Add JSON and CSV output formats.

4 files changed, 134 insertions, 82 deletions

diff --git a/NEWS b/NEWS
index 5022fa6..1c8623d 100644
--- a/NEWS
+++ b/NEWS
@@ -2,6 +2,12 @@ NEWS file for summain
 =====================
 
 
+Version 0.10, released UNRELEASED
+---------------------------------
+
+* Added support for JSON and CSV output formats, in addition to the default
+  RFC822-like one. (See `--output-format` option.)
+
 Version 0.9, released 2011-07-24
 --------------------------------
 
diff --git a/summain b/summain
index 7082594..026a2ae 100755
--- a/summain
+++ b/summain
@@ -16,12 +16,90 @@
 
 
 import cliapp
+import csv
+import json
 import os
 import sys
 
 import summainlib
 
 
+class OutputFormat(object):
+
+    keys = ['Mtime', 'Mode', 'Ino', 'Dev', 'Nlink', 'Size',
+            'Uid', 'Username', 'Gid', 'Group', 'Target']
+
+    def __init__(self, output, checksums, objects):
+        self.output = output
+        self.checksums = checksums
+        self.objects = objects
+
+    def write(self):
+        for name, o in self.objects:
+            self.write_object(name, o)
+        
+    def write_object(self, name, o):
+        raise NotImplemented()
+
+
+class Rfc822(OutputFormat):
+
+    def write_object(self, name, o):
+        keys = self.keys + self.checksums
+        values = [('Name', name)]
+        values += [(k, o[k]) for k in keys if o[k] != '']
+        record = ''.join('%s: %s\n' % (k, v) for k, v in values if v != '')
+        self.output.write('%s\n' % record)
+
+
+class CSV(OutputFormat):
+
+    def __init__(self, output, checksums, objects):
+        OutputFormat.__init__(self, output, checksums, objects)
+        self.writer = csv.writer(output)
+        self.wrote_headings = False
+
+    def write_object(self, name, o):
+        keys = self.keys + self.checksums
+
+        if not self.wrote_headings:
+            self.writer.writerow(['Name'] + keys)
+            self.wrote_headings = True
+
+        values = [name] + [o[k] or '' for k in keys]
+        self.writer.writerow(values)
+
+
+class GeneratorList(list):
+
+    def __init__(self, gen):
+        self.gen = gen
+        
+    def __len__(self):
+        return 1
+
+    def __iter__(self):
+        return iter(self.gen)
+
+
+class Json(OutputFormat):
+
+    def write(self):
+        gen = GeneratorList(self.dictify(name, o) for name, o in self.objects)
+        json.dump(gen, self.output, sort_keys=True, indent=1)
+        self.output.write('\n')
+
+    def dictify(self, name, o):
+        keys = self.keys + self.checksums
+
+        values = { 'Name': name }
+        for k in keys:
+            if o[k] != '':
+                values[k] = o[k]
+
+        return values
+
+
 class Summain(cliapp.Application):
 
     def add_settings(self):
@@ -39,6 +117,9 @@ class Summain(cliapp.Application):
                                   'MD5, SHA1, SHA224, SHA256, SHA384, SHA512; '
                                   'use once per checksum type '
                                   '(default is SHA1)')
+        self.settings.choice(['output-format', 'f'],
+                             ['rfc822', 'csv', 'json'],
+                             'choose output format (rfc822, csv, json)')
 
     def files(self, root):
         if os.path.isdir(root):
@@ -51,6 +132,12 @@ class Summain(cliapp.Application):
             yield root
 
     def process_args(self, args):
+        checksums = [x.upper() 
+                     for x in self.settings['checksum'] or ['SHA1']]
+        fmt = self.new_formatter(checksums, self.find_roots(args))
+        fmt.write()
+
+    def find_roots(self, roots):
         relative = self.settings['relative-paths']
         exclude = self.settings['exclude']
         nn = summainlib.NumberNormalizer()
@@ -58,20 +145,44 @@ class Summain(cliapp.Application):
             pn = summainlib.PathNormalizer(self.settings['secret'])
         else:
             pn = summainlib.SamePath()
-        checksums = [x.upper() 
-                     for x in self.settings['checksum'] or ['SHA1']]
-        o = summainlib.FilesystemObject('.', nn, pn, exclude, checksums)
-        for checksum in checksums:
-            try:
-                o[checksum]
-            except KeyError:
-                raise cliapp.AppException('Unknown checksum %s' % checksum)
-        for root in args:
+
+        for root in roots:
             for filename in self.files(root):
-                o = summainlib.FilesystemObject(filename, nn, pn, exclude,
-                                                checksums)
-                self.output.write(o.format(root if relative else None))
-                self.output.write('\n')
+                o = summainlib.FilesystemObject(filename, nn, pn, exclude)
+                if relative:
+                    name = self.relative_path(root, o)
+                else:
+                    name = o['Name']
+                yield name, o
+
+    def relative_path(self, root, o):
+        '''Return a path that is relative to root, if possible.
+        
+        If pathname does not start with root, then return it
+        unmodified.
+        
+        '''
+        
+        if root.endswith(os.sep):
+            root2 = root
+        else:
+            root2 = root + os.sep
+        pathname = o['Name']
+        if pathname.startswith(root2):
+            return pathname[len(root2):]
+        elif pathname == root and o.isdir():
+            return '.'
+        else:
+            return pathname
+
+    def new_formatter(self, checksums, objects):
+        table = {
+            'rfc822': Rfc822,
+            'csv': CSV,
+            'json': Json,
+        }
+        return table[self.settings['output-format']](self.output, checksums,
+                                                      objects)
 
 
 Summain(version=summainlib.__version__).run()
diff --git a/summainlib.py b/summainlib.py
index 5376e00..34dad47 100644
--- a/summainlib.py
+++ b/summainlib.py
@@ -112,13 +112,12 @@ class FilesystemObject(object):
     
     '''
     
-    def __init__(self, filename, nn, pn, exclude, checksums, 
+    def __init__(self, filename, nn, pn, exclude, 
                  stat_result=None, sha1=None, sha224=None,
                  sha256=None, sha384=None, sha512=None,
                  md5=None, open_file=None, readlink=None):
         self._filename = filename
         self._exclude = set(self._normalize_key(k) for k in exclude)
-        self._checksums = checksums
         self._pn = pn
         self._nn = nn
         self._md5 = md5 or hashlib.md5()
@@ -232,41 +231,7 @@ class FilesystemObject(object):
                 raise KeyError(key)
         return self.values.get(key, '')
 
-    def _isdir(self):
+    def isdir(self): # pragma: no cover
         '''Is this a directory?'''
-        
         return stat.S_ISDIR(int(self['Mode'], 8))
 
-    def relative_path(self, root):
-        '''Return a path that is relative to root, if possible.
-        
-        If pathname does not start with root, then return it
-        unmodified.
-        
-        '''
-        
-        if root.endswith(os.sep):
-            root2 = root
-        else:
-            root2 = root + os.sep
-        pathname = self['Name']
-        if pathname.startswith(root2):
-            return pathname[len(root2):]
-        elif pathname == root and self._isdir():
-            return '.'
-        else:
-            return pathname
-
-    def format(self, root=None): # pragma: no cover
-        if root is None:
-            name = self['Name']
-        else:
-            name = self.relative_path(root)
-        
-        keys = (['Mtime', 'Mode', 'Ino', 'Dev', 'Nlink', 'Size',
-                 'Uid', 'Username', 'Gid', 'Group', 'Target'] +
-                self._checksums)
-        values = [('Name', name)]
-        values += [(k, self[k]) for k in keys if self[k] != '']
-        return ''.join('%s: %s\n' % (k, v) for k, v in values if v != '')
-
diff --git a/summainlib_tests.py b/summainlib_tests.py
index 6015541..4ec3da5 100644
--- a/summainlib_tests.py
+++ b/summainlib_tests.py
@@ -85,13 +85,12 @@ class FilesystemObjectTests(unittest.TestCase):
         self.nn = summainlib.NumberNormalizer()
         self.pn = summainlib.SamePath()
         self.exclude = []
-        self.checksums = ['SHA1']
 
     def new(self, name, mode=None):
         if mode is not None:
             self.st.st_mode = mode
         return summainlib.FilesystemObject(name, self.nn, self.pn, 
-                                            self.exclude, self.checksums,
+                                            self.exclude,
                                             stat_result=self.st,
                                             sha1=FakeChecksummer(),
                                             sha224=FakeChecksummer(),
@@ -168,35 +167,6 @@ class FilesystemObjectTests(unittest.TestCase):
 
     def test_formats_target_correctly_for_regular_file(self):
         self.assertEqual(self.new('foo')['Target'], '')
-        
-    def test_does_not_output_size_for_directory(self):
-        fso = self.new('foo', mode=stat.S_IFDIR | 0777)
-        output = fso.format()
-        self.assert_('Size:' not in output)
-        
-    def test_relative_path_returns_path_if_not_starting_with_root(self):
-        fso = self.new('/foo/bar')
-        self.assertEqual(fso.relative_path('/yo'), '/foo/bar')
-        
-    def test_relative_path_returns_partial_path_if_starting_with_root(self):
-        fso = self.new('/foo/bar')
-        self.assertEqual(fso.relative_path('/foo'), 'bar')
-        
-    def test_relative_path_returns_dot_if_same_as_root_and_dir(self):
-        fso = self.new('/foo/bar', mode=stat.S_IFDIR)
-        self.assertEqual(fso.relative_path('/foo/bar'), '.')
-        
-    def test_relative_path_returns_path_if_same_as_root_and_not_dir(self):
-        fso = self.new('/foo/bar', mode=stat.S_IFREG)
-        self.assertEqual(fso.relative_path('/foo/bar'), '/foo/bar')
-        
-    def test_relative_path_returns_path_if_root_is_slashless_prefix(self):
-        fso = self.new('/foobar', mode=stat.S_IFREG)
-        self.assertEqual(fso.relative_path('/foo'), '/foobar')
-        
-    def test_relative_path_returns_partial_path_if_root_ends_in_slash(self):
-        fso = self.new('/foo/bar', mode=stat.S_IFREG)
-        self.assertEqual(fso.relative_path('/foo/'), 'bar')
 
     def test_excludes_unwanted_fields_from_output(self):
         self.exclude = ['mtime']
@@ -224,7 +194,7 @@ class FilesystemObjectNormalizedNumbersTests(unittest.TestCase):
                            st_uid=0, st_gid=0)
         self.ino += 1
         return summainlib.FilesystemObject(name, self.nn, self.pn, 
-                                            self.exclude, self.checksums,
+                                            self.exclude,
                                             stat_result=st,
                                             sha1=FakeChecksummer(),
                                             sha224=FakeChecksummer(),