add shape() method

pdf-raku · Jan 21, 2024 · 9f1bd04 · 9f1bd04
1 parent f78d145
commit 9f1bd04
Show file tree

Hide file tree

Showing 24 changed files with 187 additions and 111 deletions.
diff --git a/Changes b/Changes
@@ -1,5 +1,14 @@
 {{$NEXT}}
-   - Add auto-loading of type-1 AFM metrics
+   - Add auto-loading of type-1 AFM metrics. Look for a
+     sibling *.afm file when loading type-1 *.pfa or *.pfb
+     files and attach it.
+   - Add shape() method:
+     -- integrate HarfBuzz for TrueType and OpenType shaping
+     -- fallback to kerning + typographic ligature substitution
+        for other fonts
+     -- Add allocate-cid() and encode-cids() method to encoders.
+        To support HarfBuzz shaping, which may return non-unicode mapped
+        glyphs.
 
 0.7.8  2023-11-13T06:38:11+13:00
    - Further streamline CMAP encoding. Remove redundant code-point checks.

diff --git a/META6.json b/META6.json
@@ -6,9 +6,9 @@
   "build-depends": [
   ],
   "depends": [
-    "Font::AFM:ver<1.24.9+>",
-    "Font::FreeType:ver<0.4.3+>",
-    "HarfBuzz",
+    "Font::AFM:ver<1.24.10+>",
+    "Font::FreeType:ver<0.5.5+>",
+    "HarfBuzz:ver<0.1.3+>",
     "HarfBuzz::Font::FreeType",
     "Hash::int",
     "Native::Packing",

diff --git a/lib/PDF/Font/Loader/Enc.rakumod b/lib/PDF/Font/Loader/Enc.rakumod
@@ -34,7 +34,7 @@ has Bool $.encoding-updated is rw;
 has PDF::COS::Stream $.cmap is rw; # /ToUnicode CMap
 has Font::AFM $.core-metrics is rw;
 has Lock $.lock handles<protect> .= new;
-has Glyph %!glyphs{Int};
+has Glyph %.glyphs-seen{Int} is built;
 # internal font units to 1000ths
 submethod TWEAK(:$widths) {
     @!widths = .map(*.Int) with $widths;
@@ -84,7 +84,7 @@ method local-glyph-name($cid) {
 }
 
 method glyph($cid is raw) {
-    self.protect: { %!glyphs{$cid} //= self!make-glyph($cid); }
+    self.protect: { %!glyphs-seen{$cid} //= self!make-glyph($cid); }
 }
 
 multi method get-glyphs(Str:D $text) {
@@ -215,7 +215,7 @@ sub codepoint-to-hex(UInt $_) {
         !! $s;                   # 1-2 bytes
 }
 
-sub ligature-to-hex(UInt @codes) {
+sub ligature-to-hex(@codes) {
     my $buf = @codes>>.chr.join.encode("utf16");
     $buf>>.fmt('%04X').join;
 }

diff --git a/lib/PDF/Font/Loader/Enc/CMap.rakumod b/lib/PDF/Font/Loader/Enc/CMap.rakumod
@@ -290,24 +290,29 @@ my constant %PreferredEnc = do {
     %win;
 }
 has UInt $!next-cid = 0;
-has %!used-cid;
-method use-cid($_) { %!used-cid{$_}++ }
-method add-encoding($ord) {
-    my $cid := %PreferredEnc{$ord};
-    if $cid && !@!to-unicode[$cid] && !%!used-cid{$cid} && !self!skip-cid-block($cid) {
-        self.set-encoding($ord, $cid);
+has @!used-cid;
+method use-cid($_) { @!used-cid[$_] = 1 }
+method allocate-cid {
+    my UInt $cid;
+    repeat {
+        ++$!next-cid;
+    } while @!used-cid[$!next-cid] || @!to-unicode[$!next-cid] || self!skip-cid-block($!next-cid);
+    if $!next-cid >= 2 ** ($.is-wide ?? 16 !! 8)  {
+        has $!out-of-gas //= warn "CID code-range is exhausted";
     }
     else {
-        # sequential allocation
-        repeat {
-        } while %!used-cid{$!next-cid} || @!to-unicode[++$!next-cid] || self!skip-cid-block($!next-cid) ;
         $cid := $!next-cid;
-        if $cid >= 2 ** ($.is-wide ?? 16 !! 8)  {
-            has $!out-of-gas //= warn "CID code-range is exhausted";
-        }
-        else {
-            self.set-encoding($ord, $cid);
-        }
+    }
+    @!used-cid[$cid] = 1;
+    $cid;
+}
+method add-encoding($ord) {
+    my $cid = %PreferredEnc{$ord};
+    if $cid && !@!to-unicode[$cid] && !@!used-cid[$cid] && !self!skip-cid-block($cid) {
+        self.set-encoding($ord, $cid);
+    }
+    elsif $cid = self.allocate-cid {
+        self.set-encoding($ord, $cid);
     }
     $cid;
 }

diff --git a/lib/PDF/Font/Loader/Enc/Glyphic.rakumod b/lib/PDF/Font/Loader/Enc/Glyphic.rakumod
@@ -24,7 +24,7 @@ role PDF::Font::Loader::Enc::Glyphic
         }
 
         if $gid {
-            $glyph-name = $!face.glyph-name-from-index($gid) // callsame() // $ord.chr.uniname;
+            $glyph-name = $!face.glyph-name-from-index($gid) || callsame() || $ord.chr.uniname;
             # Not sure what glyph names are universally supported. This is conservative.
             $.encoding-updated = True
                 unless $glyph-name ~~ %Font::AFM::Glyphs{$ord.chr};
@@ -37,8 +37,8 @@ role PDF::Font::Loader::Enc::Glyphic
     }
 
     # Callback for unmapped glyphs
-    method cid-map-glyph($glyph-name, $cid) {
-        if $!face.index-from-glyph-name($glyph-name) -> $gid {
+    method cid-map-glyph($glyph-name, $cid, $gid = $!face.index-from-glyph-name($glyph-name)) {
+        if $gid {
             $.cid-to-gid-map[$cid] ||= $gid;
         }
     }

diff --git a/lib/PDF/Font/Loader/Enc/Identity16.rakumod b/lib/PDF/Font/Loader/Enc/Identity16.rakumod
@@ -14,14 +14,30 @@ also does PDF::Content::Font::Encoder;
 
     has Font::FreeType::Face $.face is required;
     has FT_Face $!raw;
-    has uint32 @!to-unicode;
     has UInt %.charset{UInt};
     has UInt $.min-index;
     has UInt $.max-index;
-    has atomicint $!init = 0;
+    has uint32 @.to-unicode is built;
 
     submethod TWEAK {
         $!raw = $!face.raw;
+        if self.cid-to-gid-map -> $cid-gid {
+            my $gid-uni:= $!face.index-to-unicode;
+            for $cid-gid.kv -> $cid, $gid {
+                if $gid {
+                    my $cp := $gid-uni[$gid];
+                    if $cp {
+                        @!to-unicode[$cid] = $cp
+                    }
+                    else {
+                        @!used-cid[$cid] = 1;
+                    }
+                }
+            }
+        }
+        else {
+            @!to-unicode = $!face.index-to-unicode;
+        }
     }
 
     method is-wide {True}
@@ -33,7 +49,22 @@ also does PDF::Content::Font::Encoder;
     }
 
     method add-encoding(UInt:D $ord) {
-        self.set-encoding: $ord, $!raw.FT_Get_Char_Index($ord);
+        my $gid = $!raw.FT_Get_Char_Index($ord);
+        my $cid;
+        if self.cid-to-gid-map -> $map {
+            if $ord <= $!raw.num-glyphs && !@!to-unicode[$ord] && !@!used-cid[$ord] {
+                $cid = $ord;
+            }
+            else {
+                $cid = self.allocate-cid;
+            }
+            $map[$cid] = $gid
+                if $cid;
+        }
+        else {
+            $cid = $gid;
+        }
+        self.set-encoding: $ord, $cid
     }
 
     multi method encode(Str $text, :cids($)!) {
@@ -44,6 +75,17 @@ also does PDF::Content::Font::Encoder;
         }
     }
 
+    has UInt $!next-cid = 0;
+    has uint8 @!used-cid;
+    method allocate-cid {
+        repeat {
+            $!next-cid++;
+        } while @!used-cid[$!next-cid] || @!to-unicode[$!next-cid] && $!next-cid <= $!raw.num-glyphs;
+        my $cid = $!next-cid <= $!raw.num-glyphs ?? $!next-cid !! 0;
+        @!used-cid[$cid] = 1;
+        $cid;
+    }
+
     method encode-cids(@cids is raw) {
         my blob8 $buf := pack(@cids, 16);
         $buf.decode: 'latin-1';
@@ -53,27 +95,6 @@ also does PDF::Content::Font::Encoder;
         self.encode-cids: self.encode($text, :cids);
     }
 
-    method !setup-decoding {
-        $.lock.protect: {
-            unless $!init {
-                my FT_Face $struct = $!face.raw;
-                my FT_UInt $glyph-idx;
-                my FT_ULong $char-code = $struct.FT_Get_First_Char( $glyph-idx);
-                @!to-unicode[$!face.num-glyphs] = 0;
-                while $glyph-idx {
-                    @!to-unicode[ $glyph-idx ] = $char-code;
-                    $char-code = $struct.FT_Get_Next_Char( $char-code, $glyph-idx);
-                }
-                $!init ⚛= 1;
-            }
-        }
-    }
-
-    method to-unicode {
-        $!init || self!setup-decoding();
-        @!to-unicode;
-    }
-
     multi method decode(Str $encoded, :cids($)!) {
         $encoded.ords.map: -> \hi, \lo {hi +< 8 + lo};
     }

diff --git a/lib/PDF/Font/Loader/Enc/Type1.rakumod b/lib/PDF/Font/Loader/Enc/Type1.rakumod
@@ -1,9 +1,11 @@
 #| Implements a Type1 single byte encoding scheme, such as win, mac, or std
-use PDF::Content::Font::Enc::Type1;
-use PDF::Font::Loader::Enc;
-class PDF::Font::Loader::Enc::Type1
-    is PDF::Content::Font::Enc::Type1
-    is PDF::Font::Loader::Enc {
+class PDF::Font::Loader::Enc::Type1 {
+
+    use PDF::Content::Font::Enc::Type1;
+    also is PDF::Content::Font::Enc::Type1;
+
+    use PDF::Font::Loader::Enc;
+    also is PDF::Font::Loader::Enc;
 
     use PDF::Font::Loader::Enc::Glyphic;
     also does PDF::Font::Loader::Enc::Glyphic;