RFC: move cmdline parsing to cli.py and settings RFC

dvc/__init__.py

@@ -0,0 +1 @@
+VERSION = '0.8.7'

dvc/cli.py

@@ -0,0 +1,271 @@
+from __future__ import print_function
+
+import os
+import sys
+import argparse
+import configparser
+from multiprocessing import cpu_count
+
+from dvc.command.init import CmdInit
+from dvc.command.remove import CmdRemove
+from dvc.command.run import CmdRun
+from dvc.command.repro import CmdRepro
+from dvc.command.data_sync import CmdDataSync
+from dvc.command.lock import CmdLock
+from dvc.command.gc import CmdGC
+from dvc.command.import_file import CmdImportFile
+from dvc.command.target import CmdTarget
+from dvc.command.test import CmdTest
+from dvc.config import Config
+from dvc import VERSION
+
+def parse_args(argv=None):
+    # Common args
+    parent_parser = argparse.ArgumentParser(add_help=False)
+    parent_parser.add_argument('-q',
+                        '--quiet',
+                        action='store_true',
+                        default=False,
+                        help='Be quiet.')
+    parent_parser.add_argument('-v',
+                        '--verbose',
+                        action='store_true',
+                        default=False,
+                        help='Be verbose.')
+    parent_parser.add_argument('-G',
+                        '--no-git-actions',
+                        action='store_true',
+                        default=False,
+                        help='Skip all git actions including reproducibility check and commits.')
+
+    desc = 'Data Version Control'
+    parser = argparse.ArgumentParser(description=desc, parents=[parent_parser],
+             formatter_class=argparse.RawTextHelpFormatter)
+
+    parser.add_argument('-V',
+                        '--version',
+                        action='version',
+                        version='%(prog)s ' + VERSION)
+
+    # Sub commands
+    subparsers = parser.add_subparsers(
+                        dest='cmd',
+                        help='Use dvc CMD --help for command-specific help')
+
+    # Init
+    init_parser = subparsers.add_parser(
+                        'init',
+                        parents=[parent_parser],
+                        help='Initialize dvc over a directory (should already be a git dir).')
+    init_parser.add_argument(
+                        '--data-dir',
+                        default='data',
+                        help='Data directory.')
+    init_parser.add_argument(
+                        '--cache-dir',
+                        default='.cache',
+                        help='Cache directory.')
+    init_parser.add_argument(
+                        '--state-dir',
+                        default='.state',
+                        help='State directory.')
+    init_parser.add_argument(
+                        '--target-file',
+                        default=Config.TARGET_FILE_DEFAULT,
+                        help='Target file.')
+    init_parser.set_defaults(func=CmdInit)
+
+    # Run
+    run_parser = subparsers.add_parser(
+                        'run',
+                        parents=[parent_parser],
+                        help='Run command')
+    run_parser.add_argument(
+                        '--stdout',
+                        help='Output std output to a file.')
+    run_parser.add_argument(
+                        '--stderr',
+                        help='Output std error to a file.')
+    run_parser.add_argument('-i',
+                        '--input',
+                        action='append',
+                        help='Declare input data items for reproducible cmd.')
+    run_parser.add_argument('-o',
+                        '--output',
+                        action='append',
+                        help='Declare output data items for reproducible cmd.')
+    run_parser.add_argument('-c',
+                        '--code',
+                        action='append',
+                        help='Code dependencies which produce the output.')
+    run_parser.add_argument(
+                        '--shell',
+                        action='store_true',
+                        default=False,
+                        help='Shell command')
+    run_parser.add_argument('-l',
+                        '--lock',
+                        action='store_true',
+                        default=False,
+                        help='Lock data item - disable reproduction.')
+    run_parser.add_argument(
+                        'command',
+                        nargs=argparse.REMAINDER,
+                        help='Command to execute')
+    run_parser.set_defaults(func=CmdRun)
+
+    # Sync
+    sync_parser = subparsers.add_parser(
+                        'sync',
+                        parents=[parent_parser],
+                        help='Synchronize data file with cloud (cloud settings already setup.')
+    sync_parser.add_argument(
+                        'targets',
+                        metavar='',
+                        nargs='*',
+                        help='File or directory to sync.')
+    sync_parser.add_argument('-j',
+                        '--jobs',
+                        type=int,
+                        default=cpu_count(),
+                        help='Number of jobs to run simultaneously.')
+    sync_parser.set_defaults(func=CmdDataSync)
+
+    # Repro
+    repro_parser = subparsers.add_parser(
+                        'repro',
+                        parents=[parent_parser],
+                        help='Reproduce data')
+    repro_parser.add_argument(
+                        'target',
+                        metavar='',
+                        nargs='*',
+                        help='Data items to reproduce.')
+    repro_parser.add_argument('-f',
+                        '--force',
+                        action='store_true',
+                        default=False,
+                        help='Reproduce even if dependencies were not changed.')
+    repro_parser.add_argument('-s',
+                        '--single-item',
+                        action='store_true',
+                        default=False,
+                        help='Reproduce only single data item without recursive dependencies check.')
+    repro_parser.set_defaults(func=CmdRun)
+
+    # Remove
+    remove_parser = subparsers.add_parser(
+                        'remove',
+                        parents=[parent_parser],
+                        help='Remove data item from data directory.')
+    remove_parser.add_argument('target',
+                        metavar='',
+                        nargs='*',
+                        help='Target to remove - file or directory.')
+    remove_parser.add_argument('-l',
+                        '--keep-in-cloud',
+                        action='store_true',
+                        default=False,
+                        help='Do not remove data from cloud.')
+    remove_parser.add_argument('-r',
+                        '--recursive',
+                        action='store_true',
+                        help='Remove directory recursively.')
+    remove_parser.add_argument('-c',
+                        '--keep-in-cache',
+                        action='store_true',
+                        default=False,
+                        help='Do not remove data from cache.')
+    remove_parser.set_defaults(func=CmdRemove)
+
+    # Import
+    import_parser = subparsers.add_parser(
+                        'import',
+                        parents=[parent_parser],
+                        help='Import file to data directory.')
+    import_parser.add_argument(
+                        'input',
+                        nargs='+',
+                        help='Input file/files.')
+    import_parser.add_argument(
+                        'output',
+                        help='Output file/directory.')
+    import_parser.add_argument('-l',
+                        '--lock',
+                        action='store_true',
+                        default=False,
+                        help='Lock data item - disable reproduction.')
+    import_parser.add_argument('-j',
+                        '--jobs',
+                        type=int,
+                        default=cpu_count(),
+                        help='Number of jobs to run simultaneously.')
+    import_parser.set_defaults(func=CmdImportFile)
+
+    # Lock
+    lock_parser = subparsers.add_parser(
+                        'lock',
+                        parents=[parent_parser],
+                        help='Lock')
+    lock_parser.add_argument('-l',
+                        '--lock',
+                        action='store_true',
+                        default=False,
+                        help='Lock data item - disable reproduction.')
+    lock_parser.add_argument('-u',
+                        '--unlock',
+                        action='store_true',
+                        default=False,
+                        help='Unlock data item - enable reproduction.')
+    lock_parser.add_argument(
+                        'files',
+                        metavar='',
+                        nargs='*',
+                        help='Data items to lock or unlock.')
+    lock_parser.set_defaults(func=CmdLock)
+
+    # Garbage collector
+    gc_parser = subparsers.add_parser(
+                        'gc',
+                        parents=[parent_parser],
+                        help='Collect garbage')
+    gc_parser.add_argument('target',
+                        metavar='',
+                        nargs='*',
+                        help='Target to remove - file or directory.')
+    gc_parser.add_argument('-l',
+                        '--keep-in-cloud',
+                        action='store_true',
+                        default=False,
+                        help='Do not remove data from cloud.')
+    gc_parser.add_argument('-r',
+                        '--recursive',
+                        action='store_true',
+                        help='Remove directory recursively.')
+    gc_parser.add_argument('-c',
+                        '--keep-in-cache',
+                        action='store_false',
+                        default=False,
+                        help='Do not remove data from cache.')
+    gc_parser.set_defaults(func=CmdGC)
+
+    # Target
+    target_parser = subparsers.add_parser(
+                        'target',
+                        parents=[parent_parser],
+                        help='Set default target')
+    target_parser.add_argument('target_file',
+                        metavar='',
+                        nargs='?',
+                        help='Target data item.')
+    target_parser.add_argument('-u',
+                        '--unset',
+                        action='store_true',
+                        default=False,
+                        help='Reset target.')
+    target_parser.set_defaults(func=CmdTarget)
+
+    if isinstance(argv, str):
+        argv = argv.split()
+
+    return parser.parse_args(argv)

dvc/command/base.py

@@ -40,29 +40,24 @@ class CmdBase(object):
     def __init__(self, settings):
         self._settings = settings
 
-        parser = argparse.ArgumentParser()
-        self.define_common_args(parser)
-        self.define_args(parser)
-
-        self._parsed_args, self._command_args = parser.parse_known_args(args=self.args)
-
-        self.process_common_args()
+        if settings._parsed_args.quiet and not settings._parsed_args.verbose:
+            Logger.be_quiet()
+        elif not settings._parsed_args.quiet and settings._parsed_args.verbose:
+            Logger.be_verbose()
 
     @property
     def settings(self):
         return self._settings
 
+    #NOTE: this name is really confusing. It should really be called "command" or smth,
+    # because it is only used for "command" argument from CmdRun.
     @property
     def args(self):
         return self._settings.args
 
     @property
     def parsed_args(self):
-        return self._parsed_args
-
-    @property
-    def command_args(self):
-        return self._command_args
+        return self._settings._parsed_args
 
     @property
     def config(self):
@@ -76,34 +71,6 @@ def dvc_home(self):
     def git(self):
         return self._settings.git
 
-    def define_args(self, parser):
-        pass
-
-    def set_no_git_actions(self, parser):
-        parser.add_argument('--no-git-actions', '-G', action='store_true', default=False,
-                            help='Skip all git actions including reproducibility check and commits.')
-
-    def set_lock_action(self, parser):
-        parser.add_argument('--lock', '-l', action='store_true', default=False,
-                            help='Lock data item - disable reproduction. ' +
-                                 'It can be enabled by `dvc lock` command or by forcing reproduction.')
-
-    def define_common_args(self, parser):
-        parser.add_argument('--quiet', '-q', action='store_true', default=False, help='Be quiet.')
-        parser.add_argument('--verbose', '-v', action='store_true', default=False, help='Be verbose.')
-        parser.add_argument('--jobs', '-j', type=int, default=cpu_count(),
-                            help='Number of jobs to run simultaneously.')
-
-    @staticmethod
-    def set_reset_flag(parser, short_name, long_name, message):
-        parser.add_argument(short_name, long_name, action='store_true', default=False, help=message)
-
-    def process_common_args(self):
-        if self._parsed_args.quiet and not self._parsed_args.verbose:
-            Logger.be_quiet()
-        elif not self._parsed_args.quiet and self._parsed_args.verbose:
-            Logger.be_verbose()
-
     @property
     def no_git_actions(self):
         return self.parsed_args.no_git_actions
@@ -132,24 +99,5 @@ def commit_if_needed(self, message, error=False):
     def not_committed_changes_warning():
         Logger.warn('changes were not committed to git')
 
-    def add_string_arg(self, parser, name, message, default = None,
-                       conf_section=None, conf_name=None):
-        if conf_section and conf_name:
-            section = self.config[conf_section]
-            if not section:
-                raise ConfigError("")
-            default_value = section.get(conf_section, default)
-        else:
-            default_value = default
-
-        parser.add_argument(name,
-                            metavar='',
-                            default=default_value,
-                            help=message)
-
     def run(self):
         pass
-
-    @staticmethod
-    def warning_dvc_is_busy():
-        Logger.warn('Cannot perform the cmd since DVC is busy and locked. Please retry the cmd later.')

dvc/command/data_sync.py

@@ -3,7 +3,6 @@
 
 from dvc.command.base import CmdBase, DvcLock
 from dvc.exceptions import DvcException
-from dvc.runtime import Runtime
 from dvc.system import System
 from dvc.data_cloud import DataCloud
 from dvc.utils import map_progress
@@ -18,12 +17,6 @@ class CmdDataSync(CmdBase):
     def __init__(self, settings):
         super(CmdDataSync, self).__init__(settings)
 
-    def define_args(self, parser):
-        parser.add_argument('targets',
-                            metavar='',
-                            help='File or directory to sync.',
-                            nargs='*')
-
     def run(self):
         with DvcLock(self.is_locker, self.git):
             cloud = DataCloud(self.settings)
@@ -44,6 +37,3 @@ def run(self):
 
             map_progress(cloud.sync, targets, self.parsed_args.jobs)
         pass
-
-if __name__ == '__main__':
-    Runtime.run(CmdDataSync)