Perl script to mark lines in lg/sp elements

jhellingman · Jan 5, 2025 · 12bca16 · 12bca16
1 parent 3912bda
commit 12bca16
Show file tree

Hide file tree

Showing 3 changed files with 106 additions and 2 deletions.
diff --git a/modules/divisions.xsl b/modules/divisions.xsl
@@ -100,6 +100,17 @@
     </xsl:template>
 
 
+    <xsl:function name="f:has-head" as="xs:boolean">
+        <xsl:param name="div"/>
+        <xsl:sequence select="$div/head[not(f:rend-value(./@rend, 'display') = 'image-only')]"/>
+    </xsl:function>
+
+    <xsl:function name="f:determine-output-div-element" as="xs:string">
+        <xsl:param name="div"/>
+        <xsl:sequence select="if (f:is-html5() and f:has-head($div)) then 'section' else 'div'"/>
+    </xsl:function>
+
+
     <xd:doc>
         <xd:short>Format head element of div0.</xd:short>
         <xd:detail>Format a head element for a <code>div0</code>. At this level we still set the running header (for ePub3).</xd:detail>

diff --git a/tei2imageinfo.xsl b/tei2imageinfo.xsl
@@ -71,8 +71,6 @@
     <xsl:variable name="p.element" select="'p'"/>
 
 
-
-
     <xsl:template match="/">
         <images>
             <xsl:apply-templates mode="imageinfo"/>

diff --git a/tools/tagPoetryLgSp.pl b/tools/tagPoetryLgSp.pl
@@ -0,0 +1,95 @@
+# poetry.pl -- add <l> tags on new-lines in poetry / drama, based on the presence of <lg> or <sp> tags.
+
+use strict;
+use warnings;
+
+my $lgSeen = 0;
+my $spSeen = 0;
+my $spLooksLikeProse = 0;
+my $lineNumber = 0;
+
+
+while (<>) {
+    my $line = $_;
+    $lineNumber++;
+
+    if ($line =~ /^<lg>$/) {
+        if ($spSeen > 0) {
+            print STDERR "WARNING: entering <lg> inside <sp> on line $lineNumber!\n";
+        }
+        if ($lgSeen > 0) {
+            print STDERR "WARNING: entering nested <lg> ($lgSeen) on line $lineNumber!\n";
+        }
+        $lgSeen++;
+        print "<lg>\n";
+
+    } elsif ($line =~ /^<\/lg>$/) {
+        if ($lgSeen == 0) {
+            print STDERR "ERROR: unmatched </lg> on line $lineNumber!\n";
+        } else {
+            $lgSeen--;
+        }
+        print "</lg>\n";
+
+    } elsif ($line =~ /^(<sp\b.*?>)/) {
+        my $tag = $1;
+        if ($lgSeen > 0) {
+            print STDERR "WARNING: entering $tag inside <lg> on line $lineNumber!\n";
+        }
+        if ($spSeen > 0) {
+            print STDERR "WARNING: entering $tag inside <sp> on line $lineNumber!\n";
+        } else {
+            $spLooksLikeProse = 0;
+        }
+        $spSeen++;
+        print $line;
+
+    } elsif ($line =~ /^<\/sp>$/) {
+        if ($spSeen == 0) {
+            print STDERR "ERROR: unmatched </sp> mode on line $lineNumber!\n";
+        } else {
+            $spSeen--;
+        }
+        print "</sp>\n";
+
+
+    } elsif ($lgSeen == 0 && $spSeen == 0 || ($spSeen > 0 && $spLooksLikeProse == 1)) {
+        print $line;
+
+    } elsif ($line =~ /^<speaker/) {
+        print $line;
+
+    } else {
+        if ($spSeen > 0 && $line =~ /^[a-z]/) {
+            print STDERR "WARNING: <sp> looks like prose on line $lineNumber!\n";
+            $spLooksLikeProse = 1;
+            print $line;
+        } elsif ($line =~ /^\s*$/) {
+            # blank line in poetry mode:
+
+            print "</lg>\n\n<lg>\n";
+        } elsif ($line =~ /^\s*<l>/) {  # Already tagged
+            print $line;
+        } else {
+            # count white space before
+            $line =~ /^(\s*)(.*)$/;
+            my $spaces = $1;
+            $line = $2;
+            my $n = length($spaces);
+            if ($n == 0) {
+                print "    <l>$line\n";
+            } else {
+                print "    $spaces<l rend=\"indent($n)\">$line\n";
+            }
+        }
+    }
+}
+
+if ($spSeen != 0) {
+    print STDERR "ERROR: open <sp> tags ($spSeen) at end of file!\n"
+}
+
+if ($lgSeen != 0) {
+    print STDERR "ERROR: open <lg> tags ($lgSeen) at end of file!\n"
+}
+