.
                if( $matches[2] != '' ) {
                    $string .= "  DEPRECATED";

                    if( $matches[3] === '' ) {
                        # if $matches[3] is empty
                        # i.e. there is a suggestion to use some other hooks.

                        # $matches[4] contains all the text 
                        # which is between 'DEPRECATED!' keyword and explanation of arguments. 
                        # So, to get the suggestion of other hook,
                        # split $matches[4] in description of hook and suggestion of other hook.
                        $text = preg_split( "/\.\n/", $matches[4], 2 );

                        # text[0] contains the suggestion of other hook to be used.
                        $string .= multiline( $text[0].".", $spaces );

                        # text[1] contains description of hook.
                        $string .= "\n  Description".multiline( $text[1], $spaces );
                    } else {
                        # if $matches[3] is not empty i.e. hook is deprecated 
                        # but there is no suggestion of other hook to be used in place of this hook.
                        $string .= ":\n";

                        # $matches[4] contains description of hook
                        $string .= "  Description".multiline( $matches[4], $spaces );
                      
                    }
                } else {
                    if ( $matches[4] != '' ) {

                        if (preg_match( "/^(\&\\$\w+\s*\:|\\$\w+\s*\:)/", $matches[4]) ) {
                            # This covers the case when a hook doesn't have a description.
                            # i.e. $matches[4] contains an argument.
                            $string .= "  Description:";
                            $matches[5] = $matches[4]."\n".$matches[5];
                        } else {
                            $string .= "  Description".multiline( $matches[4], $spaces );
                        }
                    }
                }
                fwrite( $handle, $string );

                # $matches[5] has all the arguments.
                splitArguments( $handle, $matches[5], $hookFile );
                fwrite( $handle, "\n\n" );
            }
        }
    }
}


# returns a string with literal block if 
# a string is multiline or has a character that 
# doesn't belong to A-Z a-z 0-9 _ . space
function multiline( $finalString, $spaces ) {
    $string = preg_match( "/\n(?!$)|[^\w\s\.]/", $finalString ) ? (": |\n") : (": ");

    if ( $string === ": " ) {
        # don't add indentation in the string if string is singal line.
        return $string.$finalString;         
    }
    $pattern = "/\n\s\s|\n(?!$)/";
    $replacement = "\n$spaces";
    $finalString = preg_replace( $pattern, $replacement, $finalString );
    return $string.$spaces.$finalString;
}

# function to split the arguments
function splitArguments( $handle, $string, $hookFile ) {
    fwrite( $handle, "\n  arguments:" );
    $pattern = "/\s*(\&\\$\w+|\\$\w+)\s*\:\s*(.*?(?:(?=\n\&\\$\w+\s*\:|\n\\$\w+\s*\:)|\Z))/msi";

    if ( preg_match_all( $pattern, $string, $matches, PREG_SET_ORDER )) {
        $spaces = "        ";
        foreach ( $matches as $match ) {
            $string = "\n    - \"".$match[1]."\"";
            $string .= multiline( $match[2], $spaces );
            fwrite( $handle, $string );
        }
    }
}

?>