Add csv_paths option

Co-authored-by: Alexandr Smirnov <[email protected]>
Desbordante · Nov 24, 2023 · 615ab74 · 615ab74
1 parent 7b5f86b
commit 615ab74
Show file tree

Hide file tree

Showing 7 changed files with 49 additions and 2 deletions.
diff --git a/src/core/algorithms/algo_factory.h b/src/core/algorithms/algo_factory.h
@@ -11,6 +11,7 @@
 #include "algorithms/create_algorithm.h"
 #include "algorithms/pipelines/typo_miner/typo_miner.h"
 #include "config/names.h"
+#include "tabular_data/input_tables_type.h"
 
 namespace algos {
 
@@ -71,12 +72,28 @@ template <typename OptionMap>
 void LoadAlgorithm(Algorithm& algorithm, OptionMap&& options) {
     ConfigureFromFunction(algorithm, [&options](std::string_view option_name) {
         using namespace config::names;
+        namespace fs = std::filesystem;
         if (option_name == kTable && options.find(std::string{kTable}) == options.end()) {
             config::InputTable parser = std::make_shared<CSVParser>(
-                    details::ExtractOptionValue<std::filesystem::path>(options, kCsvPath),
+                    details::ExtractOptionValue<fs::path>(options, kCsvPath),
                     details::ExtractOptionValue<char>(options, kSeparator),
                     details::ExtractOptionValue<bool>(options, kHasHeader));
             return boost::any{parser};
+        } else if (option_name == kTables && options.find(std::string{kTables}) == options.end()) {
+            auto paths = details::ExtractOptionValue<std::vector<fs::path>>(options, kCsvPaths);
+            auto separator = details::ExtractOptionValue<char>(options, kSeparator);
+            auto has_header = details::ExtractOptionValue<bool>(options, kHasHeader);
+
+            if (paths.empty()) {
+                throw config::ConfigurationError("Expected collection of csv paths");
+            }
+
+            config::InputTables tables;
+            tables.reserve(paths.size());
+            for (auto const& path : paths) {
+                tables.push_back(std::make_shared<CSVParser>(path, separator, has_header));
+            }
+            return boost::any{tables};
         }
         return details::GetOrEmpty(options, option_name);
     });

diff --git a/src/core/config/all_options.cpp b/src/core/config/all_options.cpp
@@ -41,7 +41,8 @@ boost::program_options::options_description GeneralOptions() {
     // clang-format off
     po::options_description general_options("General options");
     general_options.add_options()
-            (kCsvPath, po::value<std::filesystem::path>()->required(), kDCsvPath)
+            (kCsvPath, po::value<std::filesystem::path>(), kDCsvPath)
+            (kCsvPaths, po::value<std::vector<std::filesystem::path>>(), kDCsvPaths)
             (kSeparatorOpt.c_str(), po::value<char>()->default_value(','), kDSeparator)
             (kHasHeader, po::value<bool>()->default_value(true), kDHasHeader)
             (kEqualNulls, po::value<bool>(), kDEqualNulls)

diff --git a/src/core/config/descriptions.h b/src/core/config/descriptions.h
@@ -9,7 +9,9 @@
 
 namespace config::descriptions {
 constexpr auto kDTable = "table processed by the algorithm";
+constexpr auto kDTables = "table collection processed by the algorithm";
 constexpr auto kDCsvPath = "path to the CSV table";
+constexpr auto kDCsvPaths = "path to the collection of CSV tables";
 constexpr auto kDSeparator = "CSV separator";
 constexpr auto kDHasHeader = "CSV header presence flag [true|false]";
 constexpr auto kDEqualNulls = "specify whether two NULLs should be considered equal";

diff --git a/src/core/config/names.h b/src/core/config/names.h
@@ -2,7 +2,9 @@
 
 namespace config::names {
 constexpr auto kTable = "table";
+constexpr auto kTables = "tables";
 constexpr auto kCsvPath = "csv_path";
+constexpr auto kCsvPaths = "csv_paths";
 constexpr auto kSeparator = "separator";
 constexpr auto kHasHeader = "has_header";
 constexpr auto kEqualNulls = "is_null_equal_null";

diff --git a/src/core/config/tabular_data/input_tables/option.cpp b/src/core/config/tabular_data/input_tables/option.cpp
@@ -0,0 +1,8 @@
+#include "config/tabular_data/input_tables/option.h"
+
+#include "config/names_and_descriptions.h"
+
+namespace config {
+using names::kTables, descriptions::kDTables;
+extern const CommonOption<InputTables> TablesOpt{kTables, kDTables};
+}  // namespace config
diff --git a/src/core/config/tabular_data/input_tables/option.h b/src/core/config/tabular_data/input_tables/option.h
@@ -0,0 +1,8 @@
+#pragma once
+
+#include "config/common_option.h"
+#include "config/tabular_data/input_tables_type.h"
+
+namespace config {
+extern const CommonOption<InputTables> TablesOpt;
+}  // namespace config
diff --git a/src/core/config/tabular_data/input_tables_type.h b/src/core/config/tabular_data/input_tables_type.h
@@ -0,0 +1,9 @@
+#pragma once
+
+#include <vector>
+
+#include "input_table_type.h"
+
+namespace config {
+using InputTables = std::vector<InputTable>;
+}  // namespace config