Restructure string_utils into a library. (#200)

y-scope · Dec 31, 2023 · abda9a3 · abda9a3
1 parent 4124824
commit abda9a3
Show file tree

Hide file tree

Showing 19 changed files with 141 additions and 98 deletions.
diff --git a/components/core/.clang-format b/components/core/.clang-format
@@ -70,8 +70,10 @@ FixNamespaceComments: true
 IncludeBlocks: "Regroup"
 IncludeCategories:
   # NOTE: A header is grouped by first matching regex
-  # Third-party headers. Update when adding new third-party libraries.
-  - Regex: "^<(archive|boost|catch2|date|fmt|json|log_surgeon|mariadb|spdlog|sqlite3|yaml-cpp|zstd)"
+  # Library headers. Update when adding new libraries.
+  # NOTE: clang-format retains leading white-space on a line in violation of the YAML spec.
+  - Regex: "^<(archive|boost|catch2|date|fmt|json|log_surgeon|mariadb|spdlog|sqlite3|string_utils\
+|yaml-cpp|zstd)"
     Priority: 3
   # C system headers
   - Regex: "^<.+\\.h>"

diff --git a/components/core/CMakeLists.txt b/components/core/CMakeLists.txt
@@ -302,8 +302,6 @@ set(SOURCE_FILES_clp
         src/streaming_compression/zstd/Constants.hpp
         src/streaming_compression/zstd/Decompressor.cpp
         src/streaming_compression/zstd/Decompressor.hpp
-        src/string_utils.cpp
-        src/string_utils.hpp
         src/StringReader.cpp
         src/StringReader.hpp
         src/TimestampPattern.cpp
@@ -338,6 +336,7 @@ target_link_libraries(clp
         LibArchive::LibArchive
         MariaDBClient::MariaDBClient
         ${STD_FS_LIBS}
+        string_utils::string_utils
         yaml-cpp::yaml-cpp
         ZStd::ZStd
         )
@@ -449,8 +448,6 @@ set(SOURCE_FILES_clg
         src/streaming_compression/zstd/Constants.hpp
         src/streaming_compression/zstd/Decompressor.cpp
         src/streaming_compression/zstd/Decompressor.hpp
-        src/string_utils.cpp
-        src/string_utils.hpp
         src/StringReader.cpp
         src/StringReader.hpp
         src/TimestampPattern.cpp
@@ -486,6 +483,7 @@ target_link_libraries(clg
         spdlog::spdlog
         ${sqlite_LIBRARY_DEPENDENCIES}
         ${STD_FS_LIBS}
+        string_utils::string_utils
         yaml-cpp::yaml-cpp
         ZStd::ZStd
         )
@@ -583,8 +581,6 @@ set(SOURCE_FILES_clo
         src/streaming_compression/zstd/Constants.hpp
         src/streaming_compression/zstd/Decompressor.cpp
         src/streaming_compression/zstd/Decompressor.hpp
-        src/string_utils.cpp
-        src/string_utils.hpp
         src/StringReader.cpp
         src/StringReader.hpp
         src/Thread.cpp
@@ -621,6 +617,7 @@ target_link_libraries(clo
         spdlog::spdlog
         ${sqlite_LIBRARY_DEPENDENCIES}
         ${STD_FS_LIBS}
+        string_utils::string_utils
         ZStd::ZStd
         )
 target_compile_features(clo
@@ -774,9 +771,6 @@ set(SOURCE_FILES_unitTest
         src/streaming_compression/zstd/Constants.hpp
         src/streaming_compression/zstd/Decompressor.cpp
         src/streaming_compression/zstd/Decompressor.hpp
-        src/string_utils.cpp
-        src/string_utils.hpp
-        src/string_utils.inc
         src/StringReader.cpp
         src/StringReader.hpp
         src/TimestampPattern.cpp
@@ -828,6 +822,7 @@ target_link_libraries(unitTest
         spdlog::spdlog
         ${sqlite_LIBRARY_DEPENDENCIES}
         ${STD_FS_LIBS}
+        string_utils::string_utils
         yaml-cpp::yaml-cpp
         ZStd::ZStd
         )
@@ -836,3 +831,5 @@ target_compile_features(unitTest
         )
 
 include(cmake/utils.cmake)
+
+add_subdirectory(src)
diff --git a/components/core/cmake/utils.cmake b/components/core/cmake/utils.cmake
@@ -22,8 +22,6 @@ set(SOURCE_FILES_make-dictionaries-readable
         ${CMAKE_CURRENT_SOURCE_DIR}/src/streaming_compression/passthrough/Decompressor.hpp
         ${CMAKE_CURRENT_SOURCE_DIR}/src/streaming_compression/zstd/Decompressor.cpp
         ${CMAKE_CURRENT_SOURCE_DIR}/src/streaming_compression/zstd/Decompressor.hpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/src/string_utils.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/src/string_utils.hpp
         ${CMAKE_CURRENT_SOURCE_DIR}/src/Utils.cpp
         ${CMAKE_CURRENT_SOURCE_DIR}/src/Utils.hpp
         ${CMAKE_CURRENT_SOURCE_DIR}/src/utils/make_dictionaries_readable/CommandLineArguments.cpp
@@ -46,6 +44,7 @@ target_link_libraries(make-dictionaries-readable
         Boost::filesystem Boost::iostreams Boost::program_options
         log_surgeon::log_surgeon
         spdlog::spdlog
+        string_utils::string_utils
         ZStd::ZStd
         )
 target_compile_features(make-dictionaries-readable

diff --git a/components/core/src/string_utils.hpp → ...ore/include/string_utils/string_utils.hpp b/components/core/src/string_utils.hpp → ...ore/include/string_utils/string_utils.hpp
@@ -1,8 +1,10 @@
 #ifndef STRING_UTILS_HPP
 #define STRING_UTILS_HPP
 
+#include <charconv>
 #include <string>
 
+namespace string_utils {
 /**
  * Checks if the given character is an alphabet
  * @param c
@@ -122,6 +124,16 @@ bool wildcard_match_unsafe_case_sensitive(std::string_view tame, std::string_vie
 template <typename integer_t>
 bool convert_string_to_int(std::string_view raw, integer_t& converted);
 
-#include "string_utils.inc"
+template <typename integer_t>
+bool convert_string_to_int(std::string_view raw, integer_t& converted) {
+    auto raw_end = raw.cend();
+    auto result = std::from_chars(raw.cbegin(), raw_end, converted);
+    if (raw_end != result.ptr) {
+        return false;
+    } else {
+        return result.ec == std::errc();
+    }
+}
+}  // namespace string_utils
 
 #endif  // STRING_UTILS_HPP
diff --git a/components/core/src/CMakeLists.txt b/components/core/src/CMakeLists.txt
@@ -0,0 +1,12 @@
+set(
+        STRING_UTILS_HEADER_LIST
+        "${PROJECT_SOURCE_DIR}/include/string_utils/string_utils.hpp"
+)
+add_library(
+        string_utils
+        string_utils/string_utils.cpp
+        ${STRING_UTILS_HEADER_LIST}
+)
+add_library(string_utils::string_utils ALIAS string_utils)
+target_include_directories(string_utils PUBLIC "${PROJECT_SOURCE_DIR}/include")
+target_compile_features(string_utils PRIVATE cxx_std_17)
diff --git a/components/core/src/DictionaryReader.hpp b/components/core/src/DictionaryReader.hpp
@@ -5,13 +5,13 @@
 #include <vector>
 
 #include <boost/algorithm/string.hpp>
+#include <string_utils/string_utils.hpp>
 
 #include "dictionary_utils.hpp"
 #include "DictionaryEntry.hpp"
 #include "FileReader.hpp"
 #include "streaming_compression/passthrough/Decompressor.hpp"
 #include "streaming_compression/zstd/Decompressor.hpp"
-#include "string_utils.hpp"
 #include "Utils.hpp"
 
 /**
@@ -256,7 +256,12 @@ void DictionaryReader<DictionaryIdType, EntryType>::get_entries_matching_wildcar
         std::unordered_set<EntryType const*>& entries
 ) const {
     for (auto const& entry : m_entries) {
-        if (wildcard_match_unsafe(entry.get_value(), wildcard_string, false == ignore_case)) {
+        if (string_utils::wildcard_match_unsafe(
+                    entry.get_value(),
+                    wildcard_string,
+                    false == ignore_case
+            ))
+        {
             entries.insert(&entry);
         }
     }

diff --git a/components/core/src/EncodedVariableInterpreter.cpp b/components/core/src/EncodedVariableInterpreter.cpp
@@ -3,12 +3,13 @@
 #include <cassert>
 #include <cmath>
 
+#include <string_utils/string_utils.hpp>
+
 #include "Defs.h"
 #include "ffi/ir_stream/decoding_methods.hpp"
 #include "ir/LogEvent.hpp"
 #include "ir/types.hpp"
 #include "spdlog_with_specializations.hpp"
-#include "string_utils.hpp"
 #include "type_utils.hpp"
 
 using ffi::cEightByteEncodedFloatDigitsBitMask;
@@ -55,7 +56,7 @@ bool EncodedVariableInterpreter::convert_string_to_representable_integer_var(
     }
 
     int64_t result;
-    if (false == convert_string_to_int(value, result)) {
+    if (false == string_utils::convert_string_to_int(value, result)) {
         // Conversion failed
         return false;
     } else {

diff --git a/components/core/src/Grep.cpp b/components/core/src/Grep.cpp
@@ -3,6 +3,7 @@
 #include <algorithm>
 
 #include <log_surgeon/Constants.hpp>
+#include <string_utils/string_utils.hpp>
 
 #include "EncodedVariableInterpreter.hpp"
 #include "ir/parsing.hpp"
@@ -17,6 +18,10 @@ using std::vector;
 using streaming_archive::reader::Archive;
 using streaming_archive::reader::File;
 using streaming_archive::reader::Message;
+using string_utils::clean_up_wildcard_search_string;
+using string_utils::is_alphabet;
+using string_utils::is_wildcard;
+using string_utils::wildcard_match_unsafe;
 
 // Local types
 enum class SubQueryMatchabilityResult {
@@ -693,7 +698,7 @@ bool Grep::get_bounds_of_next_potential_var(
                 }
             }
 
-            if (is_decimal_digit(c)) {
+            if (string_utils::is_decimal_digit(c)) {
                 contains_decimal_digit = true;
             } else if (is_alphabet(c)) {
                 contains_alphabet = true;

diff --git a/components/core/src/Utils.cpp b/components/core/src/Utils.cpp
@@ -12,9 +12,9 @@
 #include <boost/lexical_cast.hpp>
 #include <log_surgeon/SchemaParser.hpp>
 #include <spdlog/spdlog.h>
+#include <string_utils/string_utils.hpp>
 
 #include "spdlog_with_specializations.hpp"
-#include "string_utils.hpp"
 
 using std::list;
 using std::string;

diff --git a/components/core/src/ffi/encoding_methods.inc b/components/core/src/ffi/encoding_methods.inc
@@ -3,9 +3,10 @@
 
 #include <algorithm>
 
+#include <string_utils/string_utils.hpp>
+
 #include "../ir/parsing.hpp"
 #include "../ir/types.hpp"
-#include "../string_utils.hpp"
 #include "../type_utils.hpp"
 
 namespace ffi {
@@ -326,7 +327,7 @@ bool encode_integer_string(std::string_view str, encoded_variable_t& encoded_var
     }
 
     encoded_variable_t result;
-    if (false == convert_string_to_int(str, result)) {
+    if (false == string_utils::convert_string_to_int(str, result)) {
         // Conversion failed
         return false;
     } else {
@@ -519,7 +520,7 @@ bool wildcard_query_matches_any_encoded_var(
 
             if constexpr (ir::VariablePlaceholder::Float == var_placeholder) {
                 auto decoded_var = decode_float_var(encoded_vars[encoded_vars_ix]);
-                if (wildcard_match_unsafe(decoded_var, wildcard_query)) {
+                if (string_utils::wildcard_match_unsafe(decoded_var, wildcard_query)) {
                     return true;
                 }
             }
@@ -537,7 +538,7 @@ bool wildcard_query_matches_any_encoded_var(
 
             if constexpr (ir::VariablePlaceholder::Integer == var_placeholder) {
                 auto decoded_var = decode_integer_var(encoded_vars[encoded_vars_ix]);
-                if (wildcard_match_unsafe(decoded_var, wildcard_query)) {
+                if (string_utils::wildcard_match_unsafe(decoded_var, wildcard_query)) {
                     return true;
                 }
             }
@@ -591,7 +592,11 @@ bool wildcard_match_encoded_vars(
 
             if (wildcard_var_placeholders[wildcard_var_ix] == c) {
                 auto decoded_var = decode_float_var(encoded_vars[var_ix]);
-                if (wildcard_match_unsafe(decoded_var, wildcard_var_queries[wildcard_var_ix])) {
+                if (string_utils::wildcard_match_unsafe(
+                            decoded_var,
+                            wildcard_var_queries[wildcard_var_ix]
+                    ))
+                {
                     ++wildcard_var_ix;
                     if (wildcard_var_ix == wildcard_var_queries_len) {
                         break;
@@ -612,7 +617,11 @@ bool wildcard_match_encoded_vars(
 
             if (wildcard_var_placeholders[wildcard_var_ix] == c) {
                 auto decoded_var = decode_integer_var(encoded_vars[var_ix]);
-                if (wildcard_match_unsafe(decoded_var, wildcard_var_queries[wildcard_var_ix])) {
+                if (string_utils::wildcard_match_unsafe(
+                            decoded_var,
+                            wildcard_var_queries[wildcard_var_ix]
+                    ))
+                {
                     ++wildcard_var_ix;
                     if (wildcard_var_ix == wildcard_var_queries_len) {
                         break;

diff --git a/components/core/src/ffi/search/CompositeWildcardToken.cpp b/components/core/src/ffi/search/CompositeWildcardToken.cpp
@@ -1,5 +1,7 @@
 #include "CompositeWildcardToken.hpp"
 
+#include <string_utils/string_utils.hpp>
+
 #include "../../ir/parsing.hpp"
 #include "../../ir/types.hpp"
 
@@ -28,7 +30,7 @@ CompositeWildcardToken<encoded_variable_t>::CompositeWildcardToken(
             is_escaped = false;
         } else if ('\\' == c) {
             is_escaped = true;
-        } else if (is_wildcard(c)) {
+        } else if (string_utils::is_wildcard(c)) {
             m_wildcards.emplace_back(c, i, begin_pos == i || end_pos - 1 == i);
         }
     }

diff --git a/components/core/src/ffi/search/WildcardToken.cpp b/components/core/src/ffi/search/WildcardToken.cpp
@@ -2,8 +2,9 @@
 
 #include <string_view>
 
+#include <string_utils/string_utils.hpp>
+
 #include "../../ir/types.hpp"
-#include "../../string_utils.hpp"
 #include "../../type_utils.hpp"
 #include "../encoding_methods.hpp"
 #include "QueryWildcard.hpp"
@@ -122,9 +123,9 @@ static bool could_be_static_text(string_view query, size_t begin_pos, size_t end
             is_escaped = false;
         } else if ('\\' == c) {
             is_escaped = true;
-        } else if (is_decimal_digit(c)) {
+        } else if (string_utils::is_decimal_digit(c)) {
             return false;
-        } else if (is_alphabet(c)) {
+        } else if (string_utils::is_alphabet(c)) {
             contains_alphabet = true;
         }
     }

diff --git a/components/core/src/ffi/search/query_methods.cpp b/components/core/src/ffi/search/query_methods.cpp
@@ -1,5 +1,7 @@
 #include "query_methods.hpp"
 
+#include <string_utils/string_utils.hpp>
+
 #include "../../ir/parsing.hpp"
 #include "../../ir/types.hpp"
 #include "CompositeWildcardToken.hpp"
@@ -13,6 +15,7 @@ using std::string;
 using std::string_view;
 using std::variant;
 using std::vector;
+using string_utils::is_wildcard;
 
 namespace ffi::search {
 static auto TokenGetBeginPos = [](auto const& token) { return token.get_begin_pos(); };
@@ -251,9 +254,9 @@ static void find_delimiter(
             }
         }
 
-        if (is_decimal_digit(c)) {
+        if (string_utils::is_decimal_digit(c)) {
             contains_decimal_digit = true;
-        } else if (is_alphabet(c)) {
+        } else if (string_utils::is_alphabet(c)) {
             contains_alphabet = true;
         }
     }

diff --git a/components/core/src/ir/LogEventDeserializer.cpp b/components/core/src/ir/LogEventDeserializer.cpp
@@ -3,6 +3,7 @@
 #include <vector>
 
 #include <json/single_include/nlohmann/json.hpp>
+#include <string_utils/string_utils.hpp>
 
 #include "../ffi/ir_stream/decoding_methods.hpp"
 #include "types.hpp"
@@ -56,7 +57,7 @@ auto LogEventDeserializer<encoded_variable_t>::create(ReaderInterface& reader)
         }
         auto ref_timestamp_str = ref_timestamp_iter->get_ref<nlohmann::json::string_t&>();
         epoch_time_ms_t ref_timestamp{};
-        if (false == convert_string_to_int(ref_timestamp_str, ref_timestamp)) {
+        if (false == string_utils::convert_string_to_int(ref_timestamp_str, ref_timestamp)) {
             return std::errc::protocol_error;
         }