Added --pattern to CLI, closes #30

Author: RhetTbull

README.md

@@ -32,8 +32,7 @@ The command line tool can also be run via `python -m osxmetadata`.  Running it w
 
 ```
 Usage: osxmetadata [OPTIONS] FILE
-Usage: __main__.py [OPTIONS] FILE
-
+  
   Read/write metadata from file(s).
 
 Options:
@@ -64,13 +63,25 @@ Options:
                                   tags' sets tags and keywords to same values.
   -B, --backup                    Backup FILE attributes.  Backup file
                                   '.osxmetadata.json' will be created in same
-                                  folder as FILE. Only backs up attributes 
+                                  folder as FILE. Only backs up attributes
                                   known to osxmetadata.
   -R, --restore                   Restore FILE attributes from backup file.
                                   Restore will look for backup file
                                   '.osxmetadata.json' in same folder as FILE.
   -V, --verbose                   Print verbose output.
   -f, --copyfrom SOURCE_FILE      Copy attributes from file SOURCE_FILE.
+  --files-only                    Do not apply metadata commands to
+                                  directories themselves, only files in a
+                                  directory.
+  -p, --pattern PATTERN           Only process files matching PATTERN; only
+                                  applies to --walk. If specified, only files
+                                  matching PATTERN will be processed as each
+                                  directory is walked. May be used for than
+                                  once to specify multiple patterns. For
+                                  example, tag all *.pdf files in projectdir
+                                  and subfolders with tag 'project':
+                                  osxmetadata --append tags 'project' --walk
+                                  projectdir/ --pattern '*.pdf'
 
 Valid attributes for ATTRIBUTE: Each attribute has a short name, a constant
 name, and a long constant name. Any of these may be used for ATTRIBUTE

osxmetadata/__main__.py

@@ -1,6 +1,7 @@
 # /usr/bin/env python3
 
 import datetime
+import glob
 import itertools
 import json
 import logging
@@ -274,6 +275,18 @@ def get_help(self, ctx):
     default=False,
     required=False,
 )
+PATTERN_OPTION = click.option(
+    "--pattern",
+    "-p",
+    metavar="PATTERN",
+    help="Only process files matching PATTERN; only applies to --walk. "
+    "If specified, only files matching PATTERN will be processed as each directory is walked. "
+    "May be used for than once to specify multiple patterns. "
+    "For example, tag all *.pdf files in projectdir and subfolders with tag 'project': "
+    "osxmetadata --append tags 'project' --walk projectdir/ --pattern '*.pdf'",
+    multiple=True,
+    required=False,
+)
 
 
 @click.command(cls=MyClickCommand)
@@ -297,6 +310,7 @@ def get_help(self, ctx):
 @VERBOSE_OPTION
 @COPY_FROM_OPTION
 @FILES_ONLY_OPTION
+@PATTERN_OPTION
 @click.pass_context
 def cli(
     ctx,
@@ -319,6 +333,7 @@ def cli(
     verbose,
     copyfrom,
     files_only,
+    pattern,
 ):
     """ Read/write metadata from file(s). """
 
@@ -403,30 +418,43 @@ def cli(
 
     # loop through each file, process it, then do backup or restore if needed
     for filename in files:
-        process_files(
-            ctx,
-            [filename],
-            json_,
-            set_,
-            append,
-            update,
-            remove,
-            clear,
-            get,
-            list_,
-            mirror,
-            wipe,
-            verbose,
-            copyfrom,
-            backup,
-            restore,
-            walk,
-            files_only,
-        )
+        if not all([os.path.isdir(filename), walk, pattern]):
+            process_files(
+                ctx,
+                [filename],
+                json_,
+                set_,
+                append,
+                update,
+                remove,
+                clear,
+                get,
+                list_,
+                mirror,
+                wipe,
+                verbose,
+                copyfrom,
+                backup,
+                restore,
+                walk,
+                files_only,
+            )
 
         if walk and os.path.isdir(filename):
             for root, dirnames, filenames in os.walk(filename):
-                filepaths = [os.path.join(root, fname) for fname in dirnames + filenames]
+                if pattern:
+                    # only process files matching pattern
+                    filepaths = []
+                    for dirname in dirnames:
+                        for matches in [
+                            glob.glob(os.path.join(os.path.join(root, dirname), pat))
+                            for pat in pattern
+                        ]:
+                            filepaths.extend(matches)
+                else:
+                    filepaths = [
+                        os.path.join(root, fname) for fname in dirnames + filenames
+                    ]
                 process_files(
                     ctx,
                     filepaths,
@@ -481,7 +509,7 @@ def process_files(
             if verbose:
                 click.echo(f"Skipping directory: {fpath}")
             continue
-        
+
         if verbose:
             click.echo(f"Processing file: {fpath}")
 

osxmetadata/_version.py

@@ -1 +1 @@
-__version__ = "0.99.8"
+__version__ = "0.99.9"

tests/test_cli.py

@@ -843,6 +843,36 @@ def test_cli_walk_files_only(temp_dir):
     assert not md.tags
 
 
+def test_cli_walk_pattern(temp_dir):
+    """ test --walk with --pattern """
+    import os
+    import pathlib
+    from osxmetadata import OSXMetaData, Tag
+    from osxmetadata.__main__ import cli
+
+    dirname = pathlib.Path(temp_dir)
+    os.makedirs(dirname / "temp" / "subfolder1")
+    os.makedirs(dirname / "temp" / "subfolder2")
+    create_file(dirname / "temp" / "temp1.txt")
+    create_file(dirname / "temp" / "subfolder1" / "sub1.txt")
+    create_file(dirname / "temp" / "subfolder1" / "sub1.pdf")
+
+    runner = CliRunner()
+    result = runner.invoke(
+        cli, ["--set", "tags", "FOO", "--walk", "--pattern", "*.pdf", temp_dir]
+    )
+    assert result.exit_code == 0
+
+    md = OSXMetaData(dirname / "temp" / "subfolder1" / "sub1.pdf")
+    assert md.tags == [Tag("FOO")]
+
+    md = OSXMetaData(dirname / "temp" / "subfolder1" / "sub1.txt")
+    assert not md.tags
+
+    md = OSXMetaData(dirname / "temp" / "subfolder2")
+    assert not md.tags
+
+
 def test_cli_files_only(temp_dir):
     """ test --files-only without --walk """
     import glob