restructured File Detector and Parsing

cli/CMakeLists.txt

@@ -22,14 +22,20 @@ CPMAddPackage(
             "CXXOPTS_BUILD_EXAMPLES OFF"
             "CXXOPTS_BUILD_TESTS OFF"
 )
+CPMAddPackage(
+        NAME zlib
+        GITHUB_REPOSITORY madler/zlib
+        GIT_TAG v1.3.1
+)
 
 file(GLOB GG_LIB_SOURCES "../src/*.cpp")
 file(GLOB GG_CLI_SOURCES "src/*.cpp")
 
 add_executable(genogrove_cli ${GG_LIB_SOURCES} ${GG_CLI_SOURCES})
-target_include_directories(genogrove_cli PUBLIC "${cxxopts_SOURCE_DIR}/include/")
-target_include_directories(genogrove_cli PUBLIC "${CMAKE_CURRENT_SOURCE_DIR}/../include/")
-target_link_libraries(genogrove_cli cxxopts)
+target_include_directories(genogrove_cli PUBLIC "${cxxopts_SOURCE_DIR}/include/") # cxxopts header files
+target_include_directories(genogrove_cli PUBLIC "${CMAKE_CURRENT_SOURCE_DIR}/../include/") # genogrove header files
+target_include_directories(genogrove_cli PUBLIC include) # cli header files
+target_link_libraries(genogrove_cli cxxopts zlibstatic)
 
 # set the output directory for the executable
 set_target_properties(genogrove_cli PROPERTIES

cli/include/BEDReader.hpp

@@ -1,8 +1,28 @@
-//
-// Created by Richard Albin Schaefer on 11/17/24.
-//
-
 #ifndef GENOGROVE_BEDREADER_HPP
 #define GENOGROVE_BEDREADER_HPP
 
+// Standard
+#include <string>
+#include <filesystem>
+#include <istream>
+#include <algorithm>
+
+// Class
+#include "FileReader.hpp"
+#include "genogrove/Interval.hpp"
+#include "FileEntry.hpp"
+
+class BEDReader : public FileReader {
+    public:
+        BEDReader(const std::filesystem::path&, bool gzipped);
+        bool readNext(FileEntry& entry) override;
+        bool hasNext() override;
+        std::string getErrorMessage() override;
+
+    private:
+        std::unique_ptr<std::istream> inputStream;
+        std::string errorMessage;
+        size_t lineNum;
+};
+
 #endif //GENOGROVE_BEDREADER_HPP

cli/include/FileReader.hpp

@@ -1,8 +1,16 @@
-//
-// Created by Richard Albin Schaefer on 11/17/24.
-//
-
 #ifndef GENOGROVE_FILEREADER_HPP
 #define GENOGROVE_FILEREADER_HPP
 
+// Class
+#include "genogrove/Interval.hpp"
+#include "FileEntry.hpp"
+
+class FileReader {
+    public:
+        virtual bool readNext(FileEntry& entry) = 0;
+        virtual bool hasNext() = 0;
+        virtual std::string getErrorMessage() = 0;
+        virtual ~FileReader() = default;
+};
+
 #endif //GENOGROVE_FILEREADER_HPP

cli/include/FileReaderFactory.hpp

@@ -1,8 +1,28 @@
-//
-// Created by Richard Albin Schaefer on 11/17/24.
-//
-
 #ifndef GENOGROVE_FILEREADERFACTORY_HPP
 #define GENOGROVE_FILEREADERFACTORY_HPP
 
+// Standard
+#include <filesystem>
+
+// Class
+#include "FileReader.hpp"
+#include "BEDReader.hpp"
+#include "FileTypeDetector.hpp"
+
+class FileReaderFactory {
+    public:
+        static std::unique_ptr<FileReader> create(
+                const std::filesystem::path& filepath,
+                FileType filetype,
+                bool gzipped
+                ) {
+            switch(filetype) {
+                case FileType::BED:
+                    return std::make_unique<BEDReader>(filepath, gzipped);
+                default:
+                    return nullptr;
+            }
+        }
+};
+
 #endif //GENOGROVE_FILEREADERFACTORY_HPP

cli/include/FileTypeDetector.hpp

@@ -9,21 +9,19 @@
 #include <unordered_map>
 #include <filesystem>
 
-namespace genogrove {
-    enum class FileType {
-        BED,
-        BEDGRAPH,
-        GFF,
-        GTF,
-        VCF,
-        UNKNOWN
-    };
+enum class FileType {
+    BED,
+    BEDGRAPH,
+    GFF,
+    GTF,
+    VCF,
+    UNKNOWN
+};
 
-    class FileTypeDetector {
-        public:
-            std::tuple<FileType, bool> detectFileType(const std::filesystem::path& filepath);
-    };
-}
+class FileTypeDetector {
+    public:
+        std::tuple<FileType, bool> detectFileType(const std::filesystem::path& filepath);
+};
 
 
 #endif //GENOGROVE_FILETYPEDETECTOR_HPP

cli/src/BEDReader.cpp

@@ -1,3 +1,59 @@
-//
-// Created by Richard Albin Schaefer on 11/17/24.
-//
+#include "BEDReader.hpp"
+
+
+BEDReader::BEDReader(const std::filesystem::path& filepath, bool gzipped) {
+    lineNum = 1;
+    if(gzipped) {
+        // open file gzipped
+    } else {
+//        inputStream = std::make_unique<std::ifstream>(filepath);
+    }
+}
+
+bool BEDReader::readNext(FileEntry& entry) {
+    std::string line;
+    if(!std::getline(*inputStream, line)) {
+        errorMessage = "Failed to read line at line " + std::to_string(lineNum);
+    }
+
+    std::stringstream ss(line);
+    std::string chrom, start, end;
+
+    try {
+        if(!(ss >> chrom >> start >> end)) {
+            errorMessage = "Invalid line format at line " + std::to_string(lineNum);
+            return false;
+        }
+
+        // validate integers
+        if(!std::all_of(start.begin(), start.end(), ::isdigit) ||
+            !std::all_of(end.begin(), end.end(), ::isdigit)) {
+            errorMessage = "Invalid coordinate format at line " + std::to_string(lineNum);
+            return false;
+        }
+
+        // validate and create interval object
+        size_t startNum = std::stoul(start);
+        size_t endNum = std::stoul(end);
+        if(startNum >= endNum) {
+            errorMessage = "Start coordinate must be less than end coordinate at line " + std::to_string(lineNum);
+            return false;
+        }
+        FileEntry entry(chrom, genogrove::Interval{startNum, endNum}, '\0');
+        lineNum++;
+        return true;
+
+    } catch(std::exception& e) {
+        errorMessage = "Failed to parse line at line " + std::to_string(lineNum) + ": " + line;
+        return false;
+    }
+}
+
+bool BEDReader::hasNext() {
+    return inputStream && !inputStream->eof();
+}
+
+std::string BEDReader::getErrorMessage() {
+    return errorMessage;
+}
+