|
| 1 | +from collections import OrderedDict |
| 2 | + |
| 3 | +import pytest |
| 4 | +from voluptuous import Schema as _Schema |
| 5 | + |
| 6 | +from dvc.dvcfile import PIPELINE_FILE |
| 7 | +from dvc.schema import LOCK_FILE_STAGE_SCHEMA, LOCKFILE_SCHEMA |
| 8 | +from dvc.stage import PipelineStage, create_stage |
| 9 | +from dvc.stage.serialize import DEFAULT_PARAMS_FILE, to_lockfile |
| 10 | +from dvc.stage.serialize import ( |
| 11 | + to_single_stage_lockfile as _to_single_stage_lockfile, |
| 12 | +) |
| 13 | +from dvc.stage.utils import split_params_deps |
| 14 | + |
| 15 | +kwargs = {"name": "something", "cmd": "command", "path": PIPELINE_FILE} |
| 16 | +Schema = _Schema(LOCK_FILE_STAGE_SCHEMA) |
| 17 | + |
| 18 | + |
| 19 | +def to_single_stage_lockfile(stage): |
| 20 | + """Validate schema on each serialization.""" |
| 21 | + e = _to_single_stage_lockfile(stage) |
| 22 | + assert Schema(e) |
| 23 | + return e |
| 24 | + |
| 25 | + |
| 26 | +def test_lock(dvc): |
| 27 | + stage = create_stage(PipelineStage, dvc, **kwargs) |
| 28 | + assert to_single_stage_lockfile(stage) == {"cmd": "command"} |
| 29 | + |
| 30 | + |
| 31 | +def test_lock_deps(dvc): |
| 32 | + stage = create_stage(PipelineStage, dvc, deps=["input"], **kwargs) |
| 33 | + stage.deps[0].info = {"md5": "md-five"} |
| 34 | + assert to_single_stage_lockfile(stage) == OrderedDict( |
| 35 | + [ |
| 36 | + ("cmd", "command"), |
| 37 | + ("deps", [OrderedDict([("path", "input"), ("md5", "md-five")])]), |
| 38 | + ] |
| 39 | + ) |
| 40 | + |
| 41 | + |
| 42 | +def test_lock_deps_order(dvc): |
| 43 | + stage = create_stage( |
| 44 | + PipelineStage, dvc, deps=["input1", "input0"], **kwargs |
| 45 | + ) |
| 46 | + stage.deps[0].info = {"md5": "md-one1"} |
| 47 | + stage.deps[1].info = {"md5": "md-zer0"} |
| 48 | + assert to_single_stage_lockfile(stage) == OrderedDict( |
| 49 | + [ |
| 50 | + ("cmd", "command"), |
| 51 | + ( |
| 52 | + "deps", |
| 53 | + [ |
| 54 | + OrderedDict([("path", "input0"), ("md5", "md-zer0")]), |
| 55 | + OrderedDict([("path", "input1"), ("md5", "md-one1")]), |
| 56 | + ], |
| 57 | + ), |
| 58 | + ] |
| 59 | + ) |
| 60 | + |
| 61 | + |
| 62 | +def test_lock_params(dvc): |
| 63 | + stage = create_stage( |
| 64 | + PipelineStage, dvc, params=["lorem.ipsum", "abc"], **kwargs |
| 65 | + ) |
| 66 | + stage.deps[0].info = {"lorem.ipsum": {"lorem1": 1, "lorem2": 2}, "abc": 3} |
| 67 | + assert to_single_stage_lockfile(stage)["params"][ |
| 68 | + DEFAULT_PARAMS_FILE |
| 69 | + ] == OrderedDict([("abc", 3), ("lorem.ipsum", {"lorem1": 1, "lorem2": 2})]) |
| 70 | + |
| 71 | + |
| 72 | +def test_lock_params_file_sorted(dvc): |
| 73 | + stage = create_stage( |
| 74 | + PipelineStage, |
| 75 | + dvc, |
| 76 | + params=[ |
| 77 | + "lorem.ipsum", |
| 78 | + "abc", |
| 79 | + {"myparams.yaml": ["foo", "foobar"]}, |
| 80 | + {"a-params-file.yaml": ["bar", "barr"]}, |
| 81 | + ], |
| 82 | + **kwargs |
| 83 | + ) |
| 84 | + stage.deps[0].info = {"lorem.ipsum": {"lorem1": 1, "lorem2": 2}, "abc": 3} |
| 85 | + stage.deps[1].info = {"foo": ["f", "o", "o"], "foobar": "foobar"} |
| 86 | + stage.deps[2].info = {"bar": ["b", "a", "r"], "barr": "barr"} |
| 87 | + assert to_single_stage_lockfile(stage)["params"] == OrderedDict( |
| 88 | + [ |
| 89 | + ( |
| 90 | + DEFAULT_PARAMS_FILE, |
| 91 | + OrderedDict( |
| 92 | + [("abc", 3), ("lorem.ipsum", {"lorem1": 1, "lorem2": 2})] |
| 93 | + ), |
| 94 | + ), |
| 95 | + ( |
| 96 | + "a-params-file.yaml", |
| 97 | + OrderedDict([("bar", ["b", "a", "r"]), ("barr", "barr")]), |
| 98 | + ), |
| 99 | + ( |
| 100 | + "myparams.yaml", |
| 101 | + OrderedDict([("foo", ["f", "o", "o"]), ("foobar", "foobar")]), |
| 102 | + ), |
| 103 | + ] |
| 104 | + ) |
| 105 | + |
| 106 | + |
| 107 | +def test_lock_params_no_values_filled(dvc): |
| 108 | + stage = create_stage( |
| 109 | + PipelineStage, dvc, params=["lorem.ipsum", "abc"], **kwargs |
| 110 | + ) |
| 111 | + assert to_single_stage_lockfile(stage) == {"cmd": "command"} |
| 112 | + |
| 113 | + |
| 114 | +@pytest.mark.parametrize("typ", ["plots", "metrics", "outs"]) |
| 115 | +def test_lock_outs(dvc, typ): |
| 116 | + stage = create_stage(PipelineStage, dvc, **{typ: ["input"]}, **kwargs) |
| 117 | + stage.outs[0].info = {"md5": "md-five"} |
| 118 | + assert to_single_stage_lockfile(stage) == OrderedDict( |
| 119 | + [ |
| 120 | + ("cmd", "command"), |
| 121 | + ("outs", [OrderedDict([("path", "input"), ("md5", "md-five")])]), |
| 122 | + ] |
| 123 | + ) |
| 124 | + |
| 125 | + |
| 126 | +@pytest.mark.parametrize("typ", ["plots", "metrics", "outs"]) |
| 127 | +def test_lock_outs_order(dvc, typ): |
| 128 | + stage = create_stage( |
| 129 | + PipelineStage, dvc, **{typ: ["input1", "input0"]}, **kwargs |
| 130 | + ) |
| 131 | + stage.outs[0].info = {"md5": "md-one1"} |
| 132 | + stage.outs[1].info = {"md5": "md-zer0"} |
| 133 | + assert to_single_stage_lockfile(stage) == OrderedDict( |
| 134 | + [ |
| 135 | + ("cmd", "command"), |
| 136 | + ( |
| 137 | + "outs", |
| 138 | + [ |
| 139 | + OrderedDict([("path", "input0"), ("md5", "md-zer0")]), |
| 140 | + OrderedDict([("path", "input1"), ("md5", "md-one1")]), |
| 141 | + ], |
| 142 | + ), |
| 143 | + ] |
| 144 | + ) |
| 145 | + |
| 146 | + |
| 147 | +def test_dump_appropriate_checksums(dvc): |
| 148 | + stage = create_stage( |
| 149 | + PipelineStage, dvc, deps=["s3://dvc-temp/file"], **kwargs |
| 150 | + ) |
| 151 | + stage.deps[0].info = {"etag": "is-it-etag", "md5": "or-md5?"} |
| 152 | + assert to_single_stage_lockfile(stage) == OrderedDict( |
| 153 | + [ |
| 154 | + ("cmd", "command"), |
| 155 | + ( |
| 156 | + "deps", |
| 157 | + [ |
| 158 | + OrderedDict( |
| 159 | + [ |
| 160 | + ("path", "s3://dvc-temp/file"), |
| 161 | + ("etag", "is-it-etag"), |
| 162 | + ] |
| 163 | + ) |
| 164 | + ], |
| 165 | + ), |
| 166 | + ] |
| 167 | + ) |
| 168 | + |
| 169 | + |
| 170 | +def test_order(dvc): |
| 171 | + stage = create_stage( |
| 172 | + PipelineStage, |
| 173 | + dvc, |
| 174 | + deps=["input"], |
| 175 | + outs=["output"], |
| 176 | + params=["foo-param"], |
| 177 | + **kwargs |
| 178 | + ) |
| 179 | + params, deps = split_params_deps(stage) |
| 180 | + |
| 181 | + deps[0].info = {"md5": "md-five"} |
| 182 | + params[0].info = {"foo-param": "value"} |
| 183 | + stage.outs[0].info = {"md5": "md5-output"} |
| 184 | + |
| 185 | + assert to_single_stage_lockfile(stage) == OrderedDict( |
| 186 | + [ |
| 187 | + ("cmd", "command"), |
| 188 | + ("deps", [{"path": "input", "md5": "md-five"}]), |
| 189 | + ("params", {"params.yaml": {"foo-param": "value"}}), |
| 190 | + ("outs", [{"path": "output", "md5": "md5-output"}]), |
| 191 | + ] |
| 192 | + ) |
| 193 | + |
| 194 | + |
| 195 | +def test_to_lockfile(dvc): |
| 196 | + stage = create_stage(PipelineStage, dvc, deps=["input"], **kwargs) |
| 197 | + stage.deps[0].info = {"md5": "md-five"} |
| 198 | + entry = to_lockfile(stage) |
| 199 | + assert len(entry) == 1 |
| 200 | + _Schema(LOCKFILE_SCHEMA)(entry) |
| 201 | + assert entry == { |
| 202 | + "something": OrderedDict( |
| 203 | + [ |
| 204 | + ("cmd", "command"), |
| 205 | + ("deps", [{"path": "input", "md5": "md-five"}]), |
| 206 | + ] |
| 207 | + ) |
| 208 | + } |
0 commit comments