Duplicate hostsfile utils for daemon

include/daemon/util.hpp

@@ -14,6 +14,9 @@
 #ifndef GEKKOFS_DAEMON_UTIL_HPP
 #define GEKKOFS_DAEMON_UTIL_HPP
 
+#include <vector>
+#include <string>
+
 namespace gkfs {
 namespace util {
 void
@@ -21,6 +24,9 @@ populate_hosts_file();
 
 void
 destroy_hosts_file();
+
+std::vector<std::pair<std::string, std::string>>
+read_hosts_file();
 } // namespace util
 } // namespace gkfs
 

src/daemon/util.cpp

@@ -18,6 +18,8 @@
 #include <fstream>
 #include <iostream>
 
+#include <regex>
+
 using namespace std;
 
 namespace gkfs::util {
@@ -48,4 +50,68 @@ destroy_hosts_file() {
     std::remove(GKFS_DATA->hosts_file().c_str());
 }
 
+
+// hosts file functions copied from daemon - need Modifications for LOG and hostsfile path source
+// TODO clean that up, probably best into global utilities - TBD How do we do logging there?
+/**
+ * Reads the daemon generator hosts file by a given path, returning hosts and URI addresses
+ * @param path to hosts file
+ * @return vector<pair<hosts, URI>>
+ * @throws std::runtime_error
+ */
+vector<pair<string, string>> load_hostfile(const std::string& path) {
+
+    //LOG(DEBUG, "Loading hosts file: \"{}\"", path);
+
+    ifstream lf(path);
+    if (!lf) {
+        throw runtime_error(fmt::format("Failed to open hosts file '{}': {}",
+                                        path, strerror(errno)));
+    }
+    vector<pair<string, string>> hosts;
+    const regex line_re("^(\\S+)\\s+(\\S+)$",
+                        regex::ECMAScript | regex::optimize);
+    string line;
+    string host;
+    string uri;
+    std::smatch match;
+    while (getline(lf, line)) {
+        if (!regex_match(line, match, line_re)) {
+
+            //LOG(ERROR, "Unrecognized line format: [path: '{}', line: '{}']",
+            //    path, line);
+
+            throw runtime_error(
+                    fmt::format("unrecognized line format: '{}'", line));
+        }
+        host = match[1];
+        uri = match[2];
+        hosts.emplace_back(host, uri);
+    }
+    if (hosts.empty()) {
+        throw runtime_error("Hosts file found but no suitable addresses could be extracted");
+    }
+    // ??? extract_protocol(hosts[0].second);
+    return hosts;
+}
+
+vector<pair<string, string>> read_hosts_file() {
+    string hostfile = GKFS_DATA->hosts_file();
+
+    vector<pair<string, string>> hosts;
+    try {
+        hosts = load_hostfile(hostfile);
+    } catch (const exception& e) {
+        auto emsg = fmt::format("Failed to load hosts file: {}", e.what());
+        throw runtime_error(emsg);
+    }
+
+    if (hosts.empty()) {
+        throw runtime_error(fmt::format("Hostfile empty: '{}'", hostfile));
+    }
+
+    GKFS_DATA->spdlogger()->info("Hosts pool size: {}", hosts.size());
+
+    return hosts;
+}
 } // namespace gkfs::util