def import_folder(bookpath, bookname="", pages="*.xml"):
if not bookname:
bookname = path.split(bookpath)[1]
no_pages_total = len(glob(bookpath + "/*.xml"))
try:
book = Book.query.filter_by(name=bookname).one()
except NoResultFound:
book = Book(name=bookname, no_pages_total=no_pages_total)
book.no_pages_total = no_pages_total
print('Importing book "{}"...'.format(bookname))
cnt = 0
for xmlfile in sorted(glob(bookpath + "/" + pages)):
pagename = path.split(xmlfile)[1].split(".")[0]
print("Importing page {}...".format(pagename))
try:
page = Page.query.filter_by(book_id=book.id, name=pagename).one()
except NoResultFound:
page = Page(book=book, name=pagename)
root = etree.parse(xmlfile).getroot()
ns = {"ns": root.nsmap[None]}
# convert point notation from pagexml version 2013
for c in root.xpath("//ns:Coords[not(@points)]", namespaces=ns):
cc = []
for point in c.xpath("./ns:Point", namespaces=ns):
cc.append(point.attrib["x"] + "," + point.attrib["y"])
c.remove(point)
c.attrib["points"] = " ".join(cc)
textregions = root.xpath('//ns:TextRegion', namespaces=ns)
page.no_lines_segm = int(
root.xpath("count(//ns:TextLine)", namespaces=ns))
page.no_lines_gt = int(
root.xpath('count(//ns:TextLine/ns:TextEquiv[@index="0"])',
namespaces=ns))
page.no_lines_ocr = int(
root.xpath(
'count(//ns:TextLine[count'
'(./ns:TextEquiv[@index>0])>0])',
namespaces=ns))
page.data = etree.tounicode(root.getroottree()).replace(
"http://schema.primaresearch.org/PAGE/gts/pagecontent/2010-03-19",
"http://schema.primaresearch.org/PAGE/gts/pagecontent/2017-07-15")
cnt += 1
db_session.add(book)
db_session.commit()
print('{} pages imported for book {}.'.format(cnt, bookname))
def scan_bookfolder(bookfolder, img_subdir):
""" Scan bookfolder and write book info to database. """
books = glob(bookfolder + "//*/")
for bookpath in books:
bookname = path.split(bookpath[:-1])[1]
files = set([f.split(sep=".")[0] for f in glob(bookpath + img_subdir
+ "*.png")])
no_pages_total = len(files)
book = Book.query.filter_by(name=bookname).first()
if not book:
book = Book(name=bookname, no_pages_total=no_pages_total)
db_session.add(book)
else:
book.no_pages_total = no_pages_total
db_session.commit()
def add_page(book, xmlfile, commit=True, scale=None):
""" Add page from PageXML file to book (segment lines, update version...).
scale may either be int, {"rtype": int, ... "other": int} or None
TODO: add reading order
"""
text_direction = 'horizontal-rl'\
if book.name.endswith("_ar") else 'horizontal-lr'
bookpath = path.split(xmlfile)[0]
pagename = path.split(xmlfile)[1].split(".")[0]
page = Page.query.filter_by(book_id=book.id, name=pagename).first()
if not page:
page = Page(book=book, name=pagename)
page.data, page.no_lines_segm = \
pagexmllineseg(xmlfile, bookpath, text_direction=text_direction,
scale=scale)
if commit:
db_session.add(book)
db_session.commit()
return page.no_lines_segm
def editorsettings():
if current_user.is_anonymous:
email = "user@nashi"
else:
email = current_user.email
if request.method == "GET":
try:
s = EditorSettings.query.filter_by(email=email).one()
except NoResultFound:
return jsonify(status="fail")
return jsonify(status="success", settings=json.loads(s.settings))
if request.method == "POST":
try:
s = EditorSettings.query.filter_by(email=email).one()
except NoResultFound:
s = EditorSettings(email=email)
s.settings = json.dumps(request.get_json())
db_session.add(s)
db_session.commit()
return jsonify(status="success")
"""
text_direction = 'horizontal-rl'\
if book.name.endswith("_ar") else 'horizontal-lr'
bookpath = path.split(xmlfile)[0]
pagename = path.split(xmlfile)[1].split(".")[0]
page = Page.query.filter_by(book_id=book.id, name=pagename).first()
if not page:
page = Page(book=book, name=pagename)
page.data, page.no_lines_segm = \
pagexmllineseg(xmlfile, bookpath, text_direction=text_direction,
scale=scale)
if commit:
db_session.add(book)
db_session.commit()
return page.no_lines_segm
if __name__ == '__main__':
for bookpath in [
x for x in glob(app.config["LAREX_DIR"] + "/*") if path.isdir(x)
]:
book = create_book(bookpath)
for xmlfile in sorted(glob(bookpath + "/*.xml")):
add_page(book, xmlfile, commit=False)
db_session.add(book)
db_session.commit()