plancraft 0.1.1__py3-none-any.whl → 0.1.2__py3-none-any.whl
Sign up to get free protection for your applications and to get access to all the features.
- plancraft-0.1.2.dist-info/METADATA +74 -0
- plancraft-0.1.2.dist-info/RECORD +5 -0
- {plancraft-0.1.1.dist-info → plancraft-0.1.2.dist-info}/WHEEL +1 -1
- plancraft-0.1.2.dist-info/top_level.txt +1 -0
- environments/__init__.py +0 -0
- environments/actions.py +0 -218
- environments/env_real.py +0 -316
- environments/env_symbolic.py +0 -212
- environments/items.py +0 -10
- environments/planner.py +0 -109
- environments/recipes.py +0 -542
- environments/sampler.py +0 -224
- models/__init__.py +0 -21
- models/act.py +0 -184
- models/base.py +0 -152
- models/bbox_model.py +0 -492
- models/dummy.py +0 -54
- models/few_shot_images/__init__.py +0 -16
- models/generators.py +0 -480
- models/oam.py +0 -283
- models/oracle.py +0 -265
- models/prompts.py +0 -158
- models/react.py +0 -93
- models/utils.py +0 -289
- plancraft-0.1.1.dist-info/METADATA +0 -74
- plancraft-0.1.1.dist-info/RECORD +0 -26
- plancraft-0.1.1.dist-info/top_level.txt +0 -3
- train/dataset.py +0 -187
- {plancraft-0.1.1.dist-info → plancraft-0.1.2.dist-info}/LICENSE +0 -0
@@ -0,0 +1,74 @@
|
|
1
|
+
Metadata-Version: 2.1
|
2
|
+
Name: plancraft
|
3
|
+
Version: 0.1.2
|
4
|
+
Summary: Plancraft: an evaluation dataset for planning with LLM agents
|
5
|
+
Requires-Python: >=3.9
|
6
|
+
Description-Content-Type: text/markdown
|
7
|
+
License-File: LICENSE
|
8
|
+
Requires-Dist: accelerate>=1.0.1
|
9
|
+
Requires-Dist: coloredlogs>=10.0
|
10
|
+
Requires-Dist: daemoniker>=0.2.3
|
11
|
+
Requires-Dist: datasets>=3.0.2
|
12
|
+
Requires-Dist: dill>=0.3.1.1
|
13
|
+
Requires-Dist: einops>=0.8.0
|
14
|
+
Requires-Dist: flaky>=3.8.1
|
15
|
+
Requires-Dist: hf-transfer>=0.1.8
|
16
|
+
Requires-Dist: huggingface-hub>=0.26.1
|
17
|
+
Requires-Dist: hydra-core>=1.3.2
|
18
|
+
Requires-Dist: imagehash>=4.0.0
|
19
|
+
Requires-Dist: imageio>=2.36.0
|
20
|
+
Requires-Dist: inflection>=0.3.1
|
21
|
+
Requires-Dist: ipython>=7.5.0
|
22
|
+
Requires-Dist: jinja2>=2.11.2
|
23
|
+
Requires-Dist: loguru>=0.7.2
|
24
|
+
Requires-Dist: lxml>=4.3.3
|
25
|
+
Requires-Dist: matplotlib>=3.9.2
|
26
|
+
Requires-Dist: networkx>=3.2.1
|
27
|
+
Requires-Dist: numpy<1.24,>=1.16.2
|
28
|
+
Requires-Dist: openai>=1.52.2
|
29
|
+
Requires-Dist: opencv-python>=4.1.0.25
|
30
|
+
Requires-Dist: pandas>=2.1.0
|
31
|
+
Requires-Dist: peft>=0.13.2
|
32
|
+
Requires-Dist: pillow>=8.0.0
|
33
|
+
Requires-Dist: psutil>=5.6.2
|
34
|
+
Requires-Dist: pydantic>=2.9.2
|
35
|
+
Requires-Dist: pyglet>=2.0.18
|
36
|
+
Requires-Dist: pyro4>=4.76
|
37
|
+
Requires-Dist: python-dotenv>=1.0.1
|
38
|
+
Requires-Dist: pyyaml>=6.0.2
|
39
|
+
Requires-Dist: requests>=2.20.0
|
40
|
+
Requires-Dist: seaborn>=0.13.2
|
41
|
+
Requires-Dist: setuptools>=49.2.0
|
42
|
+
Requires-Dist: tinydb>=4.8.2
|
43
|
+
Requires-Dist: torch>=2.5.0
|
44
|
+
Requires-Dist: torchvision>=0.20.0
|
45
|
+
Requires-Dist: tqdm>=4.32.2
|
46
|
+
Requires-Dist: transformers>=4.43.3
|
47
|
+
Requires-Dist: typing>=3.6.6
|
48
|
+
Requires-Dist: wandb>=0.18.5
|
49
|
+
Requires-Dist: xmltodict==0.12.0
|
50
|
+
Provides-Extra: full
|
51
|
+
Requires-Dist: gym<=0.23.1,>=0.19.0; extra == "full"
|
52
|
+
|
53
|
+
# plancraft
|
54
|
+
|
55
|
+
[![Test](https://github.com/gautierdag/plancraft/actions/workflows/test.yaml/badge.svg)](https://github.com/gautierdag/plancraft/actions/workflows/test.yaml)
|
56
|
+
![Python Version](https://img.shields.io/badge/python-3.9+-blue)
|
57
|
+
![Ruff](https://img.shields.io/badge/linter-ruff-blue)
|
58
|
+
[![PyPI Version](https://img.shields.io/pypi/v/plancraft)](https://pypi.org/project/plancraft/)
|
59
|
+
|
60
|
+
Plancraft is a minecraft environment and agent that innovates on planning LLM agents with a retriever
|
61
|
+
|
62
|
+
You can install the package by running the following command:
|
63
|
+
|
64
|
+
```bash
|
65
|
+
pip install plancraft
|
66
|
+
```
|
67
|
+
|
68
|
+
Should you need the multimodal version of the package, you will also need a custom [fork](https://github.com/gautierdag/minerl.git) of the minerl package. You can install it by running the following command:
|
69
|
+
|
70
|
+
```bash
|
71
|
+
pip install git+hhttps://github.com/gautierdag/minerl.git
|
72
|
+
```
|
73
|
+
|
74
|
+
Note that you may need to follow the same installation instructions as in the [minerl documentation](https://minerl.readthedocs.io/en/latest/tutorials/index.html).
|
@@ -0,0 +1,5 @@
|
|
1
|
+
plancraft-0.1.2.dist-info/LICENSE,sha256=YGR8ehDB4t-T-lOQKMfKNR-2zsOU7E3E5NA8t25HKE0,1070
|
2
|
+
plancraft-0.1.2.dist-info/METADATA,sha256=IDsYHFCvjx4Nhe4Jgyj0x7YsIVAR95ZyrSA-LIQBgeI,2631
|
3
|
+
plancraft-0.1.2.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
|
4
|
+
plancraft-0.1.2.dist-info/top_level.txt,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
|
5
|
+
plancraft-0.1.2.dist-info/RECORD,,
|
@@ -0,0 +1 @@
|
|
1
|
+
|
environments/__init__.py
DELETED
File without changes
|
environments/actions.py
DELETED
@@ -1,218 +0,0 @@
|
|
1
|
-
from typing import Union
|
2
|
-
|
3
|
-
from pydantic import BaseModel, field_validator, model_validator
|
4
|
-
|
5
|
-
|
6
|
-
def convert_to_slot_index(slot: str) -> int:
|
7
|
-
slot = slot.strip()
|
8
|
-
grid_map = {
|
9
|
-
"[0]": 0,
|
10
|
-
"[A1]": 1,
|
11
|
-
"[A2]": 2,
|
12
|
-
"[A3]": 3,
|
13
|
-
"[B1]": 4,
|
14
|
-
"[B2]": 5,
|
15
|
-
"[B3]": 6,
|
16
|
-
"[C1]": 7,
|
17
|
-
"[C2]": 8,
|
18
|
-
"[C3]": 9,
|
19
|
-
}
|
20
|
-
if slot in grid_map:
|
21
|
-
return grid_map[slot]
|
22
|
-
else:
|
23
|
-
return int(slot[2:-1]) + 9
|
24
|
-
|
25
|
-
|
26
|
-
def convert_from_slot_index(slot_index: int) -> str:
|
27
|
-
grid_map = {
|
28
|
-
0: "[0]",
|
29
|
-
1: "[A1]",
|
30
|
-
2: "[A2]",
|
31
|
-
3: "[A3]",
|
32
|
-
4: "[B1]",
|
33
|
-
5: "[B2]",
|
34
|
-
6: "[B3]",
|
35
|
-
7: "[C1]",
|
36
|
-
8: "[C2]",
|
37
|
-
9: "[C3]",
|
38
|
-
}
|
39
|
-
if slot_index < 10:
|
40
|
-
return grid_map[slot_index]
|
41
|
-
else:
|
42
|
-
return f"[I{slot_index-9}]"
|
43
|
-
|
44
|
-
|
45
|
-
class SymbolicMoveAction(BaseModel):
|
46
|
-
""" "Moves an item from one slot to another"""
|
47
|
-
|
48
|
-
slot_from: int
|
49
|
-
slot_to: int
|
50
|
-
quantity: int
|
51
|
-
action_type: str = "move"
|
52
|
-
|
53
|
-
@field_validator("action_type", mode="before")
|
54
|
-
def fix_action_type(cls, value) -> str:
|
55
|
-
return "move"
|
56
|
-
|
57
|
-
@field_validator("slot_from", "slot_to", mode="before")
|
58
|
-
def transform_str_to_int(cls, value) -> int:
|
59
|
-
# if value is a string like [A1] or [I1], convert it to an integer
|
60
|
-
if isinstance(value, str):
|
61
|
-
try:
|
62
|
-
return convert_to_slot_index(value)
|
63
|
-
except ValueError:
|
64
|
-
raise AttributeError(
|
65
|
-
"slot_from and slot_to must be [0] or [A1] to [C3] or [I1] to [I36]"
|
66
|
-
)
|
67
|
-
return value
|
68
|
-
|
69
|
-
@field_validator("quantity", mode="before")
|
70
|
-
def transform_quantity(cls, value) -> int:
|
71
|
-
if isinstance(value, str):
|
72
|
-
try:
|
73
|
-
return int(value)
|
74
|
-
except ValueError:
|
75
|
-
raise AttributeError("quantity must be an integer")
|
76
|
-
return value
|
77
|
-
|
78
|
-
@model_validator(mode="after")
|
79
|
-
def validate(self):
|
80
|
-
if self.slot_from == self.slot_to:
|
81
|
-
raise AttributeError("slot_from and slot_to must be different")
|
82
|
-
if self.slot_from < 0 or self.slot_from > 45:
|
83
|
-
raise AttributeError("slot_from must be between 0 and 45")
|
84
|
-
if self.slot_to < 1 or self.slot_to > 45:
|
85
|
-
raise AttributeError("slot_to must be between 1 and 45")
|
86
|
-
if self.quantity < 1 or self.quantity > 64:
|
87
|
-
raise AttributeError("quantity must be between 1 and 64")
|
88
|
-
|
89
|
-
def to_action_dict(self) -> dict:
|
90
|
-
return {
|
91
|
-
"inventory_command": [self.slot_from, self.slot_to, self.quantity],
|
92
|
-
}
|
93
|
-
|
94
|
-
|
95
|
-
class SymbolicSmeltAction(BaseModel):
|
96
|
-
"""Smelts an item and moves the result into a new slot"""
|
97
|
-
|
98
|
-
slot_from: int
|
99
|
-
slot_to: int
|
100
|
-
quantity: int
|
101
|
-
action_type: str = "smelt"
|
102
|
-
|
103
|
-
@field_validator("action_type", mode="before")
|
104
|
-
def fix_action_type(cls, value) -> str:
|
105
|
-
return "smelt"
|
106
|
-
|
107
|
-
@field_validator("slot_from", "slot_to", mode="before")
|
108
|
-
def transform_str_to_int(cls, value) -> int:
|
109
|
-
# if value is a string like [A1] or [I1], convert it to an integer
|
110
|
-
if isinstance(value, str):
|
111
|
-
try:
|
112
|
-
return convert_to_slot_index(value)
|
113
|
-
except ValueError:
|
114
|
-
raise AttributeError(
|
115
|
-
"slot_from and slot_to must be [0] or [A1] to [C3] or [I1] to [I36]"
|
116
|
-
)
|
117
|
-
return value
|
118
|
-
|
119
|
-
@field_validator("quantity", mode="before")
|
120
|
-
def transform_quantity(cls, value) -> int:
|
121
|
-
if isinstance(value, str):
|
122
|
-
try:
|
123
|
-
return int(value)
|
124
|
-
except ValueError:
|
125
|
-
raise AttributeError("quantity must be an integer")
|
126
|
-
return value
|
127
|
-
|
128
|
-
@model_validator(mode="after")
|
129
|
-
def validate(self):
|
130
|
-
if self.slot_from == self.slot_to:
|
131
|
-
raise AttributeError("slot_from and slot_to must be different")
|
132
|
-
if self.slot_from < 0 or self.slot_from > 45:
|
133
|
-
raise AttributeError("slot_from must be between 0 and 45")
|
134
|
-
if self.slot_to < 1 or self.slot_to > 45:
|
135
|
-
raise AttributeError("slot_to must be between 1 and 45")
|
136
|
-
if self.quantity < 1 or self.quantity > 64:
|
137
|
-
raise AttributeError("quantity must be between 1 and 64")
|
138
|
-
|
139
|
-
def to_action_dict(self) -> dict:
|
140
|
-
return {
|
141
|
-
"smelt": [self.slot_from, self.slot_to, self.quantity],
|
142
|
-
}
|
143
|
-
|
144
|
-
|
145
|
-
class ThinkAction(BaseModel):
|
146
|
-
"""Think about the answer before answering"""
|
147
|
-
|
148
|
-
thought: str
|
149
|
-
|
150
|
-
def to_action_dict(self) -> dict:
|
151
|
-
return {}
|
152
|
-
|
153
|
-
|
154
|
-
class SearchAction(BaseModel):
|
155
|
-
"""Searches for a relevant document in the wiki"""
|
156
|
-
|
157
|
-
search_string: str
|
158
|
-
|
159
|
-
def to_action_dict(self) -> dict:
|
160
|
-
return {
|
161
|
-
"search": self.search_string,
|
162
|
-
}
|
163
|
-
|
164
|
-
|
165
|
-
class RealActionInteraction(BaseModel):
|
166
|
-
mouse_direction_x: float = 0
|
167
|
-
mouse_direction_y: float = 0
|
168
|
-
right_click: bool = False
|
169
|
-
left_click: bool = False
|
170
|
-
|
171
|
-
@field_validator("mouse_direction_x", "mouse_direction_y")
|
172
|
-
def prevent_zero(cls, v):
|
173
|
-
if v > 10:
|
174
|
-
return 10
|
175
|
-
elif v < -10:
|
176
|
-
return -10
|
177
|
-
return v
|
178
|
-
|
179
|
-
def to_action_dict(self) -> dict:
|
180
|
-
return {
|
181
|
-
"camera": [self.mouse_direction_x, self.mouse_direction_y],
|
182
|
-
"use": int(self.right_click),
|
183
|
-
"attack": int(self.left_click),
|
184
|
-
}
|
185
|
-
|
186
|
-
|
187
|
-
class StopAction(BaseModel):
|
188
|
-
"""
|
189
|
-
Action that model can take to stop planning - decide impossible to continue
|
190
|
-
Note: also known as the "impossible" action
|
191
|
-
"""
|
192
|
-
|
193
|
-
reason: str = ""
|
194
|
-
|
195
|
-
|
196
|
-
class NoOp(SymbolicMoveAction):
|
197
|
-
"""No operation action - special instance of move"""
|
198
|
-
|
199
|
-
def __init__(self):
|
200
|
-
super().__init__(slot_from=0, slot_to=1, quantity=1)
|
201
|
-
self.slot_to = 0
|
202
|
-
|
203
|
-
def __call__(self, *args, **kwargs):
|
204
|
-
return None
|
205
|
-
|
206
|
-
def __str__(self):
|
207
|
-
return "NoOp"
|
208
|
-
|
209
|
-
|
210
|
-
# when symbolic action is true, can either move objects around or smelt
|
211
|
-
SymbolicAction = SymbolicMoveAction # | SymbolicSmeltAction
|
212
|
-
|
213
|
-
# when symbolic action is false, then need to use mouse to move things around, but can use smelt action
|
214
|
-
RealAction = RealActionInteraction | SymbolicSmeltAction
|
215
|
-
|
216
|
-
|
217
|
-
class PydanticSymbolicAction(BaseModel):
|
218
|
-
root: Union[SymbolicMoveAction, SymbolicSmeltAction]
|
environments/env_real.py
DELETED
@@ -1,316 +0,0 @@
|
|
1
|
-
from typing import Sequence, Union
|
2
|
-
|
3
|
-
import numpy as np
|
4
|
-
import json
|
5
|
-
|
6
|
-
from loguru import logger
|
7
|
-
|
8
|
-
from plancraft.environments.actions import RealAction
|
9
|
-
|
10
|
-
try:
|
11
|
-
from minerl.env import _singleagent
|
12
|
-
from minerl.herobraine.env_specs.human_controls import HumanControlEnvSpec
|
13
|
-
from minerl.herobraine.hero import handlers, mc, spaces
|
14
|
-
from minerl.herobraine.hero.handler import Handler
|
15
|
-
from minerl.herobraine.hero.handlers.agent.action import Action
|
16
|
-
from minerl.herobraine.hero.handlers.agent.start import InventoryAgentStart
|
17
|
-
from minerl.herobraine.hero.handlers.translation import TranslationHandler
|
18
|
-
|
19
|
-
class InventoryCommandAction(Action):
|
20
|
-
"""
|
21
|
-
Handler which lets agents programmatically interact with an open container
|
22
|
-
|
23
|
-
Using this - agents can move a chosen quantity of items from one slot to another.
|
24
|
-
"""
|
25
|
-
|
26
|
-
def to_string(self):
|
27
|
-
return "inventory_command"
|
28
|
-
|
29
|
-
def xml_template(self) -> str:
|
30
|
-
return str("<InventoryCommands/>")
|
31
|
-
|
32
|
-
def __init__(self):
|
33
|
-
self._command = "inventory_command"
|
34
|
-
# first argument is the slot to take from
|
35
|
-
# second is the slot to put into
|
36
|
-
# third is the count to take
|
37
|
-
super().__init__(
|
38
|
-
self.command,
|
39
|
-
spaces.Tuple(
|
40
|
-
(
|
41
|
-
spaces.Discrete(46),
|
42
|
-
spaces.Discrete(46),
|
43
|
-
spaces.Discrete(64),
|
44
|
-
)
|
45
|
-
),
|
46
|
-
)
|
47
|
-
|
48
|
-
def from_universal(self, x):
|
49
|
-
return np.array([0, 0, 0], dtype=np.int32)
|
50
|
-
|
51
|
-
class SmeltCommandAction(Action):
|
52
|
-
"""
|
53
|
-
An action handler for smelting an item
|
54
|
-
We assume smelting is immediate.
|
55
|
-
@TODO: might be interesting to explore using the smelting time as an additional planning parameter.
|
56
|
-
|
57
|
-
Using this agents can smelt items in their inventory.
|
58
|
-
"""
|
59
|
-
|
60
|
-
def __init__(self):
|
61
|
-
self._command = "smelt"
|
62
|
-
# first argument is the slot to take from
|
63
|
-
# second is the slot to put into
|
64
|
-
# third is the count to smelt
|
65
|
-
super().__init__(
|
66
|
-
self.command,
|
67
|
-
spaces.Tuple(
|
68
|
-
(
|
69
|
-
spaces.Discrete(46),
|
70
|
-
spaces.Discrete(46),
|
71
|
-
spaces.Discrete(64),
|
72
|
-
)
|
73
|
-
),
|
74
|
-
)
|
75
|
-
|
76
|
-
def to_string(self):
|
77
|
-
return "smelt"
|
78
|
-
|
79
|
-
def xml_template(self) -> str:
|
80
|
-
return str("<SmeltCommands/>")
|
81
|
-
|
82
|
-
def from_universal(self, x):
|
83
|
-
return np.array([0, 0, 0], dtype=np.int32)
|
84
|
-
|
85
|
-
class InventoryResetAction(Action):
|
86
|
-
def __init__(self):
|
87
|
-
self._command = "inventory_reset"
|
88
|
-
super().__init__(self._command, spaces.Text([1]))
|
89
|
-
|
90
|
-
def to_string(self) -> str:
|
91
|
-
return "inventory_reset"
|
92
|
-
|
93
|
-
def to_hero(self, inventory_items: list[dict]):
|
94
|
-
return "{} {}".format(self._command, json.dumps(inventory_items))
|
95
|
-
|
96
|
-
def xml_template(self) -> str:
|
97
|
-
return "<InventoryResetCommands/>"
|
98
|
-
|
99
|
-
def from_universal(self, x):
|
100
|
-
return []
|
101
|
-
|
102
|
-
MINUTE = 20 * 60
|
103
|
-
|
104
|
-
class CustomInventoryAgentStart(InventoryAgentStart):
|
105
|
-
def __init__(self, inventory: list[dict[str, Union[str, int]]]):
|
106
|
-
super().__init__({item["slot"]: item for item in inventory})
|
107
|
-
|
108
|
-
class CraftingTableOnly(Handler):
|
109
|
-
def to_string(self):
|
110
|
-
return "start_with_crafting_table"
|
111
|
-
|
112
|
-
def xml_template(self) -> str:
|
113
|
-
return "<CraftingTableOnly>true</CraftingTableOnly>"
|
114
|
-
|
115
|
-
class InventoryObservation(TranslationHandler):
|
116
|
-
"""
|
117
|
-
Handles GUI Workbench Observations for selected items
|
118
|
-
"""
|
119
|
-
|
120
|
-
def to_string(self):
|
121
|
-
return "inventory"
|
122
|
-
|
123
|
-
def xml_template(self) -> str:
|
124
|
-
return str("""<ObservationFromFullInventory flat="false"/>""")
|
125
|
-
|
126
|
-
def __init__(self, item_list, _other="other"):
|
127
|
-
item_list = sorted(item_list)
|
128
|
-
super().__init__(
|
129
|
-
spaces.Dict(
|
130
|
-
spaces={
|
131
|
-
k: spaces.Box(
|
132
|
-
low=0,
|
133
|
-
high=2304,
|
134
|
-
shape=(),
|
135
|
-
dtype=np.int32,
|
136
|
-
normalizer_scale="log",
|
137
|
-
)
|
138
|
-
for k in item_list
|
139
|
-
}
|
140
|
-
)
|
141
|
-
)
|
142
|
-
self.num_items = len(item_list)
|
143
|
-
self.items = item_list
|
144
|
-
|
145
|
-
def add_to_mission_spec(self, mission_spec):
|
146
|
-
pass
|
147
|
-
|
148
|
-
def from_hero(self, info):
|
149
|
-
return info["inventory"]
|
150
|
-
|
151
|
-
def from_universal(self, obs):
|
152
|
-
raise NotImplementedError(
|
153
|
-
"from_universal not implemented in InventoryObservation"
|
154
|
-
)
|
155
|
-
|
156
|
-
class PlancraftBaseEnvSpec(HumanControlEnvSpec):
|
157
|
-
def __init__(
|
158
|
-
self,
|
159
|
-
symbolic_action_space=False,
|
160
|
-
symbolic_observation_space=False,
|
161
|
-
max_episode_steps=2 * MINUTE,
|
162
|
-
inventory: Sequence[dict] = (),
|
163
|
-
preferred_spawn_biome: str = "plains",
|
164
|
-
resolution=[260, 180],
|
165
|
-
):
|
166
|
-
self.inventory = inventory
|
167
|
-
self.preferred_spawn_biome = preferred_spawn_biome
|
168
|
-
self.symbolic_action_space = symbolic_action_space
|
169
|
-
self.symbolic_observation_space = symbolic_observation_space
|
170
|
-
|
171
|
-
mode = "real"
|
172
|
-
if symbolic_action_space:
|
173
|
-
mode += "-symbolic-act"
|
174
|
-
else:
|
175
|
-
mode += "-real-act"
|
176
|
-
|
177
|
-
if symbolic_observation_space:
|
178
|
-
mode += "-symbolic-obs"
|
179
|
-
|
180
|
-
if symbolic_action_space:
|
181
|
-
cursor_size = 1
|
182
|
-
else:
|
183
|
-
cursor_size = 16
|
184
|
-
|
185
|
-
name = f"plancraft-{mode}-v0"
|
186
|
-
super().__init__(
|
187
|
-
name=name,
|
188
|
-
max_episode_steps=max_episode_steps,
|
189
|
-
resolution=resolution,
|
190
|
-
cursor_size_range=[cursor_size, cursor_size],
|
191
|
-
)
|
192
|
-
|
193
|
-
def create_agent_start(self) -> list[Handler]:
|
194
|
-
base_agent_start_handlers = super().create_agent_start()
|
195
|
-
return base_agent_start_handlers + [
|
196
|
-
CustomInventoryAgentStart(self.inventory),
|
197
|
-
handlers.PreferredSpawnBiome(self.preferred_spawn_biome),
|
198
|
-
handlers.DoneOnDeath(),
|
199
|
-
CraftingTableOnly(),
|
200
|
-
]
|
201
|
-
|
202
|
-
def create_observables(self) -> list[TranslationHandler]:
|
203
|
-
if self.symbolic_observation_space:
|
204
|
-
return [
|
205
|
-
handlers.POVObservation(self.resolution),
|
206
|
-
InventoryObservation([item["slot"] for item in self.inventory]),
|
207
|
-
]
|
208
|
-
return [handlers.POVObservation(self.resolution)]
|
209
|
-
|
210
|
-
def create_server_world_generators(self) -> list[Handler]:
|
211
|
-
# TODO the original biome forced is not implemented yet. Use this for now.
|
212
|
-
return [handlers.DefaultWorldGenerator(force_reset=True)]
|
213
|
-
|
214
|
-
def create_server_quit_producers(self) -> list[Handler]:
|
215
|
-
return [
|
216
|
-
handlers.ServerQuitFromTimeUp(
|
217
|
-
(self.max_episode_steps * mc.MS_PER_STEP)
|
218
|
-
),
|
219
|
-
handlers.ServerQuitWhenAnyAgentFinishes(),
|
220
|
-
]
|
221
|
-
|
222
|
-
def create_server_initial_conditions(self) -> list[Handler]:
|
223
|
-
return [
|
224
|
-
handlers.TimeInitialCondition(allow_passage_of_time=False),
|
225
|
-
handlers.SpawningInitialCondition(allow_spawning=True),
|
226
|
-
]
|
227
|
-
|
228
|
-
def create_actionables(self) -> list[TranslationHandler]:
|
229
|
-
"""
|
230
|
-
Symbolic env can move items around in the inventory using function
|
231
|
-
Real env can use camera/keyboard
|
232
|
-
"""
|
233
|
-
# Camera and mouse
|
234
|
-
if self.symbolic_action_space:
|
235
|
-
return [
|
236
|
-
InventoryCommandAction(),
|
237
|
-
SmeltCommandAction(),
|
238
|
-
InventoryResetAction(),
|
239
|
-
]
|
240
|
-
return [
|
241
|
-
handlers.KeybasedCommandAction(v, v) for k, v in mc.KEYMAP.items()
|
242
|
-
] + [
|
243
|
-
handlers.CameraAction(),
|
244
|
-
SmeltCommandAction(),
|
245
|
-
InventoryResetAction(),
|
246
|
-
]
|
247
|
-
|
248
|
-
def is_from_folder(self, folder: str) -> bool:
|
249
|
-
return False
|
250
|
-
|
251
|
-
def create_agent_handlers(self) -> list[Handler]:
|
252
|
-
return []
|
253
|
-
|
254
|
-
def create_mission_handlers(self):
|
255
|
-
return []
|
256
|
-
|
257
|
-
def create_monitors(self):
|
258
|
-
return []
|
259
|
-
|
260
|
-
def create_rewardables(self):
|
261
|
-
return []
|
262
|
-
|
263
|
-
def create_server_decorators(self) -> list[Handler]:
|
264
|
-
return []
|
265
|
-
|
266
|
-
def determine_success_from_rewards(self, rewards: list) -> bool:
|
267
|
-
return False
|
268
|
-
|
269
|
-
def get_docstring(self):
|
270
|
-
return self.__class__.__doc__
|
271
|
-
|
272
|
-
class RealPlancraft(_singleagent._SingleAgentEnv):
|
273
|
-
def __init__(
|
274
|
-
self,
|
275
|
-
inventory: list[dict],
|
276
|
-
preferred_spawn_biome="plains",
|
277
|
-
symbolic_action_space=False,
|
278
|
-
symbolic_observation_space=True,
|
279
|
-
resolution=[512, 512],
|
280
|
-
crop=True,
|
281
|
-
):
|
282
|
-
# NOTE: crop is only supported for resolution 512x512 (default)
|
283
|
-
self.crop = crop
|
284
|
-
self.resolution = resolution
|
285
|
-
env_spec = PlancraftBaseEnvSpec(
|
286
|
-
symbolic_action_space=symbolic_action_space,
|
287
|
-
symbolic_observation_space=symbolic_observation_space,
|
288
|
-
preferred_spawn_biome=preferred_spawn_biome,
|
289
|
-
inventory=inventory,
|
290
|
-
resolution=resolution,
|
291
|
-
)
|
292
|
-
super(RealPlancraft, self).__init__(env_spec=env_spec)
|
293
|
-
self.reset()
|
294
|
-
|
295
|
-
def step(self, action: RealAction | dict):
|
296
|
-
if not isinstance(action, dict):
|
297
|
-
action = action.to_action_dict()
|
298
|
-
obs, rew, done, info = super().step(action)
|
299
|
-
if "pov" in obs and self.crop and self.resolution == [512, 512]:
|
300
|
-
# crop at position x=174, y=170 with width=164 and height=173
|
301
|
-
obs["pov"] = obs["pov"][174 : 174 + 164, 170 : 168 + 173]
|
302
|
-
return obs, rew, done, info
|
303
|
-
|
304
|
-
def fast_reset(self, new_inventory: list[dict]):
|
305
|
-
super().step({"inventory_reset": new_inventory})
|
306
|
-
|
307
|
-
|
308
|
-
except ImportError:
|
309
|
-
|
310
|
-
class RealPlancraft:
|
311
|
-
def __init__(self, *args, **kwargs):
|
312
|
-
logger.warning(
|
313
|
-
"The 'minerl' package is required to use RealPlancraft. "
|
314
|
-
"Please install it using 'pip install plancraft[full]' or 'pip install minerl'."
|
315
|
-
)
|
316
|
-
raise ImportError("minerl package not found")
|