adding handling of archives

catcli/catcli.py

@@ -36,11 +36,11 @@ USAGE = """
 {0}
 
 Usage:
-    {1} index  [--catalog=<path>] [--meta=<meta>...] [-fcuV] <name> <path>
-    {1} ls     [--catalog=<path>] [-rVS] [<path>]
+    {1} index  [--catalog=<path>] [--meta=<meta>...] [-acfuV] <name> <path>
+    {1} ls     [--catalog=<path>] [-arVS] [<path>]
     {1} find   [--catalog=<path>] [-bV] <term>
     {1} rm     [--catalog=<path>] [-fV] <storage>
-    {1} tree   [--catalog=<path>] [-VS] [<path>]
+    {1} tree   [--catalog=<path>] [-aVS] [<path>]
     {1} rename [--catalog=<path>] [-fV] <storage> <name>
     {1} edit   [--catalog=<path>] [-fV] <storage>
     {1} graph  [--catalog=<path>] [-V] [<path>]
@@ -52,6 +52,7 @@ Options:
     --catalog=<path>    Path to the catalog [default: {2}].
     --meta=<meta>       Additional attribute to store [default: ].
     -u --subsize        Store size of folders [default: False].
+    -a --archive        Handle archive file [default: False].
     -f --force          Force overwrite [default: False].
     -b --script         Output script to manage found file(s) [default: False].
     -S --sortsize       Sort by size, largest first [default: False].
@@ -184,7 +185,8 @@ def main():
     banner()
 
     # init noder
-    noder = Noder(verbose=args['--verbose'], sortsize=args['--sortsize'])
+    noder = Noder(verbose=args['--verbose'], sortsize=args['--sortsize'],
+                  arc=args['--archive'])
     # init catalog
     catalog = Catalog(args['--catalog'], verbose=args['--verbose'],
                       force=args['--force'])

catcli/decomp.py

@@ -0,0 +1,52 @@
+"""
+author: deadc0de6 (https://github.com/deadc0de6)
+Copyright (c) 2017, deadc0de6
+
+Catcli generic compressed data lister
+"""
+
+import sys
+import os
+import tarfile
+import zipfile
+
+
+class Decomp:
+
+    def __init__(self):
+        self.ext = {
+            'tar': self._tar,
+            'tgz': self._tar,
+            'gz': self._tar,
+            'tar.gz': self._tar,
+            'xz': self._tar,
+            'tar.xz': self._tar,
+            'lzma': self._tar,
+            'tar.lzma': self._tar,
+            'tlz': self._tar,
+            'bz2': self._tar,
+            'tar.bz2': self._tar,
+            'zip': self._zip
+          }
+
+    def get_format(self):
+        return list(self.ext.keys())
+
+    def get_names(self, path):
+        ''' get tree of compressed archive '''
+        ext = os.path.splitext(path)[1][1:]
+        if ext in list(self.ext.keys()):
+            return self.ext[ext](path)
+        return None
+
+    def _tar(self, path):
+        if not tarfile.is_tarfile(path):
+            return None
+        tar = tarfile.open(path, "r")
+        return tar.getnames()
+
+    def _zip(self, path):
+        if not zipfile.is_zipfile(path):
+            return None
+        z = zipfile.ZipFile(path)
+        return z.namelist()

catcli/logger.py

@@ -56,6 +56,11 @@ class Logger:
         s += '{}{}{}'.format(Logger.GRAY, end, Logger.RESET)
         sys.stdout.write(s+'\n')
 
+    def arc(pre, name, archive):
+        s = '{}{}{}{}'.format(pre, Logger.YELLOW, name, Logger.RESET)
+        s += ' {}[archive:{}]{}'.format(Logger.GRAY, archive, Logger.RESET)
+        sys.stdout.write(s+'\n')
+
     ######################################################################
     # generic output
     ######################################################################

catcli/noder.py

@@ -14,6 +14,7 @@ import time
 from . import __version__ as VERSION
 import catcli.utils as utils
 from catcli.logger import Logger
+from catcli.decomp import Decomp
 
 '''
 There are 4 types of node:
@@ -31,13 +32,17 @@ class Noder:
     TYPE_TOP = 'top'  # tip top ;-)
     TYPE_FILE = 'file'
     TYPE_DIR = 'dir'
+    TYPE_ARC = 'arc'
     TYPE_STORAGE = 'storage'
     TYPE_META = 'meta'
 
-    def __init__(self, verbose=False, sortsize=False):
+    def __init__(self, verbose=False, sortsize=False, arc=False):
         self.hash = True
         self.verbose = verbose
         self.sortsize = sortsize
+        self.arc = arc
+        if self.arc:
+            self.decomp = Decomp()
 
     def set_hashing(self, val):
         self.hash = val
@@ -96,8 +101,15 @@ class Noder:
             md5 = utils.md5sum(path)
         relpath = os.path.join(os.path.basename(storagepath),
                                os.path.relpath(path, start=storagepath))
-        return self._node(name, self.TYPE_FILE, relpath, parent,
-                          size=st.st_size, md5=md5)
+
+        n = self._node(name, self.TYPE_FILE, relpath, parent,
+                       size=st.st_size, md5=md5)
+        if self.arc:
+            ext = os.path.splitext(path)[1][1:]
+            if ext in self.decomp.get_format():
+                names = self.decomp.get_names(path)
+                self.list_to_tree(n, names)
+        return n
 
     def dir_node(self, name, path, parent, storagepath):
         ''' create a new node representing a directory '''
@@ -114,6 +126,11 @@ class Noder:
         return anytree.AnyNode(name=name, type=self.TYPE_STORAGE, free=free,
                                total=total, parent=parent, attr=attr, ts=epoch)
 
+    def archive_node(self, name, path, parent, archive):
+        return anytree.AnyNode(name=name, type=self.TYPE_ARC, relpath=path,
+                               parent=parent, size=0, md5=None,
+                               archive=archive)
+
     def _node(self, name, type, relpath, parent, size=None, md5=None):
         ''' generic node creation '''
         return anytree.AnyNode(name=name, type=type, relpath=relpath,
@@ -160,6 +177,9 @@ class Noder:
             ht = utils.human(node.total)
             name = '{} (free:{}, total:{})'.format(node.name, hf, ht)
             Logger.storage(pre, name, node.attr)
+        elif node.type == self.TYPE_ARC:
+            if self.arc:
+                Logger.arc(pre, node.name, node.archive)
         else:
             Logger.err('Weird node encountered: {}'.format(node))
             # Logger.out('{}{}'.format(pre, node.name))
@@ -224,6 +244,36 @@ class Noder:
             pass
         return found
 
+    ###############################################################
+    # tree creationg
+    ###############################################################
+    def _add_entry(self, name, top, resolv):
+        ''' add an entry to the tree '''
+        entries = name.rstrip(os.sep).split(os.sep)
+        if len(entries) == 1:
+            self.archive_node(name, name, top, top.name)
+            return
+        sub = os.sep.join(entries[:-1])
+        f = entries[-1]
+        try:
+            parent = resolv.get(top, sub)
+            parent = self.archive_node(f, name, parent, top.name)
+        except anytree.resolver.ChildResolverError:
+            self.archive_node(f, name, top, top.name)
+
+    def list_to_tree(self, parent, names):
+        ''' convert list of files to a tree '''
+        if not names:
+            return
+        r = anytree.resolver.Resolver('name')
+        # dirty trick for zip lists
+        first = names[0].split(os.sep)
+        if len(first) > 1:
+            self.archive_node(first[0], first[0], parent, parent.name)
+
+        for name in names:
+            self._add_entry(name, parent, r)
+
     ###############################################################
     # diverse
     ###############################################################