dvc find command

dvc/cli.py

@@ -4,6 +4,7 @@ import sys
 import argparse
 from multiprocessing import cpu_count
 
+from dvc.command.find import CmdFind
 from dvc.command.init import CmdInit
 from dvc.command.remove import CmdRemove
 from dvc.command.run import CmdRun
@@ -420,6 +421,33 @@ 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(
+                        'target',
+                        nargs='?',
+                        help='Target metric file.')
+    find_parser.add_argument(
+                        '-r',
+                        '--branch-regex',
+                        help='Branch name regexp.')
+    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.')
+
     if isinstance(argv, str):
         argv = argv.split()
 

dvc/command/find.py

@@ -0,0 +1,62 @@
+from dvc import utils
+from dvc.command.base import CmdBase, DvcLock
+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):
+        with DvcLock(self.is_locker, self.git):
+            fname = self.parsed_args.target
+            regex = self.parsed_args.branch_regex
+
+            branches = self.git.branches(regex)
+            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/git_wrapper.py

@@ -444,3 +444,24 @@ class GitWrapper(GitWrapperI):
         git_cmd = u'git ls-tree --name-only -r {}'.format(commit)
         lines = Executor.exec_cmd_only_success(git_cmd.split()).split('\n')
         return filter(lambda s: s.startswith(ConfigI.STATE_DIR), lines)
+
+    @staticmethod
+    def branches(regexp):
+        cmd = 'git branch --format "%(refname)"'
+        lines = Executor.exec_cmd_only_success(cmd.split()).split('\n')
+        lines = map(lambda l: l.strip('"'), lines)
+
+        if not regexp or '/' not in regexp:
+            # remove remotes like 'refs/heads'
+            lines = map(lambda s: s.split('/')[-1], lines)
+
+        if not regexp:
+            return lines
+
+        reg = re.compile(regexp)
+        return filter(reg.search, lines)
+
+    @staticmethod
+    def get_file_content(fname, branch):
+        cmd = 'git show {}:{}'.format(branch, fname)
+        return Executor.exec_cmd_only_success(cmd.split())

dvc/state_file.py

@@ -4,6 +4,7 @@ import sys
 import json
 import re
 
+from dvc import utils
 from dvc.exceptions import DvcException
 from dvc.path.data_item import DataDirError
 from dvc.system import System
@@ -18,8 +19,6 @@ class StateFile(object):
     MAGIC = 'DVC-State'
     VERSION = '0.1'
 
-    FLOATS_FROM_STRING = re.compile(r'[-+]?(?:(?:\d*\.\d+)|(?:\d+\.?))(?:[Ee][+-]?\d+)?')
-
     DVC_PYTHON_FILE_NAME = 'dvc.py'
     DVC_COMMAND = 'dvc'
 
@@ -110,29 +109,13 @@ class StateFile(object):
         if target_metric:
             self.target_metrics[StateFile.TARGET_METRICS_SINGLE_METRIC] = target_metric
 
-    @staticmethod
-    def parse_target_metric(file_name):
-        fd = open(file_name, 'r')
-
-        try:
-            lines = fd.readlines(2)
-        except Exception as e:
-            raise StateFileError('Unable to parse target file')
-
-        if len(lines) != 1:
-            raise StateFileError('file contains more then one line')
-
-        # Extract float from string. I.e. from 'AUC: 0.596182'
-        nums = StateFile.FLOATS_FROM_STRING.findall(lines[0])
-        if len(nums) < 1:
-            raise StateFileError("Unable to parse metrics from the first line")
-
-        return float(nums[0])
-
     @staticmethod
     def try_parse_target_metric(file_name):
         try:
-            return StateFile.parse_target_metric(file_name)
+            metric = utils.parse_target_metric_file(file_name)
+            if not metric:
+                raise StateFileError('Unable to parse metrics from the first line of file {}'.format(file_name))
+            return metric
         except StateFileError:
             return None
 

dvc/utils.py

@@ -2,6 +2,7 @@
 Helpers for other modules.
 """
 import os
+import re
 import stat
 import shutil
 
@@ -13,6 +14,7 @@ from dvc.logger import Logger
 
 LOCAL_CHUNK_SIZE = 1024*1024*1024
 
+
 def cached_property(func):
     '''A decorator for caching properties in classes.'''
     def get(self):
@@ -82,3 +84,27 @@ def map_progress(func, targets, n_threads):
         progress.finish()
 
     return zip(targets, ret)
+
+
+def parse_target_metric_file(file_name):
+    with open(file_name, 'r') as fd:
+        try:
+            lines = fd.readlines(2)
+        except Exception:
+            return None
+        return parse_target_metric(lines)
+
+
+FLOATS_FROM_STRING = re.compile(r'[-+]?(?:(?:\d*\.\d+)|(?:\d+\.?))(?:[Ee][+-]?\d+)?')
+
+
+def parse_target_metric(lines):
+    if len(lines) != 1:
+        return None
+
+    # Extract float from string. I.e. from 'AUC: 0.596182'
+    nums = FLOATS_FROM_STRING.findall(lines[0])
+    if len(nums) < 1:
+        return None
+
+    return float(nums[0])