@fugood/llama.node 0.2.2 → 0.3.0

package/src/llama.cpp/{unicode.cpp → src/unicode.cpp}

@@ -1,3 +1,7 @@
+#if defined(_MSC_VER)
+#define _SILENCE_CXX17_CODECVT_HEADER_DEPRECATION_WARNING
+#endif
+
 #include "unicode.h"
 #include "unicode-data.h"
 
@@ -15,6 +19,12 @@
 #include <locale>
 #include <codecvt>
 
+size_t unicode_len_utf8(char src) {
+    const size_t lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
+    uint8_t highbits = static_cast<uint8_t>(src) >> 4;
+    return lookup[highbits];
+}
+
 static std::string unicode_cpts_to_utf8(const std::vector<uint32_t> & cps) {
     std::string result;
     for (size_t i = 0; i < cps.size(); ++i) {
@@ -23,7 +33,7 @@ static std::string unicode_cpts_to_utf8(const std::vector<uint32_t> & cps) {
     return result;
 }
 
-static uint32_t unicode_cpt_from_utf8(const std::string & utf8, size_t & offset) {
+uint32_t unicode_cpt_from_utf8(const std::string & utf8, size_t & offset) {
     assert(offset < utf8.size());
     if (!(utf8[offset + 0] & 0x80)) {
         auto result = utf8[offset + 0];
@@ -226,13 +236,13 @@ static std::vector<size_t> unicode_regex_split_custom_gpt2(const std::string & t
         assert(offset_end <= cpts.size());
         start = offset_end;
 
-        auto _get_cpt = [&] (const size_t pos) -> char32_t {
-            return (offset_ini <= pos && pos < offset_end) ? cpts[pos] : 0;
+        static const uint32_t OUT_OF_RANGE = 0xFFFFFFFF;
+        auto _get_cpt = [&] (const size_t pos) -> uint32_t {
+            return (offset_ini <= pos && pos < offset_end) ? cpts[pos] : OUT_OF_RANGE;
         };
 
         auto _get_flags = [&] (const size_t pos) -> codepoint_flags {
-            static const codepoint_flags undef(codepoint_flags::UNDEFINED);
-            return (offset_ini <= pos && pos < offset_end) ? unicode_cpt_flags(cpts[pos]) : undef;
+            return (offset_ini <= pos && pos < offset_end) ? unicode_cpt_flags(cpts[pos]) : codepoint_flags{};
         };
 
         size_t _prev_end = offset_ini;
@@ -253,18 +263,18 @@ static std::vector<size_t> unicode_regex_split_custom_gpt2(const std::string & t
         };
 
         for (size_t pos = offset_ini; pos < offset_end; /*pos++*/ ) {
-            const char32_t cpt = _get_cpt(pos);
+            const uint32_t cpt = _get_cpt(pos);
             const auto flags = _get_flags(pos);
 
             // regex: 's|'t|'re|'ve|'m|'ll|'d
             if (cpt == '\'' && pos+1 < offset_end) {
-                char32_t cpt_next = _get_cpt(pos+1);
+                uint32_t cpt_next = _get_cpt(pos+1);
                 if (cpt_next == 's' || cpt_next == 't' || cpt_next == 'm' || cpt_next == 'd') {
                     pos += _add_token(pos+2);
                     continue;
                 }
                 if (pos+2 < offset_end) {
-                    char32_t cpt_next_next = _get_cpt(pos+2);
+                    uint32_t cpt_next_next = _get_cpt(pos+2);
                     if ((cpt_next == 'r' && cpt_next_next == 'e') ||
                         (cpt_next == 'v' && cpt_next_next == 'e') ||
                         (cpt_next == 'l' && cpt_next_next == 'l')) {
@@ -294,9 +304,9 @@ static std::vector<size_t> unicode_regex_split_custom_gpt2(const std::string & t
                 continue;
             }
             // regex: <space>?[^\s\p{L}\p{N}]+
-            if (!(flags2.is_whitespace || flags2.is_letter || flags2.is_number || flags2.is_undefined)) {
+            if (!(flags2.is_whitespace | flags2.is_letter | flags2.is_number) && flags2.as_uint()) {
                 pos += (cpt == ' ');
-                while (!(flags2.is_whitespace || flags2.is_letter || flags2.is_number || flags2.is_undefined)) {
+                while (!(flags2.is_whitespace | flags2.is_letter | flags2.is_number) && flags2.as_uint()) {
                     flags2 = _get_flags(++pos);
                 }
                 _add_token(pos);
@@ -309,7 +319,7 @@ static std::vector<size_t> unicode_regex_split_custom_gpt2(const std::string & t
             }
 
             // regex: \s+(?!\S)
-            if (num_whitespaces > 1 && _get_cpt(pos+num_whitespaces) != 0) {
+            if (num_whitespaces > 1 && _get_cpt(pos+num_whitespaces) != OUT_OF_RANGE) {
                 pos += num_whitespaces - 1;
                 _add_token(pos);
                 continue;
@@ -344,13 +354,13 @@ static std::vector<size_t> unicode_regex_split_custom_llama3(const std::string &
         assert(offset_end <= cpts.size());
         start = offset_end;
 
-        auto _get_cpt = [&] (const size_t pos) -> char32_t {
-            return (offset_ini <= pos && pos < offset_end) ? cpts[pos] : 0;
+        static const uint32_t OUT_OF_RANGE = 0xFFFFFFFF;
+        auto _get_cpt = [&] (const size_t pos) -> uint32_t {
+            return (offset_ini <= pos && pos < offset_end) ? cpts[pos] : OUT_OF_RANGE;
         };
 
         auto _get_flags = [&] (const size_t pos) -> codepoint_flags {
-            static const codepoint_flags undef(codepoint_flags::UNDEFINED);
-            return (offset_ini <= pos && pos < offset_end) ? unicode_cpt_flags(cpts[pos]) : undef;
+            return (offset_ini <= pos && pos < offset_end) ? unicode_cpt_flags(cpts[pos]) : codepoint_flags{};
         };
 
         size_t _prev_end = offset_ini;
@@ -371,18 +381,18 @@ static std::vector<size_t> unicode_regex_split_custom_llama3(const std::string &
         };
 
         for (size_t pos = offset_ini; pos < offset_end; /*pos++*/ ) {
-            const char32_t cpt = _get_cpt(pos);
+            const uint32_t cpt = _get_cpt(pos);
             const auto flags = _get_flags(pos);
 
             // regex: (?i:'s|'t|'re|'ve|'m|'ll|'d) // case insensitive
             if (cpt == '\'' && pos+1 < offset_end) {
-                char32_t cpt_next = unicode_tolower(_get_cpt(pos+1));
+                uint32_t cpt_next = unicode_tolower(_get_cpt(pos+1));
                 if (cpt_next == 's' || cpt_next == 't' || cpt_next == 'm' || cpt_next == 'd') {
                     pos += _add_token(pos+2);
                     continue;
                 }
                 if (pos+2 < offset_end) {
-                    char32_t cpt_next_next = unicode_tolower(_get_cpt(pos+2));
+                    uint32_t cpt_next_next = unicode_tolower(_get_cpt(pos+2));
                     if ((cpt_next == 'r' && cpt_next_next == 'e') ||
                         (cpt_next == 'v' && cpt_next_next == 'e') ||
                         (cpt_next == 'l' && cpt_next_next == 'l')) {
@@ -392,8 +402,8 @@ static std::vector<size_t> unicode_regex_split_custom_llama3(const std::string &
                 }
             }
 
-            // regex: [^\r\n\p{L}\p{N}]?\p{L}+  //####FIXME: the first \p{L} is correct?
-            if (!(cpt == '\r' || cpt == '\n' || /*flags.is_letter |*/ flags.is_number)) {
+            // regex: [^\r\n\p{L}\p{N}]?\p{L}+
+            if (!(cpt == '\r' || cpt == '\n' || flags.is_number)) {
                 if (flags.is_letter || _get_flags(pos+1).is_letter) {  // one or more letters
                     pos++;
                     while (_get_flags(pos).is_letter) {
@@ -419,12 +429,12 @@ static std::vector<size_t> unicode_regex_split_custom_llama3(const std::string &
 
             // regex: <space>?[^\s\p{L}\p{N}]+[\r\n]*
             auto flags2 = (cpt == ' ' ? _get_flags(pos+1) : flags);
-            if (!(flags2.is_whitespace || flags2.is_letter || flags2.is_number || flags2.is_undefined)) {
+            if (!(flags2.is_whitespace | flags2.is_letter | flags2.is_number) && flags.as_uint()) {
                 pos += (cpt == ' ');
-                while (!(flags2.is_whitespace || flags2.is_letter || flags2.is_number || flags2.is_undefined)) {
+                while (!(flags2.is_whitespace | flags2.is_letter | flags2.is_number) && flags2.as_uint()) {
                     flags2 = _get_flags(++pos);
                 }
-                char32_t cpt2 = _get_cpt(pos);
+                uint32_t cpt2 = _get_cpt(pos);
                 while (cpt2 == '\r' || cpt2 == '\n') {
                     cpt2 = _get_cpt(++pos);
                 }
@@ -435,7 +445,7 @@ static std::vector<size_t> unicode_regex_split_custom_llama3(const std::string &
             size_t num_whitespaces = 0;
             size_t last_end_r_or_n = 0;
             while (_get_flags(pos+num_whitespaces).is_whitespace) {
-                char32_t cpt2 = _get_cpt(pos+num_whitespaces);
+                uint32_t cpt2 = _get_cpt(pos+num_whitespaces);
                 if (cpt2 == '\r' || cpt2 == '\n') {
                     last_end_r_or_n = pos + num_whitespaces + 1;
                 }
@@ -450,7 +460,7 @@ static std::vector<size_t> unicode_regex_split_custom_llama3(const std::string &
             }
 
             // regex: \s+(?!\S)
-            if (num_whitespaces > 1 && _get_cpt(pos+num_whitespaces) != 0) {
+            if (num_whitespaces > 1 && _get_cpt(pos+num_whitespaces) != OUT_OF_RANGE) {
                 pos += num_whitespaces - 1;
                 _add_token(pos);
                 continue;
@@ -594,6 +604,7 @@ std::vector<uint32_t> unicode_cpts_normalize_nfd(const std::vector<uint32_t> & c
 
 std::vector<uint32_t> unicode_cpts_from_utf8(const std::string & utf8) {
     std::vector<uint32_t> result;
+    result.reserve(utf8.size());
     size_t offset = 0;
     while (offset < utf8.size()) {
         result.push_back(unicode_cpt_from_utf8(utf8, offset));
@@ -626,7 +637,7 @@ uint8_t unicode_utf8_to_byte(const std::string & utf8) {
     return map.at(utf8);
 }
 
-char32_t unicode_tolower(char32_t cp) {
+uint32_t unicode_tolower(uint32_t cp) {
     auto it = unicode_map_lowercase.find(cp);
     return it == unicode_map_lowercase.end() ? cp : it->second;
 }
@@ -679,10 +690,14 @@ std::vector<std::string> unicode_regex_split(const std::string & text, const std
                 continue;
             }
 
-            const int cpt_flag = unicode_cpt_flags(cpts[i]).category_flag();
+            const auto flags = unicode_cpt_flags(cpts[i]);
 
-            if (k_ucat_cpt.find(cpt_flag) != k_ucat_cpt.end()) {
-                text_collapsed[i] = k_ucat_cpt.at(cpt_flag);
+            if (flags.is_whitespace) {
+                //NOTE: C++ std::regex \s does not mach 0x85, Rust and Python regex does.
+                //text_collapsed[i] = (char) 0x85;  // <Next Line> as whitespace fallback
+                text_collapsed[i] = (char) 0x0B;    // <vertical tab> as whitespace fallback
+            } else if (k_ucat_cpt.find(flags.category_flag()) != k_ucat_cpt.end()) {
+                text_collapsed[i] = k_ucat_cpt.at(flags.category_flag());
             } else {
                 text_collapsed[i] = (char) 0xD0; // fallback
             }
@@ -766,9 +781,16 @@ std::vector<std::string> unicode_regex_split(const std::string & text, const std
                 bpe_offsets = unicode_regex_split_stl(text_collapsed, regex_expr_collapsed, bpe_offsets);
             } else {
                 // no unicode category used, we can use std::wregex directly
-                const std::wstring wtext       = unicode_wstring_from_utf8(text);
                 const std::wstring wregex_expr = unicode_wstring_from_utf8(regex_expr);
 
+                // std::wregex \s does not mach non-ASCII whitespaces, using 0x0B as fallback
+                std::wstring wtext(cpts.begin(), cpts.end());
+                for (size_t i = 0; i < wtext.size(); ++i) {
+                    if (wtext[i] > 0x7F && unicode_cpt_flags(wtext[i]).is_whitespace) {
+                        wtext[i] = 0x0B;
+                    }
+                }
+
                 //printf("text: %s\n", text.c_str());
                 //printf("regex_expr: %s\n", regex_expr.c_str());
                 bpe_offsets = unicode_regex_split_stl(wtext, wregex_expr, bpe_offsets);

package/src/llama.cpp/{unicode.h → src/unicode.h}

@@ -4,6 +4,8 @@
 #include <string>
 #include <vector>
 
+// TODO: prefix all symbols with "llama_"
+
 struct codepoint_flags {
     enum {
         UNDEFINED       = 0x0001,
@@ -46,8 +48,10 @@ struct codepoint_flags {
     }
 };
 
+size_t unicode_len_utf8(char src);
 
 std::string unicode_cpt_to_utf8(uint32_t cp);
+uint32_t unicode_cpt_from_utf8(const std::string & utf8, size_t & offset);
 std::vector<uint32_t> unicode_cpts_from_utf8(const std::string & utf8);
 
 std::vector<uint32_t> unicode_cpts_normalize_nfd(const std::vector<uint32_t> & cpts);
@@ -58,6 +62,6 @@ codepoint_flags unicode_cpt_flags(const std::string & utf8);
 std::string unicode_byte_to_utf8(uint8_t byte);
 uint8_t unicode_utf8_to_byte(const std::string & utf8);
 
-char32_t unicode_tolower(char32_t cp);
+uint32_t unicode_tolower(uint32_t cp);
 
 std::vector<std::string> unicode_regex_split(const std::string & text, const std::vector<std::string> & regex_exprs);

package/src/llama.cpp/tests/CMakeLists.txt

@@ -70,21 +70,19 @@ add_executable(test-tokenizer-0 test-tokenizer-0.cpp)
 target_link_libraries(test-tokenizer-0 PRIVATE common)
 install(TARGETS test-tokenizer-0 RUNTIME)
 
-llama_test(test-tokenizer-0 NAME test-tokenizer-0-llama-spm         ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama-spm.gguf)
-llama_test(test-tokenizer-0 NAME test-tokenizer-0-llama-bpe         ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama-bpe.gguf)
-llama_test(test-tokenizer-0 NAME test-tokenizer-0-phi-3             ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-phi-3.gguf)
-llama_test(test-tokenizer-0 NAME test-tokenizer-0-falcon            ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-falcon.gguf)
 llama_test(test-tokenizer-0 NAME test-tokenizer-0-bert-bge          ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-bert-bge.gguf)
-# TODO: enable when fixed
-#       https://github.com/ggerganov/llama.cpp/pull/7036
-#llama_test(test-tokenizer-0 NAME test-tokenizer-0-mpt               ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-mpt.gguf)
-#llama_test(test-tokenizer-0 NAME test-tokenizer-0-deepseek-llm      ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-deepseek-llm.gguf)
-#llama_test(test-tokenizer-0 NAME test-tokenizer-0-deepseek-coder    ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-deepseek-coder.gguf)
-llama_test(test-tokenizer-0 NAME test-tokenizer-0-starcoder         ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-starcoder.gguf)
-llama_test(test-tokenizer-0 NAME test-tokenizer-0-gpt-2             ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-gpt-2.gguf)
-llama_test(test-tokenizer-0 NAME test-tokenizer-0-refact            ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-refact.gguf)
 llama_test(test-tokenizer-0 NAME test-tokenizer-0-command-r         ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-command-r.gguf)
+llama_test(test-tokenizer-0 NAME test-tokenizer-0-deepseek-coder    ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-deepseek-coder.gguf)
+llama_test(test-tokenizer-0 NAME test-tokenizer-0-deepseek-llm      ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-deepseek-llm.gguf)
+llama_test(test-tokenizer-0 NAME test-tokenizer-0-falcon            ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-falcon.gguf)
+llama_test(test-tokenizer-0 NAME test-tokenizer-0-gpt-2             ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-gpt-2.gguf)
+llama_test(test-tokenizer-0 NAME test-tokenizer-0-llama-bpe         ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama-bpe.gguf)
+llama_test(test-tokenizer-0 NAME test-tokenizer-0-llama-spm         ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama-spm.gguf)
+llama_test(test-tokenizer-0 NAME test-tokenizer-0-mpt               ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-mpt.gguf)
+llama_test(test-tokenizer-0 NAME test-tokenizer-0-phi-3             ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-phi-3.gguf)
 llama_test(test-tokenizer-0 NAME test-tokenizer-0-qwen2             ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-qwen2.gguf)
+llama_test(test-tokenizer-0 NAME test-tokenizer-0-refact            ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-refact.gguf)
+llama_test(test-tokenizer-0 NAME test-tokenizer-0-starcoder         ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-starcoder.gguf)
 
 # build test-tokenizer-1-bpe target once and add many tests
 add_executable(test-tokenizer-1-bpe test-tokenizer-1-bpe.cpp)
@@ -92,16 +90,14 @@ target_link_libraries(test-tokenizer-1-bpe PRIVATE common)
 install(TARGETS test-tokenizer-1-bpe RUNTIME)
 
 # TODO: disabled due to slowness
-#llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-llama-bpe ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama-bpe.gguf --ignore-merges)
-#llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-falcon    ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-falcon.gguf)
 #llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-aquila    ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-aquila.gguf)
-#llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-mpt       ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-mpt.gguf)
-#llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-stablelm  ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-stablelm.gguf)
+#llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-falcon    ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-falcon.gguf)
+#llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-gpt-2     ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-gpt-2.gguf)
 #llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-gpt-neox  ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-gpt-neox.gguf)
+#llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-llama-bpe ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama-bpe.gguf --ignore-merges)
+#llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-mpt       ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-mpt.gguf)
 #llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-refact    ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-refact.gguf)
 #llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-starcoder ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-starcoder.gguf)
-#llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-gpt2      ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-gpt2.gguf)
-#llama_test(test-tokenizer-1-bpe NAME test-tokenizer-1-bloom     ARGS ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-bloom.gguf)
 
 # build test-tokenizer-1-spm target once and add many tests
 add_executable(test-tokenizer-1-spm test-tokenizer-1-spm.cpp)
@@ -129,8 +125,11 @@ llama_target_and_test(test-rope.cpp)
 llama_target_and_test(test-model-load-cancel.cpp  LABEL "model")
 llama_target_and_test(test-autorelease.cpp        LABEL "model")
 
-llama_target_and_test(test-json-schema-to-grammar.cpp   WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/..)
-target_include_directories(test-json-schema-to-grammar PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/../examples/server)
+# TODO: disabled on loongarch64 because the ggml-ci node lacks Python 3.8
+if (NOT ${CMAKE_SYSTEM_PROCESSOR} MATCHES "loongarch64")
+    llama_target_and_test(test-json-schema-to-grammar.cpp   WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/..)
+    target_include_directories(test-json-schema-to-grammar PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/../examples/server)
+endif()
 
 # dummy executable - not installed
 get_filename_component(TEST_TARGET test-c.c NAME_WE)