return ' '.join(divided_list) if __name__ == "__main__": from IO import IO ioer = IO() raw_novel_path = "../../BackUpSource/RawNovel.txt" raw_novel = ioer.ReadFile(raw_novel_path) raw_novel = "".join(raw_novel) text_procer = TextProc() chapters = text_procer.Divide2Chapter(raw_novel) for i in range(len(chapters)): chapters[i] = text_procer.RmPuntuation(chapters[i]) chapters[i] = text_procer.RmEnter(chapters[i]) chapters[i] = text_procer.Divide2Word(chapters[i]) ioer.WriteFile("\n".join(chapters), "../../BackUpSource/chapter_in_line") ioer.WriteFiles(chapters, "../../BackUpSource/Chapters/", "chapter", 1, 120) chapters = ioer.ReadFiles("../../BackUpSource/Chapters/", 'chapter', 1, 120) for i in chapters: print(i[:1])
class Generator(): #constructor def __init__(self, sizesFilename): self.io = IO() dataLines = self.io.ReadFile(sizesFilename) self.sizes = Sizes(dataLines) # generates more brands data def GenerateBrands(self, number, outFilename): clotheSizeCategories = self.sizes.getSizeCats() clotheSizeTypes = self.sizes.getSizeTypes() self.io.WriteSizesHeader(outFilename) for i in range(number): brand = 'brand_'+ str(i) url = 'url_' + str(i) offset = round(random.random(), 2) brandLines = self.PopulateBrand(brand, url, clotheSizeTypes, clotheSizeCategories, offset) self.io.WriteFile(outFilename, brandLines, 'a') # generates more people data def GeneratePeople(self, numPeople, outFilename): sizeTypesList = self.sizes.getSizeTypesList() peopleLines = self.PopulatePeople(numPeople, sizeTypesList) self.io.WritePeopleHeader(outFilename, sizeTypesList) self.io.WriteFile(outFilename, peopleLines, 'a') def PopulatePeople(self, numPeople, sizeTypesList): lines = [] sizesBounds = self.sizes.getSizesBounds() for i in range(numPeople): line = "" line = str(i) for size in sizeTypesList: line += '\t' + str( random.uniform(sizesBounds[size][0], sizesBounds[size][1]) ) lines.append(line) return lines def PopulateBrand(self, brand, url, catSizeTypes, catSizeCategories, offset): outLines = [] for clotheCat in catSizeTypes: sizeTypes = catSizeTypes[clotheCat] sizeCats = catSizeCategories[clotheCat] for sizeCat in sizeCats: for sizeType in sizeTypes: sizesList = self.sizes.getSizes(clotheCat, sizeCat, sizeType) for size in sizesList: ss = size.split('-') newSize = str( float(ss[0])+float(offset) ) + '-' + str( float(ss[1])+float(offset) ) outLines.append( sizeType + '\t' + newSize + '\t' + '-' + '\t' + brand + '\t' + url + '\t' + clotheCat + '\t' + sizeCat ) return outLines