added checksum checking

files/image_list.py

@@ -29,6 +29,8 @@ def setup_options():
                       help="Do not add new files [%(default)s]")
     db.add_argument("-c",action="store_true",dest="changed",default=False,
                      help="Modify changed files [%(default)s]")
+    db.add_argument("--check",action="store_true",dest="check",default=False,
+                     help="Check md5sums of files. Limit check with -s.")
     db.add_argument("-d",action="store_true",dest="delete",default=False,
                      help="Delete non-existing entries [%(default)s]")
     db.add_argument("-D",action="store_true",dest="delete_data",default=False,
@@ -98,6 +100,52 @@ def setup_options():
         options.add=False
     return options
 
+def checkdb(options):
+    
+    if options.search:
+        needle='%'+options.search+'%'
+    else:
+        needle='%'
+    conn=sqlite3.connect(options.sqlfile)
+    conn.text_factory=str
+    db=conn.cursor()
+    db.execute("SELECT file,hash,size,date FROM list WHERE file LIKE ? ORDER BY file",(needle,))
+    missing=[]
+    differing=[]
+    OK_count=0
+    for row in db:
+        status='OK'
+        sys.stdout.write("\r%s"%(row[0],))
+        if os.path.exists(row[0]):
+            md5f=get_md5(row[0])
+            if row[1]!=md5f:
+                status='Checksum-difference'
+                differing.append(row)
+        else:
+            status='Not-found'
+            missing.append(row)
+        sys.stdout.write("\r%s %s\n"%(row[0],status))
+        if status=='OK':
+            OK_count+=1
+    if len(differing)>0:
+        print_stderr("----\nDiffering files:")
+        pad=str(max([len(x[0]) for x in differing]))
+        for f in differing:
+            print_stderr(("%-"+pad+"s (%s %7s => %s %7s)")%(f[0],humanize_date(f[3]),humanize_size(f[2]),
+                                                humanize_date(os.path.getmtime(f[0])),
+                                                humanize_size(os.path.getsize(f[0]))))
+    if len(missing)>0:
+        print_stderr("----\nMissing files:")
+        pad=str(max([len(x[0]) for x in missing]))
+        for f in missing:
+            print_stderr(("%-"+pad+"s (%s %7s)")%(f[0],humanize_date(f[3]),humanize_size(f[2])))
+    
+    print_stderr("----\nFile check summary:")
+    print_stderr("Database modified: %s"%(humanize_date(os.path.getmtime(options.sqlfile)),))
+    print_stderr("Checksum matches : %d"%(OK_count,))
+    print_stderr("Checksum mismatch: %d"%(len(differing),))
+    print_stderr("Files missing    : %d"%(len(missing),))
+
 def createdb(sqlfile):
     conn=sqlite3.connect(sqlfile)
     db=conn.cursor()
@@ -824,6 +872,11 @@ def disk_used(options):
                          humanize_size(entry[0]).rjust(8), 
                          entry[1]]))
 
+def print_stderr(s):
+    sys.stderr.write(s)
+    sys.stderr.write("\n")
+    sys.stderr.flush()
+    
 def print_structure(files):
     for hash in files:
         #print(hash[0])
@@ -1020,7 +1073,7 @@ def main():
     if options.delete_data:
         print('Deleting metadata...') 
         delete_data(options.sqlfile)
-    if options.search:
+    if options.search and not options.check:
         print_structure(searchdb(options.sqlfile,options.search))
     if options.measure:
         append_data(options)
@@ -1075,6 +1128,8 @@ def main():
     if options.export_descriptions:
         print("Export descriptions")
         export_descriptions(options)
+    if options.check:
+        checkdb(options)
     sys.exit(0)
 
 if __name__ == "__main__":