Handle symlinks

lib/private/modify_mtree.awk

@@ -45,5 +45,75 @@
     if (package_dir != "") {
         sub(/^/, package_dir "/")
     }
-    print;
+    if (preserve_symlinks != "") {
+        # By default Bazel reports symlinks as regular file/dir therefore mtree_spec has no way of knowing that a file
+        # is a symlink. This is a problem when we want to preserve symlinks especially for symlink sensitive applications
+        # such as nodejs with pnpm. To work around this we need to determine if a file a symlink and if so, we need to
+        # determine where the symlink points to by calling readlink repeatedly until we get the final destination.
+        #
+        # We then need to decide if it's a symlink based on how many times we had to call readlink and where we ended up.
+        #
+        # Unlike Bazels own symlinks, which points out of the sandbox symlinks, symlinks created by ctx.actions.symlink
+        # stays within the bazel sandbox so it's possible to detect those.
+        #
+        # See https://github.com/bazelbuild/rules_pkg/pull/609
+
+        symlink = ""
+        if ($0 ~ /type=file/ && $0 ~ /content=/) {
+            match($0, /content=[^ ]+/)
+            content_field = substr($0, RSTART, RLENGTH)
+            split(content_field, parts, "=")
+            path = parts[2]
+	    # Store paths for look up
+	    symlink_map[path] = $1
+	    # Resolve the symlink if it exists
+	    resolved_path = ""
+	    cmd = "readlink -f " path
+	    cmd | getline resolved_path
+	    close(cmd)
+
+	    if (resolved_path) {
+		if (resolved_path ~ bin_dir) {
+		    # Strip down the resolved path to start from bin_dir
+		    sub("^.*" bin_dir, bin_dir, resolved_path)
+		    if (path != resolved_path) {
+			# Replace the content field with the new path
+		        symlink = resolved_path
+		    }
+		}
+	    }
+        }
+	if (symlink != "") {
+	  line_array[NR] = $1 SUBSEP resolved_path
+	  }
+	else {
+	    line_array[NR] = $0  # Store other lines too, with an empty path
+	}
+    }
+
+    else {
+
+      print;  # Print immediately if symlinks are not preserved
+
+    }
+}
+END {
+    if (preserve_symlinks != "") {
+        # Process symlinks if needed
+        for (i = 1; i <= NR; i++) {
+            line = line_array[i]
+            if (index(line, SUBSEP) > 0) {  # Check if this path was a symlink
+	        split(line, fields, SUBSEP)
+		field0 = fields[1]
+		resolved_path = fields[2]
+                linked_to = symlink_map[resolved_path]
+                # Adjust the line for symlink using the map we created
+                new_line = field0 " type=link link=" linked_to
+                print new_line
+            } else {
+                # Print the original line if no symlink adjustment was needed
+                print line
+            }
+        }
+    }
 }

lib/private/tar.bzl

@@ -446,71 +446,6 @@ def _mtree_impl(ctx):
 
     return DefaultInfo(files = depset([out]), runfiles = ctx.runfiles([out]))
 
-def _mtree_mutate_impl(ctx):
-    srcs_runfiles = [
-        src[DefaultInfo].default_runfiles.files
-        for src in ctx.attr.srcs
-    ]
-    args = ctx.actions.args()
-    bsdtar = ctx.toolchains[TAR_TOOLCHAIN_TYPE]
-    mtree_generator = ctx.executable.mtree_generator.path
-
-    out_mtree = ctx.outputs.out
-    args.add("--input", ctx.file.mtree)
-    args.add("--output", out_mtree)
-    args.add("--bin_dir", ctx.bin_dir.path)
-
-    if ctx.attr.owner:
-        args.add("--owner", ctx.attr.owner)
-    if ctx.attr.ownername:
-        args.add("--ownername", ctx.attr.ownername)
-    if ctx.attr.strip_prefix:
-        args.add("--strip_prefix", ctx.attr.strip_prefix)
-    if ctx.attr.package_dir:
-        args.add("--package_dir", ctx.attr.package_dir)
-    if ctx.attr.mtime:
-        args.add("--mtime", ctx.attr.mtime)
-
-    #executable = bsdtar.tarinfo.binary,
-    inputs = ctx.files.srcs[:]
-    inputs.append(ctx.file.mtree)
-    ctx.actions.run(
-        executable = mtree_generator,
-        arguments = [args],
-        inputs = depset(
-            direct = inputs,
-            transitive = srcs_runfiles + [
-                ctx.attr.mtree_generator.default_runfiles.files,
-            ],
-        ),
-        outputs = [out_mtree],
-    )
-
-    return [DefaultInfo(files = depset([out_mtree]))]
-
-mtree_mutate = rule(
-    implementation = _mtree_mutate_impl,
-    attrs = {
-        "mtree": attr.label(allow_single_file = True),
-        "awk_script": attr.label(allow_single_file = True, default = "@aspect_bazel_lib//lib/private:modify_mtree.awk"),
-        "srcs": attr.label_list(allow_files = True),
-        "strip_prefix": attr.string(),
-        "package_dir": attr.string(),
-        "mtime": attr.string(),
-        "owner": attr.string(),
-        "ownername": attr.string(),
-        "out": attr.output(),
-        "mtree_generator": attr.label(
-            default = Label("//tools/mtree:mtree"),
-            executable = True,
-            cfg = "exec",
-        ),
-    },
-    toolchains = [
-        TAR_TOOLCHAIN_TYPE,
-        "@aspect_bazel_lib//lib:coreutils_toolchain_type",
-    ],
-)
 tar_lib = struct(
     attrs = _tar_attrs,
     implementation = _tar_impl,

lib/tar.bzl

@@ -55,7 +55,7 @@ TODO:
 load("@bazel_skylib//lib:types.bzl", "types")
 load("//lib:expand_template.bzl", "expand_template")
 load("//lib:utils.bzl", "propagate_common_rule_attributes")
-load("//lib/private:tar.bzl", _mutate_mtree = "mtree_mutate", _tar = "tar", _tar_lib = "tar_lib")
+load("//lib/private:tar.bzl", _tar = "tar", _tar_lib = "tar_lib")
 
 mtree_spec = rule(
     doc = "Create an mtree specification to map a directory hierarchy. See https://man.freebsd.org/cgi/man.cgi?mtree(8)",
@@ -137,6 +137,8 @@ def tar(name, mtree = "auto", stamp = 0, **kwargs):
 def mtree_mutate(
         name,
         mtree,
+        srcs = None,
+        preserve_symlinks = False,
         strip_prefix = None,
         package_dir = None,
         mtime = None,
@@ -148,24 +150,50 @@ def mtree_mutate(
 
     Args:
         name: name of the target, output will be `[name].mtree`.
+        srcs: source files to be used when resolving symlinks. required if `preserve_symlinks` is set to True.
+        preserve_symlinks: preserve symlinks
         mtree: input mtree file, typically created by `mtree_spec`.
         strip_prefix: prefix to remove from all paths in the tar. Files and directories not under this prefix are dropped.
         package_dir: directory prefix to add to all paths in the tar.
         mtime: new modification time for all entries.
         owner: new uid for all entries.
         ownername: new uname for all entries.
         awk_script: may be overridden to change the script containing the modification logic.
+
         **kwargs: additional named parameters to genrule
     """
-    _mutate_mtree(
+    vars = []
+    if strip_prefix:
+        vars.append("-v strip_prefix='{}'".format(strip_prefix))
+    if package_dir:
+        vars.append("-v package_dir='{}'".format(package_dir))
+    if mtime:
+        vars.append("-v mtime='{}'".format(mtime))
+    if owner:
+        vars.append("-v owner='{}'".format(owner))
+    if ownername:
+        vars.append("-v ownername='{}'".format(ownername))
+    if preserve_symlinks:
+        vars.append("-v preserve_symlinks=1")
+        if not srcs:
+            fail("preserve_symlinks requires srcs to be set in order to resolve symlinks")
+
+    # Check if srcs is of type list
+    if srcs and not types.is_list(srcs):
+        srcs = [srcs]
+
+    native.genrule(
         name = name,
-        mtree = mtree,
-        strip_prefix = strip_prefix,
-        package_dir = package_dir,
-        mtime = str(mtime) if mtime else None,
-        owner = owner,
-        ownername = ownername,
-        awk_script = awk_script,
-        out = "{}.mtree".format(name),
+        srcs = [mtree] + (srcs or []),
+        outs = [name + ".mtree"],
+        cmd = """
+            awk {variables} \
+            -v bin_dir=$(BINDIR) \
+            -f $(execpath {awk_script}) $(execpath {mtree_spec}) >$@""".format(
+            variables = " ".join(vars),
+            awk_script = awk_script,
+            mtree_spec = mtree,
+        ),
+        tools = [awk_script],
         **kwargs
     )

lib/tests/tar/BUILD.bazel

@@ -549,6 +549,7 @@ mtree_mutate(
         ":e17_node_modules",
     ],
     mtree = ":mtree17",
+    preserve_symlinks = True,
 )
 
 assert_tar_listing(