From 560ee0c05e32e068ca59f9110f45a0413dc751f3 Mon Sep 17 00:00:00 2001
From: Tim Wojtulewicz <tim@corelight.com>
Date: Mon, 6 Jul 2020 11:24:34 -0700
Subject: [PATCH] GH-1041: Move compress_path to a bif that uses normalize_path

---
 scripts/base/utils/paths.zeek                 | 39 -------------------
 src/zeek.bif                                  | 11 ++++++
 testing/btest/Baseline/bifs.compress_path/out |  2 +
 testing/btest/bifs/compress_path.zeek         |  9 +++++
 4 files changed, 22 insertions(+), 39 deletions(-)
 create mode 100644 testing/btest/Baseline/bifs.compress_path/out
 create mode 100644 testing/btest/bifs/compress_path.zeek

diff --git a/scripts/base/utils/paths.zeek b/scripts/base/utils/paths.zeek
index fdc9bd5d3d..76b7061ec6 100644
--- a/scripts/base/utils/paths.zeek
+++ b/scripts/base/utils/paths.zeek
@@ -21,45 +21,6 @@ function extract_path(input: string): string
 	return parts[1];
 	}
 
-## Compresses a given path by removing '..'s and the parent directory it
-## references and also removing dual '/'s and extraneous '/./'s.
-##
-## dir: a path string, either relative or absolute.
-##
-## Returns: a compressed version of the input path.
-function compress_path(dir: string): string
-	{
-	const cdup_sep = /((\/)*([^\/]|\\\/)+)?((\/)+\.\.(\/)*)/;
-
-	local parts = split_string_n(dir, cdup_sep, T, 1);
-	if ( |parts| > 1 )
-		{
-		# reaching a point with two parent dir references back-to-back means
-		# we don't know about anything higher in the tree to pop off
-		if ( parts[1] == "../.." )
-			return join_string_vec(parts, "");
-		if ( sub_bytes(parts[1], 0, 1) == "/" )
-			parts[1] = "/";
-		else
-			parts[1] = "";
-		dir = join_string_vec(parts, "");
-		return compress_path(dir);
-		}
-
-	const multislash_sep = /(\/\.?){2,}/;
-	parts = split_string_all(dir, multislash_sep);
-	for ( i in parts )
-		if ( i % 2 == 1 )
-			parts[i] = "/";
-	dir = join_string_vec(parts, "");
-
-	# remove trailing slashes from path
-	if ( |dir| > 1 && sub_bytes(dir, |dir|, 1) == "/" )
-		dir = sub_bytes(dir, 0, |dir| - 1);
-
-	return dir;
-	}
-
 ## Constructs a path to a file given a directory and a file name.
 ##
 ## dir: the directory in which the file lives.
diff --git a/src/zeek.bif b/src/zeek.bif
index 79ded154ba..d374698a8e 100644
--- a/src/zeek.bif
+++ b/src/zeek.bif
@@ -5157,3 +5157,14 @@ function to_json%(val: any, only_loggable: bool &default=F, field_escape_pattern
 	%{
 	return val->ToJSON(only_loggable, field_escape_pattern);
 	%}
+
+## Compresses a given path by removing '..'s and the parent directory it
+## references and also removing dual '/'s and extraneous '/./'s.
+##
+## dir: a path string, either relative or absolute.
+##
+## Returns: a compressed version of the input path.
+function compress_path%(dir: string%): string
+	%{
+	return zeek::make_intrusive<zeek::StringVal>(normalize_path(dir->ToStdString()));
+	%}
\ No newline at end of file
diff --git a/testing/btest/Baseline/bifs.compress_path/out b/testing/btest/Baseline/bifs.compress_path/out
new file mode 100644
index 0000000000..ad5fdd8dc0
--- /dev/null
+++ b/testing/btest/Baseline/bifs.compress_path/out
@@ -0,0 +1,2 @@
+../foo
+../foo
diff --git a/testing/btest/bifs/compress_path.zeek b/testing/btest/bifs/compress_path.zeek
new file mode 100644
index 0000000000..a21c7349e7
--- /dev/null
+++ b/testing/btest/bifs/compress_path.zeek
@@ -0,0 +1,9 @@
+#
+# @TEST-EXEC: zeek -b %INPUT >out
+# @TEST-EXEC: btest-diff out
+
+event zeek_init()
+	{
+	print compress_path("./../foo");
+	print compress_path("././../foo");
+	}