<script src="displayHandles.js"></script>
</head>
<body>"""
# try multiple simplification
for s in range(nbSimple): # nb de simplification
subprocess.call([
"cp", outputFolder + "/" + iName + "-simpl" + str(s) + ".svg",
outputFolder + "/" + iName + "-simpl" + str(s + 1) + ".svg"
])
# print("cp","Glyphes/vectors/"+iName+"-simpl"+str(s)+".svg","Glyphes/vectors/"+iName+"-simpl"+str(s+1)+".svg")
subprocess.call([
"inkscape", "--verb=EditSelectAll", "--verb=SelectionSimplify",
"--verb=FileSave", "--verb=FileQuit",
outputFolder + "/" + iName + "-simpl" + str(s + 1) + ".svg"
])
# print("inkscape","--verb=EditSelectAll","--verb=SelectionSimplify","--verb=FileSave","--verb=FileQuit","Glyphes/vectors/"+iName+"-simpl"+str(s+1)+".svg")
# subprocess.call(["inkscape","--verb=EditSelectAll","--verb=SelectionSimplify","--export-plain-svg='Glyphes/vectors/"+iName+"-simpl"+str(s+1)+".svg'","Glyphes/vectors/"+iName+"-simpl"+str(s)+".svg"])
html += "\n<object class='obj' data='{}' width='200px' height='200px' onload='draw(this)'></object>".format(
iName + "-simpl" + str(s + 1) + ".svg")
html += """
</body>
</html>
"""
if ifHTML:
with open(outputFolder + "/" + iName + ".html", 'wb') as f:
f.write(html.encode('utf8'))
Bar.update()
area = extracthocr.zoning(imgPage, n, margin)
outputName = glyphStr + "-" + str(int(confidenceValue)) + "-" + str(pageNumber) + "-" + str(word_id) + "-" + str(globalAreaCounter) + ".png"
if glyphStr == ".": # to fix "." name
if not os.path.isdir(outputFolderFamily + ".point"):
os.mkdir(outputFolderFamily + ".point")
area.save(outputFolderFamily + ".point/" + outputName)
else:
if not os.path.isdir(outputFolderFamily + glyphStr):
os.mkdir(outputFolderFamily + glyphStr)
area.save(outputFolderFamily + "/" + glyphStr + "/" + outputName)
# area.save(outputFolder + outputName)
globalAreaCounter += 1
BarByPage.update()
# coordsCorpList.append(coordCrop)
if mode == "word":
words = firstPage.find_all(attrs={"class": u"ocrx_word"})
for word in words:
fontFamily = getFontFamily(word, stylised_nodes, fontStyles)
if len(specificStyles) > 0 and not fontFamily in specificStyles:
continue
word_id = word.get('id')
wordStr = word.findAll(text=True, recursive=False)
if len(wordStr) > 0:
wordStr = wordStr[0]
