Add support for computing SHA-1 of files. Currently assumes it is

a regular file.

1 files changed, 17 insertions, 2 deletions

diff --git a/summainlib.py b/summainlib.py
index 4d0a89c..509bc50 100644
--- a/summainlib.py
+++ b/summainlib.py
@@ -15,6 +15,7 @@
 
 
 import grp
+import hashlib
 import os
 import pwd
 import time
@@ -28,12 +29,15 @@ class FilesystemObject(object):
     Responsible for gathering information and formatting it for
     reporting.
     
-    The optional stat_result argument is intended for unit tests.
+    The optional arguments are intended for unit tests.
     
     '''
     
-    def __init__(self, filename, stat_result=None):
+    def __init__(self, filename, stat_result=None, sha1=None,
+                 open_file=None):
         stat_result = stat_result or os.lstat(filename)
+        sha1 = sha1 or hashlib.sha1()
+        self.open_file = open_file or self.open_file
         self.values = dict()
         self['Name'] = filename
         self['Mtime'] = self.format_time(stat_result.st_mtime)
@@ -46,6 +50,7 @@ class FilesystemObject(object):
         self['Username'] = self.lookup_username(stat_result.st_uid)
         self['Gid'] = '%d' % stat_result.st_gid
         self['Group'] = self.lookup_group(stat_result.st_gid)
+        self['Sha-1'] = self.compute_sha1(filename, sha1)
 
     def format_time(self, timestamp):
         return time.strftime('%Y-%m-%d %H:%M:%S +0000', 
@@ -57,6 +62,16 @@ class FilesystemObject(object):
     def lookup_group(self, gid):
         return grp.getgrgid(gid).gr_name
 
+    def compute_sha1(self, filename, sha1):
+        f = self.open_file(filename)
+        while True:
+            data = f.read(64*1024) # 64 KiB seems reasonable.
+            if not data:
+                break
+            sha1.update(data)
+        f.close()
+        return sha1.hexdigest()
+
     def hook_name(self, value):
         return urllib.quote(value)