Merge branch 'main' into sqlbox

Author: davemarco

components/clp-py-utils/clp_py_utils/compression.py

@@ -3,28 +3,12 @@
 
 import Levenshtein
 
+from clp_py_utils.core import FileMetadata
+
 # Constants
 FILE_GROUPING_MIN_LEVENSHTEIN_RATIO = 0.6
 
 
-class FileMetadata:
-    __slots__ = ("path", "size", "estimated_uncompressed_size")
-
-    def __init__(self, path: pathlib.Path, size: int):
-        self.path = path
-        self.size = size
-        self.estimated_uncompressed_size = size
-
-        filename = path.name
-        if any(filename.endswith(extension) for extension in [".gz", ".gzip", ".tgz", ".tar.gz"]):
-            self.estimated_uncompressed_size *= 13
-        elif any(
-            filename.endswith(extension)
-            for extension in [".zstd", ".zstandard", ".tar.zstd", ".tar.zstandard"]
-        ):
-            self.estimated_uncompressed_size *= 8
-
-
 class FilesPartition:
     def __init__(self):
         self.__files = []

components/clp-py-utils/clp_py_utils/core.py

@@ -4,6 +4,24 @@
 from yaml.parser import ParserError
 
 
+class FileMetadata:
+    __slots__ = ("path", "size", "estimated_uncompressed_size")
+
+    def __init__(self, path: pathlib.Path, size: int):
+        self.path = path
+        self.size = size
+        self.estimated_uncompressed_size = size
+
+        filename = path.name
+        if any(filename.endswith(extension) for extension in [".gz", ".gzip", ".tgz", ".tar.gz"]):
+            self.estimated_uncompressed_size *= 13
+        elif any(
+            filename.endswith(extension)
+            for extension in [".zstd", ".zstandard", ".tar.zstd", ".tar.zstandard"]
+        ):
+            self.estimated_uncompressed_size *= 8
+
+
 def get_config_value(config, key):
     """
     Gets a value from the given dictionary using a dot-separated configuration

components/clp-py-utils/clp_py_utils/s3_utils.py

@@ -22,7 +22,7 @@
     StorageType,
     WEBUI_COMPONENT_NAME,
 )
-from clp_py_utils.compression import FileMetadata
+from clp_py_utils.core import FileMetadata
 
 # Constants
 AWS_ENDPOINT = "amazonaws.com"

components/core/cmake/Options/options.cmake

@@ -198,6 +198,7 @@ endfunction()
 
 function(validate_clp_s_archivereader_dependencies)
     validate_clp_dependencies_for_target(CLP_BUILD_CLP_S_ARCHIVEREADER
+        CLP_BUILD_CLP_STRING_UTILS
         CLP_BUILD_CLP_S_CLP_DEPENDENCIES
         CLP_BUILD_CLP_S_IO
         CLP_BUILD_CLP_S_TIMESTAMPPATTERN
@@ -300,6 +301,7 @@ endfunction()
 
 function(validate_clp_s_search_dependencies)
     validate_clp_dependencies_for_target(CLP_BUILD_CLP_S_SEARCH
+        CLP_BUILD_CLP_STRING_UTILS
         CLP_BUILD_CLP_S_ARCHIVEREADER
         CLP_BUILD_CLP_S_CLP_DEPENDENCIES
         CLP_BUILD_CLP_S_SEARCH_AST

components/core/src/clp/SQLitePreparedStatement.hpp

@@ -1,6 +1,7 @@
 #ifndef CLP_SQLITEPREPAREDSTATEMENT_HPP
 #define CLP_SQLITEPREPAREDSTATEMENT_HPP
 
+#include <cstdint>
 #include <string>
 
 #include <sqlite3/sqlite3.h>

components/core/src/clp_s/CMakeLists.txt

@@ -302,6 +302,7 @@ if(CLP_BUILD_CLP_S_ARCHIVEREADER)
                 clp_s_archive_reader
                 PUBLIC
                 absl::flat_hash_map
+                clp::string_utils
                 clp_s::io
                 msgpack-cxx
                 nlohmann_json::nlohmann_json

components/core/src/clp_s/DictionaryReader.hpp

@@ -6,6 +6,7 @@
 #include <unordered_set>
 
 #include <boost/algorithm/string/case_conv.hpp>
+#include <string_utils/string_utils.hpp>
 
 #include "ArchiveReaderAdaptor.hpp"
 #include "DictionaryEntry.hpp"
@@ -191,7 +192,12 @@ void DictionaryReader<DictionaryIdType, EntryType>::get_entries_matching_wildcar
         std::unordered_set<EntryType const*>& entries
 ) const {
     for (auto const& entry : m_entries) {
-        if (StringUtils::wildcard_match_unsafe(entry.get_value(), wildcard_string, !ignore_case)) {
+        if (clp::string_utils::wildcard_match_unsafe(
+                    entry.get_value(),
+                    wildcard_string,
+                    !ignore_case
+            ))
+        {
             entries.insert(&entry);
         }
     }