dvc: add basic 'dvc status'

(dvc) ➜  myrepo git:(master) ✗ dvc status
foo.dvc
        outs
                changed:  foo
foo2.dvc
        outs
                changed:  foo2
        deps
                changed:  foo1
foo1.dvc
        outs
                changed:  foo1
        deps
                changed:  foo

(dvc) ➜  myrepo git:(master) ✗ dvc status --cloud

        new:      c1/57a79031e1c40f85931829bc5fc552

Signed-off-by: Ruslan Kuprieiev <[email protected]>

dvc/cli.py

@@ -163,7 +163,12 @@ def parse_args(argv=None):
     status_parser = subparsers.add_parser(
                         'status',
                         parents=[parent_cache_parser],
-                        help='Show mismatches between local cache and cloud cache')
+                        help='Show the project status')
+    status_parser.add_argument('-c',
+                        '--cloud',
+                        action='store_true',
+                        default=False,
+                        help='Show status of a local cache compared to a cloud')
     status_parser.set_defaults(func=CmdDataStatus)
 
     # Repro

dvc/command/status.py

@@ -6,27 +6,30 @@ from dvc.command.data_sync import CmdDataBase
 
 
 class CmdDataStatus(CmdDataBase):
-    def _show(self, status):
-        for s in status:
-            target, ret = s
+    STATUS_LEN = 10
+    STATUS_INDENT = '\t'
 
-            if ret == cloud.STATUS_UNKNOWN or ret == cloud.STATUS_OK:
-                continue
+    def _normalize(self, s):
+        s+=':'
+        if len(s) >= self.STATUS_LEN:
+            return s
+        return s + (self.STATUS_LEN - len(s))*' '
 
-            prefix_map = {
-                cloud.STATUS_DELETED  : 'deleted: ',
-                cloud.STATUS_MODIFIED : 'modified:',
-                cloud.STATUS_NEW      : 'new file:',
-            }
+    def _show(self, status, indent=0):
+        ind = indent * self.STATUS_INDENT
 
-            path = os.path.relpath(target, self.project.cache.cache_dir)
-
-            self.project.logger.info('\t{}\t{}'.format(prefix_map[ret], path))
+        for key, value in status.items():
+            if isinstance(value, dict):
+                self.project.logger.info('{}{}'.format(ind, key))
+                self._show(value, indent+1)
+            else:
+                self.project.logger.info('{}{}{}'.format(ind, self._normalize(value), key))
 
     def do_run(self, target=None):
+        indent = 1 if self.args.cloud else 0
         try:
-            status = self.project.status(target=target, jobs=self.args.jobs)
-            self._show(status)
+            status = self.project.status(target=target, jobs=self.args.jobs, cloud=self.args.cloud)
+            self._show(status, indent)
         except Exception as exc:
             self.project.logger.error('Failed to obtain data status', exc)
             return 1

dvc/output.py

@@ -75,6 +75,12 @@ class Dependency(object):
     def changed(self):
         return self._changed_md5()
 
+    def status(self):
+        if self.changed():
+            #FIXME better msgs
+            return {self.rel_path: 'changed'}
+        return {}
+
     @staticmethod
     def is_dir_cache(cache):
         return cache.endswith(Output.MD5_DIR_SUFFIX)

dvc/project.py

@@ -2,6 +2,8 @@ import os
 import stat
 import networkx as nx
 
+import dvc.cloud.base as cloud
+
 from dvc.logger import Logger
 from dvc.exceptions import DvcException
 from dvc.stage import Stage, Output
@@ -209,8 +211,41 @@ class Project(object):
         self.checkout()
         return ret
 
-    def status(self, target=None, jobs=1):
-        return self.cloud.status(self._used_cache(target), jobs)
+    def _local_status(self, target=None):
+        status = {}
+
+        if target:
+            stages = [Stage.load(self, target)]
+        else:
+            stages = self.stages()
+
+        for stage in self.stages():
+            status.update(stage.status())
+
+        return status
+
+    def _cloud_status(self, target=None, jobs=1):
+        status = {}
+        for target, ret in self.cloud.status(self._used_cache(target), jobs):
+            if ret == cloud.STATUS_UNKNOWN or ret == cloud.STATUS_OK:
+                continue
+
+            prefix_map = {
+                cloud.STATUS_DELETED: 'deleted',
+                cloud.STATUS_MODIFIED: 'modified',
+                cloud.STATUS_NEW: 'new',
+            }
+
+            path = os.path.relpath(target, self.cache.cache_dir)
+
+            status[path] = prefix_map[ret]
+
+        return status
+
+    def status(self, target=None, jobs=1, cloud=False):
+        if cloud:
+            return self._cloud_status(target, jobs)
+        return self._local_status(target)
 
     def graph(self):
         G = nx.DiGraph()

dvc/stage.py

@@ -183,3 +183,24 @@ class Stage(object):
     def checkout(self):
         for out in self.outs:
             out.checkout()
+
+    def _status(self, entries, name):
+        ret = {}
+
+        for entry in entries:
+            ret.update(entry.status())
+
+        if ret:
+            return {name: ret}
+
+        return {}
+
+    def status(self):
+        ret = {}
+        ret.update(self._status(self.deps, 'deps'))
+        ret.update(self._status(self.outs, 'outs'))
+
+        if ret:
+            return {self.relpath: ret}
+
+        return {}