From 2a4e3b9ca03e5d751ce51549408c774e71defcb6 Mon Sep 17 00:00:00 2001
From: Milind Changire <mchangir@redhat.com>
Date: Fri, 3 Jul 2015 12:02:11 +0530
Subject: [PATCH 202/212] tools/glusterfind: RENAME and MODIFY issues

If Modification happens before RENAME, GFID to Path Conversion
converts it into New Path. Delete Modify Entry and insert again
So that MODIFY <NEW NAME> comes after RENAME.

Default value of pgfids and basenames changed to "" instead of NULL
Also fixed RENAME issue of displaying "RENAME <NEW NAME> <NEW NAME>".
Also fixed RENAME followed by missing MODIFY

Change-Id: Ie67219cc049a427e95ab6426bb890b02ffc1a853
BUG: 1228247
Reviewed-On: http://review.gluster.org/#/c/11443/
Signed-off-by: Milind Changire <mchangir@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/52285
Reviewed-by: Kotresh Hiremath Ravishankar <khiremat@redhat.com>
Reviewed-by: Venky Shankar <vshankar@redhat.com>
Tested-by: Venky Shankar <vshankar@redhat.com>
---
 tools/glusterfind/src/changelogdata.py |   39 +++++++++++++++++++++++---------
 tools/glusterfind/src/main.py          |    3 +-
 2 files changed, 30 insertions(+), 12 deletions(-)

diff --git a/tools/glusterfind/src/changelogdata.py b/tools/glusterfind/src/changelogdata.py
index c42aa2a..08d711b 100644
--- a/tools/glusterfind/src/changelogdata.py
+++ b/tools/glusterfind/src/changelogdata.py
@@ -100,10 +100,10 @@ class ChangelogData(object):
             ts     VARCHAR,
             type   VARCHAR,
             gfid   VARCHAR(40),
-            pgfid1 VARCHAR(40),
-            bn1    VARCHAR(500),
-            pgfid2 VARCHAR(40),
-            bn2    VARCHAR(500),
+            pgfid1 VARCHAR(40) DEFAULT '',
+            bn1    VARCHAR(500) DEFAULT '',
+            pgfid2 VARCHAR(40) DEFAULT '',
+            bn2    VARCHAR(500) DEFAULT '',
             path1  VARCHAR DEFAULT '',
             path2  VARCHAR DEFAULT ''
         )
@@ -283,7 +283,7 @@ class ChangelogData(object):
 
     def append_path1(self, path, inode):
         # || is for concatenate in SQL
-        query = """UPDATE gfidpath SET path1 = ',' || ?
+        query = """UPDATE gfidpath SET path1 = path1 || ',' || ?
         WHERE gfid IN (SELECT gfid FROM inodegfid WHERE inode = ?)"""
         self.cursor.execute(query, (path, inode))
 
@@ -344,16 +344,32 @@ class ChangelogData(object):
                                   "pgfid1": pgfid1, "bn1": bn1})
         elif self.gfidpath_exists({"gfid": data[1], "type": "RENAME",
                                    "pgfid2": pgfid1, "bn2": bn1}):
-            # If <OLD_PGFID>/<BNAME> is same as <PGFID2>/<BN2>(may be previous
-            # RENAME) then UPDATE <NEW_PGFID>/<BNAME> as <PGFID2>/<BN2>
-            self.gfidpath_update({"pgfid2": pgfid2, "bn2": bn2},
-                                 {"gfid": data[1], "type": "RENAME",
-                                 "pgfid2": pgfid1, "bn2": bn1})
+            # If we are renaming file back to original name then just
+            # delete the entry since it will effectively be a no-op
+            if self.gfidpath_exists({"gfid": data[1], "type": "RENAME",
+                                     "pgfid2": pgfid1, "bn2": bn1,
+                                     "pgfid1": pgfid2, "bn1": bn2}):
+                self.gfidpath_delete({"gfid": data[1], "type": "RENAME",
+                                      "pgfid2": pgfid1, "bn2": bn1})
+            else:
+                # If <OLD_PGFID>/<BNAME> is same as <PGFID2>/<BN2>
+                # (may be previous RENAME)
+                # then UPDATE <NEW_PGFID>/<BNAME> as <PGFID2>/<BN2>
+                self.gfidpath_update({"pgfid2": pgfid2, "bn2": bn2},
+                                     {"gfid": data[1], "type": "RENAME",
+                                      "pgfid2": pgfid1, "bn2": bn1})
         else:
             # Else insert as RENAME
             self.gfidpath_add(changelogfile, RecordType.RENAME, data[1],
                               pgfid1, bn1, pgfid2, bn2)
 
+        if self.gfidpath_exists({"gfid": data[1], "type": "MODIFY"}):
+            # If MODIFY exists already for that GFID, remove it and insert
+            # again so that MODIFY entry comes after RENAME entry
+            # Output will have MODIFY <NEWNAME>
+            self.gfidpath_delete({"gfid": data[1], "type": "MODIFY"})
+            self.gfidpath_add(changelogfile, RecordType.MODIFY, data[1])
+
     def when_link_symlink(self, changelogfile, data):
         # E <GFID> <LINK|SYMLINK> <PGFID>/<BASENAME>
         # Add as New record in Db as Type NEW
@@ -366,7 +382,8 @@ class ChangelogData(object):
 
     def when_data_meta(self, changelogfile, data):
         # If GFID row exists, Ignore else Add to Db
-        if not self.gfidpath_exists({"gfid": data[1]}):
+        if not self.gfidpath_exists({"gfid": data[1], "type": "NEW"}) and \
+           not self.gfidpath_exists({"gfid": data[1], "type": "MODIFY"}):
             self.gfidpath_add(changelogfile, RecordType.MODIFY, data[1])
 
     def when_unlink_rmdir(self, changelogfile, data):
diff --git a/tools/glusterfind/src/main.py b/tools/glusterfind/src/main.py
index 8d32295..9a6a2ac 100644
--- a/tools/glusterfind/src/main.py
+++ b/tools/glusterfind/src/main.py
@@ -433,7 +433,8 @@ def mode_pre(session_dir, args):
                 # Multiple paths in case of Hardlinks
                 paths = row[1].split(",")
                 for p in paths:
-                    if p == "":
+                    if p == "" or p.replace("%2F%2F","%2F") == \
+                       row[2].replace("%2F%2F","%2F"):
                         continue
                     f.write("%s %s %s\n" % (row[0], p, row[2]))
 
-- 
1.7.1