From d5a3b6f4f69082623f8421499e2b32eec2fd848d Mon Sep 17 00:00:00 2001 From: Dmitry Petrov Date: Sun, 26 Jan 2020 02:50:42 -0800 Subject: [PATCH] Summon dataframe support --- dvc/api.py | 52 ++++++++++++++++++++++++++++++------------ tests/func/test_api.py | 8 +++---- 2 files changed, 41 insertions(+), 19 deletions(-) diff --git a/dvc/api.py b/dvc/api.py index 4736b3fa7..d22c70091 100644 --- a/dvc/api.py +++ b/dvc/api.py @@ -14,12 +14,14 @@ from dvc.exceptions import DvcException, NotDvcRepoError from dvc.external_repo import external_repo DEF_SUMMON = "Summon.yaml" +DOBJ_SECTION = "d-objects" SUMMON_FILE_SCHEMA = Schema( { - Required("objects"): [ + Required(DOBJ_SECTION): [ { Required("name"): str, + "description": str, "meta": dict, Required("summon"): { Required("type"): str, @@ -142,8 +144,9 @@ def summon(name, repo=None, rev=None, summon_file=DEF_SUMMON, args=None): class SummonDesc(object): def __init__(self, repo_obj, summon_file=DEF_SUMMON): self.repo = repo_obj + self.filename = summon_file self.path = os.path.join(self.repo.root_dir, summon_file) - self.summon_content = self._read_summon_content() + self.content = self._read_summon_content() def _read_summon_content(self): try: @@ -156,6 +159,16 @@ class SummonDesc(object): except Invalid as exc: raise SummonError(str(exc)) from exc + def _write_summon_content(self): + try: + with builtin_open(self.path, "w") as fobj: + content = SUMMON_FILE_SCHEMA(self.content) + ruamel.yaml.serialize_all(content, fobj) + except ruamel.yaml.YAMLError as exc: + raise SummonError("Summon file schema error") from exc + except Exception as exc: + raise SummonError(str(exc)) from exc + @staticmethod @contextmanager def prepare_summon(repo=None, rev=None, summon_file=DEF_SUMMON): @@ -198,17 +211,20 @@ class SummonDesc(object): self.repo.cloud.pull(out.get_used_cache()) out.checkout() - # def to_abs_paths(self, paths): - # return [self.repo.find_out_by_relpath(d) for d in paths] + def push(self, dobj): + paths = self.deps_abs_paths(dobj) - def get_dobject(self, name): + with self.repo.state: + for path in paths: + self.repo.add(path) + self.repo.add(path) + + def get_dobject(self, name, default=False): """ Given a summonable object's name, search for it on the given content and return its description. """ - objects = [ - x for x in self.summon_content["objects"] if x["name"] == name - ] + objects = [x for x in self.content[DOBJ_SECTION] if x["name"] == name] if not objects: raise SummonErrorNoObjectFound( @@ -221,18 +237,24 @@ class SummonDesc(object): return objects[0] - def set_dobject(self, obj_new, overwrite=False): + def update_dobj(self, new_dobj, overwrite=False): try: - name = obj_new["name"] - obj = self.get_dobject(name) + name = new_dobj["name"] + dobj = self.get_dobject(name) if overwrite: - idx = self.summon_content["objects"].index(obj) - self.summon_content["objects"][idx] = obj_new + idx = self.content[DOBJ_SECTION].index(dobj) + self.content[DOBJ_SECTION][idx] = new_dobj else: - raise SummonError("Object '{}' already exist".format(name)) + raise SummonError( + "D-object '{}' already exist in '{}'".format( + name, self.filename + ) + ) except SummonErrorNoObjectFound: - self.summon_content["objects"].append(obj_new) + self.content[DOBJ_SECTION].append(new_dobj) + + self._write_summon_content() @wrap_with(threading.Lock()) diff --git a/tests/func/test_api.py b/tests/func/test_api.py index a9d665075..ad078fed4 100644 --- a/tests/func/test_api.py +++ b/tests/func/test_api.py @@ -6,7 +6,7 @@ import ruamel.yaml import pytest from dvc import api -from dvc.api import SummonError, UrlNotDvcRepoError, DEF_SUMMON +from dvc.api import SummonError, UrlNotDvcRepoError, DEF_SUMMON, DOBJ_SECTION from dvc.compat import fspath from dvc.exceptions import FileMissingError from dvc.main import main @@ -145,7 +145,7 @@ def test_open_not_cached(dvc): def test_summon(tmp_dir, dvc, erepo_dir): objects = { - "objects": [ + DOBJ_SECTION: [ { "name": "sum", "meta": {"description": "Add to "}, @@ -160,10 +160,10 @@ def test_summon(tmp_dir, dvc, erepo_dir): } other_objects = copy.deepcopy(objects) - other_objects["objects"][0]["summon"]["args"]["x"] = 100 + other_objects[DOBJ_SECTION][0]["summon"]["args"]["x"] = 100 dup_objects = copy.deepcopy(objects) - dup_objects["objects"] *= 2 + dup_objects[DOBJ_SECTION] *= 2 with erepo_dir.chdir(): erepo_dir.dvc_gen("number", "100", commit="Add number.dvc")