path info: replace path_info dicts with PathInfo class

dvc/dependency/http.py

@@ -1,5 +1,6 @@
 from __future__ import unicode_literals
 
+from dvc.path.http import HTTPPathInfo
 from dvc.utils.compat import urlparse, urljoin
 from dvc.output.base import OutputBase
 from dvc.remote.http import RemoteHTTP
@@ -16,5 +17,4 @@ def __init__(self, stage, path, info=None, remote=None):
         if path.startswith("remote"):
             path = urljoin(self.remote.cache_dir, urlparse(path).path)
 
-        self.path_info["scheme"] = urlparse(path).scheme
-        self.path_info["path"] = path
+        self.path_info = HTTPPathInfo(url=self.url, path=path)

dvc/output/__init__.py

@@ -2,6 +2,7 @@
 
 import schema
 
+from dvc.path import Schemes
 from dvc.utils.compat import urlparse, str
 
 from dvc.output.base import OutputBase
@@ -25,11 +26,11 @@
 ]
 
 OUTS_MAP = {
-    "hdfs": OutputHDFS,
-    "s3": OutputS3,
-    "gs": OutputGS,
-    "ssh": OutputSSH,
-    "local": OutputLOCAL,
+    Schemes.HDFS: OutputHDFS,
+    Schemes.S3: OutputS3,
+    Schemes.GS: OutputGS,
+    Schemes.SSH: OutputSSH,
+    Schemes.LOCAL: OutputLOCAL,
 }
 
 # NOTE: currently there are only 3 possible checksum names:

dvc/output/base.py

@@ -2,6 +2,8 @@
 
 import re
 import logging
+from copy import copy
+
 from schema import Or, Optional
 
 from dvc.exceptions import DvcException
@@ -84,8 +86,6 @@ def __init__(
                 )
             )
 
-        self.path_info = {"scheme": self.scheme, "url": self.url}
-
     def __repr__(self):
         return "{class_name}: '{url}'".format(
             class_name=type(self).__name__, url=(self.url or "No url")
@@ -131,7 +131,7 @@ def scheme(self):
 
     @property
     def path(self):
-        return self.path_info["path"]
+        return self.path_info.path
 
     @property
     def cache_path(self):
@@ -241,7 +241,7 @@ def commit(self):
             self.cache.save(self.path_info, self.info)
 
     def dumpd(self):
-        ret = self.info.copy()
+        ret = copy(self.info)
         ret[self.PARAM_PATH] = self.url
 
         if self.IS_DEPENDENCY:
@@ -302,7 +302,7 @@ def move(self, out):
 
         self.remote.move(self.path_info, out.path_info)
         self.url = out.url
-        self.path_info = out.path_info.copy()
+        self.path_info = copy(out.path_info)
         self.save()
         self.commit()
 

dvc/output/hdfs.py

@@ -2,6 +2,7 @@
 
 import posixpath
 
+from dvc.path.hdfs import HDFSPathInfo
 from dvc.utils.compat import urlparse
 from dvc.output.base import OutputBase
 from dvc.remote.hdfs import RemoteHDFS
@@ -34,5 +35,4 @@ def __init__(
         if remote:
             path = posixpath.join(remote.url, urlparse(path).path.lstrip("/"))
         user = remote.user if remote else self.group("user")
-        self.path_info["user"] = user
-        self.path_info["path"] = path
+        self.path_info = HDFSPathInfo(user=user, url=self.url, path=path)

dvc/output/local.py

@@ -3,6 +3,7 @@
 import os
 import logging
 
+from dvc.path.local import LocalPathInfo
 from dvc.utils.compat import urlparse
 from dvc.istextfile import istextfile
 from dvc.exceptions import DvcException
@@ -49,7 +50,7 @@ def __init__(
             p = os.path.join(stage.wdir, p)
         p = os.path.abspath(os.path.normpath(p))
 
-        self.path_info["path"] = p
+        self.path_info = LocalPathInfo(url=self.url, path=p)
 
     def __str__(self):
         return self.rel_path
@@ -64,7 +65,7 @@ def assign_to_stage_file(self, stage):
         from dvc.repo import Repo
 
         fullpath = os.path.abspath(stage.wdir)
-        self.path_info["path"] = os.path.join(fullpath, self.stage_path)
+        self.path_info.path = os.path.join(fullpath, self.stage_path)
 
         self.repo = Repo(self.path)
 

dvc/output/s3.py

@@ -2,6 +2,7 @@
 
 import posixpath
 
+from dvc.path.utils import PathInfo
 from dvc.remote.s3 import RemoteS3
 from dvc.utils.compat import urlparse
 from dvc.output.base import OutputBase
@@ -36,5 +37,6 @@ def __init__(
         if remote:
             path = posixpath.join(remote.prefix, path)
 
-        self.path_info["bucket"] = bucket
-        self.path_info["path"] = path
+        self.path_info = PathInfo(
+            self.scheme, bucket=bucket, path=path, url=self.url
+        )

dvc/output/ssh.py

@@ -3,6 +3,7 @@
 import getpass
 import posixpath
 
+from dvc.path.ssh import SSHPathInfo
 from dvc.utils.compat import urlparse
 from dvc.output.base import OutputBase
 from dvc.remote.ssh import RemoteSSH
@@ -48,7 +49,6 @@ def __init__(
         else:
             path = parsed.path
 
-        self.path_info["host"] = host
-        self.path_info["port"] = port
-        self.path_info["user"] = user
-        self.path_info["path"] = path
+        self.path_info = SSHPathInfo(
+            host=host, user=user, port=port, url=self.url, path=path
+        )

dvc/path/__init__.py

@@ -0,0 +1,34 @@
+from dvc.utils.compat import urlunsplit
+
+
+class Schemes:
+    SSH = "ssh"
+    HDFS = "hdfs"
+    S3 = "s3"
+    AZURE = "azure"
+    HTTP = "http"
+    GS = "gs"
+    LOCAL = "local"
+    OSS = "oss"
+
+
+class BasePathInfo(object):
+    scheme = None
+
+    def __init__(self, url=None, path=None):
+        self.url = url
+        self.path = path
+
+    def __str__(self):
+        return self.url
+
+
+class DefaultCloudPathInfo(BasePathInfo):
+    def __init__(self, bucket, url=None, path=None):
+        super(DefaultCloudPathInfo, self).__init__(url, path)
+        self.bucket = bucket
+
+    def __str__(self):
+        if not self.url:
+            return urlunsplit((self.scheme, self.bucket, self.path, "", ""))
+        return self.url

dvc/path/azure.py

@@ -0,0 +1,5 @@
+from dvc.path import Schemes, DefaultCloudPathInfo
+
+
+class AzurePathInfo(DefaultCloudPathInfo):
+    scheme = Schemes.AZURE

dvc/path/gs.py

@@ -0,0 +1,5 @@
+from dvc.path import Schemes, DefaultCloudPathInfo
+
+
+class GSPathInfo(DefaultCloudPathInfo):
+    scheme = Schemes.GS

dvc/path/hdfs.py

@@ -0,0 +1,15 @@
+from dvc.utils.compat import urlunsplit
+from dvc.path import BasePathInfo, Schemes
+
+
+class HDFSPathInfo(BasePathInfo):
+    scheme = Schemes.HDFS
+
+    def __init__(self, user, url=None, path=None):
+        super(HDFSPathInfo, self).__init__(url, path)
+        self.user = user
+
+    def __str__(self):
+        if not self.url:
+            return urlunsplit((self.scheme, self.user, self.path, "", ""))
+        return self.url

dvc/path/http.py

@@ -0,0 +1,16 @@
+from dvc.path import BasePathInfo, Schemes
+from dvc.utils.compat import urlparse, urlunsplit
+
+
+class HTTPPathInfo(BasePathInfo):
+    @property
+    def scheme(self):
+        if self.path:
+            return urlparse(self.path).scheme
+        else:
+            return Schemes.HTTP
+
+    def __str__(self):
+        if not self.url:
+            return urlunsplit((self.scheme, self.path, "", "", ""))
+        return self.url

dvc/path/local.py

@@ -0,0 +1,10 @@
+import os
+
+from dvc.path import BasePathInfo, Schemes
+
+
+class LocalPathInfo(BasePathInfo):
+    scheme = Schemes.LOCAL
+
+    def __str__(self):
+        return os.path.relpath(self.path)

dvc/path/oss.py

@@ -0,0 +1,5 @@
+from dvc.path import Schemes, DefaultCloudPathInfo
+
+
+class OSSPathInfo(DefaultCloudPathInfo):
+    scheme = Schemes.OSS

dvc/path/s3.py

@@ -0,0 +1,5 @@
+from dvc.path import Schemes, DefaultCloudPathInfo
+
+
+class S3PathInfo(DefaultCloudPathInfo):
+    scheme = Schemes.S3

dvc/path/ssh.py

@@ -0,0 +1,25 @@
+from dvc.utils.compat import urlunsplit
+from dvc.path import BasePathInfo, Schemes
+
+
+class SSHPathInfo(BasePathInfo):
+    scheme = Schemes.SSH
+
+    def __init__(self, host, user, port, url=None, path=None):
+        super(SSHPathInfo, self).__init__(url, path)
+        self.host = host
+        self.user = user
+        self.port = port
+
+    def __str__(self):
+        if not self.url:
+            return urlunsplit(
+                (
+                    self.scheme,
+                    "{}@{}:{}".format(self.user, self.host, self.port),
+                    self.path,
+                    "",
+                    "",
+                )
+            )
+        return self.url

dvc/path/utils.py

@@ -0,0 +1,25 @@
+from dvc.path import Schemes
+from dvc.path.azure import AzurePathInfo
+from dvc.path.gs import GSPathInfo
+from dvc.path.hdfs import HDFSPathInfo
+from dvc.path.http import HTTPPathInfo
+from dvc.path.local import LocalPathInfo
+from dvc.path.oss import OSSPathInfo
+from dvc.path.s3 import S3PathInfo
+from dvc.path.ssh import SSHPathInfo
+
+PATH_MAP = {
+    Schemes.SSH: SSHPathInfo,
+    Schemes.HDFS: HDFSPathInfo,
+    Schemes.S3: S3PathInfo,
+    Schemes.AZURE: AzurePathInfo,
+    Schemes.HTTP: HTTPPathInfo,
+    Schemes.GS: GSPathInfo,
+    Schemes.LOCAL: LocalPathInfo,
+    Schemes.OSS: OSSPathInfo,
+}
+
+
+def PathInfo(scheme, *args, **kwargs):
+    cls = PATH_MAP[scheme]
+    return cls(*args, **kwargs)