From 680c5392e6130f68e78f810a8232b09d49e4b1e6 Mon Sep 17 00:00:00 2001
From: Guoyi Zhang <kuoi@bioarchlinux.org>
Date: Fri, 5 Jul 2024 11:35:53 +1000
Subject: [PATCH] polish: sortdiamond modularity

---
 sortdiamond.cpp | 128 +++++++++++++++++++++++++-----------------------
 1 file changed, 68 insertions(+), 60 deletions(-)

diff --git a/sortdiamond.cpp b/sortdiamond.cpp
index 2ac5077..3141dac 100644
--- a/sortdiamond.cpp
+++ b/sortdiamond.cpp
@@ -64,6 +64,70 @@ string revcomp(const string &seq) {
 	return revseq;
 }
 
+void readInputFile(const string &filename,
+		   map<string, pair<double, string>> &max_map) {
+	ifstream infile(filename);
+	if (!infile) {
+		cerr << "Error opening input file: " << filename << endl;
+		return;
+	}
+
+	string line;
+	while (getline(infile, line)) {
+		istringstream iss(line);
+		string fields[20];
+		int i = 0;
+		while (iss >> fields[i] && i < 20) {
+			i++;
+		}
+		string key = fields[1];
+		double value = stod(fields[11]);
+		// Check if the key already exists in the map
+		if (max_map.find(key) != max_map.end()) {
+			// If the new value is greater, update the map
+			if (value > max_map[key].first) {
+				max_map[key] = make_pair(value, line);
+			}
+		} else {
+			// If the key does not exist, insert the new key-value
+			// pair
+			max_map[key] = make_pair(value, line);
+		}
+	}
+	infile.close();
+}
+
+void processMap(const map<string, pair<double, string>> &max_map,
+		vector<pair<string, string>> &result) {
+	for (const auto &entry : max_map) {
+		istringstream iss(entry.second.second);
+		string fields[20];
+		int i = 0;
+		while (iss >> fields[i] && i < 20) {
+			i++;
+		}
+		if (stoi(fields[6]) > stoi(fields[7])) {
+			fields[17] = revcomp(fields[17]);
+		}
+		result.push_back(make_pair(">" + fields[1], fields[17]));
+	}
+	sort(result.begin(), result.end());
+}
+
+void writeOutputFile(const string &filename,
+		     const vector<pair<string, string>> &result) {
+	ofstream outfile(filename);
+	if (!outfile) {
+		cerr << "Error opening output file: " << filename << endl;
+		return;
+	}
+
+	for (const auto &entry : result) {
+		outfile << entry.first << "\n" << entry.second << "\n";
+	}
+	outfile.close();
+}
+
 int main(int argc, char *argv[]) {
 	if (argc != 3) {
 		cerr << "Usage: " << argv[0] << " <input_file> <output_file>"
@@ -74,69 +138,13 @@ int main(int argc, char *argv[]) {
 	string in_name = argv[1];
 	string ot_name = argv[2];
 
-	ifstream infile(in_name);
-	if (!infile) {
-		cerr << "Error opening input file: " << in_name << endl;
-		return 1;
-	}
-
-	ofstream outfile(ot_name);
-	if (!outfile) {
-		cerr << "Error opening output file: " << ot_name << endl;
-		return 1;
-	}
-
-	map<string, pair<double, string>>
-	    max_map;  // Key: sseqid, Value: pair<score, line>
-	string line;
-
-	while (getline(infile, line)) {
-		istringstream iss(line);
-		string fields[20];  // Adjust size if needed
-		int i = 0;
-
-		while (iss >> fields[i] && i < 20) {
-			i++;
-		}
-
-		string key = fields[1];
-		double value = stod(fields[11]);
-
-		if (max_map.find(key) != max_map.end()) {
-			if (value > max_map[key].first) {
-				max_map[key] = make_pair(value, line);
-			}
-		} else {
-			max_map[key] = make_pair(value, line);
-		}
-	}
+	map<string, pair<double, string>> max_map;
+	readInputFile(in_name, max_map);
 
 	vector<pair<string, string>> result;
+	processMap(max_map, result);
 
-	for (const auto &entry : max_map) {
-		istringstream iss(entry.second.second);
-		string fields[20];  // Adjust size if needed
-		int i = 0;
-
-		while (iss >> fields[i] && i < 20) {
-			i++;
-		}
-
-		if (stoi(fields[6]) > stoi(fields[7])) {
-			fields[17] = revcomp(fields[17]);
-		}
-
-		result.push_back(make_pair(">" + fields[1], fields[17]));
-	}
-
-	sort(result.begin(), result.end());
-
-	for (const auto &entry : result) {
-		outfile << entry.first << "\n" << entry.second << endl;
-	}
-
-	infile.close();
-	outfile.close();
+	writeOutputFile(ot_name, result);
 
 	return 0;
 }