def convert_legacy_folder_to_tkbs_format(src_path, dst_path):
try:
p = Document()
p.load_legacy_data(src_path)
p.export_tkbs_format(dst_path)
except Exception as e:
print("ERROR in convert_legacy_folder_to_tkbs_format with src_path " +
src_path)
print(e)
def make_pxml(res=None, f1=None, f2=None):
log(0,"pxml")
p = Document()
for f in factors:
p.set_factors(f[0], f[1], f[2])
if res is not None:
p.set_factors(res,f1,f2)
# directory containing TOC.xml
p.load_legacy_data(paper)
p.export_tkbs_format(os.path.join(paper, config['pxml_dir']))
log(1,"pxml")
pass
v = True
infolder = r'C:\_test_\in_0105' #CHANGE THIS
outfolder = r'C:\_test_\out' #CHANGE THIS
v and print("--- CREATING DATA to upload ---")
p = Document()
#p.set_factors(150, 1.7238, 0.67)
p.load_legacy_data(infolder)
exportdir = os.path.join(outfolder, "pagexml_for_upload")
prep_dir(exportdir)
p.export_tkbs_format(exportdir)
v and print("--- CONNECTING to server ---")
user = "******" #CHANGE THIS
key = "<password>" #CHANGE THIS
collec = "17989" #CHANGE THIS
tkbs = TranskribusClient(sServerUrl="https://transkribus.eu/TrpServer")
tkbs.auth_login(user, key, True)
#HTRmodelname = 'Test'
HTRmodelid = "10168" #CHANGE THIS
#dictName = "Hebrew_Test.dict" #CHANGE THIS
#print("session id: " + tkbs.getSessionId() + "\n=================")
v and print("--- UPLOADING data to server ---")
docid = upload(collec, exportdir, p.img_names_by_pgnum(),
p.pxml_names_by_pgnum(), p.title, user, "pipeline test", tkbs)
def convert_legacy_folder_to_tkbs_format(src_path, dst_path):
p = Document()
p.load_legacy_data(src_path)
p.export_tkbs_format(dst_path)
v and print("--- CREATING DATA to upload ---")
p = Document()
#p.set_factors(150, 1.7238, 0.67)
p.load_legacy_data(infolder)
teifolder = os.path.join(exportfolder, 'tei')
teifiles = glob.glob(teifolder + r'\*' + p.doc_title + r'*_tei.xml')
if len(teifiles) > 0:
v and print("TEI found, Skipping document " + p.doc_title)
continue
uniquename = p.doc_title + "_" + start
firstuploadtopdir = prep_dir(
os.path.join(workfolder, r'pagexml_for_upload'))
firstexportdir = prep_dir(os.path.join(firstuploadtopdir, uniquename))
p.export_tkbs_format(firstexportdir)
v and print("--- UPLOADING data to server ---")
docid = upload(collec, firstexportdir, p.img_names_by_pgnum(),
p.pxml_names_by_pgnum(), p.title, user, "pipeline test",
tkbs)
if docid <= 0:
print("ERROR - document failed to upload " + p.title)
continue
v and print("--- DOWNLOADING-1 doc for page ids ---")
tempdowndir = prep_dir(os.path.join(workfolder, "tempdowndir"))
target_dir = os.path.join(
tempdowndir, p.title + "_" + str(collec) + "_" + str(docid))
docjson = download(collec, str(docid), target_dir, tkbs,
p.tkbs_meta_filename)