WycliffeAssociates
/
en_ulb_tagged
9
0
Fork 2
Files Issues 4 Pull Requests Releases Wiki Activity See in Reader

Updated the buld scripts to Stephen's local setup

This commit is contained in:
stephendillard 2021-04-02 11:06:18 -04:00
parent 3c524d09ec
commit e1fac9168b
3 changed files with 11 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
Build_ULB_XML_for_Tagging.pl Normal file → Executable file
View File

@ -9,10 +9,10 @@ use utf8;
use open IO => ":utf8";
open(LOG, ">Logs/Log.txt") or die "$!";
open(OUT, ">/Users/Henry/Documents/WACS/en_ulb_tagged/ULB_xml/ULB.xml") or die "$!";
open(OUT, ">/Users/dillardfam/Documents/WA/WACS/fork/ULB_xml/ULB.xml") or die "$!";
say OUT "<xml>";
my ($topDir, $outDir) = ("/Users/Henry/Documents/WACS/en_ulb", "/Users/Henry/Documents/WACS/en_ulb_tagged/ULB_xml");
my ($topDir, $outDir) = ("/Users/dillardfam/Documents/WA/WACS/en_ulb", "/Users/dillardfam/Documents/WA/WACS/fork/ULB_xml");
my @filesToRun = ();
my $filePattern = '\.usfm' ;

9
Check_ULB.pl
View File

@ -14,7 +14,7 @@ open(OUT2, ">Output/Tagged.txt") or die "$!";
my ($ULBxml, $taggedULBDir) = ("/Users/dillardfam/Documents/WA/WACS/fork/ULB_xml/ULB.xml", "/Users/dillardfam/Documents/WA/WACS/fork/Manual_Tagging");
my (@filesToRun) = ();
my %fullName;
my $filePattern = "\.xml" ;
my $filePattern = "\55-1TI.xml" ;
my $file;
my $xmlText = read_file("$ULBxml", binmode => 'utf8');
@ -37,7 +37,6 @@ sub GetGist {
my ($verseRef, $standard, $tagged);
say LOG "|$fileName|, |$fullName{$fileName}|";
# while ($wholeTaggedText =~ /<verse name="($fullName{$fileName} \d+:\d+)">((.|\n)*?)<preText>(.*?)<\/preText>((.|\n)*?)\n\t+((<w ((.|\n)*?)\n)*)\t+<\/verse>/sg) {
# while ($wholeTaggedText =~ /<verse name="($fullName{$fileName} \d+:\d+)">((.|\n)*?)<preText>(.*?)<\/preText>((.|\n)*?)\n\t+((<w(>| )((.|\n)*?)\n)*)\t+<\/verse>/sg) {
while ($wholeTaggedText =~ /<verse name="($fullName{$fileName} \d+:\d+)">((.|\n)*?)<preText>(.*?)<\/preText>((.|\n)*?)\n\t+((<(w|usfm)(>| )((.|\n)*?)\n)*)\t+<\/verse>/sg) {
my ($preText, $gist) = ($4, $7);
$gist =~ s/<comment>.*?<\/comment>//sg;
@ -78,9 +77,9 @@ sub Untag {
$txt =~ s/(√|<[^<>]*>)//g;
$txt = $pre . " " . $txt;
$txt =~ s/ {2,}/ /g;
$txt =~ s/— /—/g;
$txt =~ s/^(.+[^ ])(\\)/$1 $2/g;
$txt =~ s/ +$//;
$txt =~ s/— +/—/g;
# say LOG "<1>\t$txt";
return $txt;
}
@ -117,7 +116,7 @@ say "\nDone.";
__DATA__
#41-MAT Matthew
#42-MRK Mark
43-LUK Luke
#43-LUK Luke
#44-JHN John
#45-ACT Acts
#46-ROM Romans
@ -125,7 +124,7 @@ __DATA__
#48-2CO 2 Corinthians
#49-GAL Galatians
#50-EPH Ephesians
51-PHP Philippians
#51-PHP Philippians
#52-COL Colossians
#53-1TH 1 Thessalonians
#54-2TH 2 Thessalonians

10
Construct_auto-tagged_ULB_XML_files_from_unified_ULB_XML_and_tWs_and_OGNT.pl
View File

@ -17,11 +17,11 @@ $" = "\n";
mkdir "Logs";
open(LOG, ">Logs/tW_pairs.txt") or die "$!";
my $ULBfile = "/Users/Henry/Documents/WACS/en_ulb_tagged/ULB_xml/ULB.xml";
my $topDirOGNT = "/Users/Henry/Documents/WACS/OGNT";
#my $topDirOGNT = "/Users/Henry/Documents/WACS/en_ulb_tagged/Tag_test";
my $topDirtW = "/Users/Henry/Documents/WACS/en_tw/bible";
my ($outDir, $outFile) = ("/Users/Henry/Documents/WACS/en_ulb_tagged/Auto-tagged", "");
my $ULBfile = "/Users/dillardfam/Documents/WACS/fork/ULB_xml/ULB.xml";
my $topDirOGNT = "/Users/dillardfam/Documents/WACS/OGNT";
#my $topDirOGNT = "/Users/dillardfam/Documents/WACS/fork/Tag_test";
my $topDirtW = "/Users/dillardfam/Documents/WACS/en_tw/bible";
my ($outDir, $outFile) = ("/Users/dillardfam/Documents/WACS/fork/Auto-tagged", "");
my ($ULBText, $workText, $language);
my ($file);
my (%ULBtextThisVerse, %ULBpreTextThisVerse, %SNsThisVerse, %entriesThisSN, %longName);