Merge pull request #374 from AsherGlick/abstract_file_iteration

Abstract File Iteration

xml_converter/src/file_helper.cpp

@@ -1,7 +1,19 @@
 #include "file_helper.hpp"
 
+#include <dirent.h>
+
+#include <algorithm>
+#include <filesystem>
 #include <fstream>
+#include <iostream>
+#include <iterator>
+#include <memory>
+#include <sstream>
 #include <string>
+#include <utility>
+#include <vector>
+
+#include "string_helper.hpp"
 
 using namespace std;
 
@@ -12,3 +24,85 @@ void copy_file(string path, string new_path) {
     ofstream outfile(new_path, ios::binary);
     outfile << infile.rdbuf();
 }
+
+////////////////////////////////////////////////////////////////////////////////
+// MarkerPackFile::MarkerPackFile
+//
+// Constructor for the MarkerPackFile class
+////////////////////////////////////////////////////////////////////////////////
+MarkerPackFile::MarkerPackFile(std::string base, std::string relative_filepath) {
+    this->base = base;
+    this->relative_filepath = relative_filepath;
+}
+
+// Temporary compatibility function mimicing the old string filepath
+// functionality until it is removed.
+const string MarkerPackFile::tmp_get_path() const {
+    return join_file_paths(this->base, this->relative_filepath);
+}
+
+////////////////////////////////////////////////////////////////////////////////
+// marker_pack_file_comp
+//
+// Helper function to compare the full file paths of marker pack files so that
+// they can be sorted deterministically.
+////////////////////////////////////////////////////////////////////////////////
+bool marker_pack_file_comp(const MarkerPackFile& a, const MarkerPackFile& b) {
+    if (a.base == b.base) {
+        return lowercase(a.relative_filepath) < lowercase(b.relative_filepath);
+    }
+    return lowercase(a.base) < lowercase(b.base);
+}
+
+////////////////////////////////////////////////////////////////////////////////
+// get_files_by_suffix
+//
+// Gets all the marker pack files with a given `suffix` inside of a marker pack
+// filepath `base`
+////////////////////////////////////////////////////////////////////////////////
+vector<MarkerPackFile> get_files_by_suffix(
+    const string& base,
+    const string& suffix,
+    const string& subpath) {
+    vector<MarkerPackFile> files;
+
+    if (!filesystem::exists(base)) {
+        cout << "Error: " << base << " does not exist" << endl;
+        return vector<MarkerPackFile>();
+    }
+    // If it is a directory, call open_directory_file_to_read
+    else if (filesystem::is_directory(base)) {
+        DIR* dir = opendir(join_file_paths(base, subpath).c_str());
+        struct dirent* entry;
+        while ((entry = readdir(dir)) != NULL) {
+            string filename = entry->d_name;
+            if (filename == ".") {
+                continue;
+            }
+            else if (filename == "..") {
+                continue;
+            }
+
+            if (entry->d_type == DT_DIR) {
+                string new_subpath = join_file_paths(subpath, filename);
+                vector<MarkerPackFile> subfiles = get_files_by_suffix(base, suffix, new_subpath);
+                files.reserve(files.size() + subfiles.size());
+                move(begin(subfiles), end(subfiles), back_inserter(files));
+            }
+            else if (has_suffix(filename, suffix)) {
+                MarkerPackFile new_file(base, join_file_paths(subpath, filename));
+                files.push_back(new_file);
+            }
+        }
+        closedir(dir);
+    }
+    // else if (filesystem::is_regular_file(file.base)) {
+    // }
+
+    sort(files.begin(), files.end(), marker_pack_file_comp);
+    return files;
+}
+
+vector<MarkerPackFile> get_files_by_suffix(const string& base, const string& suffix) {
+    return get_files_by_suffix(base, suffix, "");
+}

xml_converter/src/file_helper.hpp

@@ -1,4 +1,17 @@
 #pragma once
 #include <string>
+#include <vector>
+
+class MarkerPackFile {
+ public:
+    MarkerPackFile(std::string base, std::string relative_filepath);
+
+    const std::string tmp_get_path() const;
+
+    std::string base;
+    std::string relative_filepath;
+};
 
 void copy_file(std::string path, std::string new_path);
+
+std::vector<MarkerPackFile> get_files_by_suffix(const std::string& base, const std::string& suffix);

xml_converter/src/string_helper.cpp

@@ -292,6 +292,13 @@ bool has_suffix(std::string const& fullString, std::string const& ending) {
 }
 
 string join_file_paths(const string& path_a, const string& path_b) {
+    if (path_a.empty()) {
+        return path_b;
+    }
+    if (path_b.empty()) {
+        return path_a;
+    }
+
     string output = path_a;
     if (!has_suffix(path_a, "/")) {
         output += "/";

xml_converter/src/xml_converter.cpp

@@ -32,37 +32,6 @@
 
 using namespace std;
 
-bool filename_comp(string a, string b) {
-    return lowercase(a) < lowercase(b);
-}
-
-// Searchs for files within a directory with a suffix and returns their relative paths.
-vector<string> get_files_by_suffix(string directory, string suffix) {
-    vector<string> files;
-    DIR* dir = opendir(directory.c_str());
-    struct dirent* entry;
-    while ((entry = readdir(dir)) != NULL) {
-        string filename = entry->d_name;
-        if (filename != "." && filename != "..") {
-            string path = join_file_paths(directory, filename);
-            if (entry->d_type == DT_DIR) {
-                vector<string> subfiles = get_files_by_suffix(path, suffix);
-                // Default: markerpacks have all xml files in the first directory
-                for (string subfile : subfiles) {
-                    cout << subfile << " found in subfolder" << endl;
-                    files.push_back(join_file_paths(filename, subfile));
-                }
-            }
-            else if (has_suffix(filename, suffix)) {
-                files.push_back(filename);
-            }
-        }
-    }
-    closedir(dir);
-    std::sort(files.begin(), files.end(), filename_comp);
-    return files;
-}
-
 map<string, vector<string>> read_taco_directory(
     string input_path,
     map<string, Category>* marker_categories,
@@ -73,10 +42,10 @@ map<string, vector<string>> read_taco_directory(
     }
     else if (filesystem::is_directory(input_path)) {
         string directory_name = filesystem::path(input_path).filename();
-        vector<string> xml_files = get_files_by_suffix(input_path, ".xml");
-        for (const string& path : xml_files) {
-            set<string> top_level_category_names = parse_xml_file(join_file_paths(input_path, path), input_path, marker_categories, parsed_pois);
-            string relative_path = join_file_paths(directory_name, path);
+        vector<MarkerPackFile> xml_files = get_files_by_suffix(input_path, ".xml");
+        for (const MarkerPackFile& path : xml_files) {
+            set<string> top_level_category_names = parse_xml_file(path.tmp_get_path(), input_path, marker_categories, parsed_pois);
+            string relative_path = join_file_paths(directory_name, path.relative_filepath);
             for (set<string>::iterator it = top_level_category_names.begin(); it != top_level_category_names.end(); it++) {
                 top_level_category_file_locations[*it].push_back(relative_path);
             }
@@ -95,10 +64,10 @@ map<string, vector<string>> read_burrito_directory(
     }
     else if (filesystem::is_directory(input_path)) {
         string directory_name = filesystem::path(input_path).filename();
-        vector<string> burrito_files = get_files_by_suffix(input_path, ".guildpoint");
-        for (const string& path : burrito_files) {
-            set<string> top_level_category_names = read_protobuf_file(join_file_paths(input_path, path), input_path, marker_categories, parsed_pois);
-            string relative_path = join_file_paths(directory_name, path);
+        vector<MarkerPackFile> burrito_files = get_files_by_suffix(input_path, ".guildpoint");
+        for (const MarkerPackFile& path : burrito_files) {
+            set<string> top_level_category_names = read_protobuf_file(path.tmp_get_path(), input_path, marker_categories, parsed_pois);
+            string relative_path = join_file_paths(directory_name, path.relative_filepath);
             for (set<string>::iterator it = top_level_category_names.begin(); it != top_level_category_names.end(); it++) {
                 top_level_category_file_locations[*it].push_back(relative_path);
             }