Programmiererforum

Code: Select all

def procesar_imagenes_html(doc):

script_dir = os.path.dirname(os.path.abspath(__file__))
attachments_dir = os.path.join(script_dir, ".attachments")

if not os.path.isdir(attachments_dir):
os.makedirs(attachments_dir)
print(f"{attachments_dir}")
else:
print(f"Directorio .attachments encontrado: {attachments_dir}")
img_pattern = re.compile(r'\(a href="[^"]+"\)\(img src="([^"]+)"[^>]*\/\)\(\/a\)')
paragraphs = list(doc.Paragraphs)

for paragraph in paragraphs:
match = img_pattern.search(paragraph.Range.Text)
if match:
img_url = match.group(1)
print(f"{img_url}")

try:
response = requests.get(img_url)
if response.status_code == 200:
img_name = unquote(os.path.basename(img_url))
img_name = img_name.replace(' ', '_')
img_path = os.path.join(attachments_dir, img_name)

with open(img_path, 'wb') as img_file:
img_file.write(response.content)
print(f"{img_path}")
match_range = paragraph.Range.Duplicate
match_range.Start = paragraph.Range.Start + match.start()
match_range.End = paragraph.Range.Start + match.end()
match_range.Delete()
paragraph.Range.InsertParagraphBefore()
table = doc.Tables.Add(paragraph.Range, 3, 1)
table.Borders.Enable = False  # Sin bordes

cell_sup = table.Cell(1, 1)
cell_sup.Range.Text = "|"
cell_sup.Range.ParagraphFormat.Alignment = win32.constants.wdAlignParagraphLeft
cell_sup.Range.ParagraphFormat.SpaceBefore = 0
cell_sup.Range.ParagraphFormat.SpaceAfter = 0

cell_img = table.Cell(2, 1)
image = cell_img.Range.InlineShapes.AddPicture(
FileName=img_path,
LinkToFile=False,
SaveWithDocument=True
)
if image.Height > max_height:
image.Height = max_height
cell_img.Range.ParagraphFormat.Alignment = win32.constants.wdAlignParagraphCenter

cell_inf = table.Cell(3, 1)
cell_inf.Range.Text = "|"
cell_inf.Range.ParagraphFormat.Alignment = win32.constants.wdAlignParagraphLeft
cell_inf.Range.ParagraphFormat.SpaceBefore = 0
cell_inf.Range.ParagraphFormat.SpaceAfter = 0

table.Columns.AutoFit()

print(f"Imagen insertada: {img_name}")
else:
print(f"Error al descargar la imagen: {img_url}")
except Exception as e:
print(f"Error al procesar la imagen {img_url}: {e}")

procesar_imagenes_html(doc)