Merge pull request #48 from KoslickiLab/fix-archive-path

Fix archive path (resolves #41)

Author: mahmudhera

.github/workflows/ci.yml

@@ -69,7 +69,7 @@ jobs:
     # 8. Test correctness of tar.gz index
     - name: Test correctness of tar.gz index
       run: |
-        ./bin/index test/data/filelist_100.txt test/output/archived_index_100 -t 4 -n 1000 -s
+        ./bin/index test/data/filelist_100.txt test/output/archived_index_100 -t 4 -n 1000 -s -a test/output/archived_index_100.tar.gz
         rm -r test/output/archived_index_100
         bin/compare test/data/filelist_100.txt test/output/archived_index_100.tar.gz test/output/working_directory test/output/compare_100_v_100_tar -c 0.0 -t 2 -n 500 -k 51
         python test/compare_multisearch_results.py test/data/multisearch_100_v_100 test/output/compare_100_v_100_tar

src/MultiSketchIndex.cpp

@@ -123,6 +123,7 @@ bool MultiSketchIndex::write_to_file(std::string directory_name,
                                     int num_threads, 
                                     std::vector<SketchInfo> info_of_sketches,
                                     bool store_archive,
+                                    std::string archive_name,
                                     bool force_write) {
     // check if the directory exists, if not then create it
     struct stat info;
@@ -193,20 +194,22 @@ bool MultiSketchIndex::write_to_file(std::string directory_name,
     }
 
     output_file.close();
+    std::cout << "Index written to " << directory_name << std::endl;
 
-    if (store_archive) {
-        std::cout << "Storing archive..." << std::endl;
-        std::string archive_name = directory_name + ".tar.gz";
-        std::string command = "tar -czf " + archive_name + " -C " + directory_name + " .";
-        std::cout << command << std::endl;
-        if (system(command.c_str()) != 0) {
-            std::cout << "Error storing archive." << std::endl;
-            return false;
-        }
-        std::cout << "Archive stored to " << archive_name << std::endl;
+    if (!store_archive) {
+        return true;
     }
 
+    std::cout << "Storing archive..." << std::endl;
+    std::string command = "tar -czf " + archive_name + " -C " + directory_name + " .";
+    std::cout << command << std::endl;
+    if (system(command.c_str()) != 0) {
+        std::cout << "Error storing archive." << std::endl;
+        return false;
+    }
+    std::cout << "Archive stored to " << archive_name << std::endl;
     return true;
+    
 }
 
 

src/MultiSketchIndex.h

@@ -136,6 +136,7 @@ class MultiSketchIndex {
                                     int num_threads, 
                                     std::vector<SketchInfo> info_of_sketches,
                                     bool store_archive,
+                                    std::string archive_name,
                                     bool force_write);
 
         /**

src/index.cpp

@@ -17,6 +17,7 @@ struct Arguments {
     int number_of_threads;
     int num_hashtables;
     bool store_archive;
+    string archive_name;
     bool force_write;
 };
 
@@ -35,7 +36,7 @@ void parse_args(int argc, char** argv, Arguments &arguments) {
         .store_into(arguments.filelist_sketches);
 
     parser.add_argument("index_directory_name")
-        .help("The directory where the index will be stored (needs to be empty)")
+        .help("The directory where the index will be stored (must be a directory)")
         .required()
         .store_into(arguments.index_directory_name);
 
@@ -57,6 +58,11 @@ void parse_args(int argc, char** argv, Arguments &arguments) {
         .implicit_value(true)
         .store_into(arguments.store_archive);
 
+    parser.add_argument("-a", "--archive-name")
+        .help("The name of the archive (will be discarded if store-archive is false)")
+        .default_value("index.tar.gz")
+        .store_into(arguments.archive_name);
+
     parser.add_argument("-f", "--force-write")
         .help("Force write the index to the directory")
         .default_value(false)
@@ -83,6 +89,7 @@ void show_arguments(Arguments &arguments) {
     cout << "*  number_of_threads: " << arguments.number_of_threads << endl;
     cout << "*  num_hashtables: " << arguments.num_hashtables << endl;
     cout << "*  store_archive: " << arguments.store_archive << endl;
+    cout << "*  archive_name: " << (arguments.store_archive ? arguments.archive_name : "N/A") << endl;
     cout << "*  force_write: " << arguments.force_write << endl;
     cout << "* " << endl;
     cout << "*********************************" << endl;
@@ -115,8 +122,8 @@ int main(int argc, char** argv) {
     MultiSketchIndex multi_sketch_index(arguments.num_hashtables);
     compute_index_from_sketches(sketches, multi_sketch_index, arguments.number_of_threads);
     cout << "Index built." << endl;
-    cout << "Some index stats:" << endl;
-    multi_sketch_index.show_index_stats();
+    //cout << "Some index stats:" << endl;
+    //multi_sketch_index.show_index_stats();
 
 
     cout << "Writing index to file..." << endl;
@@ -129,6 +136,7 @@ int main(int argc, char** argv) {
                                             arguments.number_of_threads, 
                                             info_of_sketches,
                                             arguments.store_archive,
+                                            arguments.archive_name,
                                             arguments.force_write);
     if (!success) {
         cout << "Error writing index to file." << endl;