File size: 9,336 Bytes
d6a54fa |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 |
import os
import sys
import argparse
import subprocess
from pathlib import Path
from concurrent.futures import (
ProcessPoolExecutor,
as_completed,
)
from zipnn_decompress_file import (
decompress_file,
)
sys.path.append(
os.path.abspath(
os.path.join(
os.path.dirname(__file__),
"..",
)
)
)
RED = "\033[91m"
YELLOW = "\033[93m"
GREEN = "\033[92m"
RESET = "\033[0m"
def check_and_install_zipnn():
try:
import zipnn
except ImportError:
print("zipnn not found. Installing...")
subprocess.check_call(
[
sys.executable,
"-m",
"pip",
"install",
"zipnn",
]
)
import zipnn
def replace_in_file(file_path, old: str, new: str) -> None:
"""Given a file_path, replace all occurrences of `old` with `new` inpalce."""
with open(file_path, 'r') as file:
file_data = file.read()
file_data = file_data.replace(old, new)
with open(file_path, 'w') as file:
file.write(file_data)
def decompress_znn_files(
path=".",
delete=False,
force=False,
max_processes=1,
hf_cache=False,
model="",
branch="main",
):
import zipnn
overwrite_first=True
if model:
if not hf_cache:
raise ValueError(
"Must specify --hf_cache when using --model"
)
try:
from huggingface_hub import scan_cache_dir
except ImportError:
raise ImportError(
"huggingface_hub not found. Please pip install huggingface_hub."
)
cache = scan_cache_dir()
repo = next((repo for repo in cache.repos if repo.repo_id == model), None)
if repo is not None:
print(f"Found repo {model} in cache")
# Get the latest revision path
hash = ''
try:
with open(os.path.join(repo.repo_path, 'refs', branch), "r") as ref:
hash = ref.read()
except FileNotFoundError:
raise FileNotFoundError(f"Branch {branch} not found in repo {model}")
path = os.path.join(repo.repo_path, 'snapshots', hash)
file_list = []
directories_to_search = [
(
path,
[],
os.listdir(path),
)
]
for (
root,
_,
files,
) in directories_to_search:
for file_name in files:
if file_name.endswith(".znn"):
decompressed_path = file_name[:-4]
if not force and os.path.exists(
decompressed_path
):
#
if overwrite_first:
overwrite_first=False
user_input = (
input(
f"Decompressed files already exists; Would you like to overwrite them all (y/n)? "
)
.strip()
.lower()
)
if user_input not in (
"y",
"yes",
):
print(
f"No forced overwriting."
)
else:
print(
f"Overwriting all decompressed files."
)
force=True
#
if not force:
user_input = (
input(
f"{decompressed_path} already exists; overwrite (y/n)? "
)
.strip()
.lower()
)
if user_input not in (
"y",
"yes",
):
print(
f"Skipping {file_name}..."
)
continue
full_path = os.path.join(
root,
file_name,
)
file_list.append(full_path)
if file_list and hf_cache:
try:
from transformers.utils import (
SAFE_WEIGHTS_INDEX_NAME,
WEIGHTS_INDEX_NAME
)
except ImportError:
raise ImportError(
"Transformers not found. Please pip install transformers."
)
suffix = file_list[0].split('/')[-1].split('.')[-2] # get the one before .znn
if os.path.exists(os.path.join(path, SAFE_WEIGHTS_INDEX_NAME)):
print(f"{YELLOW}Fixing Hugging Face model json...{RESET}")
blob_name = os.path.join(path, os.readlink(os.path.join(path, SAFE_WEIGHTS_INDEX_NAME)))
replace_in_file(
file_path=blob_name,
old=f"{suffix}.znn",
new=f"{suffix}"
)
elif os.path.exists(os.path.join(path, WEIGHTS_INDEX_NAME)):
print(f"{YELLOW}Fixing Hugging Face model json...{RESET}")
blob_name = os.path.join(path, os.readlink(os.path.join(path, WEIGHTS_INDEX_NAME)))
replace_in_file(
file_path=blob_name,
old=f"{suffix}.znn",
new=f"{suffix}"
)
with ProcessPoolExecutor(
max_workers=max_processes
) as executor:
for file in file_list[:max_processes]:
future_to_file = {
executor.submit(
decompress_file,
file,
delete,
True,
hf_cache,
): file
for file in file_list[
:max_processes
]
}
file_list = file_list[max_processes:]
while future_to_file:
for future in as_completed(
future_to_file
):
file = future_to_file.pop(
future
)
try:
future.result()
except Exception as exc:
print(
f"{RED}File {file} generated an exception: {exc}{RESET}"
)
if file_list:
next_file = file_list.pop(
0
)
future_to_file[
executor.submit(
decompress_file,
next_file,
delete,
True,
hf_cache,
)
] = next_file
#
print(f"{GREEN}All files decompressed{RESET}")
if __name__ == "__main__":
check_and_install_zipnn()
parser = argparse.ArgumentParser(
description="Compresses all .znn files."
)
parser.add_argument(
"--path",
type=str,
help="Path to folder of files to decompress. If left empty, checks current folder.",
)
parser.add_argument(
"--delete",
action="store_true",
help="A flag that triggers deletion of a single compressed file instead of decompression",
)
parser.add_argument(
"--force",
action="store_true",
help="A flag that forces overwriting when decompressing.",
)
parser.add_argument(
"--max_processes",
type=int,
help="The amount of maximum processes.",
)
parser.add_argument(
"--hf_cache",
action="store_true",
help="A flag that indicates if the file is in the Hugging Face cache. Must either specify --model or --path to the model's snapshot cache.",
)
parser.add_argument(
"--model",
type=str,
help="Only when using --hf_cache, specify the model name or path. E.g. 'ibm-granite/granite-7b-instruct'",
)
parser.add_argument(
"--model_branch",
type=str,
default="main",
help="Only when using --model, specify the model branch. Default is 'main'",
)
args = parser.parse_args()
optional_kwargs = {}
if args.path is not None:
optional_kwargs["path"] = args.path
if args.delete:
optional_kwargs["delete"] = args.delete
if args.force:
optional_kwargs["force"] = args.force
if args.max_processes:
optional_kwargs["max_processes"] = (
args.max_processes
)
if args.hf_cache:
optional_kwargs["hf_cache"] = args.hf_cache
if args.model:
optional_kwargs["model"] = args.model
if args.model_branch:
optional_kwargs[
"branch"
] = args.model_branch
decompress_znn_files(**optional_kwargs)
|