Updated cld3

app/src/main/jni/cld_3/src/embedding_network.cc

@@ -167,6 +167,7 @@ EmbeddingNetwork::EmbeddingNetwork(const EmbeddingNetworkParams *model)
   for (int i = 0; i < model_->embedding_dim_size(); ++i) {
     CLD3_DCHECK(offset_sum == model_->concat_offset(i));
     offset_sum += model_->embedding_dim(i) * model_->embedding_num_features(i);
+    (void)offset_sum;  // Avoid compiler warning for "unused" variable.
     embedding_matrices_.emplace_back(model_->GetEmbeddingMatrix(i));
   }
 

app/src/main/jni/cld_3/src/language_identifier_features_test.cc

@@ -16,6 +16,7 @@ limitations under the License.
 #include <cmath>
 #include <iostream>
 #include <vector>
+#include <set>
 
 #include "base.h"
 #include "feature_extractor.h"

app/src/main/jni/cld_3/src/nnet_lang_id_test.cc

@@ -209,6 +209,34 @@ bool TestMultipleLanguagesInInput() {
                 << result.proportion << std::endl;
       return false;
     }
+
+    // Skip over undefined language.
+    if (result.language == "und")
+      continue;
+    if (result.byte_ranges.size() != 1) {
+      std::cout << " Should only detect one span containing " << result.language
+                << std::endl;
+      return false;
+    }
+    // Check that specified byte ranges for language are correct.
+    int start_index = result.byte_ranges[0].start_index;
+    int end_index = result.byte_ranges[0].end_index;
+    std::string byte_ranges_text = text.substr(start_index, end_index - start_index);
+    if (result.language == "bg") {
+      if (byte_ranges_text.compare("Този текст е на Български.") != 0) {
+        std::cout << " Incorrect byte ranges returned for Bulgarian " << std::endl;
+        return false;
+      }
+    } else if (result.language == "en") {
+      if (byte_ranges_text.compare("This piece of text is in English. ") != 0) {
+        std::cout << " Incorrect byte ranges returned for English " << std::endl;
+        return false;
+      }
+    } else {
+      std::cout << " Got language other than English or Bulgarian "
+                << std::endl;
+      return false;
+    }
   }
   std::cout << "  Success!" << std::endl;
   return true;

app/src/main/jni/cld_3/src/nnet_language_identifier.cc

@@ -47,6 +47,9 @@ struct LangChunksStats {
 
   // Number chunks corresponding to the language.
   int num_chunks = 0;
+
+  // Specifies the byte ranges that language applies to.
+  std::vector<NNetLanguageIdentifier::SpanInfo> byte_ranges;
 };
 
 // Compares two pairs based on their values.
@@ -281,8 +284,6 @@ NNetLanguageIdentifier::FindTopNMostFreqLangs(const string &text,
   CLD2::LangSpan script_span;
   std::unordered_map<string, LangChunksStats> lang_stats;
   int total_num_bytes = 0;
-  Result result;
-  string language;
   int chunk_size = 0;  // Use the default.
   while (ss.GetOneScriptSpanLower(&script_span)) {
     const int num_original_span_bytes = script_span.text_bytes;
@@ -298,12 +299,16 @@ NNetLanguageIdentifier::FindTopNMostFreqLangs(const string &text,
     total_num_bytes += num_original_span_bytes;
 
     const string selected_text = SelectTextGivenScriptSpan(script_span);
-    result = FindLanguageOfValidUTF8(selected_text);
-    language = result.language;
+
+    Result result = FindLanguageOfValidUTF8(selected_text);
+    string language = result.language;
     lang_stats[language].byte_sum += num_original_span_bytes;
     lang_stats[language].prob_sum +=
         result.probability * num_original_span_bytes;
     lang_stats[language].num_chunks++;
+    // Add SpanInfo. Start and end indices are relative to original input.
+    lang_stats[language].byte_ranges.push_back(SpanInfo(
+        ss.MapBack(0), ss.MapBack(script_span.text_bytes), result.probability));
   }
 
   // Sort the languages based on the number of bytes associated with them.
@@ -329,6 +334,7 @@ NNetLanguageIdentifier::FindTopNMostFreqLangs(const string &text,
     result.probability = stats.prob_sum / stats.byte_sum;
     result.proportion = stats.byte_sum / byte_sum;
     result.is_reliable = ResultIsReliable(language, result.probability);
+    result.byte_ranges = stats.byte_ranges;
     results.push_back(result);
   }
 
@@ -348,7 +354,7 @@ string NNetLanguageIdentifier::SelectTextGivenBeginAndSize(
     const char *text_begin, int text_size) {
   string output_text;
 
-  // If the size of the input is greater than the maxium number of bytes needed
+  // If the size of the input is greater than the maximum number of bytes needed
   // for a prediction, then concatenate snippets that are equally spread out
   // throughout the input.
   if (text_size > max_num_bytes_) {

app/src/main/jni/cld_3/src/nnet_language_identifier.h

@@ -44,6 +44,19 @@ class LanguageIdEmbeddingFeatureExtractor
 // Class for detecting the language of a document.
 class NNetLanguageIdentifier {
  public:
+  // Holds probability that Span, specified by start/end indices, is a given
+  // language. The langauge is not stored here; it can be found in Result, which
+  // holds a vector of SpanInfo.
+  struct SpanInfo {
+    SpanInfo(int start_index_val, int end_index_val, float probability_val)
+        : start_index(start_index_val),
+          end_index(end_index_val),
+          probability(probability_val) {}
+    int start_index = -1;
+    int end_index = -1;
+    float probability = 0.0;
+  };
+
   // Information about a predicted language.
   struct Result {
     string language = kUnknown;
@@ -53,6 +66,9 @@ class NNetLanguageIdentifier {
     // Proportion of bytes associated with the language. If FindLanguage is
     // called, this variable is set to 1.
     float proportion = 0.0;
+
+    // Specifies the byte ranges that |language| applies to.
+    std::vector<SpanInfo> byte_ranges;
   };
 
   NNetLanguageIdentifier();

app/src/main/jni/cld_3/src/script_span/getonescriptspan.cc

@@ -878,7 +878,6 @@ bool ScriptScanner::GetOneScriptSpan(LangSpan* span) {
   // copying letters to buffer with single spaces for each run of non-letters
   while (take < byte_length_) {
     // Copy run of letters in same script (&LS | LS)*
-    int letter_count = 0;              // Keep track of word length
     bool need_break = false;
 
     while (take < byte_length_) {
@@ -963,7 +962,6 @@ bool ScriptScanner::GetOneScriptSpan(LangSpan* span) {
         map2original_.Delete(tlen - plen);
       }
 
-      ++letter_count;
       if (put >= kMaxScriptBytes) {
         // Buffer is full
         span->truncated = true;

app/src/main/jni/cld_3/src/script_span/getonescriptspan.h

@@ -33,14 +33,14 @@ static const int kMaxScriptBytes = kMaxScriptBuffer - 32;   // Leave some room
 static const int kWithinScriptTail = 32;    // Stop at word space in last
                                             // N bytes of script buffer
 
-typedef struct {
+struct LangSpan {
   char* text = nullptr;   // Pointer to the span, somewhere
   int text_bytes = 0;     // Number of bytes of text in the span
   int offset = 0;         // Offset of start of span in original input buffer
   ULScript ulscript = UNKNOWN_ULSCRIPT; // Unicode Letters Script of this span
   bool truncated = false; // true if buffer filled up before a
                           // different script or EOF was found
-} LangSpan;
+};
 
 static inline bool IsContinuationByte(char c) {
   return static_cast<signed char>(c) < -64;
@@ -93,7 +93,7 @@ class ScriptScanner {
   // again with the first byte of the following range.
   int MapBack(int text_offset);
 
-  const char* GetBufferStart() {return start_byte_;};
+  const char* GetBufferStart() {return start_byte_;}
 
  private:
   // Skip over tags and non-letters

app/src/main/jni/cld_3/src/script_span/utf8statetable.cc

@@ -158,6 +158,20 @@ static const int kHtmlPlaintextFlag = 0x80;    // Bit in add byte to distinguish
  *
 **/
 
+// All intentional fallthroughs in breakpad are in this file, so define
+// this macro locally.
+// If you ever move this to a .h file, make sure it's defined in a
+// private header file: clang suggests the first macro expanding to
+// [[clang::fallthrough]] in its diagnostics, so if BP_FALLTHROUGH
+// is visible in code depending on breakpad, clang would suggest
+// BP_FALLTHROUGH for code depending on breakpad, instead of the
+// client code's own fallthrough macro.
+#if defined(__clang__)
+#define CLD_FALLTHROUGH [[clang::fallthrough]]
+#else
+#define CLD_FALLTHROUGH
+#endif
+
 // Return true if current Tbl pointer is within state0 range
 // Note that unsigned compare checks both ends of range simultaneously
 static inline bool InStateZero(const UTF8ScanObj* st, const uint8* Tbl) {
@@ -715,10 +729,10 @@ static int UTF8GenericReplaceInternal(const UTF8ReplaceObj* st,
       goto Do_state_table;
     case kExitReplace3:    // update 3 bytes to change
       dst[-3] = (unsigned char)Tbl[c + (nEntries * 3)];
-      // Fall into next case
+      CLD_FALLTHROUGH;
     case kExitReplace2:    // update 2 bytes to change
       dst[-2] = (unsigned char)Tbl[c + (nEntries * 2)];
-      // Fall into next case
+      CLD_FALLTHROUGH;
     case kExitReplace1:    // update 1 byte to change
       dst[-1] = (unsigned char)Tbl[c + (nEntries * 1)];
       total_changed++;
@@ -736,7 +750,7 @@ static int UTF8GenericReplaceInternal(const UTF8ReplaceObj* st,
       } else {
         offset += ((unsigned char)Tbl[c + (nEntries * 2)] << 8);
       }
-      // Fall into next case
+      CLD_FALLTHROUGH;
     case kExitSpecial:      // Apply special fixups [read: hacks]
     case kExitReplaceOffset1:
       if ((nEntries != 256) && InStateZero(st, Tbl)) {
@@ -986,10 +1000,10 @@ static int UTF8GenericReplaceInternalTwoByte(const UTF8ReplaceObj_2* st,
       goto Do_state_table_2;
     case kExitReplace3_2:    // update 3 bytes to change
       dst[-3] = (unsigned char)(Tbl[c + (nEntries * 2)] & 0xff);
-      // Fall into next case
+      CLD_FALLTHROUGH;
     case kExitReplace2_2:    // update 2 bytes to change
       dst[-2] = (unsigned char)(Tbl[c + (nEntries * 1)] >> 8 & 0xff);
-      // Fall into next case
+      CLD_FALLTHROUGH;
     case kExitReplace1_2:    // update 1 byte to change
       dst[-1] = (unsigned char)(Tbl[c + (nEntries * 1)] & 0xff);
       total_changed++;
@@ -1007,7 +1021,7 @@ static int UTF8GenericReplaceInternalTwoByte(const UTF8ReplaceObj_2* st,
       } else {
         offset += ((unsigned char)(Tbl[c + (nEntries * 1)] >> 8 & 0xff) << 8);
       }
-      // Fall into next case
+      CLD_FALLTHROUGH;
     case kExitReplaceOffset1_2:
       if ((nEntries != 256) && InStateZero_2(st, Tbl)) {
         // For space-optimized table, we need multiples of 256 bytes

app/src/main/jni/cld_3/src/sentence_features.cc

@@ -19,11 +19,11 @@ limitations under the License.
 
 namespace chrome_lang_id {
 
-// Declare registry for the whole Sentence feature functions.  NOTE: this is not
+// Define registry for the whole Sentence feature functions.  NOTE: this is not
 // yet set to anything meaningful.  It will be set so in NNetLanguageIdentifier
 // constructor, *before* we use any feature.
 template <>
-WholeSentenceFeature::Registry
-    *RegisterableClass<WholeSentenceFeature>::registry_ = nullptr;
+WholeSentenceFeature::Registry*
+    RegisterableClass<WholeSentenceFeature>::registry_ = nullptr;
 
 }  // namespace chrome_lang_id

app/src/main/jni/cld_3/src/sentence_features.h

@@ -26,9 +26,19 @@ limitations under the License.
 namespace chrome_lang_id {
 
 // Feature function that extracts features for the full Sentence.
-typedef FeatureFunction<Sentence> WholeSentenceFeature;
-
-typedef FeatureExtractor<Sentence> WholeSentenceExtractor;
+using WholeSentenceFeature = FeatureFunction<Sentence>;
+
+using WholeSentenceExtractor = FeatureExtractor<Sentence>;
+
+// Declare registry for the whole Sentence feature functions.  This is required
+// for clang's -Wundefined-var-template.  However, MSVC has a bug which treats
+// this declaration as a definition, leading to multiple definition errors, so
+// omit this on MSVC.
+#if !defined(COMPILER_MSVC)
+template <>
+WholeSentenceFeature::Registry
+    *RegisterableClass<WholeSentenceFeature>::registry_;
+#endif
 
 }  // namespace chrome_lang_id