Merge pull request #419 from efiop/master

Remove ml-specific code and fix gitignore

dvc/cli.py

@@ -5,7 +5,6 @@ import sys
 import argparse
 from multiprocessing import cpu_count
 
-from dvc.command.find import CmdFind
 from dvc.command.fsck import CmdFsck
 from dvc.command.init import CmdInit
 from dvc.command.remove import CmdRemove
@@ -38,17 +37,6 @@ def parse_args(argv=None):
                         default=False,
                         help='Be verbose.')
 
-    parent_parser.add_argument(
-                        '-b',
-                        '--branch',
-                        metavar='BRANCH',
-                        help='Execute a command in the branch.')
-    parent_parser.add_argument(
-                        '-n',
-                        '--new-branch',
-                        metavar='BRANCH',
-                        help='Create a new branch and execute a command in the branch.')
-
     # Main parser
     desc = 'Data Version Control'
     parser = argparse.ArgumentParser(
@@ -272,33 +260,7 @@ def parse_args(argv=None):
                         help='Checkout')
     checkout_parser.set_defaults(func=CmdCheckout)
 
-    # Find
-    find_parser = subparsers.add_parser(
-                        'find',
-                        parents=[parent_parser],
-                        help='Find branch name')
-    find_parser.set_defaults(func=CmdFind)
-    find_parser.add_argument(
-                        'branch_name',
-                        nargs='?',
-                        help='Branch name regexp.')
-    find_parser.add_argument(
-                        'target',
-                        nargs='?',
-                        help='Target metric file.')
-    find_parser.add_argument(
-                        '-c',
-                        '--criteria',
-                        help='Search criteria. By default it finds max.',
-                        choices=['max', 'min', 'all'],
-                        default='max')
-    find_parser.add_argument(
-                        '-s',
-                        '--show-value',
-                        action='store_true',
-                        default=False,
-                        help='Show metrics value.')
-
+    # Fsck
     fsck_parser = subparsers.add_parser(
                         'fsck',
                         parents=[parent_parser],

dvc/command/common/base.py

@@ -27,7 +27,6 @@ class CmdBase(object):
 
     def run_cmd(self):
         with self.project.lock:
-            with self.project.scm.brancher(self.args.branch, self.args.new_branch):
                 return self.run()
 
     # Abstract methods that have to be implemented by any inheritance class

dvc/command/common/branch_changer.py

@@ -1,31 +0,0 @@
-from dvc.command.common.common_error import CmdCommonError
-
-
-class BranchChangerError(CmdCommonError):
-    def __init__(self, branch, msg):
-        super(CmdCommonError, self).__init__('Error in changing branch \'{}\': {}'.format(branch, msg))
-
-
-class BranchChanger(object):
-    def __init__(self, scm, branch, new_branch):
-        if branch and new_branch:
-            raise BranchChangerError("Commands conflict: --branch and --new-branch cannot be used at the same command")
-
-        self.perform_action = branch or new_branch
-
-        self.branch = branch if branch else new_branch
-        self.create_new = new_branch is not None
-        self.scm = scm
-
-    def __enter__(self):
-        if self.perform_action:
-            code, _, err = self.scm.checkout(self.branch, self.create_new)
-            if code != 0:
-                raise BranchChangerError(self.branch, err)
-        return self
-
-    def __exit__(self, type, value, traceback):
-        if self.perform_action:
-            code, _, err = self.scm.checkout('-')
-            if code != 0:
-                raise BranchChangerError(self.branch, err)

dvc/command/find.py

@@ -1,61 +0,0 @@
-from dvc import utils
-from dvc.command.common.base import CmdBase
-from dvc.executor import ExecutorError
-from dvc.logger import Logger
-
-
-class CmdFind(CmdBase):
-    def __init__(self, settings):
-        super(CmdFind, self).__init__(settings)
-
-    def run(self):
-        fname = self.parsed_args.target
-        branch_name = self.parsed_args.branch_name
-
-        branches = self.git.branches(branch_name)
-        metrics = [self.read_metrics(fname, b) for b in branches]
-
-        zipped = zip(metrics, branches)
-        res = self.filter(zipped, self.parsed_args.criteria)
-        res = filter(lambda x: x[0] is not None, res)
-
-        for (val, branch) in res:
-            if self.parsed_args.show_value:
-                print('{}: {}'.format(branch, val))
-            else:
-                print(branch)
-        return 0
-
-    @staticmethod
-    def filter(zipped, criteria):
-        if criteria == 'max':
-            res = [max(zipped, key=lambda t: t[0])]
-        elif criteria == 'min':
-            res = [min(zipped, key=lambda t: t[0])]
-        else:
-            res = zipped
-        return res
-
-    def read_metrics(self, fname, branch):
-        try:
-            lines = self.git.get_file_content(fname, branch).split('\n')
-        except ExecutorError as ex:
-            msg = 'Unable to read metrics file from branch {}: {}'
-            data_item = self.settings.path_factory.data_item(fname)
-
-            try:
-                self.git.get_file_content(data_item.state.relative, branch)
-                Logger.error(msg.format(branch, 'this is data file, not metric file'))
-            except ExecutorError:
-                Logger.error(msg.format(branch, 'file does not exist in this branch'))
-                return None
-
-            return None
-
-        metric = utils.parse_target_metric(lines)
-        if not metric:
-            msg = 'Unable to parse metrics from the first line of file {} in branch {}'
-            Logger.error(msg.format(fname, branch))
-            return None
-
-        return metric

dvc/project.py

@@ -109,7 +109,6 @@ class Project(object):
 
         ignore_file = os.path.join(dvc_dir, scm.ignore_file())
         scm.add([config.config_file, ignore_file])
-        scm.commit('DVC init')
 
         return Project(root_dir)
 

dvc/scm.py

@@ -2,7 +2,6 @@ import os
 import git
 
 from dvc.exceptions import DvcException
-from dvc.command.common.branch_changer import BranchChanger
 
 
 class SCMError(DvcException):
@@ -45,9 +44,6 @@ class Base(object):
     def branch(self, branch):
         pass
 
-    def brancher(self, branch, new_branch):
-        return BranchChanger(self, branch, new_branch)
-
     def untracked_files(self):
         pass
 
@@ -84,11 +80,12 @@ class Git(Base):
         ignore_list = []
         if os.path.exists(gitignore):
             ignore_list = open(gitignore, 'r').readlines()
-            if entry in ignore_list:
+            filtered = list(filter(lambda x: x.strip() == entry.strip(), ignore_list))
+            if len(filtered) != 0:
                 return
 
         content = entry
-        if ignore_list:
+        if len(ignore_list) > 0:
             content = '\n' + content
 
         open(gitignore, 'a').write(content)

tests/test_cli.py

@@ -12,7 +12,6 @@ from dvc.command.config import CmdConfig
 from dvc.command.show_pipeline import CmdShowPipeline
 from dvc.command.show_workflow import CmdShowWorkflow
 from dvc.command.checkout import CmdCheckout
-from dvc.command.find import CmdFind
 from dvc.command.fsck import CmdFsck
 
 from tests.basic_env import TestDvc
@@ -184,26 +183,6 @@ class TestCheckout(TestDvc):
         self.assertIsInstance(args.func(args), CmdCheckout)
 
 
-class TestFind(TestDvc):
-    def test(self):
-        branch = 'branch'
-        target = 'target'
-        criteria = 'max'
-
-        args = parse_args(['find',
-                           branch,
-                           target,
-                           '-c', criteria,
-                           '--criteria', criteria,
-                           '-s', '--show-value'])
-
-        self.assertIsInstance(args.func(args), CmdFind)
-        self.assertEqual(args.branch_name, branch)
-        self.assertEqual(args.target, target)
-        self.assertEqual(args.criteria, criteria)
-        self.assertTrue(args.show_value)
-
-
 class TestFsck(TestDvc):
     def test(self):
         target1 = 'target1'

tests/test_scm.py

@@ -20,13 +20,25 @@ class TestSCMGit(TestGit):
     def test_is_repo(self):
         self.assertTrue(Git.is_repo(os.curdir))
 
-    def test_ignore(self):
-        Git(self._root_dir).ignore('foo')
-        self.assertTrue(os.path.isfile(Git.GITIGNORE))
-        self.assertTrue('foo' in open(Git.GITIGNORE, 'r').readlines())
-
     def test_commit(self):
         G = Git(self._root_dir)
         G.add(['foo'])
         G.commit('add')
         self.assertTrue('foo' in self.git.git.ls_files())
+
+
+class TestIgnore(TestGit):
+    def _count_gitignore(self):
+        with open(Git.GITIGNORE, 'r') as fd:
+            lines = fd.readlines()
+            return len(list(filter(lambda x: x.strip() == 'foo', lines)))
+
+    def test(self):
+        git = Git(self._root_dir)
+
+        git.ignore('foo')
+        self.assertTrue(os.path.isfile(Git.GITIGNORE))
+        self.assertEqual(self._count_gitignore(), 1)
+
+        git.ignore('foo')
+        self.assertEqual(self._count_gitignore(), 1)