When brotli quality is set to 0 estimate compressed size instead of doing the compression.

Author: garretrieger

common/font_helper.cc

@@ -10,13 +10,14 @@
 #include "common/hb_set_unique_ptr.h"
 #include "common/indexed_data_reader.h"
 #include "common/int_set.h"
+#include "common/try.h"
 #include "hb-ot.h"
 #include "hb-subset.h"
 #include "hb.h"
-#include "common/try.h"
 
 using absl::btree_set;
 using absl::flat_hash_map;
+using absl::flat_hash_set;
 using absl::Status;
 using absl::StatusOr;
 using absl::StrCat;
@@ -48,7 +49,6 @@ bool FontHelper::HasWideGvar(const hb_face_t* face) {
   return (((uint8_t)gvar.str()[gvar_flags_offset]) & 0x01);
 }
 
-
 absl::StatusOr<string_view> FontHelper::GlyfData(const hb_face_t* face,
                                                  uint32_t gid) {
   auto loca = Loca(face);
@@ -126,8 +126,9 @@ FontData FontHelper::Cff2Data(hb_face_t* face, uint32_t gid) {
   return data;
 }
 
-StatusOr<uint32_t> FontHelper::TotalGlyphData(hb_face_t* face, const GlyphSet& gids) {
-  auto tags = FontHelper::GetTags(face);
+StatusOr<uint32_t> FontHelper::TotalGlyphData(hb_face_t* face,
+                                              const GlyphSet& gids) {
+  flat_hash_set<hb_tag_t> tags = FontHelper::GetTags(face);
 
   uint32_t total = 0;
   for (uint32_t gid : gids) {

common/font_helper.h

@@ -147,7 +147,8 @@ class FontHelper {
 
   // Counts up the total size of all glyph data (gvar, glyf, cff, cff2)
   // for the provided set of gids.
-  static absl::StatusOr<uint32_t> TotalGlyphData(hb_face_t* face, const GlyphSet& gids);
+  static absl::StatusOr<uint32_t> TotalGlyphData(hb_face_t* face,
+                                                 const GlyphSet& gids);
 
   static absl::Status Cff2GetCharstrings(hb_face_t* face,
                                          FontData& non_charstrings,

common/font_helper_test.cc

@@ -619,42 +619,42 @@ TEST_F(FontHelperTest, GlyfData_ShortOverflowSynthetic) {
 }
 
 TEST_F(FontHelperTest, TotalGlyphData_GlyfGvar) {
-  auto size = FontHelper::TotalGlyphData(roboto_vf.get(), GlyphSet {78, 83, 95});
+  auto size = FontHelper::TotalGlyphData(roboto_vf.get(), GlyphSet{78, 83, 95});
   ASSERT_TRUE(size.ok()) << size.status();
 
-  uint32_t expected =
-    FontHelper::GlyfData(roboto_vf.get(), 78)->size() +
-    FontHelper::GlyfData(roboto_vf.get(), 83)->size() +
-    FontHelper::GlyfData(roboto_vf.get(), 95)->size() +
-    FontHelper::GvarData(roboto_vf.get(), 78)->size() +
-    FontHelper::GvarData(roboto_vf.get(), 83)->size() +
-    FontHelper::GvarData(roboto_vf.get(), 95)->size();
+  uint32_t expected = FontHelper::GlyfData(roboto_vf.get(), 78)->size() +
+                      FontHelper::GlyfData(roboto_vf.get(), 83)->size() +
+                      FontHelper::GlyfData(roboto_vf.get(), 95)->size() +
+                      FontHelper::GvarData(roboto_vf.get(), 78)->size() +
+                      FontHelper::GvarData(roboto_vf.get(), 83)->size() +
+                      FontHelper::GvarData(roboto_vf.get(), 95)->size();
 
   ASSERT_GT(*size, 0);
   ASSERT_EQ(*size, expected);
 }
 
 TEST_F(FontHelperTest, TotalGlyphData_Cff) {
-  auto size = FontHelper::TotalGlyphData(noto_sans_jp_otf.get(), GlyphSet {78, 83, 95});
+  auto size =
+      FontHelper::TotalGlyphData(noto_sans_jp_otf.get(), GlyphSet{78, 83, 95});
   ASSERT_TRUE(size.ok()) << size.status();
 
-  uint32_t expected =
-    FontHelper::CffData(noto_sans_jp_otf.get(), 78).size() +
-    FontHelper::CffData(noto_sans_jp_otf.get(), 83).size() +
-    FontHelper::CffData(noto_sans_jp_otf.get(), 95).size();
+  uint32_t expected = FontHelper::CffData(noto_sans_jp_otf.get(), 78).size() +
+                      FontHelper::CffData(noto_sans_jp_otf.get(), 83).size() +
+                      FontHelper::CffData(noto_sans_jp_otf.get(), 95).size();
 
   ASSERT_GT(*size, 0);
   ASSERT_EQ(*size, expected);
 }
 
 TEST_F(FontHelperTest, TotalGlyphData_Cff2) {
-  auto size = FontHelper::TotalGlyphData(noto_sans_vf_jp_otf.get(), GlyphSet {34, 35, 46});
+  auto size = FontHelper::TotalGlyphData(noto_sans_vf_jp_otf.get(),
+                                         GlyphSet{34, 35, 46});
   ASSERT_TRUE(size.ok()) << size.status();
 
   uint32_t expected =
-    FontHelper::Cff2Data(noto_sans_vf_jp_otf.get(), 34).size() +
-    FontHelper::Cff2Data(noto_sans_vf_jp_otf.get(), 35).size() +
-    FontHelper::Cff2Data(noto_sans_vf_jp_otf.get(), 46).size();
+      FontHelper::Cff2Data(noto_sans_vf_jp_otf.get(), 34).size() +
+      FontHelper::Cff2Data(noto_sans_vf_jp_otf.get(), 35).size() +
+      FontHelper::Cff2Data(noto_sans_vf_jp_otf.get(), 46).size();
 
   ASSERT_GT(*size, 0);
   ASSERT_EQ(*size, expected);

ift/encoder/BUILD

@@ -63,6 +63,8 @@ cc_library(
         "candidate_merge.cc",
         "candidate_merge.h",
         "patch_size_cache.h",
+        "estimated_patch_size_cache.h",
+        "estimated_patch_size_cache.cc",
         "merge_strategy.h",
         "merge_strategy.cc",
         "merger.cc",
@@ -253,4 +255,17 @@ cc_test(
         "@abseil-cpp//absl/types:span",
         "@googletest//:gtest_main",
     ],
+)
+
+cc_test(
+    name = "estimated_patch_size_cache_test",
+    srcs = ["estimated_patch_size_cache_test.cc"],
+    data = [
+        "//common:testdata",
+    ],
+    deps = [
+        ":segmentation_context",
+        "//common",
+        "@googletest//:gtest_main",
+    ],
 )

ift/encoder/estimated_patch_size_cache.cc

@@ -0,0 +1,60 @@
+#include "ift/encoder/estimated_patch_size_cache.h"
+
+#include "common/int_set.h"
+#include "common/try.h"
+
+using absl::StatusOr;
+using absl::flat_hash_set;
+using common::GlyphSet;
+
+namespace ift::encoder {
+
+StatusOr<uint32_t> EstimatedPatchSizeCache::GetPatchSize(const GlyphSet& gids) {
+  auto it = cache_.find(gids);
+  if (it != cache_.end()) {
+    return it->second;
+  }
+
+  flat_hash_set<hb_tag_t> tags = common::FontHelper::GetTags(face_.get());
+  uint32_t table_count = (tags.contains(common::FontHelper::kCFF) ? 1 : 0) +
+                         (tags.contains(common::FontHelper::kCFF2) ? 1 : 0) +
+                         (tags.contains(common::FontHelper::kGlyf) ? 1 : 0) +
+                         (tags.contains(common::FontHelper::kGvar) ? 1 : 0);
+
+  uint32_t gid_width = (gids.size() > 255) ? 3 : 2;
+
+  uint32_t header_size = 1 + 7 * 4;
+  uint32_t uncompressed_stream_size =
+      5 + gids.size() * gid_width +         // glyph ids
+      4 * table_count +                     // table tags
+      4 * (gids.size() * table_count + 1);  // data offsets
+
+  uncompressed_stream_size +=
+      TRY(common::FontHelper::TotalGlyphData(face_.get(), gids));
+
+  uint32_t size = header_size + (uint32_t)((double)uncompressed_stream_size *
+                                           compression_ratio_);
+  cache_[gids] = size;
+  return size;
+}
+
+StatusOr<double> EstimatedPatchSizeCache::EstimateCompressionRatio(
+    hb_face_t* original_face) {
+  PatchSizeCacheImpl patch_sizes(original_face, 11);
+
+  uint32_t glyph_count = hb_face_get_glyph_count(original_face);
+  if (glyph_count == 0) {
+    return 0.0;
+  }
+
+  common::GlyphSet gids;
+  gids.insert_range(0, glyph_count - 1);
+
+  double uncompressed_size =
+      TRY(common::FontHelper::TotalGlyphData(original_face, gids));
+  double compressed_size = TRY(patch_sizes.GetPatchSize(gids));
+
+  return compressed_size / uncompressed_size;
+}
+
+}  // namespace ift::encoder

ift/encoder/estimated_patch_size_cache.h

@@ -0,0 +1,47 @@
+#ifndef IFT_ENCODER_ESTIMATED_PATCH_SIZE_CACHE_H_
+#define IFT_ENCODER_ESTIMATED_PATCH_SIZE_CACHE_H_
+
+#include <memory>
+#include "absl/status/statusor.h"
+#include "common/font_data.h"
+#include "common/int_set.h"
+#include "ift/encoder/patch_size_cache.h"
+
+namespace ift::encoder {
+
+// Estimates the size of a glyph keyed patch using a fixed compression ratio.
+// Does not actually run the brotli compression.
+//
+// The fixed compression ratio is determined by looking at the compression ratio
+// of glyph data in the provided original_face.
+class EstimatedPatchSizeCache : public PatchSizeCache {
+ public:
+  static absl::StatusOr<std::unique_ptr<PatchSizeCache>> New(hb_face_t* face) {
+    double compression_ratio = TRY(EstimateCompressionRatio(face));
+    return std::unique_ptr<PatchSizeCache>(new EstimatedPatchSizeCache(face, compression_ratio));
+  }
+
+  absl::StatusOr<uint32_t> GetPatchSize(const common::GlyphSet& gids) override;
+
+  double CompressionRatio() const {
+    return compression_ratio_;
+  }
+
+ private:
+  explicit EstimatedPatchSizeCache(hb_face_t* original_face,
+                                   double compression_ratio)
+      : face_(common::make_hb_face(hb_face_reference(original_face))),
+        compression_ratio_(compression_ratio),
+        cache_() {}
+
+  static absl::StatusOr<double> EstimateCompressionRatio(
+      hb_face_t* original_face);
+
+  common::hb_face_unique_ptr face_;
+  double compression_ratio_;
+  absl::flat_hash_map<common::GlyphSet, uint32_t> cache_;
+};
+
+}  // namespace ift::encoder
+
+#endif  // IFT_ENCODER_ESTIMATED_PATCH_SIZE_CACHE_H_

ift/encoder/estimated_patch_size_cache_test.cc

@@ -0,0 +1,46 @@
+#include "ift/encoder/estimated_patch_size_cache.h"
+
+#include "common/font_helper.h"
+#include "common/int_set.h"
+#include "gtest/gtest.h"
+
+#include "common/font_data.h"
+
+using common::hb_face_unique_ptr;
+using common::make_hb_face;
+using common::hb_blob_unique_ptr;
+using common::make_hb_blob;
+using common::FontHelper;
+using common::GlyphSet;
+
+namespace ift::encoder {
+
+class EstimatedPatchSizeCacheTest : public ::testing::Test {
+ protected:
+  EstimatedPatchSizeCacheTest() : roboto(make_hb_face(nullptr)) {
+    hb_blob_unique_ptr blob = make_hb_blob(
+        hb_blob_create_from_file("common/testdata/Roboto-Regular.ttf"));
+    roboto = make_hb_face(hb_face_create(blob.get(), 0));
+  }
+
+
+  double CompressionRatio(GlyphSet gids, double expected_compression_ratio) {
+    uint32_t raw_outline_size =
+      *FontHelper::TotalGlyphData(roboto.get(), gids);
+    double fixed_size = 1 + 7 * 4; // header
+    fixed_size += (double) (5 + gids.size() * 2 + 4 + (gids.size() + 1)*4) * expected_compression_ratio; // glyph patches header
+    auto estimated = *EstimatedPatchSizeCache::New(roboto.get());
+    uint32_t compressed_size = *estimated->GetPatchSize(gids);
+    return (double) (compressed_size - fixed_size) / (double) raw_outline_size;
+  }
+
+  hb_face_unique_ptr roboto;
+};
+
+TEST_F(EstimatedPatchSizeCacheTest, PatchSize) {
+  // There should be a consistent compression ratio between patches.
+  ASSERT_NEAR(this->CompressionRatio(GlyphSet {44, 47, 49}, 0.457), 0.46, 0.01);
+  ASSERT_NEAR(CompressionRatio(GlyphSet {45, 48, 50, 51, 52, 53}, 0.457), 0.46, 0.01);
+}
+
+}  // namespace ift::encoder

ift/encoder/patch_size_cache.h

@@ -2,6 +2,7 @@
 #define IFT_ENCODER_PATCH_SIZE_CACHE_H_
 
 #include <cstdint>
+#include <memory>
 
 #include "absl/container/flat_hash_map.h"
 #include "absl/status/statusor.h"

ift/encoder/segmentation_context.h

@@ -8,6 +8,7 @@
 #include "common/font_data.h"
 #include "common/int_set.h"
 #include "common/try.h"
+#include "ift/encoder/estimated_patch_size_cache.h"
 #include "ift/encoder/glyph_closure_cache.h"
 #include "ift/encoder/glyph_condition_set.h"
 #include "ift/encoder/glyph_groupings.h"
@@ -43,9 +44,9 @@ class SegmentationContext {
                       const std::vector<Segment>& segments,
                       uint32_t brotli_quality,
                       uint32_t init_font_brotli_quality)
-      : patch_size_cache(new PatchSizeCacheImpl(face, brotli_quality)),
+      : patch_size_cache(NewPatchSizeCache(face, brotli_quality)),
         patch_size_cache_for_init_font(
-            new PatchSizeCacheImpl(face, init_font_brotli_quality)),
+            NewPatchSizeCache(face, init_font_brotli_quality)),
         glyph_closure_cache(face),
         original_face(common::make_hb_face(hb_face_reference(face))),
         segmentation_info_(segments, initial_segment, glyph_closure_cache),
@@ -175,6 +176,16 @@ class SegmentationContext {
   // too small to be worthwhile.
   absl::StatusOr<segment_index_t> ComputeSegmentCutoff() const;
 
+  static std::unique_ptr<PatchSizeCache> NewPatchSizeCache(hb_face_t* face, uint32_t brotli_quality) {
+    if (brotli_quality == 0) {
+      auto cache = EstimatedPatchSizeCache::New(face);
+      if (cache.ok()) {
+        return std::move(*cache);
+      }
+    }
+    return std::unique_ptr<PatchSizeCache>(new PatchSizeCacheImpl(face, brotli_quality));
+  }
+
  public:
   // Caches and logging
   std::unique_ptr<PatchSizeCache> patch_size_cache;

util/segmenter_config.proto

@@ -29,6 +29,10 @@ message SegmenterConfig {
   // When generating compressed patches (to evaluate their size) this is the brotli quality
   // level used. Segmentation is typically bottle necked on brotli compression so higher values
   // increase segmentation times, but yield more accurate results.
+  //
+  // If quality is set to '0' this disables brotli compression and instead estimates the
+  // affect of compression using a fixed compression ratio calculated based on how well
+  // the glyph data in the input font compresses.
   uint32 brotli_quality = 5 [default = 8];
 
   // During processing to determine which segments to move into the initial font this is