def getFileLanguage(workspace, filegroup, stopwordspath, index=0):
fgrppath = workspace + '/' + filegroup
_, _, files = os.walk(fgrppath).__next__()
inputfiles = []
for file in files:
if 'xml' in file:
inputfiles.append(file)
fgrp = defaultdict(int)
for input_file in inputfiles:
alignurl = fgrppath + '/' + input_file
pcgts = parse(alignurl, True)
page = pcgts.get_Page()
regions = page.get_TextRegion()
pagetext = ''
for region in regions:
pagetext += region.get_TextEquiv()[index].Unicode + ' '
lang = detect_language(pagetext, stopwordspath)
fgrp[lang] += 1
return max(fgrp, key=lambda k: fgrp[k])
def process(self):
"""
Performs the (text) recognition.
"""
mainIndex = self.parameter['mainIndex']
for (n, input_file) in enumerate(self.input_files):
alignurl = input_file.url
pcgts = parse(alignurl, True)
page = pcgts.get_Page()
regions = page.get_TextRegion()
pagecontent = ''
for region in regions:
regioncontent = ''
lines = region.get_TextLine()
for line in lines:
linecontent = ''
words = line.get_Word()
for word in words:
wordunicode = word.get_TextEquiv()[mainIndex].Unicode
word.add_TextEquiv(TextEquivType(Unicode=wordunicode))
linecontent += ' ' + wordunicode
line.add_TextEquiv(TextEquivType(Unicode=regioncontent))
regioncontent += '\n' + linecontent
region.add_TextEquiv(TextEquivType(Unicode=regioncontent))
pagecontent += '\n' + regioncontent
page.add_TextEquiv(TextEquivType(Unicode=pagecontent))
ID = concat_padded(self.output_file_grp, n)
self.log.info('creating file id: %s, name: %s, file_grp: %s', ID,
input_file.basename, self.output_file_grp)
# Use the input file's basename for the new file
# this way the files retain the same basenames.
out = self.workspace.add_file(
ID=ID,
file_grp=self.output_file_grp,
pageId=input_file.pageId,
basename=self.output_file_grp + '-' + input_file.basename,
mimetype=MIMETYPE_PAGE,
content=to_xml(pcgts),
)
self.log.info('created file %s', out)
def negative2zero(inputfile, outputfile):
print("Setting negative coords to zero..")
pcgts = parse(inputfile, silence=True)
page = pcgts.get_Page()
for attr in dir(page):
if "get_" in attr and "Region" in attr:
for regiondata in getattr(page,attr)():
if attr == "get_TextRegion":
for textline in regiondata.get_TextLine():
textcoords = textline.get_Coords()
textcoords.set_points(update_points(textcoords.get_points()))
regcoords = regiondata.get_Coords()
regcoords.set_points(update_points(regcoords.get_points()))
content = to_xml(pcgts)
with open(outputfile,"w") as fout:
fout.write(content)
def get_imagefilename(inputfile):
pcgts = parse(inputfile, silence=True)
page = pcgts.get_Page()
print(page.get_imageFilename())
def process(self):
"""
Performs the (text) recognition.
"""
d = dict()
# if "gt" not in d:
# d["gt"] = 0
inputfiles = self.input_files
for input_file in inputfiles:
# index = input_file.url.rfind('/')
# alignurl = input_file.url[:index] + '/' + self.input_file_grp + input_file.url[index:]
alignurl = input_file.url
pcgts = parse(alignurl, True)
page = pcgts.get_Page()
regions = page.get_TextRegion()
# find index of GT
gti = self.get_gt_index(regions)
if gti == -1:
# sigh. just give up for this file
continue
if gti not in d:
d['gt'] = 0
for region in regions:
lines = region.get_TextLine()
for line in lines:
if len(line.get_TextEquiv()) <= gti:
continue
gtline = line.get_TextEquiv()[gti].Unicode
d['gt'] += len(gtline)
#Type = line.get_TextEquiv()[0].dataType[9:]
# for i in range(2, len(line.get_TextEquiv())):
for i in range(0, gti):
# print("%d: %s".format(i, line.get_
# OCRType = line.get_TextEquiv()[i].dataType
# lindex = OCRType.find('OCR-D-')
# rindex = OCRType.find(Type)
# model = OCRType[lindex:rindex-1]
if i not in d:
d[i] = 0
# print(line.get_TextEquiv()[2].dataType)
unicodeline = line.get_TextEquiv()[i].Unicode
d[i] += distance(gtline, unicodeline)
# words = line.get_Word()
# for word in words:
# for ocr in word.get_TextEquiv():
# print(ocr.Unicode)
print(json.dumps(d))
def process(self):
"""
Performs the (text) recognition.
"""
linelang = defaultdict(int)
wordlang = defaultdict(int)
linefont = defaultdict(int)
wordfont = defaultdict(int)
inputfiles = self.input_files
for input_file in inputfiles:
alignurl = input_file.url
pcgts = parse(alignurl, True)
page = pcgts.get_Page()
regions = page.get_TextRegion()
for region in regions:
lines = region.get_TextLine()
for line in lines:
try:
llang = line.primaryLanguage
linelang[llang] += 1
except TypeError:
pass
try:
lfont = line.fontFamily
linefont[lfont] += 1
except TypeError:
pass
words = line.get_Word()
for word in words:
try:
wlang = word.language
wordlang[wlang] += 1
except TypeError:
pass
try:
wfont = word.get_TextStyle().fontFamily
wordfont[wfont] += 1
except TypeError:
pass
#predominant language
try:
lang = max(linelang, key=lambda k: linelang[k])
except TypeError:
try:
lang = max(wordlang, key=lambda k: wordlang[k])
except TypeError:
lang = 'German'
#predominant font
try:
font = max(linefont, key=lambda k: linefont[k])
except TypeError:
try:
font = max(wordfont, key=lambda k: wordfont[k])
except TypeError:
font = 'Antiqua'
print(lang)
print(font)