Added filename parsing functions

3 files changed, 68 insertions, 1 deletions

diff --git a/.github/workflows/build-and-package.yaml b/.github/workflows/build-and-package.yaml
index 3368220..37add71 100644
--- a/.github/workflows/build-and-package.yaml
+++ b/.github/workflows/build-and-package.yaml
@@ -22,7 +22,7 @@ jobs:
         uses: actions/checkout@v2
       - id: vars
         run: |
-          set -e -x
+          set -x -e
 
           echo "github event ref is ${{ github.ref }}"
 
diff --git a/bin/q.py b/bin/q.py
index c6d282b..653b8e5 100755
--- a/bin/q.py
+++ b/bin/q.py
@@ -147,6 +147,28 @@ def sqrt(data):
 def power(data,p):
     return data**p
 
+def file_ext(data):
+    if data is None:
+        return None
+
+    return os.path.splitext(data)[1]
+
+def file_folder(data):
+    if data is None:
+        return None
+    return os.path.split(data)[0]
+
+def file_basename(data):
+    if data is None:
+        return None
+    return os.path.split(data)[1]
+    
+def file_basename_no_ext(data):
+    if data is None:
+        return None
+
+    return os.path.split(os.path.splitext(data)[0])[-1]
+
 def percentile(l, p):
     # TODO Alpha implementation, need to provide multiple interpolation methods, and add tests
     if not l:
@@ -276,6 +298,26 @@ user_functions = [
                     "Raise expr1 to the power of expr2",
                     power,
                     2),
+    UserFunctionDef(FunctionType.REGULAR,
+                    "file_ext","file_ext(<expr>) = <filename-extension-or-empty-string>",
+                    "Get the extension of a filename",
+                    file_ext,
+                    1),
+    UserFunctionDef(FunctionType.REGULAR,
+                    "file_folder","file_folder(<expr>) = <folder-name-of-filename>",
+                    "Get the folder part of a filename",
+                    file_folder,
+                    1),
+    UserFunctionDef(FunctionType.REGULAR,
+                    "file_basename","file_basename(<expr>) = <basename-of-filename-including-extension>",
+                    "Get the basename of a filename, including extension if any",
+                    file_basename,
+                    1),
+    UserFunctionDef(FunctionType.REGULAR,
+                    "file_basename_no_ext","file_basename_no_ext(<expr>) = <basename-of-filename-without-extension>",
+                    "Get the basename of a filename, without the extension if there is one",
+                    file_basename_no_ext,
+                    1),
     UserFunctionDef(FunctionType.AGG,
                     "percentile","percentile(<expr>,<percentile-in-the-range-0-to-1>) = <percentile-value>",
                     "Calculate the strict percentile of a set of a values.",
diff --git a/test/test_suite.py b/test/test_suite.py
index fa8f2ea..aaa7476 100755
--- a/test/test_suite.py
+++ b/test/test_suite.py
@@ -4332,6 +4332,31 @@ class UserFunctionTests(AbstractQTestCase):
         self.assertEqual(o[3],six.b('32.0'))
         self.assertEqual(o[4],six.b('55.9016994375'))
 
+    def test_file_functions(self):
+        filenames = [
+            "file1",
+            "file2.csv",
+            "/var/tmp/file3",
+            "/var/tmp/file4.gz",
+            ""
+        ]
+        data = "\n".join(filenames)
+
+        cmd = 'echo "%s" | %s -c 1 -d , "select file_folder(c1),file_ext(c1),file_basename(c1),file_basename_no_ext(c1) from -"' % (data,Q_EXECUTABLE)
+        retcode, o, e = run_command(cmd)
+
+        self.assertEqual(retcode,0)
+        self.assertEqual(len(o),5)
+        self.assertEqual(len(e),0)
+        self.assertEqual(o,[
+            b',,file1,file1',
+            b',.csv,file2.csv,file2',
+            b'/var/tmp,,file3,file3',
+            b'/var/tmp,.gz,file4.gz,file4',
+            b',,,'
+        ])
+
+
     def test_sha1_function(self):
         cmd = 'seq 1 4 | %s -c 1 -d , "select c1,sha1(c1) from -"' % Q_EXECUTABLE
         retcode, o, e = run_command(cmd)