Better support various em space characters

lib/PDF/Content/Text/Box.rakumod

@@ -136,7 +136,7 @@ method content-height returns Numeric { @!lines».height.sum * $.leading; }
 
 my grammar Text {
     token nbsp   { <[ \c[NO-BREAK SPACE] \c[NARROW NO-BREAK SPACE] \c[WORD JOINER] ]> }
-    token space  { [\s <!after <nbsp> > | "\c[ZERO WIDTH SPACE]"]+ }
+    token space  { [\s <!after <nbsp> > | \c[ZERO WIDTH SPACE] ]+ }
     token hyphen { <[ \c[HYPHEN] \c[HYPHEN-MINUS] \c[HYPHENATION POINT] ]> }
     token word   { [ <!hyphen> <!space> . ]+ <[ \c[HYPHEN] \c[HYPHEN-MINUS] ]>? | <.hyphen> }
 }
@@ -218,7 +218,8 @@ method !layup(@atoms is copy) {
     my Int $i = 0;
     my Int $line-start = 0;
     my Int $n = +@atoms;
-    my UInt $preceding-spaces = self!flush-spaces: @atoms, $i;
+    my $em-spaces = self!word-gap($!style.scale: 1000) / self!word-gap;
+    my Numeric $preceding-spaces = self!flush-spaces: $em-spaces, @atoms, $i;
     my $word-gap := self!word-gap;
     my $height := $!style.font-size;
     my Numeric $hyphen-width;
@@ -317,7 +318,7 @@ method !layup(@atoms is copy) {
             if $height > $line.height;
 
         $prev-soft-hyphen = $soft-hyphen;
-        $preceding-spaces = self!flush-spaces(@atoms, $i);
+        $preceding-spaces = self!flush-spaces($em-spaces, @atoms, $i);
     }
 
     if $preceding-spaces {
@@ -360,29 +361,40 @@ method !height-exceeded {
     $!height && self.content-height > $!height;
 }
 
-method !flush-spaces(@words is raw, $i is rw) returns UInt {
+my constant %SpaceWidth = %(
+    "\c[EN SPACE]"  => .5,
+    "\c[EM SPACE]"  => 1,
+    "\c[THREE-PER-EM SPACE]" => 3,
+    "\c[FOUR-PER-EM SPACE]" => 4,
+    "\c[SIX-PER-EM SPACE]" => 6,
+    "\c[THIN SPACE]" => .2,
+    "\c[HAIR SPACE]" => .1,
+    "\c[ZERO WIDTH SPACE]" => 0,
+);
+
+method !flush-spaces($em-spaces is rw, @words is raw, $i is rw) returns Numeric:D {
     my $n = 0; # space count for padding purposes
     with  @words[$i] {
         when /<Text::space>/ {
-            $n = .chars;
             if $!verbatim && (my $last-nl = .rindex("\n")).defined {
                 # count spaces after last new-line
-                $n -= $last-nl + 1;
-                $n = 0 if $!squish;
+                $n = .substr($last-nl+1).comb.map({do with %SpaceWidth{$_} { $_ * $em-spaces } // 1}).sum
+                    unless $!squish;
             }
             else {
                 $i++;
-                $n = 1 if $!squish;
-                $n = 0 if .contains("\c[ZERO WIDTH SPACE]");
+                $n = .comb.map({do with %SpaceWidth{$_} { $_ * $em-spaces } // 1}).sum;
+                $n = 1 if $n > 1 && $!squish;
+dd [.uniname, :$n, :$i];
             }
         }
     }
     $n;
 }
 
 # calculates actual spacing between words
-method !word-gap returns Numeric {
-    my $word-gap = $.space-width + $.WordSpacing + $.CharSpacing;
+method !word-gap($space = $.space-width) returns Numeric {
+    my $word-gap = $space + $.WordSpacing + $.CharSpacing;
     $word-gap * $.HorizScaling / 100;
 }
 

lib/PDF/Content/Text/Line.rakumod

@@ -49,7 +49,7 @@ has Numeric $.word-width is rw = 0;
 has Numeric $.word-gap is rw = 0;
 has Numeric $.indent is rw = 0;
 has Numeric $.align = 0;
-has UInt @.spaces;
+has Numeric @.spaces;
 
 method content-width returns Numeric {
     $!indent  +  $!word-width  +  @!spaces.sum * $!word-gap;
@@ -153,9 +153,12 @@ method content(:$font!, Numeric :$font-size!, :$space-pad = 0, :$TextRise = 0.0)
     for ^+@!encoded -> $i {
         my $spaces := @!spaces[$i];
         if $spaces {
-            @line.push: $font.encode(Space x $spaces);
-            @line.push: $space-pad * $spaces
-                unless $space-pad =~= 0;
+            my UInt $whole-spaces = $spaces.floor;
+            my $part-spaces = $spaces - $whole-spaces;
+            @line.push: $font.encode(Space x $whole-spaces);
+            my Int $pad = round($space-pad * $spaces + -1000 * $part-spaces * $!word-gap / $font-size);
+            @line.push: $pad
+                if $pad;
         }
         @line.append: @!encoded[$i].list;
     }

lib/PDF/Content/Text/Style.rakumod

@@ -70,19 +70,19 @@ multi method baseline-shift(Baseline $_ --> Numeric) {
 #| get/set a numeric font vertical alignment offset
 multi method baseline-shift is rw { $!TextRise }
 
+method scale($v) { $v * $!font-size / $!units-per-EM; }
+
 #| return the scaled width of spaces
-method space-width {
-    $!space-width * $!font-size / $!units-per-EM;
-}
+method space-width { self.scale: $!space-width; }
 
 #| return the scaled underline position
 method underline-position {
-    ($!font.underline-position // -100) * $!font-size / $!units-per-EM;
+    self.scale: ($!font.underline-position // -100)
 }
 
 #| return the scaled underline thickness
 method underline-thickness {
-    ($!font.underline-thickness // 50) * $!font-size / $!units-per-EM;
+    self.scale: ($!font.underline-thickness // 50)
 }
 
 #| return the scaled font height

t/text-box.t

@@ -1,6 +1,6 @@
 use v6;
 use Test;
-plan 14;
+plan 15;
 use lib 't';
 use PDF::Grammar::Test :is-json-equiv;
 use PDF::Content::Text::Box;
@@ -152,6 +152,17 @@ subtest 'zero width spaces', {
     }
 }
 
+subtest 'variable spaces', {
+    $gfx.text: {
+        $text = "Spaces:en-space\c[EN SPACE]space tab\tem-space\c[EM SPACE]em-quad\c[EM QUAD]three\c[THREE-PER-EM SPACE]four\c[FOUR-PER-EM SPACE]six\c[SIX-PER-EM SPACE]thin\c[THIN SPACE]hair\c[HAIR SPACE]zero\c[ZERO WIDTH SPACE]. " x 2;
+        my $width = 400;
+        $height = 100;
+        $text-box .= new( :$text, :$font, :$font-size, :$width, :$height );
+        .text-position = 100, 250;
+        .say: $text-box;
+    }
+}
+
 subtest 'font loading from content stream', {
     if (try require PDF::Font::Loader) === Nil {
         skip 'PDF::Font::Loader is needed for this test';