DicControls.pm

#!/usr/bin/perl

package DicControls;

use warnings;
use strict;

use DicGlobals;
use Dic2Screen;
use DicPrepare;
use DicConversion;
use DicHelpUtils;

## Controls from module Dic2Screen
# Control the verbosity debug messages
( $isDebug, $isDebugVerbose, $isDebugVeryVerbose )       = ( 1, 0, 0 );  
# Control the verbosity info messages
( $isInfo, $isInfoVerbose, $isInfoVeryVerbose )          = ( 1, 0, 0 );  
# Control for sub die2()
$isRealDead = 1; # Some errors should kill the program. However, somtimes you just want to convert.

## Controls from module DicPrepare

# Controls for cleanseAr
$DebugKeyWordCleanseAr = '<k>φλέως</k>'; # In cleanseAr only extensive debug messages for this entry are shown. E.g. '<k>φλέως</k>'
$isCutDoneWithTidyXML             = 0; # Enables or disables the cutting of a line for the pocketbook dictionary in cleanseAr. Still experimental, so disable.
$isRemoveWaveReferences           = 1; # Removes all the references to wav-files Could be encoded in Base64 now.
# This controls the maximum article length. If set too large, the old converter will crash and the new will truncate the entry.
$max_article_length = 64000;
# This controls the maximum line length. If set too large, the converter wil complain about bad XML syntax and exit.
$max_line_length = 4000; # In bytes as generated by: length( encode('UTF-8', $line) )
$remove_color_tags = 0; # Not all viewers can handle color/grayscale. Removing them reduces the article size considerably. Relevant for pocketbook dictionary.

# Controls for loadXDXF
$isLoadFromPseudoFileName = 0; # Controls whether to skip converting to XDXF if an XDXF-file is already found prepared.

# Control variable makeKoreaderReady
# Sometimes koreader want something extra. E.g. create css- and/or lua-file, convert <c color="red"> tags to <span style="color:red;">
$isMakeKoreaderReady                         = 1 ; 
$isMakeKoreaderReady_SpanColor2Style         = 0 ;
$isMakeKoreaderReady_SpanWidth2Style         = 0 ;
$isMakeKoreaderReady_SpanStyleWidht2Padding  = 0 ;
$isMakeKoreaderReady_MergeStyles             = 0 ;
$isChangeTable2Div4Koreader                  = 1 ; # Adds lines to lua-file

# Controls for reconstructXDXF
# Controls manual input: 0 disables.
( $lang_from, $lang_to, $format ) = ( "eng", "eng" ,"" ); # Default settings for manual input of xdxf tag.
$reformat_full_name  = 1 ; # Value 1 demands user input for full_name tag.
$reformat_xdxf       = 1 ; # Value 1 demands user input for xdxf tag.

## Controls from module DicConversion

# Control variables for convertABBYY2XDXF, the conversion of ABBYY-generated HTML.
$isABBYYWordlistNeeded   = 1; # Controls creation of an ABBYYWordlist.txt file.
$isABBYYAllCleared       = 0; # Controls creation of a hash-file.
$isABBYYConverterReuse   = 0; # Controls the check for already generated xdxf-file
$isABBYConverted         = 0; # Global variable that gets set to 1 if convertABBYY2XDXF returns an xdxf-array.
# Conversion pauses during keywords
@ABBYYConverterPauseFor = (
# E.g.,
    # 'égard',
    # 'ète',
);
# Manual overrule. Conversion checks whether keyword is allowed and passes it without further tests.
@ABBYYConverterAllowedKeys = (
    q~corbeille-d’argent~,
    q~crespelé, e~,
    q~cul-rond~,
    q~desquels, desquelles~,
    q~duquel~,
    q~fœhn~,
    q~giboyeux, euse~,
    q~glacial, e, als~,
    q~hydro-. V~,
    q~inaliénablement~,
    q~in aliéné, e~,
    q~laquelle~,
    q~melliflu, e~,
    q~peu chère~,
    q~pick-nick n.m.~,
);

# Controls for convertCVStoXDXF
# Deliminator for CSV files, usually ",",";" or "\t"(tab).
$CVSDeliminator = ",";

# Controls for convertHTML2XDXF
$DebugKeyWordConvertHTML2XDXF = "Gewirr"; # In convertHTML2XDXF only debug messages from this entry are shown. E.g. "Gewirr"
$isConvertDiv2SpaninHTML2DXDF= 0 ;
$isConvertFont2Small         = 0 ;
$isConvertFont2Span          = 0 ;
$isConvertMMCFullText2Span   = 1 ;

# Controls for convertImage2Base64
$isConvertGIF2PNG            = 0 ; # Creates a dependency on Imagemagick "convert".

# Controls for convertIMG2Text, the conversion of scanned text images to text.
$isManualValidation = 1; # Manually validate OCRed images.
$isRemoveUnSubstitutedImageString = 1; # If not substitution for the imagestring was made, it can be removed or not.
$isRemoveUnSourcedImageStrings    = 1; # Remove Imagestrings without a src file.

# Controls for generateXDXFTagBased
$HigherFrequencyTags = 10 ; # Tags below this frequency, e.g. 10 times, are considered lower frequency.
$isDeleteLowerFrequencyTagsinFilterTagsHash = 0 ; # And the consequeces of that can be toggled, too.
$isExcludeImgTags    = 1 ; # <img.../>-tags are removed if toggle is positive.
$isgatherSetsVerbose = 0 ;    # Controls verbosity of tag functions
$isRemoveMpbAndBodyTags = 0 ; # <mbp...> and <body>-tags are removed if toggle is positive.
@KnownStylingTags = ( 'a', 'i', 'b', 'font'); # tags to be skipped if $isSkipKnownStylingTags = 1.
$isSkipKnownStylingTags = 1 ; # <b>-, <i>-tags and such are usually not relevant for structuring lemma/definition pairs. However, <font...>-tags sometimes are. So check.
$MinimumSetPercentage = 80 ; # A tag-set should be at least this percentage to be considered the outer tags for an article.
$LowFrequencyCriterium = 100; # Used in sub gatherSets

## Controls from module DicHelpUtils

# Controls escapeHTMLString and unEscapeHTMLString
$isEscapeHTMLCharacters             = 1;
$unEscapeHTML                     = 1;

# Controls for convertMobiAltCodes
$isConvertMobiAltCodes = 0; # Apparently, characters in the range of 1-31 are displayed as alt-codes in mobireader.

## Shortcuts to Collection of settings.

# If you select both settings, they will be ignored.
our $Just4Koreader   = 0 ;
our $Just4PocketBook = 1 ;

if( $Just4Koreader and !$Just4PocketBook){
    # Controls for Stardict dictionary creation and Koreader stardict compatabiltiy
    $isCreateStardictDictionary = 1; # Turns on Stardict text and binary dictionary creation.
    $SameTypeSequence = "h"; # Either "h" or "m" or "x".
    $updateSameTypeSequence = 1; # If the Stardict files give a sametypesequence value, update the initial value.
    $isConvertColorNamestoHexCodePoints = 1; # Converting takes time.
    $isMakeKoreaderReady = 1; # Sometimes koreader want something extra. E.g. create css- and/or lua-file, convert <c color="red"> tags to <span style="color:red;">

    # Controls for Pocketbook conversion
    $isCreatePocketbookDictionary = 0; # Controls conversion to Pocketbook Dictionary dic-format
    $remove_color_tags = 0; # Not all viewers can handle color/grayscale. Removing them reduces the article size considerably. Relevant for pocketbook dictionary.
    $max_article_length = 640000;
    $max_line_length = 8000;

    # Controls for recoding or deleting images and sounds.
    $isRemoveWaveReferences = 1; # Removes all the references to wav-files Could be encoded in Base64 now.
    $isCodeImageBase64 = 0; # Some dictionaries contain images. Encoding them as Base64 allows coding them inline. Only implemented with convertHTML2XDXF.
    $isConvertGIF2PNG = 0; # Creates a dependency on Imagemagick "convert".

    $unEscapeHTML = 0;
    $ForceConvertNumberedSequencesToChar = 1;
    $ForceConvertBlockquote2Div = 0;
    $isEscapeHTMLCharacters = 0;
}
if( $Just4PocketBook and !$Just4Koreader){
    # Controls for Stardict dictionary creation and Koreader stardict compatabiltiy
    $isCreateStardictDictionary = 0; # Turns on Stardict text and binary dictionary creation.
    $SameTypeSequence = "h"; # Either "h" or "m" or "x".
    $updateSameTypeSequence = 1; # If the Stardict files give a sametypesequence value, update the initial value.
    $isConvertColorNamestoHexCodePoints = 0; # Converting takes time and space
    $isMakeKoreaderReady = 0; # Sometimes koreader want something extra. E.g. create css- and/or lua-file, convert <c color="red"> tags to <span style="color:red;">

    # Controls for Pocketbook conversion
    $isCreatePocketbookDictionary = 1; # Controls conversion to Pocketbook Dictionary dic-format
    $remove_color_tags = 1; # Not all viewers can handle color/grayscale. Removing them reduces the article size considerably. Relevant for pocketbook dictionary.
    $max_article_length = 64000;
    $max_line_length = 4000;
    
    # Controls for recoding or deleting images and sounds.
    $isRemoveWaveReferences = 1; # Removes all the references to wav-files Could be encoded in Base64 now.
    $isCodeImageBase64 = 1; # Some dictionaries contain images. Encoding them as Base64 allows coding them inline. Only implemented with convertHTML2XDXF.
    $isConvertGIF2PNG = 0; # Creates a dependency on Imagemagick "convert".

    $unEscapeHTML = 1;
    $ForceConvertNumberedSequencesToChar = 1;
    $ForceConvertBlockquote2Div = 1;
    $isEscapeHTMLCharacters = 0;
}
1;