File size: 8,227 Bytes
e41d92c |
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 |
import os
import os.path as osp
import json
import random
import transformers
from tqdm import tqdm
from multiprocessing import Pool
from functools import partial
from data_class import domNode, nodeBbox
MAX_TOKEN_LEN = 800
MIN_TOKEN_LEN = 200
MAX_IoU = 0
HTML2BBOX_TEMPLATE = [
"Where is the given dom tree?",
# Other template questions here...
]
BBOX2HTML_TEMPLATE = [
"Can you give me a description of the region <objs> in image?",
# Other template questions here...
]
class NodewithHtml(domNode):
def __init__(self, id, info=None, children=None, father=None, tokenizer=None, img_w=None, img_h=None, task=""):
super().__init__(id, info, children, father, tokenizer, img_w=img_w, img_h=img_h, task="")
self.dom_xml = self.get_dom_xml(mode="dom_xml")
self.dom_xml_with_bbox = self.get_dom_xml(mode="dom_xml_with_bbox")
def bbox2str(self) -> str:
x1_rel = round(max(0, (self.info.bbox.x1 / IMG_W)) * 1000)
y1_rel = round(max(0, (self.info.bbox.y1 / IMG_H)) * 1000)
x2_rel = round(min(0.999, (self.info.bbox.x2 / IMG_W)) * 1000)
y2_rel = round(min(0.999, (self.info.bbox.y2 / IMG_H)) * 1000)
coords = [x1_rel, y1_rel, x2_rel, y2_rel]
bbox_str = ["{:03}".format(coord) for coord in coords]
bbox_str = "[" + ", ".join(bbox_str) + "]"
return bbox_str
def center2str(self) -> str:
cx = round(max(0, min(0.999, self.info.bbox.cx / IMG_W)) * 1000)
cy = round(max(0, min(0.999, self.info.bbox.cy / IMG_H)) * 1000)
point = [cx, cy]
point_str = ["{:03}".format(p) for p in point]
point_str = "[" + ", ".join(point_str) + "]"
return point_str
def node2xml_withbbox(self) -> str:
text = self.info.text
bbox_str = self.bbox2str()
if self.is_text() or self.has_single_text_child():
pseudo_html_line = f"<text box={bbox_str} content=\"{text}\">"
elif self.is_img():
if self.class_for_caption == 'img':
pseudo_html_line = f"<img box={bbox_str} alt={text}>"
elif self.class_for_caption == 'svg':
pseudo_html_line = f"<svg box={bbox_str} alt={text}>"
else:
return ""
elif self.info.func == 'type':
pseudo_html_line = f"<input box={bbox_str}, content=\"{text}\">"
else:
if self.is_leaf():
return ""
else:
pseudo_html_line = f"<plain box={bbox_str}>"
return pseudo_html_line
def node2xml(self) -> str:
text = self.info.text
bbox_str = self.bbox2str()
center_str = self.center2str()
if self.is_text() or self.has_single_text_child():
pseudo_html_line = f"<text content=\"{text}\">"
elif self.is_img():
if self.class_for_caption == 'img':
pseudo_html_line = f"<img alt={text}>"
elif self.class_for_caption == 'svg':
pseudo_html_line = f"<svg alt={text}>"
else:
return ""
elif self.info.func == 'type':
pseudo_html_line = f"<input center={center_str}, content=\"{text}\">"
else:
if self.is_leaf():
return ""
else:
pseudo_html_line = f"<plain>"
return pseudo_html_line
def get_dom_xml(self, mode: str):
xml_func = self.node2xml if mode == 'dom_xml' else self.node2xml_withbbox
self_dom_xml = xml_func()
if not self_dom_xml:
return ""
dom_xml_list = [self_dom_xml]
if not self.is_func_leaf():
for child in self.children:
child_xml = getattr(child, mode)
if not child_xml:
return ""
child_xml_list = child_xml.strip().split('\n')
child_xml_list = ['\t' + line for line in child_xml_list]
child_xml_fmt = '\n'.join(child_xml_list)
dom_xml_list.append(child_xml_fmt)
dom_xml = '\n'.join(dom_xml_list)
return dom_xml
tokenizer = transformers.AutoTokenizer.from_pretrained(
"path/to/your/tokenizer",
trust_remote_code=True,
)
def get_str_token_len(str_: str) -> int:
return len(tokenizer.encode(str_))
def collect_nodes(dom: NodewithHtml, node_list: list[NodewithHtml]) -> list:
if dom.is_valid:
node_list.append(dom)
for child in dom.children:
collect_nodes(child, node_list)
return node_list
def selece_node(xml: str, bbox: nodeBbox, selected_dom_list: list[nodeBbox]) -> bool:
if not xml:
return False
try:
token_len = get_str_token_len(xml)
except:
return False
if token_len > MAX_TOKEN_LEN or token_len < MIN_TOKEN_LEN:
return False
if selected_dom_list:
max_IoUs = bbox.get_max_IoU(selected_dom_list)
if max_IoUs > MAX_IoU:
return False
return True
def make_qa(node_list: list[NodewithHtml], img_path: str):
random.shuffle(node_list)
dom_xml_list = []
dom_xml_with_bbox_list = []
ann_list = []
for node in node_list:
if selece_node(node.dom_xml_with_bbox, node.info.bbox, dom_xml_with_bbox_list):
bbox2html_prompt = random.choice(BBOX2HTML_TEMPLATE)
prompt = bbox2html_prompt.replace("<objs>", node.bbox2str())
caption = node.dom_xml_with_bbox
ann = {
"image": img_path,
"conversations": [
{
"from": "human",
"value": f"<image>\n{prompt}"
},
{
"from": "gpt",
"value": caption
}
]
}
ann_list.append(ann)
dom_xml_with_bbox_list.append(node.info.bbox)
return ann_list
def write_ann(ann_list, ann_path):
ann_dir = osp.dirname(ann_path)
if not osp.exists(ann_dir):
os.mkdir(ann_dir)
with open(ann_path, 'a', encoding='utf-8') as f:
for i in range(len(ann_list)):
ann_list[i]['id'] = i
ann_str = json.dumps(ann_list[i], ensure_ascii=False)
f.write(ann_str + '\n')
def single_proc(data_dir, language, sub_dir):
sub_path = osp.join(data_dir, sub_dir)
screen_shot_dir = osp.join(sub_path, 'screenshot')
dom_sub_dir = 'dom_svg' if language == 'zh' else 'dom_svg_en'
dom_dir = osp.join(sub_path, dom_sub_dir)
html_dir = osp.join(sub_path, 'html')
html_path = osp.join(html_dir, 'html_0.html')
if not osp.exists(html_path):
return []
html_content = open(html_path).readline().strip()
if html_content in ['https://www.qq.com/', 'https://music.163.com/']:
return []
all_annotations = []
for img_file in os.listdir(screen_shot_dir):
if 'tmp' in img_file:
continue
file_ids = img_file.split('.')[0].split('_')[1]
json_path = osp.join(dom_dir, f'dom_{file_ids}.json')
img_path = osp.join(screen_shot_dir, img_file)
if not osp.exists(json_path):
continue
dom_data = json.load(open(json_path))
try:
dom_node = NodewithHtml(tokenizer=tokenizer, img_w=IMG_W, img_h=IMG_H, task='dom', **dom_data)
except ValueError as e:
print(f"Json data error: {e}\n{json_path}")
continue
node_list = collect_nodes(dom_node, [])
annotations = make_qa(node_list, img_path)
all_annotations.extend(annotations)
return all_annotations
def main_multi(data_dir, dst_path, language):
pool = Pool(processes=16)
sub_dir_list = os.listdir(data_dir)
single_proc_partial = partial(single_proc, data_dir, language)
num = 0
for res in tqdm(pool.imap_unordered(single_proc_partial, sub_dir_list), total=len(sub_dir_list)):
write_ann(res, dst_path)
num += len(res)
IMG_W = 1920
IMG_H = 1080
data_dir = 'data_20240624'
dst_path = 'xxx.jsonl'
if __name__ == '__main__':
main_multi(data_dir, dst_path, 'en')
|