def testConvertToText(self):
        rep = TextRepresentation(TestTextRepresentation.pdfFile,
                                 TEMPDIRUNITTEST, LOGDIR)

        rep.convertToText()
        MyFile.checkFileExists(TestTextRepresentation.tmpPdfFile)

        rep = TextRepresentation(TestTextRepresentation.textFile,
                                 TEMPDIRUNITTEST, LOGDIR)

        rep.convertToText()
        MyFile.checkFileExists(TestTextRepresentation.tmpTextFile)
    def testLoadTextFile(self):
        rep = TextRepresentation(TestTextRepresentation.textFile,
                                 TEMPDIRUNITTEST, LOGDIR)

        rep.convertToText()
        rep.loadTextFile()

        self.assertEqual(6, len(rep.sentencesList))
Esempio n. 3
0
 def convertToText(sourcePath, destinationPath, logDir):
     """Extract the textual information from a
        pdf.
     """
     tr = TextRepresentation(sourcePath, destinationPath, logDir)
     return tr.convertToText()