spyoungtech · January 28, 2025 16:10
diff --git a/0.example.py b/0.example.py
 from parse_repo import parse
 with open('6.0.5.md') as f:
  text = f.read()
 images = parse(text)
 for image in images:
  print('Image Digest:', image.digest)
  print('Layers:')
  for layer in image.layers:
    print('\t', "Digest:", layer.digest)
    print('\t', "Size:", layer.size)
    print('\t', "Last Modified:", layer.modified_date)
diff --git a/0.parse_repo.py b/0.parse_repo.py
 import re
 from typing import Sequence
 from dataclasses import dataclass
 import datetime

 @dataclass
 class Layer:
    digest: str
    modified_date: datetime.datetime
    size: int
    mime: str


 @dataclass
 class Image:
    digest: str
    layers: Sequence[Layer]


 def parse_iter(text):
    image_digest_pattern = r'Image ID\:\s+\`(sha256\:[a-f0-9]+)\`'
    layers_pattern = r'\s+(\-\s+\`(sha256\:[a-f0-9]+)\`\s+Last Modified\:\s+((?:.*)GMT)\s+Size\:\s+((?:.*)\(?\d+(?:\.\d+)? (?:bytes|B)\)?)\s+MIME:\s+(.*)$)'
    layer_parts = text.split('Layers:')[1:]
    layer_groups = []
    layer_dt_format = '%a, %d %b %Y %H:%M:%S GMT'
    for part in layer_parts:
        part_layers = []
        for layer_group in re.findall(layers_pattern, part):
            _, layer_digest, layer_modified_date, size, mime = layer_group
            layer_modified_date = datetime.datetime.strptime(layer_modified_date, layer_dt_format)
            if ' B' in size:
                size_pattern = r'(\d+)\.\d+ B'
            else:
                size_pattern = r'(?:.*)\((\d+) bytes\)'
            size = int(re.match(size_pattern, size).groups()[0])
            layer = Layer(digest=layer_digest,
                          modified_date=layer_modified_date,
                          size=size,
                          mime=mime
                          )
            part_layers.append(layer)
        layer_groups.append(part_layers)
    image_digests = []
    for image_digest in re.findall(image_digest_pattern, text):
        image_digests.append(image_digest)
    assert len(image_digests) == len(layer_groups)
    for image_digest, layer_groups in zip(image_digests, layer_groups):
        yield Image(digest=image_digest, layers=layer_groups)


 def parse(*args, **kwargs):
    return list(parse_iter(*args, **kwargs))
diff --git a/6.0.5.md b/6.0.5.md
	from parse_repo import parse
	with open('6.0.5.md') as f:
	text = f.read()
	images = parse(text)
	for image in images:
	print('Image Digest:', image.digest)
	print('Layers:')
	for layer in image.layers:
	print('\t', "Digest:", layer.digest)
	print('\t', "Size:", layer.size)
	print('\t', "Last Modified:", layer.modified_date)
	import re
	from typing import Sequence
	from dataclasses import dataclass
	import datetime

	@dataclass
	class Layer:
	digest: str
	modified_date: datetime.datetime
	size: int
	mime: str


	@dataclass
	class Image:
	digest: str
	layers: Sequence[Layer]


	def parse_iter(text):
	image_digest_pattern = r'Image ID\:\s+\`(sha256\:[a-f0-9]+)\`'
	layers_pattern = r'\s+(\-\s+\`(sha256\:[a-f0-9]+)\`\s+Last Modified\:\s+((?:.)GMT)\s+Size\:\s+((?:.)\(?\d+(?:\.\d+)? (?:bytes\|B)\)?)\s+MIME:\s+(.*)$)'
	layer_parts = text.split('Layers:')[1:]
	layer_groups = []
	layer_dt_format = '%a, %d %b %Y %H:%M:%S GMT'
	for part in layer_parts:
	part_layers = []
	for layer_group in re.findall(layers_pattern, part):
	_, layer_digest, layer_modified_date, size, mime = layer_group
	layer_modified_date = datetime.datetime.strptime(layer_modified_date, layer_dt_format)
	if ' B' in size:
	size_pattern = r'(\d+)\.\d+ B'
	else:
	size_pattern = r'(?:.*)\((\d+) bytes\)'
	size = int(re.match(size_pattern, size).groups()[0])
	layer = Layer(digest=layer_digest,
	modified_date=layer_modified_date,
	size=size,
	mime=mime
	)
	part_layers.append(layer)
	layer_groups.append(part_layers)
	image_digests = []
	for image_digest in re.findall(image_digest_pattern, text):
	image_digests.append(image_digest)
	assert len(image_digests) == len(layer_groups)
	for image_digest, layer_groups in zip(image_digests, layer_groups):
	yield Image(digest=image_digest, layers=layer_groups)


	def parse(args, *kwargs):
	return list(parse_iter(args, *kwargs))