Manager: Change exclude path logic to /.gitignore*/ style

src/dotfile.cpp

@@ -4,6 +4,7 @@
  * SPDX-License-Identifier: MIT
  */
 
+#include <cassert> // assert
 #include <cctype>  // tolower
 #include <cstddef> // size_t
 #include <cstdio>  // fprintf, printf, stderr
@@ -362,30 +363,133 @@ void Dotfile::forEachDotfile(const std::vector<std::string>& targets, const std:
 		if (path.is_directory() || filter(path)) {
 			continue;
 		}
-		if (!targets.empty() && !include(path.path().string(), targets)) {
+		if (!targets.empty() && !include(path.path(), targets)) {
 			continue;
 		}
 		callback(path, index++);
 	}
 }
 
-bool Dotfile::filter(const std::filesystem::path& path)
+bool Dotfile::filter(const std::filesystem::directory_entry& path)
 {
-	for (auto& excludePath : Config::the().excludePaths()) {
+	std::string pathString = path.path().string();
-		if (excludePath.second == "file") {
+	assert(pathString.front() == '/');
-			if (path.string() == Config::the().workingDirectory() / excludePath.first) {
+
+	// Cut off working directory
+	size_t cutFrom = pathString.find(Config::the().workingDirectory()) == 0 ? Config::the().workingDirectorySize() : 0;
+	pathString = pathString.substr(cutFrom);
+
+	for (const auto& excludePathMapEntry : Config::the().excludePaths()) {
+		const auto& excludePath = excludePathMapEntry.first;
+
+		if (pathString == excludePath) {
 			return true;
 		}
+
+		// If starts with '/', only match in the working directory root
+		bool onlyMatchInRoot = false;
+		if (excludePath.front() == '/') {
+			onlyMatchInRoot = true;
 		}
-		else if (excludePath.second == "directory") {
+
-			if (path.string().find(Config::the().workingDirectory() / excludePath.first) == 0) {
+		// If ends with '/', only match directories
-				return true;
+		bool onlyMatchDirectories = false;
+		if (excludePath.back() == '/') {
+			onlyMatchDirectories = true;
+		}
+
+		// Parsing
+
+		bool tryPatternState = true;
+
+		size_t pathIterator = 0;
+		size_t excludeIterator = 0;
+
+		if (!onlyMatchInRoot) {
+			pathIterator++;
 		}
+
+		// Current path charter 'x' == next ignore pattern characters '*x'
+		// Example, iterator at []: [.]log/output.txt
+		//                          [*].log
+		if (pathIterator < pathString.length()
+		    && excludeIterator < excludePath.length() - 1
+		    && excludePath.at(excludeIterator) == '*'
+		    && pathString.at(pathIterator) == excludePath.at(excludeIterator + 1)) {
+			excludeIterator++;
+		}
+
+		for (; pathIterator < pathString.length() && excludeIterator < excludePath.length();) {
+			char character = pathString.at(pathIterator);
+			pathIterator++;
+
+			if (!tryPatternState && character == '/') {
+				tryPatternState = true;
+				continue;
+			}
+
+			if (!tryPatternState) {
+				continue;
+			}
+
+			if (character == excludePath.at(excludeIterator)) {
+				// Fail if the final match hasn't reached the end of the ignore pattern
+				// Example, iterator at []: doc/buil[d]
+				//                          buil[d]/
+				if (pathIterator == pathString.length() && excludeIterator < excludePath.length() - 1) {
+					break;
 				}
-		else if (excludePath.second == "endsWith") {
+
-			if (path.string().find(excludePath.first) == path.string().size() - excludePath.first.size()) {
+				// Next path character 'x' == next ignore pattern characters '*x', skip the '*'
+				// Example, iterator at []: /includ[e]/header.h
+				//                          /includ[e]*/
+				if (pathIterator < pathString.length()
+				    && excludeIterator < excludePath.length() - 2
+				    && excludePath.at(excludeIterator + 1) == '*'
+				    && pathString.at(pathIterator) == excludePath.at(excludeIterator + 2)) {
+					excludeIterator++;
+				}
+
+				excludeIterator++;
+				continue;
+			}
+
+			if (excludePath.at(excludeIterator) == '*') {
+				// Fail if we're entering a subdirectory and we should only match in the root
+				// Example, iterator at []: /src[/]include/header.h
+				//                          /[*]include/
+				if (onlyMatchInRoot && character == '/') {
+					break;
+				}
+
+				// Next path character == next ignore pattern character
+				if (pathIterator < pathString.length()
+				    && excludeIterator + 1 < excludePath.length()
+				    && pathString.at(pathIterator) == excludePath.at(excludeIterator + 1)) {
+					excludeIterator++;
+				}
+
+				continue;
+			}
+
+			// Reset filter pattern if it hasnt been completed at this point
+			// Example, iterator at []: /[s]rc/include/header.h
+			//                          /[i]nclude*/
+			if (excludeIterator < excludePath.length() - 1) {
+				excludeIterator = 0;
+			}
+
+			tryPatternState = false;
+		}
+
+		if (excludeIterator == excludePath.length()) {
 			return true;
 		}
+		if (excludePath.back() == '*' && excludeIterator == excludePath.length() - 1) {
+			return true;
+		}
+		if (onlyMatchDirectories && excludeIterator == excludePath.length() - 1) {
+			return true;
 		}
 	}
 

src/dotfile.h

@@ -39,7 +39,7 @@ private:
 	void selectivelyCommentOrUncomment(const std::string& path);
 
 	void forEachDotfile(const std::vector<std::string>& targets, const std::function<void(const std::filesystem::directory_entry&, size_t)>& callback);
-	bool filter(const std::filesystem::path& path);
+	bool filter(const std::filesystem::directory_entry& path);
 	bool include(const std::filesystem::path& path, const std::vector<std::string>& targets);
 	bool isSystemTarget(const std::string& target);
 };

test/unit/testdotfile.cpp

@@ -213,8 +213,8 @@ TEST_CASE(PushDotfilesWithExcludePath)
 
 	Config::the().setExcludePaths({
 		{ "__test-file-1", "file" },
-		{ "__subdir", "directory" },
+		{ "__subdir/", "directory" },
-		{ ".test", "endsWith" },
+		{ "*.test", "endsWith" },
 	});
 	Dotfile::the().push(fileNames);
 	Config::the().setExcludePaths({});