Better

cmake/find/nlp.cmake

@@ -36,4 +36,4 @@ endif ()
 
 set (USE_NLP 1)
 
-message (STATUS "Using Libraries for NLP functions: contrib/wordnet-blast, contrib/libstemmer_c, contrib/lemmagen-c")
+message (STATUS "Using Libraries for NLP functions: contrib/wordnet-blast, contrib/libstemmer_c, contrib/lemmagen-c, contrib/cld2")

contrib/nlp-data

@@ -1 +1 @@
-Subproject commit 3bc8aef8440b66823186f47a74996cbdad66c04f
+Subproject commit 5591f91f5e748cba8fb9ef81564176feae774853

contrib/nlp-data-cmake/CMakeLists.txt

@@ -0,0 +1,14 @@
+include(${ClickHouse_SOURCE_DIR}/cmake/embed_binary.cmake)
+
+set(LIBRARY_DIR "${ClickHouse_SOURCE_DIR}/contrib/nlp-data")
+
+add_library (nlp_data INTERFACE)
+
+clickhouse_embed_binaries(
+    TARGET nlp_dicts
+    RESOURCE_DIR "${LIBRARY_DIR}"
+    RESOURCES charset.zst tonality_ru.zst programming.zst
+)
+
+add_dependencies(nlp_data nlp_dicts)
+target_link_libraries(nlp_data INTERFACE "-Wl,${WHOLE_ARCHIVE} $<TARGET_FILE:nlp_dicts> -Wl,${NO_WHOLE_ARCHIVE}")

src/Common/FrequencyHolder.h

@@ -1,22 +1,20 @@
 #pragma once
+
+#include <Common/Arena.h>
+#include <Common/getResource.h>
+#include <Common/HashTable/HashMap.h>
 #include <Common/StringUtils/StringUtils.h>
 #include <IO/ReadBufferFromFile.h>
 #include <IO/ReadBufferFromString.h>
 #include <IO/ReadHelpers.h>
 #include <IO/readFloatText.h>
-#include <IO/Operators.h>
 #include <IO/ZstdInflatingReadBuffer.h>
 
-#include <Common/Arena.h>
 #include <base/StringRef.h>
-#include <Common/HashTable/HashMap.h>
+#include <base/logger_useful.h>
 
 #include <string_view>
-#include <string>
-#include <cstring>
 #include <unordered_map>
-#include <base/logger_useful.h>
-#include <Common/getResource.h>
 
 namespace DB
 {
@@ -26,6 +24,14 @@ namespace ErrorCodes
     extern const int FILE_DOESNT_EXIST;
 }
 
+/// FrequencyHolder class is responsible for storing and loading dictionaries
+/// needed for text classification functions:
+///
+/// 1. detectLanguageUnknown
+/// 2. detectCharset
+/// 3. detectTonality
+/// 4. detectProgrammingLanguage
+
 class FrequencyHolder
 {
 
@@ -39,6 +45,7 @@ public:
     struct Encoding
     {
         String name;
+        String lang;
         HashMap<UInt16, Float64> map;
     };
 
@@ -54,14 +61,13 @@ public:
         return instance;
     }
 
-
     void loadEncodingsFrequency()
     {
         Poco::Logger * log = &Poco::Logger::get("EncodingsFrequency");
 
         LOG_TRACE(log, "Loading embedded charset frequencies");
 
-        auto resource = getResource("charset_freq.txt.zst");
+        auto resource = getResource("charset.zst");
             if (resource.empty())
                 throw Exception(ErrorCodes::FILE_DOESNT_EXIST, "There is no embedded charset frequencies");
 
@@ -71,12 +77,12 @@ public:
         String charset_name;
 
         auto buf = std::make_unique<ReadBufferFromMemory>(resource.data(), resource.size());
-        std::unique_ptr<ReadBuffer> in = std::make_unique<ZstdInflatingReadBuffer>(std::move(buf));
+        ZstdInflatingReadBuffer in(std::move(buf));
 
-        while (!in->eof())
+        while (!in.eof())
         {
-            readString(line, *in);
-            ++in->position();
+            readString(line, in);
+            in.ignore();
 
             if (line.empty())
                 continue;
@@ -84,13 +90,21 @@ public:
             ReadBufferFromString buf_line(line);
 
             // Start loading a new charset
-            if (line.starts_with("//"))
+            if (line.starts_with("// "))
             {
+                // Skip "// "
                 buf_line.ignore(3);
                 readString(charset_name, buf_line);
 
+                /* In our dictionary we have lines with form: <Language>_<Charset>
+                * If we need to find language of data, we return <Language>
+                * If we need to find charset of data, we return <Charset>.
+                */
+                size_t sep = charset_name.find('_');
+
                 Encoding enc;
-                enc.name = charset_name;
+                enc.lang = charset_name.substr(0, sep);
+                enc.name = charset_name.substr(sep + 1);
                 encodings_freq.push_back(std::move(enc));
             }
             else
@@ -109,9 +123,9 @@ public:
     void loadEmotionalDict()
     {
         Poco::Logger * log = &Poco::Logger::get("EmotionalDict");
-        LOG_TRACE(log, "Loading embedded emotional dictionary (RU)");
+        LOG_TRACE(log, "Loading embedded emotional dictionary");
 
-        auto resource = getResource("emotional_dictionary_rus.txt.zst");
+        auto resource = getResource("tonality_ru.zst");
             if (resource.empty())
                 throw Exception(ErrorCodes::FILE_DOESNT_EXIST, "There is no embedded emotional dictionary");
 
@@ -121,12 +135,12 @@ public:
         size_t count = 0;
 
         auto buf = std::make_unique<ReadBufferFromMemory>(resource.data(), resource.size());
-        std::unique_ptr<ReadBuffer> in = std::make_unique<ZstdInflatingReadBuffer>(std::move(buf));
+        ZstdInflatingReadBuffer in(std::move(buf));
 
-        while (!in->eof())
+        while (!in.eof())
         {
-            readString(line, *in);
-            ++in->position();
+            readString(line, in);
+            in.ignore();
 
             if (line.empty())
                 continue;
@@ -151,7 +165,7 @@ public:
 
         LOG_TRACE(log, "Loading embedded programming languages frequencies loading");
 
-        auto resource = getResource("prog_freq.txt.zst");
+        auto resource = getResource("programming.zst");
             if (resource.empty())
                 throw Exception(ErrorCodes::FILE_DOESNT_EXIST, "There is no embedded programming languages frequencies");
 
@@ -161,12 +175,12 @@ public:
         String programming_language;
 
         auto buf = std::make_unique<ReadBufferFromMemory>(resource.data(), resource.size());
-        std::unique_ptr<ReadBuffer> in = std::make_unique<ZstdInflatingReadBuffer>(std::move(buf));
+        ZstdInflatingReadBuffer in(std::move(buf));
 
-        while (!in->eof())
+        while (!in.eof())
         {
-            readString(line, *in);
-            ++in->position();
+            readString(line, in);
+            in.ignore();
 
             if (line.empty())
                 continue;
@@ -174,8 +188,9 @@ public:
             ReadBufferFromString buf_line(line);
 
             // Start loading a new language
-            if (line.starts_with("//"))
+            if (line.starts_with("// "))
             {
+                // Skip "// "
                 buf_line.ignore(3);
                 readString(programming_language, buf_line);
 

src/Functions/FunctionsCharsetClassification.cpp

@@ -1,18 +1,18 @@
-#include <Functions/FunctionsTextClassification.h>
 #include <Common/FrequencyHolder.h>
 #include <Functions/FunctionFactory.h>
-#include <Common/UTF8Helpers.h>
-#include <IO/ReadBufferFromString.h>
-#include <IO/ReadHelpers.h>
+#include <Functions/FunctionStringToString.h>
 
-#include <cstring>
-#include <cmath>
-#include <unordered_map>
 #include <memory>
-#include <utility>
+#include <unordered_map>
 
 namespace DB
 {
+
+namespace ErrorCodes
+{
+    extern const int ILLEGAL_COLUMN;
+}
+
 /* Determine language and charset of text data. For each text, we build the distribution of bigrams bytes.
  * Then we use marked-up dictionaries with distributions of bigram bytes of various languages ​​and charsets.
  * Using a naive Bayesian classifier, find the most likely charset and language and return it
@@ -21,10 +21,6 @@ namespace DB
 template <size_t N, bool detect_language>
 struct CharsetClassificationImpl
 {
-
-    using ResultType = String;
-    using CodePoint = UInt8;
-
     /* We need to solve zero-frequency problem for Naive Bayes Classifier
      * If the bigram is not found in the text, we assume that the probability of its meeting is 1e-06.
      * 1e-06 is minimal value in our marked-up dictionary.
@@ -34,28 +30,22 @@ struct CharsetClassificationImpl
     /// If the data size is bigger than this, behaviour is unspecified for this function.
     static constexpr size_t max_string_size = 1u << 15;
 
-    /// Default padding to read safely.
-    static constexpr size_t default_padding = 16;
-
-    /// Max codepoints to store at once. 16 is for batching usage and PODArray has this padding.
-    static constexpr size_t simultaneously_codepoints_num = default_padding + N - 1;
-
-
-    static ALWAYS_INLINE inline Float64 naiveBayes(const FrequencyHolder::EncodingMap & standard,
-        std::unordered_map<UInt16, Float64> & model,
+    static ALWAYS_INLINE inline Float64 naiveBayes(
+        const FrequencyHolder::EncodingMap & standard,
+        const HashMap<UInt16, Float64> & model,
         Float64 max_result)
     {
         Float64 res = 0;
-        for (auto & el : model)
+        for (const auto & el : model)
         {
             /// Try to find bigram in the dictionary.
-            auto it = standard.find(el.first);
+            const auto * it = standard.find(el.getKey());
             if (it != standard.end())
             {
-                res += el.second * log(it->getMapped());
+                res += el.getMapped() * log(it->getMapped());
             } else
             {
-                res += el.second * log(zero_frequency);
+                res += el.getMapped() * log(zero_frequency);
             }
             /// If at some step the result has become less than the current maximum, then it makes no sense to count it fully.
             if (res < max_result)
@@ -66,95 +56,21 @@ struct CharsetClassificationImpl
         return res;
     }
 
-
-    static ALWAYS_INLINE size_t readCodePoints(CodePoint * code_points, const char *& pos, const char * end)
-    {
-        constexpr size_t padding_offset = default_padding - N + 1;
-        memcpy(code_points, code_points + padding_offset, roundUpToPowerOfTwoOrZero(N - 1) * sizeof(CodePoint));
-        memcpy(code_points + (N - 1), pos, default_padding * sizeof(CodePoint));
-        pos += padding_offset;
-        if (pos > end)
-            return default_padding - (pos - end);
-        return default_padding;
-    }
-
     /// Сount how many times each bigram occurs in the text.
-    static ALWAYS_INLINE inline size_t calculateStats(
-        const char * data,
+    static ALWAYS_INLINE inline void calculateStats(
+        const UInt8 * data,
         const size_t size,
-        size_t (*read_code_points)(CodePoint *, const char *&, const char *),
-        std::unordered_map<UInt16, Float64>& model)
+        HashMap<UInt16, Float64> & model)
     {
-
-        const char * start = data;
-        const char * end = data + size;
-        CodePoint cp[simultaneously_codepoints_num] = {};
-        /// read_code_points returns the position of cp where it stopped reading codepoints.
-        size_t found = read_code_points(cp, start, end);
-        /// We need to start for the first time here, because first N - 1 codepoints mean nothing.
-        size_t i = N - 1;
-        size_t len = 0;
-        do
-        {
-            for (; i + N <= found; ++i)
-            {
-                UInt32 hash = 0;
-                for (size_t j = 0; j < N; ++j)
+        UInt16 hash = 0;
+        for (size_t i = 0; i < size; ++i)
         {
             hash <<= 8;
-                    hash += *(cp + i + j);
-                }
-                if (model[hash] == 0)
-                {
-                    model[hash] = 1;
-                    ++len;
-                }
+            hash += *(data + i);
             ++model[hash];
         }
-            i = 0;
-        } while (start < end && (found = read_code_points(cp, start, end)));
-
-        return len;
     }
 
-
-    static void constant(String data, String & res)
-    {
-        const auto & encodings_freq = FrequencyHolder::getInstance().getEncodingsFrequency();
-
-        std::unordered_map<UInt16, Float64> model;
-        calculateStats(data.data(), data.size(), readCodePoints, model);
-
-        Float64 max_result = log(zero_frequency) * (max_string_size);
-        String poss_ans;
-        /// Go through the dictionary and find the charset with the highest weight
-        for (auto& item : encodings_freq)
-        {
-            Float64 score = naiveBayes(item.map, model, max_result);
-            if (max_result < score)
-            {
-                poss_ans = item.name;
-                max_result = score;
-            }
-        }
-
-        /* In our dictionary we have lines with form: <Language>_<Charset>
-         * If we need to find language of data, we return <Language>
-         * If we need to find charset of data, we return <Charset>.
-         */
-
-        size_t sep = poss_ans.find('_');
-        if (detect_language)
-        {
-            res = poss_ans.erase(0, sep + 1);
-        }
-        else
-        {
-            res = poss_ans.erase(sep, poss_ans.size() - sep);
-        }
-    }
-
-
     static void vector(
         const ColumnString::Chars & data,
         const ColumnString::Offsets & offsets,
@@ -163,64 +79,53 @@ struct CharsetClassificationImpl
     {
         const auto & encodings_freq = FrequencyHolder::getInstance().getEncodingsFrequency();
 
-        res_data.reserve(1024);
+        if (detect_language)
+            /// 2 chars for ISO code + 1 zero byte
+            res_data.reserve(offsets.size() * 3);
+        else
+            /// Mean charset length is 8
+            res_data.reserve(offsets.size() * 8);
+
         res_offsets.resize(offsets.size());
 
-        size_t prev_offset = 0;
         size_t res_offset = 0;
 
         for (size_t i = 0; i < offsets.size(); ++i)
         {
-            const char * haystack = reinterpret_cast<const char *>(&data[prev_offset]);
-            String str = haystack;
+            const UInt8 * str = data.data() + offsets[i - 1];
+            const size_t str_len = offsets[i] - offsets[i - 1] - 1;
 
-            String poss_ans;
+            std::string_view res;
 
-            std::unordered_map<UInt16, Float64> model;
-            calculateStats(str.data(), str.size(), readCodePoints, model);
+            HashMap<UInt16, Float64> model;
+            calculateStats(str, str_len, model);
 
+            /// Go through the dictionary and find the charset with the highest weight
             Float64 max_result = log(zero_frequency) * (max_string_size);
-           for (auto& item : encodings_freq)
+            for (const auto & item : encodings_freq)
             {
                 Float64 score = naiveBayes(item.map, model, max_result);
                 if (max_result < score)
                 {
                     max_result = score;
-                    poss_ans = item.name;
+                    res = detect_language ? item.lang : item.name;
                 }
             }
 
-            size_t sep = poss_ans.find('_');
-            String ans_str;
+            res_data.resize(res_offset + res.size() + 1);
+            memcpy(&res_data[res_offset], res.data(), res.size());
 
-            if (detect_language)
-            {
-                ans_str = poss_ans.erase(0, sep + 1);
-            }
-            else
-            {
-                ans_str = poss_ans.erase(sep, poss_ans.size() - sep);
-            }
-
-            ans_str = poss_ans;
-
-            const auto res = ans_str.c_str();
-            size_t cur_offset = offsets[i];
-
-            size_t ans_size = strlen(res);
-            res_data.resize(res_offset + ans_size + 1);
-            memcpy(&res_data[res_offset], res, ans_size);
-            res_offset += ans_size;
-
-            res_data[res_offset] = 0;
-            ++res_offset;
+            res_data[res_offset + res.size()] = 0;
+            res_offset += res.size() + 1;
 
             res_offsets[i] = res_offset;
-            prev_offset = cur_offset;
         }
     }
 
-
+    [[noreturn]] static void vectorFixed(const ColumnString::Chars &, size_t, ColumnString::Chars &)
+    {
+        throw Exception("Cannot apply function detectProgrammingLanguage to fixed string.", ErrorCodes::ILLEGAL_COLUMN);
+    }
 };
 
 
@@ -235,8 +140,8 @@ struct NameLanguageDetect
 };
 
 
-using FunctionCharsetDetect = FunctionsTextClassification<CharsetClassificationImpl<2, true>, NameCharsetDetect>;
-using FunctionLanguageDetect = FunctionsTextClassification<CharsetClassificationImpl<2, false>, NameLanguageDetect>;
+using FunctionCharsetDetect = FunctionStringToString<CharsetClassificationImpl<2, false>, NameCharsetDetect, false>;
+using FunctionLanguageDetect = FunctionStringToString<CharsetClassificationImpl<2, true>, NameLanguageDetect, false>;
 
 void registerFunctionsCharsetClassification(FunctionFactory & factory)
 {

src/Functions/FunctionsLanguageClassification.cpp

@@ -4,7 +4,7 @@
 
 #if USE_NLP
 
-#include <Functions/FunctionsTextClassification.h>
+#include <Functions/FunctionStringToString.h>
 #include <Functions/FunctionFactory.h>
 
 #include <DataTypes/DataTypeMap.h>
@@ -34,9 +34,7 @@ extern const int ILLEGAL_COLUMN;
 
 struct LanguageClassificationImpl
 {
-    using ResultType = String;
-
-    static String codeISO(std::string_view code_string)
+    static std::string_view codeISO(std::string_view code_string)
     {
         if (code_string.ends_with("-Latn"))
             code_string.remove_suffix(code_string.size() - 5);
@@ -61,54 +59,44 @@ struct LanguageClassificationImpl
         if (code_string.size() != 2)
             return "other";
 
-        return String(code_string);
+        return code_string;
     }
 
-    static void constant(const String & data, String & res)
-    {
-        bool is_reliable = true;
-        const char * str = data.c_str();
-        auto lang = CLD2::DetectLanguage(str, strlen(str), true, &is_reliable);
-        res = codeISO(LanguageCode(lang));
-    }
-
-
     static void vector(
         const ColumnString::Chars & data,
         const ColumnString::Offsets & offsets,
         ColumnString::Chars & res_data,
         ColumnString::Offsets & res_offsets)
     {
-        res_data.reserve(1024);
+        /// Constant 3 is based on the fact that in general we need 2 characters for ISO code + 1 zero byte
+        res_data.reserve(offsets.size() * 3);
         res_offsets.resize(offsets.size());
 
-        size_t prev_offset = 0;
+        bool is_reliable = true;
         size_t res_offset = 0;
 
         for (size_t i = 0; i < offsets.size(); ++i)
         {
-            const char * str = reinterpret_cast<const char *>(&data[prev_offset]);
-            String res;
-            bool is_reliable = true;
+            const char * str = reinterpret_cast<const char *>(data.data() + offsets[i - 1]);
+            const size_t str_len = offsets[i] - offsets[i - 1] - 1;
 
-            auto lang = CLD2::DetectLanguage(str, strlen(str), true, &is_reliable);
-            res = codeISO(LanguageCode(lang));
-
-            size_t cur_offset = offsets[i];
+            auto lang = CLD2::DetectLanguage(str, str_len, true, &is_reliable);
+            auto res = codeISO(LanguageCode(lang));
 
             res_data.resize(res_offset + res.size() + 1);
             memcpy(&res_data[res_offset], res.data(), res.size());
-            res_offset += res.size();
 
-            res_data[res_offset] = 0;
-            ++res_offset;
+            res_data[res_offset + res.size()] = 0;
+            res_offset += res.size() + 1;
 
             res_offsets[i] = res_offset;
-            prev_offset = cur_offset;
         }
     }
 
-
+    [[noreturn]] static void vectorFixed(const ColumnString::Chars &, size_t, ColumnString::Chars &)
+    {
+        throw Exception("Cannot apply function detectProgrammingLanguage to fixed string.", ErrorCodes::ILLEGAL_COLUMN);
+    }
 };
 
 class LanguageClassificationMixedDetect : public IFunction
@@ -116,6 +104,9 @@ class LanguageClassificationMixedDetect : public IFunction
 public:
     static constexpr auto name = "detectLanguageMixed";
 
+    /// Number of top results
+    static constexpr auto top_N = 3;
+
     static FunctionPtr create(ContextPtr) { return std::make_shared<LanguageClassificationMixedDetect>(); }
 
     String getName() const override { return name; }
@@ -132,7 +123,7 @@ public:
             throw Exception(
                 "Illegal type " + arguments[0]->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT);
 
-        return std::make_shared<DataTypeMap>(std::make_shared<DataTypeString>(), std::make_shared<DataTypeInt32>());
+        return std::make_shared<DataTypeMap>(std::make_shared<DataTypeString>(), std::make_shared<DataTypeFloat32>());
     }
 
     ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override
@@ -145,8 +136,8 @@ public:
                 "Illegal columns " + arguments[0].column->getName() + " of arguments of function " + getName(),
                 ErrorCodes::ILLEGAL_COLUMN);
 
-        auto & input_data = col->getChars();
-        auto & input_offsets = col->getOffsets();
+        const auto & input_data = col->getChars();
+        const auto & input_offsets = col->getOffsets();
 
         /// Create and fill the result map.
 
@@ -158,15 +149,15 @@ public:
         MutableColumnPtr values_data = value_type->createColumn();
         MutableColumnPtr offsets = DataTypeNumber<IColumn::Offset>().createColumn();
 
-        size_t total_elements = input_rows_count * 3;
+        size_t total_elements = input_rows_count * top_N;
         keys_data->reserve(total_elements);
         values_data->reserve(total_elements);
         offsets->reserve(input_rows_count);
 
         bool is_reliable = true;
-        CLD2::Language result_lang_top3[3];
-        int32_t pc[3];
-        int bytes[3];
+        CLD2::Language result_lang_top3[top_N];
+        int32_t pc[top_N];
+        int bytes[top_N];
 
         IColumn::Offset current_offset = 0;
         for (size_t i = 0; i < input_rows_count; ++i)
@@ -176,16 +167,16 @@ public:
 
             CLD2::DetectLanguageSummary(str, str_len, true, result_lang_top3, pc, bytes, &is_reliable);
 
-            for (size_t j = 0; j < 3; ++j)
+            for (size_t j = 0; j < top_N; ++j)
             {
                 auto res_str = LanguageClassificationImpl::codeISO(LanguageCode(result_lang_top3[j]));
-                int32_t res_int = static_cast<int>(pc[j]);
+                Float32 res_float = static_cast<Float32>(pc[j]) / 100;
 
                 keys_data->insertData(res_str.data(), res_str.size());
-                values_data->insertData(reinterpret_cast<const char *>(&res_int), sizeof(res_int));
+                values_data->insertData(reinterpret_cast<const char *>(&res_float), sizeof(res_float));
             }
 
-            current_offset += 3;
+            current_offset += top_N;
             offsets->insert(current_offset);
         }
 
@@ -203,7 +194,7 @@ struct NameLanguageUTF8Detect
 };
 
 
-using FunctionLanguageUTF8Detect = FunctionsTextClassification<LanguageClassificationImpl, NameLanguageUTF8Detect>;
+using FunctionLanguageUTF8Detect = FunctionStringToString<LanguageClassificationImpl, NameLanguageUTF8Detect, false>;
 
 void registerFunctionLanguageDetectUTF8(FunctionFactory & factory)
 {

src/Functions/FunctionsProgrammingClassification.cpp

@@ -1,12 +1,18 @@
-#include <Functions/FunctionsTextClassification.h>
 #include <Common/FrequencyHolder.h>
 #include <Functions/FunctionFactory.h>
-#include <IO/ReadHelpers.h>
+#include <Functions/FunctionStringToString.h>
 
 #include <unordered_map>
+#include <string_view>
 
 namespace DB
 {
+
+namespace ErrorCodes
+{
+    extern const int ILLEGAL_COLUMN;
+}
+
 /**
   * Determine the programming language from the source code.
   * We calculate all the unigrams and bigrams of commands in the source code.
@@ -15,16 +21,16 @@ namespace DB
   */
 struct ProgrammingClassificationImpl
 {
-
-    using ResultType = String;
     /// Calculate total weight
-    static ALWAYS_INLINE inline Float64 stateMachine(const FrequencyHolder::Map & standard, std::unordered_map<String, Float64> & model)
+    static ALWAYS_INLINE inline Float64 stateMachine(
+        const FrequencyHolder::Map & standard,
+        const std::unordered_map<String, Float64> & model)
     {
         Float64 res = 0;
-        for (auto & el : model)
+        for (const auto & el : model)
         {
             /// Try to find each n-gram in dictionary
-            auto it = standard.find(el.first);
+            const auto * it = standard.find(el.first);
             if (it != standard.end())
             {
                 res += el.second * it->getMapped();
@@ -33,104 +39,44 @@ struct ProgrammingClassificationImpl
         return res;
     }
 
-
-    static void constant(String data, String & res)
-    {
-        auto & programming_freq = FrequencyHolder::getInstance().getProgrammingFrequency();
-        std::unordered_map<String, Float64> data_freq;
-
-        String prev_command;
-        String command;
-        /// Select all commands from the string
-        for (size_t i = 0; i < data.size();)
-        {
-            /// Assume that all commands are split by spaces
-            if (!isspace(data[i]))
-            {
-                command.push_back(data[i]);
-                ++i;
-
-                while ((i < data.size()) && (!isspace(data[i])))
-                {
-                    command.push_back(data[i]);
-                    ++i;
-                }
-                if (prev_command == "")
-                {
-                    prev_command = command;
-                }
-                else
-                {
-                    data_freq[prev_command + command] += 1;
-                    data_freq[prev_command] += 1;
-                    prev_command = command;
-                }
-                command = "";
-            }
-            else
-            {
-                ++i;
-            }
-        }
-
-        String most_liked;
-        Float64 max_result = 0;
-        /// Iterate over all programming languages ​​and find the language with the highest weight
-        for (auto& item : programming_freq)
-        {
-            Float64 result = stateMachine(item.map, data_freq);
-            if (result > max_result)
-            {
-                max_result = result;
-                most_liked = item.name;
-            }
-        }
-        /// If all weights are zero, then we assume that the language is undefined
-        if (most_liked == "")
-        {
-            most_liked = "Undefined";
-        }
-        res = most_liked;
-    }
-
-
     static void vector(
         const ColumnString::Chars & data,
         const ColumnString::Offsets & offsets,
         ColumnString::Chars & res_data,
         ColumnString::Offsets & res_offsets)
     {
-        auto & programming_freq = FrequencyHolder::getInstance().getProgrammingFrequency();
+        const auto & programming_freq = FrequencyHolder::getInstance().getProgrammingFrequency();
 
-        res_data.reserve(1024);
+        /// Constant 5 is arbitrary
+        res_data.reserve(offsets.size() * 5);
         res_offsets.resize(offsets.size());
 
-        size_t prev_offset = 0;
         size_t res_offset = 0;
 
         for (size_t i = 0; i < offsets.size(); ++i)
         {
-            const char * haystack = reinterpret_cast<const char *>(&data[prev_offset]);
+            const UInt8 * str = data.data() + offsets[i - 1];
+            const size_t str_len = offsets[i] - offsets[i - 1] - 1;
+
             std::unordered_map<String, Float64> data_freq;
-            String str_data = haystack;
 
             String prev_command;
             String command;
             /// Select all commands from the string
-            for (size_t ind = 0; ind < str_data.size();)
+            for (size_t ind = 0; ind < str_len;)
             {
                 /// Assume that all commands are split by spaces
-                if (!isspace(str_data[ind]))
+                if (!isspace(str[ind]))
                 {
-                    command.push_back(str_data[ind]);
+                    command.push_back(str[ind]);
                     ++ind;
 
-                    while ((ind < str_data.size()) && (!isspace(str_data[ind])))
+                    while ((ind < str_len) && (!isspace(str[ind])))
                     {
-                        command.push_back(str_data[ind]);
+                        command.push_back(str[ind]);
                         ++ind;
                     }
-                    if (prev_command == "")
+                    if (prev_command.empty())
                     {
                         prev_command = command;
                     }
@@ -148,39 +94,36 @@ struct ProgrammingClassificationImpl
                 }
             }
 
-            String most_liked;
+            String res;
             Float64 max_result = 0;
             /// Iterate over all programming languages ​​and find the language with the highest weight
-            for (auto& item : programming_freq)
+            for (const auto & item : programming_freq)
             {
                 Float64 result = stateMachine(item.map, data_freq);
                 if (result > max_result)
                 {
                     max_result = result;
-                    most_liked = item.name;
+                    res = item.name;
                 }
             }
             /// If all weights are zero, then we assume that the language is undefined
-            if (most_liked == "")
-            {
-                most_liked = "Undefined";
-            }
+            if (res.empty())
+                res = "Undefined";
 
-            const auto res = most_liked.c_str();
-            size_t cur_offset = offsets[i];
-            size_t ans_size = strlen(res);
-            res_data.resize(res_offset + ans_size + 1);
-            memcpy(&res_data[res_offset], res, ans_size);
-            res_offset += ans_size;
+            res_data.resize(res_offset + res.size() + 1);
+            memcpy(&res_data[res_offset], res.data(), res.size());
 
-            res_data[res_offset] = 0;
-            ++res_offset;
+            res_data[res_offset + res.size()] = 0;
+            res_offset += res.size() + 1;
 
             res_offsets[i] = res_offset;
-            prev_offset = cur_offset;
         }
     }
 
+    [[noreturn]] static void vectorFixed(const ColumnString::Chars &, size_t, ColumnString::Chars &)
+    {
+        throw Exception("Cannot apply function detectProgrammingLanguage to fixed string.", ErrorCodes::ILLEGAL_COLUMN);
+    }
 };
 
 struct NameGetProgramming
@@ -189,7 +132,7 @@ struct NameGetProgramming
 };
 
 
-using FunctionGetProgramming = FunctionsTextClassification<ProgrammingClassificationImpl, NameGetProgramming>;
+using FunctionGetProgramming = FunctionStringToString<ProgrammingClassificationImpl, NameGetProgramming, false>;
 
 void registerFunctionsProgrammingClassification(FunctionFactory & factory)
 {

src/Functions/FunctionsTextClassification.h

@@ -1,86 +0,0 @@
-#pragma once
-
-#include <Columns/ColumnConst.h>
-#include <Columns/ColumnString.h>
-#include <Columns/ColumnVector.h>
-#include <DataTypes/DataTypesNumber.h>
-#include <Functions/FunctionHelpers.h>
-#include <Functions/IFunction.h>
-#include <Interpreters/Context_fwd.h>
-
-namespace DB
-{
-/** Functions for text classification:
-  *
-  * detectCharset(string data) - detect charset of data.
-  * Returns string name of most likely charset.
-  *
-  * detectLanguage(string data) - detect language of data in various encodings (not UTF-8)
-  *
-  * getTonality(string data) - defines the emotional coloring of the text.
-  * Returns NEG if text is negative, POS if text is positive or NEUT if text is neutral.
-  *
-  * getProgrammingLanguage(string data) - detect programming language
-  */
-namespace ErrorCodes
-{
-extern const int ILLEGAL_TYPE_OF_ARGUMENT;
-extern const int ILLEGAL_COLUMN;
-}
-
-template <typename Impl, typename Name>
-class FunctionsTextClassification : public IFunction
-{
-public:
-    static constexpr auto name = Name::name;
-
-    static FunctionPtr create(ContextPtr) { return std::make_shared<FunctionsTextClassification>(); }
-
-    String getName() const override { return name; }
-
-    size_t getNumberOfArguments() const override { return 1; }
-
-    bool isSuitableForShortCircuitArgumentsExecution(const DataTypesWithConstInfo & /*arguments*/) const override { return true; }
-
-    DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override
-    {
-        if (!isString(arguments[0]))
-            throw Exception(
-                "Illegal type " + arguments[0]->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT);
-        return arguments[0];
-    }
-
-    ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t /*input_rows_count*/) const override
-    {
-        using ResultType = typename Impl::ResultType;
-
-        const ColumnPtr & column = arguments[0].column;
-
-        const ColumnConst * col_const = typeid_cast<const ColumnConst *>(&*column);
-
-        if (col_const)
-        {
-            ResultType res;
-            Impl::constant(col_const->getValue<String>(), res);
-            return result_type->createColumnConst(col_const->size(), toField(res));
-        }
-
-
-        if (const ColumnString * col = checkAndGetColumn<ColumnString>(column.get()))
-        {
-            auto col_res = ColumnString::create();
-            ColumnString::Chars & vec_res = col_res->getChars();
-            ColumnString::Offsets & offsets_res = col_res->getOffsets();
-            Impl::vector(col->getChars(), col->getOffsets(), vec_res, offsets_res);
-            return col_res;
-        }
-        else
-        {
-            throw Exception(
-                "Illegal columns " + arguments[0].column->getName() + " of arguments of function " + getName(),
-                ErrorCodes::ILLEGAL_COLUMN);
-        }
-    }
-};
-
-}

src/Functions/FunctionsTonalityClassification.cpp

@@ -1,163 +1,115 @@
-#include <Functions/FunctionsTextClassification.h>
-#include <Common/StringUtils/StringUtils.h>
 #include <Common/FrequencyHolder.h>
+#include <Common/StringUtils/StringUtils.h>
 #include <Functions/FunctionFactory.h>
-#include <Common/UTF8Helpers.h>
-#include <IO/ReadBufferFromString.h>
-#include <IO/ReadHelpers.h>
+#include <Functions/FunctionStringOrArrayToT.h>
 
 #include <unordered_map>
 
 namespace DB
 {
+
+namespace ErrorCodes
+{
+    extern const int ILLEGAL_TYPE_OF_ARGUMENT;
+}
+
 /**
   * Determines the sentiment of text data.
-  * Uses a marked-up sentiment dictionary, each word has a tonality ranging from -3 to 3.
+  * Uses a marked-up sentiment dictionary, each word has a tonality ranging from -12 to 6.
   * For each text, calculate the average sentiment value of its words and return NEG, POS or NEUT
   */
 struct TonalityClassificationImpl
 {
-
-    using ResultType = String;
-
-
-    static String get_tonality(const Float64 & tonality_level)
+    static Float32 detectTonality(const UInt8 * str, const size_t str_len, const FrequencyHolder::Map & emotional_dict)
     {
-        if (tonality_level < 0.15) { return "NEG"; }
-        if (tonality_level > 0.45) { return "POS"; }
-        return "NEUT";
-    }
-
-    static void constant(String data, String & res)
-    {
-        const auto & emotional_dict = FrequencyHolder::getInstance().getEmotionalDict();
-
         Float64 weight = 0;
-        Float64 count_words = 0;
+        UInt64 count_words = 0;
 
-        String answer;
         String word;
         /// Select all Russian words from the string
-        for (size_t i = 0; i < data.size();)
-        {
-            /// Assume that all non-Ascii characters are Russian letters
-            if (!isASCII(data[i]))
-            {
-                word.push_back(data[i]);
-                ++i;
-
-                while ((i < data.size()) && (!isASCII(data[i])))
-                {
-                    word.push_back(data[i]);
-                    ++i;
-                }
-                /// Try to find a russian word in the tonality dictionary
-                auto it = emotional_dict.find(word);
-                if (it != emotional_dict.end())
-                {
-                    count_words += 1;
-                    weight += it->getMapped();
-                }
-                word = "";
-            }
-            else
-            {
-                ++i;
-            }
-        }
-        /// Calculate average value of tonality
-        Float64 total_tonality = weight / count_words;
-        res += get_tonality(total_tonality);
-    }
-
-
-    static void vector(
-        const ColumnString::Chars & data,
-        const ColumnString::Offsets & offsets,
-        ColumnString::Chars & res_data,
-        ColumnString::Offsets & res_offsets)
-    {
-        const auto & emotional_dict = FrequencyHolder::getInstance().getEmotionalDict();
-
-        res_data.reserve(1024);
-        res_offsets.resize(offsets.size());
-
-        size_t prev_offset = 0;
-        size_t res_offset = 0;
-
-        for (size_t i = 0; i < offsets.size(); ++i)
-        {
-            const char * haystack = reinterpret_cast<const char *>(&data[prev_offset]);
-            String str = haystack;
-
-            String buf;
-
-            Float64 weight = 0;
-            Float64 count_words = 0;
-
-
-            String answer;
-            String word;
-            /// Select all Russian words from the string
-            for (size_t ind = 0; ind < str.size();)
+        for (size_t ind = 0; ind < str_len;)
         {
+            /// Assume that all non-ASCII characters are Russian letters
             if (!isASCII(str[ind]))
             {
                 word.push_back(str[ind]);
                 ++ind;
 
-                    while ((ind < str.size()) && (!isASCII(str[ind])))
+                while ((ind < str_len) && (!isASCII(str[ind])))
                 {
                     word.push_back(str[ind]);
                     ++ind;
                 }
                 /// Try to find a russian word in the tonality dictionary
-                    auto it = emotional_dict.find(word);
+                const auto * it = emotional_dict.find(word);
                 if (it != emotional_dict.end())
                 {
                     count_words += 1;
                     weight += it->getMapped();
                 }
-                    word = "";
+                word.clear();
             }
             else
             {
                 ++ind;
             }
         }
-            /// Calculate average value of tonality
-            Float64 total_tonality = weight / count_words;
-            buf = get_tonality(total_tonality);
+        /// Calculate average value of tonality.
+        /// Convert values -12..6 to -1..1
+        return std::max(weight / count_words / 6, -1.0);
+    }
 
-            const auto res = buf.c_str();
-            size_t cur_offset = offsets[i];
-            size_t ans_size = strlen(res);
-            res_data.resize(res_offset + ans_size + 1);
-            memcpy(&res_data[res_offset], res, ans_size);
-            res_offset += ans_size;
+    /// If the function will return constant value for FixedString data type.
+    static constexpr auto is_fixed_to_constant = false;
 
-            res_data[res_offset] = 0;
-            ++res_offset;
+    static void vector(
+        const ColumnString::Chars & data,
+        const ColumnString::Offsets & offsets,
+        PaddedPODArray<Float32> & res)
+    {
+        const auto & emotional_dict = FrequencyHolder::getInstance().getEmotionalDict();
 
-            res_offsets[i] = res_offset;
-            prev_offset = cur_offset;
+        size_t size = offsets.size();
+        size_t prev_offset = 0;
+        for (size_t i = 0; i < size; ++i)
+        {
+            res[i] = detectTonality(data.data() + prev_offset, offsets[i] - 1 - prev_offset, emotional_dict);
+            prev_offset = offsets[i];
         }
     }
 
+    static void vectorFixedToConstant(const ColumnString::Chars & /*data*/, size_t /*n*/, Float32 & /*res*/) {}
 
+    static void vectorFixedToVector(const ColumnString::Chars & data, size_t n, PaddedPODArray<Float32> & res)
+    {
+        const auto & emotional_dict = FrequencyHolder::getInstance().getEmotionalDict();
+
+        size_t size = data.size() / n;
+        for (size_t i = 0; i < size; ++i)
+            res[i] = detectTonality(data.data() + i * n, n, emotional_dict);
+    }
+
+    [[noreturn]] static void array(const ColumnString::Offsets &, PaddedPODArray<Float32> &)
+    {
+        throw Exception("Cannot apply function detectTonality to Array argument", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT);
+    }
+
+    [[noreturn]] static void uuid(const ColumnUUID::Container &, size_t &, PaddedPODArray<Float32> &)
+    {
+        throw Exception("Cannot apply function detectTonality to UUID argument", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT);
+    }
 };
 
-struct NameGetTonality
+struct NameDetectTonality
 {
     static constexpr auto name = "detectTonality";
 };
 
-
-using FunctionGetTonality = FunctionsTextClassification<TonalityClassificationImpl, NameGetTonality>;
+using FunctionDetectTonality = FunctionStringOrArrayToT<TonalityClassificationImpl, NameDetectTonality, Float32>;
 
 void registerFunctionsTonalityClassification(FunctionFactory & factory)
 {
-    factory.registerFunction<FunctionGetTonality>();
+    factory.registerFunction<FunctionDetectTonality>();
 }
 
 }

src/Interpreters/CMakeLists.txt

@@ -2,9 +2,9 @@ if (ENABLE_TESTS)
     add_subdirectory(tests)
 endif()
 
-# if (ENABLE_EXAMPLES)
+if (ENABLE_EXAMPLES)
     add_subdirectory(examples)
-# endif()
+endif()
 
 if (ENABLE_FUZZING)
     add_subdirectory(fuzzers)

tests/queries/0_stateless/02133_classification.reference

@@ -3,10 +3,10 @@ en
 fr
 ja
 zh
-{'ja':62,'fr':36,'un':0}
+{'ja':0.62,'fr':0.36,'un':0}
 ISO-8859-1
-English
-POS
-NEG
-POS
+en
+0.465
+-0.57647777
+0.050505556
 C++

tests/queries/0_stateless/02133_classification.sql

@@ -11,8 +11,8 @@ SELECT detectLanguageMixed('二兎を追う者は一兎をも得ず二兎を追
 SELECT detectCharset('Plain English');
 SELECT detectLanguageUnknown('Plain English');
 
-SELECT detectTonality('Милая кошка');
-SELECT detectTonality('Злой человек');
-SELECT detectTonality('Обычная прогулка по ближайшему парку');
+SELECT detectTonality('милая кошка');
+SELECT detectTonality('ненависть к людям');
+SELECT detectTonality('обычная прогулка по ближайшему парку');
 
 SELECT detectProgrammingLanguage('#include <iostream>');