LibPDF: Override Type 1 character mappings by encoding in font dict

If the font dictionary includes an "Encoding" entry, it will be used
instead of the PS1FontProgram's built-in encoding.

3 files changed, 24 insertions, 16 deletions

diff --git a/Userland/Libraries/LibPDF/Fonts/PS1FontProgram.cpp b/Userland/Libraries/LibPDF/Fonts/PS1FontProgram.cpp
index b91158e4cc..ef328d9a2c 100644
--- a/Userland/Libraries/LibPDF/Fonts/PS1FontProgram.cpp
+++ b/Userland/Libraries/LibPDF/Fonts/PS1FontProgram.cpp
@@ -44,7 +44,7 @@ enum ExtendedCommand {
     SetCurrentPoint = 33,
 };
 
-PDFErrorOr<void> PS1FontProgram::parse(ReadonlyBytes const& bytes, size_t cleartext_length, size_t encrypted_length)
+PDFErrorOr<void> PS1FontProgram::create(ReadonlyBytes const& bytes, RefPtr<Encoding> encoding, size_t cleartext_length, size_t encrypted_length)
 {
     Reader reader(bytes);
     if (reader.remaining() == 0)
@@ -57,22 +57,28 @@ PDFErrorOr<void> PS1FontProgram::parse(ReadonlyBytes const& bytes, size_t cleart
     if (!seek_name(reader, CommonNames::Encoding))
         return error("Missing encoding array");
 
-    if (TRY(parse_word(reader)) == "StandardEncoding") {
-        m_encoding = Encoding::standard_encoding();
+    if (encoding) {
+        // 9.6.6.2 Encodings for Type 1 Fonts:
+        // An Encoding entry may override a Type 1 font’s mapping from character codes to character names.
+        m_encoding = encoding;
     } else {
-        HashMap<u16, CharDescriptor> descriptors;
+        if (TRY(parse_word(reader)) == "StandardEncoding") {
+            m_encoding = Encoding::standard_encoding();
+        } else {
+            HashMap<u16, CharDescriptor> descriptors;
 
-        while (reader.remaining()) {
-            auto word = TRY(parse_word(reader));
-            if (word == "readonly") {
-                break;
-            } else if (word == "dup") {
-                u32 char_code = TRY(parse_int(reader));
-                auto name = TRY(parse_word(reader));
-                descriptors.set(char_code, { name.starts_with('/') ? name.substring_view(1) : name.view(), char_code });
+            while (reader.remaining()) {
+                auto word = TRY(parse_word(reader));
+                if (word == "readonly") {
+                    break;
+                } else if (word == "dup") {
+                    u32 char_code = TRY(parse_int(reader));
+                    auto name = TRY(parse_word(reader));
+                    descriptors.set(char_code, { name.starts_with('/') ? name.substring_view(1) : name.view(), char_code });
+                }
             }
+            m_encoding = TRY(Encoding::create(descriptors));
         }
-        m_encoding = TRY(Encoding::create(descriptors));
     }
 
     bool found_font_matrix = seek_name(reader, "FontMatrix");
diff --git a/Userland/Libraries/LibPDF/Fonts/PS1FontProgram.h b/Userland/Libraries/LibPDF/Fonts/PS1FontProgram.h
index 91291473fa..b576bdc31c 100644
--- a/Userland/Libraries/LibPDF/Fonts/PS1FontProgram.h
+++ b/Userland/Libraries/LibPDF/Fonts/PS1FontProgram.h
@@ -18,7 +18,7 @@ class Encoding;
 
 class PS1FontProgram : public RefCounted<PS1FontProgram> {
 public:
-    PDFErrorOr<void> parse(ReadonlyBytes const&, size_t cleartext_length, size_t encrypted_length);
+    PDFErrorOr<void> create(ReadonlyBytes const&, RefPtr<Encoding>, size_t cleartext_length, size_t encrypted_length);
 
     RefPtr<Gfx::Bitmap> rasterize_glyph(u32 char_code, float width);
     Gfx::Path build_char(u32 char_code, float width);
diff --git a/Userland/Libraries/LibPDF/Fonts/Type1Font.cpp b/Userland/Libraries/LibPDF/Fonts/Type1Font.cpp
index bc4533bcc1..a49fac75e3 100644
--- a/Userland/Libraries/LibPDF/Fonts/Type1Font.cpp
+++ b/Userland/Libraries/LibPDF/Fonts/Type1Font.cpp
@@ -31,8 +31,10 @@ PDFErrorOr<Type1Font::Data> Type1Font::parse_data(Document* document, NonnullRef
         auto length2 = font_file_dict->get_value(CommonNames::Length2).get<int>();
 
         data.font_program = adopt_ref(*new PS1FontProgram());
-        TRY(data.font_program->parse(font_file_stream->bytes(), length1, length2));
-        data.encoding = data.font_program->encoding();
+        TRY(data.font_program->create(font_file_stream->bytes(), data.encoding, length1, length2));
+
+        if (!data.encoding)
+            data.encoding = data.font_program->encoding();
     }
 
     return data;