def storeMPXML(self, lFiles):
"""
store files in lFiles as mpxml
"""
docDir = os.path.join(self.coldir + os.sep + 'col', self.docid)
doc = MultiPageXml.makeMultiPageXml(lFiles)
sMPXML = docDir + ".mpxml"
# print sMPXML
doc.write(sMPXML,
encoding="UTF-8",
pretty_print=True,
xml_declaration=True)
# trace("\t\t- validating the MultiPageXml ...")
# if not MultiPageXml.validate(doc):
# traceln(" *** WARNING: XML file is invalid against the schema: '%s'"%self.outputFileName)
# traceln(" Ok!")
return doc, sMPXML
def storeMPXML(self,lFiles):
"""
store files in lFiles as mpxml
"""
docDir = os.path.join(self.coldir+os.sep+'col',self.docid)
doc = MultiPageXml.makeMultiPageXml(lFiles)
## add cornerNode
lCells = PageXml.getChildByName(doc.getRootElement(),'TableCell')
for domNode in lCells:
cornerNode = libxml2.newNode('CornerPts')
cornerNode.setContent("0 1 2 3")
cornerNode.setNs(doc.getRootElement().ns())
domNode.addChild(cornerNode)
sMPXML = docDir+".mpxml"
doc.saveFormatFileEnc(sMPXML,"UTF-8",True)
return doc, sMPXML
def processParameters(self):
"""
what to do with the parameters provided by the command line
"""
if self.colid is None:
print('collection id missing!')
sys.exit(1)
self.bFullCol = self.docid != None
if self.bRegenerateMPXML and self.docid is not None:
l = glob.glob(os.path.join(self.coldir, sCOL, self.docid,
"*.pxml"))
doc = MultiPageXml.makeMultiPageXml(l)
outputFileName = os.path.join(
self.coldir, sCOL,
self.docid + TableProcessing.sMPXMLExtension)
doc.write(outputFileName,
xml_declaration=True,
encoding="UTF-8",
pretty_print=True)
return doc
return None
def mergeBaselineCells(self, coldir, colid, docid):
"""
Take a file (pxml) with stuff processed on Transkribus
Tale the CVL template tool xml (xml)
merge them
regenerate a mpxml
"""
xmlpath = os.path.abspath(os.path.join(coldir, sCOL, docid))
# print (xmlpath)
mpxml = xmlpath + ".mpxml"
mpxmldoc = etree.parse(mpxml)
lxml = glob.glob(os.path.join(xmlpath, "*.xml"))
pxmldoc = MultiPageXml.makeMultiPageXml(lxml)
lhtrxml = glob.glob(os.path.join(xmlpath, "*.pxml"))
mpxmldoc = MultiPageXml.makeMultiPageXml(lhtrxml)
lPXMLPage = PageXml.getChildByName(mpxmldoc.getroot(), 'Page')
lXMLPage = PageXml.getChildByName(pxmldoc.getroot(), 'Page')
assert len(lXMLPage) == len(lPXMLPage)
for i, cvlpage in enumerate(lXMLPage):
## remove TextRegion from xcvlpage
lTextRegions = PageXml.getChildByName(cvlpage, 'TextRegion')
for tr in lTextRegions:
tr.getparent().remove(tr)
pxmlpage = lPXMLPage[i]
lTL = []
lTextRegions = PageXml.getChildByName(pxmlpage, 'TextRegion')
for x in lTextRegions:
lTL.extend(PageXml.getChildByName(x, 'TextLine'))
ltable = PageXml.getChildByName(cvlpage, 'TableRegion')
if len(ltable) == 0:
raise "NO TABLE"
lCells = PageXml.getChildByName(ltable[0], 'TableCell')
lC = [Polygon(PageXml.getPointList(c)) for c in lCells]
lT = [Polygon(PageXml.getPointList(t)) for t in lTL]
for i, tl in enumerate(lT):
## normalization
lCoordsPoints = PageXml.getChildByName(lTL[i], 'Coords')
lCoordsB = PageXml.getChildByName(lTL[i], 'Baseline')
coordB = lCoordsB[0]
coord = lCoordsPoints[0]
iHeight = 30 # in pixel
x1, y1, x2, y2 = Polygon(
PageXml.getPointList(coordB)).getBoundingBox()
if coord is not None:
coord.set(
'points', "%d,%d %d,%d %d,%d %d,%d" %
(x1, y1 - iHeight, x2, y1 - iHeight, x2, y2, x1, y2))
tl = Polygon(PageXml.getPointList(coordB))
lOverlap = []
for _, c in enumerate(lC):
# print (lCells[j].get('row'),lCells[j].get('col'), self.signedOverlap(c,tl),tl.getBoundingBox(),c.getBoundingBox())
lOverlap.append(self.signedOverlap(
c, tl)) #.getBoundingBox()))
## region of the same size as the textline
# print (j,max(lOverlap),lOverlap.index(max(lOverlap)))
if max(lOverlap) == 0:
region = PageXml.createPageXmlNode('TextRegion')
cvlpage.append(region)
region.append(lTL[i])
else:
cell = lCells[lOverlap.index(max(lOverlap))]
cell.append(lTL[i])
# print (cell.get('row'),cell.get('col'),''.join(lTL[i].itertext()))
pxmldoc.write(mpxml)
