From 02add3c977f2efcc15ddc6b785d52ade934f5b6a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jarno=20Sepp=C3=A4nen?= <jseppanen@iki.fi>
Date: Wed, 7 May 2014 22:59:54 +0300
Subject: [PATCH] Store metadata in an sqlite database

---
 gpgfs.py    | 505 +++++++++++++++++++++++-----------------------------
 gpgstore.py |   3 +
 2 files changed, 225 insertions(+), 283 deletions(-)

diff --git a/gpgfs.py b/gpgfs.py
index ec2a262..3a9ded6 100755
--- a/gpgfs.py
+++ b/gpgfs.py
@@ -6,12 +6,10 @@ import stat
 import os
 import sys
 import logging
-import struct
 import time
-from cStringIO import StringIO
 import gpgstore
-
-magic = 'GPGFS1\n'
+import sqlite3
+from contextlib import contextmanager
 
 log = logging.getLogger('gpgfs')
 
@@ -23,82 +21,6 @@ class Entry:
         for k,v in kwargs.iteritems():
             setattr(self, k, v)
 
-# entry types:
-ENT_FILE = 0
-ENT_DIR = 1
-
-def read_index(store, path):
-    data = store.get(path)
-    buf = StringIO(data)
-    if buf.read(len(magic)) != magic:
-        raise IOError, 'index parse error: %s' % path
-    read_atom(buf)
-    root = Entry(**read_dict(buf))
-    return root
-
-def write_index(store, path, root):
-    buf = StringIO()
-    buf.write(magic)
-    header = ''
-    write_atom(buf, header)
-    write_dict(buf, root)
-    store.put(buf.getvalue(), path=path)
-
-def write_dict(fd, dct):
-    # breadth-first
-    children = []
-    buf = StringIO()
-    if not isinstance(dct, dict):
-        dct = dct.__dict__
-    for key in dct:
-        write_atom(buf, key.encode('utf8'))
-        val = dct[key]
-        if isinstance(val, dict):
-            buf.write('D')
-            children.append(val)
-        elif isinstance(val, Entry):
-            buf.write('E')
-            children.append(val)
-        elif isinstance(val, (int, long)):
-            buf.write('I')
-            buf.write(struct.pack('<I', val))
-        elif isinstance(val, str):
-            buf.write('S')
-            write_atom(buf, val)
-        elif isinstance(val, unicode):
-            buf.write('U')
-            write_atom(buf, val.encode('utf8'))
-        else:
-            raise TypeError, type(val)
-    write_atom(fd, buf.getvalue())
-    for c in children:
-        write_dict(fd, c)
-
-def read_dict(fd):
-    dct = {}
-    buf = read_atom(fd)
-    buflen = len(buf)
-    buf = StringIO(buf)
-    while buf.tell() < buflen:
-        key = read_atom(buf).decode('utf8')
-        tag = buf.read(1)
-        if tag == 'D':    val = read_dict(fd)
-        elif tag == 'E':  val = Entry(**read_dict(fd))
-        elif tag == 'I':  val = struct.unpack('<I', buf.read(4))[0]
-        elif tag == 'S':  val = read_atom(buf)
-        elif tag == 'U':  val = read_atom(buf).decode('utf8')
-        else:             raise TypeError, tag
-        dct[key] = val
-    return dct
-
-def write_atom(fd, atom):
-    assert isinstance(atom, str)
-    fd.write(struct.pack('<I', len(atom)))
-    fd.write(atom)
-
-def read_atom(fd):
-    return fd.read(struct.unpack('<I', fd.read(4))[0])
-
 class LoggingMixIn:
 
     def __call__(self, op, path, *args):
@@ -120,50 +42,80 @@ class LoggingMixIn:
                 rtxt = rtxt[:10]
             log.debug('<- %s %s', op, rtxt)
 
+@contextmanager
+def transaction(cur, active=True):
+    if not active:
+        yield
+        return
+    cur.execute('BEGIN EXCLUSIVE')
+    try:
+        yield
+    except:
+        cur.execute('ROLLBACK')
+        raise
+    try:
+        cur.execute('COMMIT')
+    except sqlite3.OperationalError:
+        log.exception("transaction failed")
+        raise FuseOSError(errno.EIO)
+
 class GpgFs(LoggingMixIn, Operations):
 #class GpgFs(Operations):
 
-    def __init__(self, encroot, keyid):
+    def __init__(self, encroot, mountpoint, keyid):
         '''
         :param encroot: Encrypted root directory
         '''
         self.encroot = encroot.rstrip('/')
         assert os.path.exists(self.encroot)
         assert os.path.isdir(self.encroot)
-        #self.cache = cache
         self.store = gpgstore.GpgStore(self.encroot, keyid)
         self.index_path = 'index'
-        if os.path.exists(self.encroot + '/' + self.index_path):
-            self.root = read_index(self.store, self.index_path)
-        else:
-            self.root = Entry(type=ENT_DIR, children={},
-                              st_mode=0755,
-                              st_mtime=int(time.time()),
-                              st_ctime=int(time.time()))
-            self._write_index()
-            log.info('created %s', self.index_path)
+        self.dbpath = '.gpgfs.db'
+        self.mountpoint = mountpoint
         self.fd = 0
         self._clear_write_cache()
 
-    def _write_index(self):
-        write_index(self.store, self.index_path, self.root)
-
-    def _find(self, path, parent=False):
+    def _find(self, path, parent=False, **kwargs):
         assert path.startswith('/')
-        if path == '/':
-            return self.root
-        node = self.root
-        path = path[1:].split('/')
+        names = path[1:].split('/')
         if parent:
-            basename = path[-1]
-            path = path[:-1]
-        for name in path:
-            if name not in node.children:
-                raise FuseOSError(errno.ENOENT)
-            node = node.children[name]
+            basename = names[-1]
+            path = names[:-1]
+        sql = 'JOIN entry e{i} ON e{i}.parent_id=e{j}.id AND e{i}.name=?'
+        joins = [{i:i+1, j:i} for i in range(len(names))]
+        joins = '\n'.join(sql.format(j) for j in joins)
+        sql = """
+          SELECT e{i}.* FROM entry e0
+          {joins}
+          WHERE e0.name='' AND e0.parent_id=0
+        """.format(joins=joins, i=len(names)-1)
+        cur = self.db.execute(sql, names)
+        if cur.rowcount != 1:
+            if 'default' in kwargs:
+                return kwargs['default']
+            raise FuseOSError(errno.ENOENT)
+        ent = cur.fetchone()
         if parent:
-            return node, basename
-        return node
+            return ent, basename
+        return ent
+
+    def _put(self, path, data, transaction_=True):
+        if path[1:] == self.dbpath:
+            encpath = self.store.put(data, self.index_path)
+        else:
+            encpath = self.store.put(data)
+        with transaction(self.db, transaction_):
+            try:
+                ent = self._find(path)
+                sql = "UPDATE entry SET size=?, encpath=?, mtime=? WHERE id=?"
+                self.db.execute(sql, [len(data), encpath, time.time(), ent.id])
+            except:
+                self.store.delete(encpath)
+                raise
+        if ent.encpath != None:
+            self.store.delete(ent.encpath)
+        return encpath
 
     def _clear_write_cache(self):
         self.write_path = None
@@ -171,45 +123,82 @@ class GpgFs(LoggingMixIn, Operations):
         self.write_len = 0
         self.write_dirty = False
 
+    def _init_db(self, db):
+        sql = """
+          CREATE TABLE entry (
+            id INT PRIMARY KEY,
+            name TEXT NOT NULL,
+            parent_id INT NOT NULL,
+            encpath TEXT UNIQUE,
+            mode INT NOT NULL,
+            nlink INT,
+            size INT,
+            mtime FLOAT,
+            ctime FLOAT,
+            UNIQUE (name, parent_id),
+            FOREIGN KEY(parent_id) REFERENCES entry(id)
+          )"""
+        db.execute(sql)
+        db.execute('BEGIN EXCLUSIVE')
+        sql = """
+          INSERT INTO entry (id, name, parent_id, mode,
+                             nlink, size, mtime, ctime)
+          VALUES (?,?,?,?,?,?,?,?)"""
+        now = time.time()
+        db.execute(sql, [0, '', 0, stat.S_IFDIR | 0755,
+                         3, 0, now, now])
+        db.execute('COMMIT')
+
+    def init(self, path):
+        init = not self.store.exists(self.index_path)
+        path = self.mountpoint + '/' + self.dbpath
+        log.debug('opening %s', path)
+        self.dbconn = sqlite3.connect(path, isolation_level=None)
+        self.dbconn.row_factory = sqlite3.Row
+        self.db = self.dbconn.cursor()
+        if init:
+            self._init_db(self.db)
+            log.info('created %s', path)
+
+    def destroy(self, path):
+        self.db.close()
+
+    def access(self, path, amode):
+        self._find(path)
+        return 0
+
     def chmod(self, path, mode):
         # sanitize mode (clear setuid/gid/sticky bits)
         mode &= 0777
-        ent = self._find(path)
-        if ent.type == ENT_DIR:
-            prev_mode = ent.st_mode
-            ent.st_mode = mode
-            try:
-                self._write_index()
-            except:
-                ent.st_mode = prev_mode
-                raise
-        else:
-            encpath = self.encroot + '/' + ent.path
-            os.chmod(encpath, mode)
+        with transaction(self.db):
+            ent = self._find(path)
+            mode |= (ent.mode & 0170000)
+            self.db.execute('UPDATE entry SET mode=? WHERE id=?', [mode, ent.id])
+            if not self.db.rowcount:
+                raise FuseOSError(errno.ENOENT)
 
     def chown(self, path, uid, gid):
         raise FuseOSError(errno.ENOSYS)
 
     def create(self, path, mode):
-        dir, path = self._find(path, parent=True)
-        if path in dir.children:
-            raise FuseOSError(errno.EEXIST)
-        # FIXME mode
-        encpath = self.store.put('')
-        prev_mtime = dir.st_mtime
-        dir.children[path] = Entry(type=ENT_FILE, path=encpath, st_size=0)
-        log.debug('new path %s => %s', path, encpath)
-        dir.st_mtime = int(time.time())
-        try:
-            self._write_index()
-        except:
-            try: self.store.delete(encpath)
-            except: pass
-            del dir.children[path]
-            dir.st_mtime = prev_mtime
-            raise
-        self.fd += 1
-        return self.fd
+        mode &= 0777
+        mode |= stat.S_IFREG
+        with transaction(self.db):
+            parent, name = self._find(path, parent=True)
+            sql = """
+              INSERT INTO entry (name, parent_id, mode, nlink, ctime)
+              VALUES (?,?,?,?,?)
+            """
+            now = time.time()
+            try:
+                self.db.execute(sql, [name, parent.id, mode, 1, now])
+            except sqlite3.IntegrityError:
+                raise FuseOSError(errno.EEXIST)
+            self._put(path, '', transaction_=False)
+            sql = "UPDATE entry SET mtime=? WHERE id=?"
+            self.db.execute(sql, [now, parent.id])
+            self.fd += 1
+            return self.fd
 
     def flush(self, path, fh):
         if not self.write_dirty:
@@ -217,19 +206,7 @@ class GpgFs(LoggingMixIn, Operations):
             return 0
         buf = ''.join(self.write_buf)
         self.write_buf = [buf]
-        ent = self._find(self.write_path)
-        prev_size = ent.st_size
-        prev_path = ent.path
-        ent.st_size = len(buf)
-        ent.path = self.store.put(buf)
-        try:
-            self._write_index()
-        except:
-            self.store.delete(ent.path)
-            ent.st_size = prev_size
-            ent.path = prev_path
-            raise
-        self.store.delete(prev_path)
+        self._put(self.write_path, buf)
         self.write_dirty = False
         log.debug('flushed %d bytes to %s', len(buf), self.write_path)
         return 0
@@ -239,19 +216,11 @@ class GpgFs(LoggingMixIn, Operations):
         return 0
 
     def getattr(self, path, fh = None):
-        ent = self._find(path)
-        if ent.type == ENT_DIR:
-            return dict(st_mode = stat.S_IFDIR | ent.st_mode,
-                        st_size = len(ent.children),
-                        st_ctime = ent.st_ctime, st_mtime = ent.st_mtime,
-                        st_atime = 0, st_nlink = 3)
-        # ensure st_size is up-to-date
-        self.flush(path, 0)
-        encpath = self.encroot + '/' + ent.path
-        s = os.stat(encpath)
-        return dict(st_mode = s.st_mode, st_size = ent.st_size,
-                    st_atime = s.st_atime, st_mtime = s.st_mtime,
-                    st_ctime = s.st_ctime, st_nlink = s.st_nlink)
+        with transaction(self.db):
+            ent = self._find(path)
+            return dict(st_mode = ent.mode, st_size = ent.size,
+                        st_ctime = ent.ctime, st_mtime = ent.mtime,
+                        st_atime = 0, st_nlink = ent.nlink)
 
     def getxattr(self, path, name, position = 0):
         raise FuseOSError(errno.ENODATA) # ENOATTR
@@ -260,21 +229,23 @@ class GpgFs(LoggingMixIn, Operations):
         return []
 
     def mkdir(self, path, mode):
-        dir, path = self._find(path, parent=True)
-        if path in dir.children:
-            raise FuseOSError(errno.EEXIST)
-        prev_mtime = dir.st_mtime
-        dir.children[path] = Entry(type=ENT_DIR, children={},
-                                   st_mode=(mode & 0777),
-                                   st_mtime=int(time.time()),
-                                   st_ctime=int(time.time()))
-        dir.st_mtime = int(time.time())
-        try:
-            self._write_index()
-        except:
-            del dir.children[path]
-            dir.st_mtime = prev_mtime
-            raise
+        mode &= 0777
+        mode |= stat.S_IFDIR
+        with transaction(self.db):
+            parent, name = self._find(path, parent=True)
+            sql = """
+              INSERT INTO entry
+                (name, type, parent_id, mode, nlink, size, mtime, ctime)
+              VALUES (?,?,?,?,?,?,?,?)
+            """
+            now = time.time()
+            try:
+                self.db.execute(sql, [name, parent.id,
+                                      mode, 2, 0, now, now])
+            except sqlite3.IntegrityError:
+                raise FuseOSError(errno.EEXIST)
+            sql = "UPDATE entry SET mtime=? WHERE id=?"
+            self.db.execute(sql, [now, parent.id])
 
     def open(self, path, flags):
         return 0
@@ -282,13 +253,18 @@ class GpgFs(LoggingMixIn, Operations):
     def read(self, path, size, offset, fh):
         self.flush(path, 0)
         ent = self._find(path)
-        assert ent.type == ENT_FILE
-        data = self.store.get(ent.path)
+        assert ent.mode & stat.S_IFREG
+        try:
+            data = self.store.get(ent.encpath)
+        except IOError:
+            raise FuseOSError(errno.ENOENT)
         return data[offset:offset + size]
 
     def readdir(self, path, fh):
-        dir = self._find(path)
-        return ['.', '..'] + list(dir.children)
+        dirent = self._find(path)
+        sql = "SELECT name FROM entry WHERE parent_id=?"
+        self.db.execute(sql, [dirent.id])
+        return ['.', '..'] + [name for name, in self.db]
 
     def readlink(self, path):
         raise FuseOSError(errno.ENOSYS)
@@ -301,53 +277,43 @@ class GpgFs(LoggingMixIn, Operations):
         self._clear_write_cache()
         if new.startswith(old):
             raise FuseOSError(errno.EINVAL)
-        old_dir, old_name = self._find(old, parent=True)
-        if old_name not in old_dir.children:
-            raise FuseOSError(errno.ENOENT)
-        new_dir, new_name = self._find(new, parent=True)
-        prev_ent = new_dir.children.get(new_name)
-        if prev_ent:
-            if prev_ent.type == ENT_DIR:
-                if old_dir[old_name].type != ENT_DIR:
-                    raise FuseOSError(errno.EISDIR)
-                if prev_ent.children:
-                    raise FuseOSError(errno.ENOTEMPTY)
-            elif old_dir[old_name].type == ENT_DIR:
-                raise FuseOSError(errno.ENOTDIR)
-        prev_old_mtime = old_dir.st_mtime
-        prev_new_mtime = new_dir.st_mtime
-        new_dir.children[new_name] = old_dir.children.pop(old_name)
-        old_dir.st_mtime = new_dir.st_mtime = int(time.time())
-        try:
-            self._write_index()
-        except:
-            old_dir.children[old_name] = new_dir.children.pop(new_name)
-            if prev_ent:
-                new_dir.children[new_name] = prev_ent
-            old_dir.st_mtime = prev_old_mtime
-            new_dir.st_mtime = prev_new_mtime
-            raise
-        if prev_ent and prev_ent.type == ENT_FILE:
-            os.remove(self.encroot + '/' + prev_ent.path)
+        with transaction(self.db):
+            old_ent = self._find(old)
+            new_ent = self._find(new, default=None)
+            old_parent, old_name = self._find(old, parent=True)
+            new_parent, new_name = self._find(new, parent=True)
+            if new_ent != None:
+                if new_ent.mode & stat.S_IFDIR:
+                    if not old_ent.mode & stat.S_IFDIR:
+                        raise FuseOSError(errno.EISDIR)
+                    sql = "SELECT COUNT(*) FROM entry WHERE parent_id=?"
+                    self.db.execute(sql, [new_ent.id])
+                    if self.db.fetchone()[0]:
+                        raise FuseOSError(errno.ENOTEMPTY)
+                elif old_ent.mode & stat.S_IFDIR:
+                    raise FuseOSError(errno.ENOTDIR)
+                sql = "DELETE FROM entry WHERE id=?"
+                self.db.execute(sql, [new_ent.id])
+            sql = "UPDATE entry SET parent_id=? WHERE id=?"
+            self.db.execute(sql, [new_parent.id, old_ent.id])
+            sql = "UPDATE entry SET mtime=? WHERE id IN (?,?)"
+            self.db.execute(sql, [time.time(), old_parent.id, new_parent.id])
+        if new_ent != None and new_ent.mode & stat.S_IFREG:
+            self.store.delete(new_ent.encpath)
 
     def rmdir(self, path):
-        parent, path = self._find(path, parent=True)
-        if path not in parent.children:
-            raise FuseOSError(errno.ENOENT)
-        ent = parent.children[path]
-        if ent.type != ENT_DIR:
-            raise FuseOSError(errno.ENOTDIR)
-        if ent.children:
-            raise FuseOSError(errno.ENOTEMPTY)
-        prev_mtime = parent.st_mtime
-        del parent.children[path]
-        parent.st_mtime = int(time.time())
-        try:
-            self._write_index()
-        except:
-            parent.children[path] = ent
-            parent.st_mtime = prev_mtime
-            raise
+        with transaction(self.db):
+            ent = self._find(path)
+            if not ent.mode & stat.S_IFDIR:
+                raise FuseOSError(errno.ENOTDIR)
+            sql = "SELECT COUNT(*) FROM entry WHERE parent_id=?"
+            self.db.execute(sql, [ent.id])
+            if self.db.fetchone()[0]:
+                raise FuseOSError(errno.ENOTEMPTY)
+            sql = "DELETE FROM entry WHERE id=?"
+            self.db.execute(sql, [ent.id])
+            sql = "UPDATE entry SET mtime=? WHERE id=?"
+            self.db.execute(sql, [time.time(), ent.parent_id])
 
     def setxattr(self, path, name, value, options, position = 0):
         raise FuseOSError(errno.ENOSYS)
@@ -361,69 +327,42 @@ class GpgFs(LoggingMixIn, Operations):
     def truncate(self, path, length, fh = None):
         self.flush(path, 0)
         self._clear_write_cache()
-        ent = self._find(path)
-        if length == 0:
-            buf = ''
-        else:
-            buf = self.store.get(ent.path)
-            buf = buf[:length]
-        prev_size = ent.st_size
-        prev_path = ent.path
-        ent.st_size = length
-        ent.path = self.store.put(buf)
-        try:
-            self._write_index()
-        except:
-            os.remove(ent.path)
-            ent.st_size = prev_size
-            ent.path = prev_path
-            raise
-        self.store.delete(prev_path)
+        with transaction(self.db):
+            ent = self._find(path)
+            if length == 0:
+                buf = ''
+            else:
+                buf = self.store.get(ent.encpath)
+                buf = buf[:length]
+            self._put(path, buf, transaction_=False)
 
     def unlink(self, path):
-        if self.write_path == path:
-            # no need to flush afterwards
-            self._clear_write_cache()
-        dir, name = self._find(path, parent=True)
-        if name not in dir.children:
-            raise FuseOSError(errno.ENOENT)
-        ent = dir.children[name]
-        encpath = self.encroot + '/' + ent.path
-        del dir.children[name]
-        prev_mtime = dir.st_mtime
-        dir.st_mtime = int(time.time())
-        try:
-            self._write_index()
-        except:
-            dir.children[name] = ent
-            dir.st_mtime = prev_mtime
-            raise
-        os.remove(encpath)
+        with transaction(self.db):
+            if self.write_path == path:
+                # no need to flush afterwards
+                self._clear_write_cache()
+            ent = self._find(path)
+            sql = "DELETE FROM entry WHERE id=?"
+            self.db.execute(sql, [ent.id])
+            sql = "UPDATE entry SET mtime=? WHERE id=?"
+            self.db.execute(sql, [time.time(), ent.parent_id])
+        self.store.delete(ent.encpath)
 
     def utimens(self, path, times = None):
-        ent = self._find(path)
-        if ent.type == ENT_DIR:
-            prev_mtime = ent.st_mtime
-            if times is None:
-                ent.st_mtime = int(time.time())
-            else:
-                ent.st_mtime = times[1]
-            try:
-                self._write_index()
-            except:
-                ent.st_mtime = prev_mtime
-                raise
+        if times is None:
+            mtime = time.time()
         else:
-            # flush may mess with mtime
-            self.flush(path, 0)
-            encpath = self.encroot + '/' + ent.path
-            os.utime(encpath, times)
+            mtime = times[1]
+        with transaction(self.db):
+            ent = self._find(path)
+            sql = "UPDATE entry SET mtime=? WHERE id=?"
+            self.db.execute(sql, [mtime, ent.id])
 
     def write(self, path, data, offset, fh):
-        ent = self._find(path)
         if path != self.write_path:
             self.flush(self.write_path, None)
-            buf = self.store.get(ent.path)
+            ent = self._find(path)
+            buf = self.store.get(ent.encpath)
             self.write_buf = [buf]
             self.write_len = len(buf)
         self.write_path = path
@@ -445,5 +384,5 @@ if __name__ == '__main__':
     logpath = os.path.join(os.path.dirname(__file__), 'gpgfs.log')
     log.addHandler(logging.FileHandler(logpath, 'w'))
     log.setLevel(logging.DEBUG)
-    fs = GpgFs(sys.argv[2], sys.argv[1])
+    fs = GpgFs(sys.argv[2], sys.argv[3], sys.argv[1])
     FUSE(fs, sys.argv[3], foreground=True)
diff --git a/gpgstore.py b/gpgstore.py
index 568606a..cb2304d 100644
--- a/gpgstore.py
+++ b/gpgstore.py
@@ -60,3 +60,6 @@ class GpgStore(object):
     def delete(self, path):
         os.remove(self.encroot + '/' + path)
         log.debug('deleted %s' % path)
+
+    def exists(self, path):
+        return os.path.exists(self.encroot + '/' + path)