New path matching feature to handle exclude/ignore rules similar to .gitignore

Author: jjlauer

jsync-engine/src/main/java/com/fizzed/jsync/engine/DefaultJsyncEventHandler.java

@@ -34,8 +34,13 @@ public void willExcludePath(VirtualPath sourcePath) {
     }
 
     @Override
-    public void willIgnorePath(VirtualPath sourcePath) {
-        log.debug("Ignoring path {}", sourcePath);
+    public void willIgnoreSourcePath(VirtualPath sourcePath) {
+        log.debug("Ignoring source path {}", sourcePath);
+    }
+
+    @Override
+    public void willIgnoreTargetPath(VirtualPath targetPath) {
+        log.debug("Ignoring target path {}", targetPath);
     }
 
     @Override

jsync-engine/src/main/java/com/fizzed/jsync/engine/JsyncEngine.java

@@ -2,6 +2,7 @@
 
 import com.fizzed.jsync.vfs.*;
 import com.fizzed.jsync.vfs.util.Permissions;
+import com.fizzed.jsync.vfs.util.VirtualPathMatchers;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
@@ -31,9 +32,10 @@ public class JsyncEngine {
     private List<String> ignores;
     // when running a sync
     private Checksum negotiatedChecksum;
-    private List<VirtualPath> excludePaths;
-    private List<VirtualPath> ignoreSourcePaths;
-    private List<VirtualPath> ignoreTargetPaths;
+    private VirtualPathMatchers excludeMatchers;
+    private VirtualPathMatchers ignoreMatchers;
+    private VirtualPath sourceRootPath;
+    private VirtualPath targetRootPath;
 
     public JsyncEngine() {
         this.eventHandler = new DefaultJsyncEventHandler();
@@ -225,6 +227,9 @@ public JsyncResult sync(VirtualFileSystem sourceVfs, String sourcePath, VirtualF
         final VirtualPath sourcePathAbsFinal = sourcePathAbs.normalize();
         final VirtualPath targetPathAbsFinal = targetPathAbs.normalize();
 
+        this.sourceRootPath = sourcePathAbsFinal;
+        this.targetRootPath = targetPathAbsFinal;
+
 
         //
         // Negotiate checksum methods between source and target filesystems if necessary
@@ -236,29 +241,9 @@ public JsyncResult sync(VirtualFileSystem sourceVfs, String sourcePath, VirtualF
         log.debug("Source filesystem stat mode: {}", sourceVfs.getStatModel());
         log.debug("Target filesystem stat mode: {}", targetVfs.getStatModel());
 
-        // build exclude and ignore paths
-        if (this.excludes != null) {
-            this.excludePaths = this.excludes.stream()
-                .map(VirtualPath::parse)
-                .map(sourcePathAbsFinal::resolve)
-                .collect(toList());
-        } else {
-            this.excludePaths = Collections.emptyList();
-        }
-
-        if (this.ignores != null) {
-            this.ignoreSourcePaths = this.ignores.stream()
-                .map(VirtualPath::parse)
-                .map(sourcePathAbsFinal::resolve)
-                .collect(toList());
-            this.ignoreTargetPaths = this.ignores.stream()
-                .map(VirtualPath::parse)
-                .map(targetPathAbsFinal::resolve)
-                .collect(toList());
-        } else {
-            this.ignoreSourcePaths = Collections.emptyList();
-            this.ignoreTargetPaths = Collections.emptyList();
-        }
+        // build exclude and ignore matchers
+        this.excludeMatchers = VirtualPathMatchers.compile(this.excludes);
+        this.ignoreMatchers = VirtualPathMatchers.compile(this.ignores);
 
 
         final long now = System.currentTimeMillis();
@@ -272,17 +257,6 @@ public JsyncResult sync(VirtualFileSystem sourceVfs, String sourcePath, VirtualF
         final List<VirtualPathPair> deferredFiles = new ArrayList<>();
 
         if (sourcePathAbsFinal.isDirectory()) {
-            // any excludes, let's resolve them against pwd of the source to make it easier to exclude them
-            final List<VirtualPath> excludePaths;
-            if (this.excludes != null) {
-                excludePaths = this.excludes.stream()
-                    .map(VirtualPath::parse)
-                    .map(sourcePathAbsFinal::resolve)
-                    .collect(toList());
-            } else {
-                excludePaths = Collections.emptyList();
-            }
-
             // as we process files, only a subset may require more advanced methods of detecting whether they were modified
             // since that process could be "expensive", we keep a list of files on source/target that we will defer processing
             // until we have a chance to do some bulk processing of checksums, etc.
@@ -409,23 +383,19 @@ protected void syncDirectory(int level, JsyncResult result, List<VirtualPathPair
 
 
         // we need a list of files in both directories, so we can see what to add/delete
-        List<VirtualPath> sourceChildPaths = sourceVfs.ls(sourcePath).stream()
+        final List<VirtualPath> sourceChildPaths = sourceVfs.ls(sourcePath).stream()
             // apply filter to source files if they are on the exclude list
             .filter(v -> {
-                for (VirtualPath p : this.excludePaths) {
-                    if (v.startsWith(p)) {
-                        this.eventHandler.willExcludePath(v);
-                        return false;
-                    }
+                if (this.excludeMatchers.matches(this.sourceRootPath, v)) {
+                    this.eventHandler.willExcludePath(v);
+                    return false;
                 }
                 return true;
             })
             .filter(v -> {
-                for (VirtualPath p : this.ignoreSourcePaths) {
-                    if (v.startsWith(p)) {
-                        this.eventHandler.willIgnorePath(v);
-                        return false;
-                    }
+                if (this.ignoreMatchers.matches(this.sourceRootPath, v)) {
+                    this.eventHandler.willIgnoreSourcePath(v);
+                    return false;
                 }
                 return true;
             })
@@ -446,10 +416,9 @@ protected void syncDirectory(int level, JsyncResult result, List<VirtualPathPair
 
         final List<VirtualPath> targetChildPaths = targetVfs.ls(targetPath).stream()
             .filter(v -> {
-                for (VirtualPath p : this.ignoreTargetPaths) {
-                    if (v.startsWith(p)) {
-                        return false;
-                    }
+                if (this.ignoreMatchers.matches(this.targetRootPath, v)) {
+                    this.eventHandler.willIgnoreTargetPath(v);
+                    return false;
                 }
                 return true;
             })

jsync-engine/src/main/java/com/fizzed/jsync/engine/JsyncEventHandler.java

@@ -16,9 +16,11 @@ public interface JsyncEventHandler {
 
     void willEnd(VirtualFileSystem sourceVfs, VirtualPath sourcePath, VirtualFileSystem targetVfs, VirtualPath targetPath, JsyncResult result, long timeMillis);
 
-    void willExcludePath(VirtualPath targetPath);
+    void willExcludePath(VirtualPath sourcePath);
 
-    void willIgnorePath(VirtualPath targetPath);
+    void willIgnoreSourcePath(VirtualPath sourcePath);
+
+    void willIgnoreTargetPath(VirtualPath targetPath);
 
     void willCreateDirectory(VirtualPath targetPath, boolean recursively);
 

jsync-vfs/src/main/java/com/fizzed/jsync/vfs/util/VirtualPathMatcher.java

@@ -0,0 +1,93 @@
+package com.fizzed.jsync.vfs.util;
+
+import com.fizzed.jsync.vfs.VirtualPath;
+
+import java.nio.file.FileSystems;
+import java.nio.file.PathMatcher;
+import java.nio.file.Paths;
+
+public class VirtualPathMatcher {
+
+    private final PathMatcher matcher;
+
+    public VirtualPathMatcher(PathMatcher matcher) {
+        this.matcher = matcher;
+    }
+
+    public boolean matches(VirtualPath rootPath, VirtualPath currentPath) {
+        final String relativePath;
+
+        // resolve the current path against the root path, so we're left with the relative path we're matching against
+        if (currentPath.isAbsolute()) {
+            String rootFullPath = rootPath.toFullPath();
+            String currentFullPath = currentPath.toFullPath();
+            int pathStartPos = currentFullPath.indexOf(rootFullPath);
+            if (pathStartPos >= 0) {
+                // remove the leading path PLUS the file separator
+                relativePath = currentFullPath.substring(pathStartPos + rootFullPath.length() + 1);
+            } else {
+                relativePath = currentFullPath;
+            }
+        } else {
+            relativePath = currentPath.toString();
+        }
+
+        return this.matcher.matches(Paths.get(relativePath));
+    }
+
+    static public VirtualPathMatcher compile(String rule) {
+        String glob = rule.trim();
+        boolean isDirectory = false;
+        boolean isRooted = false;
+
+        // 1. Check for Directory marker
+        if (glob.endsWith("/")) {
+            isDirectory = true;
+            glob = glob.substring(0, glob.length() - 1); // Strip trailing slash
+        }
+
+        // 2. Check for Root anchor
+        if (glob.startsWith("/")) {
+            isRooted = true;
+            glob = glob.substring(1); // Strip leading slash
+        }
+
+        // FIX: Handle "/**/" usually found in the middle of paths
+        // Git: "docs/**/*.md" -> Zero or more dirs
+        // Java: "docs/**/*.md" -> One or more dirs (fails on docs/file.md)
+        // Solution: Replace "/**/" with "/{,**/}" which means "Empty OR /**/"
+//        if (glob.contains("/**/")) {
+//            glob = glob.replace("/**/", "/{,**/}");
+//        }
+
+        // 3. Build the Glob
+        // We need to construct a robust brace expansion {A,B,C...}
+        StringBuilder finalGlob = new StringBuilder();
+        finalGlob.append("glob:{");
+
+        if (isRooted) {
+            // Rule: /target/ or /target
+            finalGlob.append(glob); // Matches "target" at root
+            if (isDirectory) {
+                finalGlob.append(",").append(glob).append("/**"); // Matches "target/..." at root
+            }
+        } else {
+            // Rule: target/ or target
+            finalGlob.append(glob);              // Matches "target" at root
+            finalGlob.append(",**/").append(glob); // Matches "src/target" (nested)
+
+            if (isDirectory) {
+                // If it's a directory, we must ALSO match the contents
+                finalGlob.append(",").append(glob).append("/**");       // Matches "target/file.txt" (root)
+                finalGlob.append(",**/").append(glob).append("/**");    // Matches "src/target/file.txt" (nested)
+            }
+        }
+
+        finalGlob.append("}");
+
+        PathMatcher matcher = FileSystems.getDefault().getPathMatcher(finalGlob.toString());
+
+        return new  VirtualPathMatcher(matcher);
+    }
+
+}

jsync-vfs/src/main/java/com/fizzed/jsync/vfs/util/VirtualPathMatchers.java

@@ -1,37 +1,35 @@
 package com.fizzed.jsync.vfs.util;
 
-import java.nio.file.FileSystems;
-import java.nio.file.PathMatcher;
+import com.fizzed.jsync.vfs.VirtualPath;
+
+import java.util.ArrayList;
+import java.util.List;
 
 public class VirtualPathMatchers {
 
-    
+    private final List<VirtualPathMatcher> matchers = new ArrayList<>();
 
-    static public PathMatcher compileRule(String rule) {
-        String glob = rule.trim();
+    public VirtualPathMatchers(List<VirtualPathMatcher> matchers) {
+        this.matchers.addAll(matchers);
+    }
 
-        // 1. Handle directory-only rules (e.g., "build/")
-        // Git implies "everything inside this directory"
-        if (glob.endsWith("/")) {
-            glob = glob + "**";
+    public boolean matches(VirtualPath rootPath, VirtualPath path) {
+        for  (VirtualPathMatcher matcher : matchers) {
+            if (matcher.matches(rootPath, path)) {
+                return true;
+            }
         }
+        return false;
+    }
 
-        // 2. Handle "rooted" vs "anywhere" rules
-        // If it starts with '/', it matches from the root only.
-        // If NOT, it matches anywhere (e.g., "*.log" -> "** /*.log")
-        if (glob.startsWith("/")) {
-            // Remove leading slash for Java PathMatcher consistency on relative paths
-            glob = glob.substring(1);
-        } else {
-            // If it's not rooted, allow it to match deep in the tree
-            // Example: "tmp" becomes "**/tmp"
-            if (!glob.startsWith("**/") && !glob.equals("*")) {
-                glob = "**/" + glob;
+    static public VirtualPathMatchers compile(List<String> rules) {
+        List<VirtualPathMatcher> matchers = new ArrayList<>();
+        if (rules != null) {
+            for (String rule : rules) {
+                matchers.add(VirtualPathMatcher.compile(rule));
             }
         }
-
-        // Create the matcher using "glob" syntax
-        return FileSystems.getDefault().getPathMatcher("glob:" + glob);
+        return new VirtualPathMatchers(matchers);
     }
 
 }