# -*- coding: utf-8 -*-
"""
Module implementing umd_MainWindow.
"""
import sys, os, re, time,sip
from PyQt4 import QtGui, QtCore
from dict4ini import DictIni
from Ui_ebook_txt import Ui_umd_MainWindow
class eBookException(Exception):
pass
class ebookParseWorker(QtCore.QThread):
'''QT中子线程内不能操作GUI界面,切记切记'''
def __init__(self, parent=None):
QtCore.QThread.__init__(self, parent)
self.exiting = False
self.isWait=False
self.data={}
def setVar(self, name, value):
self.data[name]=value
def __del__(self):
self.exiting = True
self.wait()
def buildRegx(self):
#self.alert(regx)
if self.data['chmChapterRegx'] is None:
regx='pages\[(\d+)\]\s*=\s*\[(.*)\];'
try:
self.chapterInfo=re.compile(self.data['chmChapterRegx'])
except:
self.alert('列表处理正则表达式不正确')
return False
self.chapterList=[]
return True
def getChapterList(self):
if os.path.exists(unicode(self.data['chmChapterJs'])) ==False:
self.alert('章节列表JS文件%s不存在'%(self.data['chmChapterJs']))
return False
if self.data['inputCharset'] is None or self.data['inputCharset'].strip()=='':
self.data['inputCharset']='gbk'
m=self.chapterInfo.findall(open(unicode(self.data['chmChapterJs']),'rb').read())
vol=''
for x in m:
oldChapterInfo=x[1].split(',')
newChapterInfo={}
if len(oldChapterInfo)==4:
if oldChapterInfo[3].strip('"').strip("'")[0:5]!='<img alt="" /><1:
self.alert('章节列表小于1,请检查列表正则表达式是否正确')
return False
return self.chapterList
def run(self):
#是否触发错误
error=False
self.alert("<span><strong>开始处理章节内容</strong>")
if self.data['inputCharset'] is None or self.data['inputCharset'].strip()=='':
self.data['inputCharset']='gbk'
if self.data['outputCharset'] is None or self.data['outputCharset'].strip()=='':
self.data['outputCharset']='gbk'
if self.data['oldChapterTxtDir'] is None or self.data['oldChapterTxtDir'].strip()=='' :
self.alert('原始文件存放位置为空')
error=True
if self.data['newChapterTxtDir'] is None or self.data['newChapterTxtDir'].strip()=='':
self.data['newChapterTxtDir']=self.data['oldChapterTxtDir']
if os.path.exists(unicode(self.data['newChapterTxtDir'])) ==False:
os.makedirs(unicode(self.data['newChapterTxtDir']))
i=0
vol=''
volname=''
tmpVolName=''
currentNum=0
self.emit(QtCore.SIGNAL("setProcegressBar(int)"),len(self.chapterList))
if self.data['chapterTxtRegx'] is None:
regx=u""
for x in self.data['chapterTxtRegx'].split("\n"):
#print x
try:
re.compile(x)
except:
self.alert(unicode('内容处理正则表达式[ %s ]部分不正确'%(x)))
error=True
#self.alert('开始处理')
if error==False:
for x in self.chapterList:
while self.isWait:
self.sleep(1)
#self.wait(1)#使用wait的时候控制台会有输出
currentNum+=1
#self.alert(x['chapterName'])
#return
filename= self.data['oldChapterTxtDir']+'/'+x['filename']+self.data['oldChapterTxtExt']
if x['volName']!=tmpVolName:
chapterCount=1
i+=1
vol='卷 '+str(i)+' '+x['volName']
if i>1:
self.alert(unicode('<strong><span style="color: red;">处理(%s)完毕!</span><strong>'%(str(i-1).zfill(2)+'卷 '+tmpVolName)))
tmpVolName=x['volName']
#volname=x['volName'].replace(u':',u'-')#必须去掉:否则可能出现文件名被截断的情况
self.alert(unicode('开始处理处理:%s……'%(str(i).zfill(2)+'卷 '+volname)))
else:
chapterCount+=1
if os.path.isfile(unicode(filename)):
content=open(unicode(filename)).read().decode(self.data['inputCharset']).encode('utf-8')
for regxSub in self.data['chapterTxtRegx'].split("\n"):
content=re.sub(regxSub,' ',content)
#去掉document.write('
content=re.sub("\s*document\.write\s*\(\s*['|\"]\s*", '',content)
#去掉");这种标签
content=re.sub("\s*['|\"]\s*\)\s*[;]?", "\r\n\r\n",content)
content=content.replace("\n","\r\n\r\n").replace("\");","\r\n\r\n")
content="\r\n\r\n "+content.replace("')",'').replace("
","\r\n\r\n")[4:]
#去掉多余的类似<strong>或</strong>的标签
content=re.sub("<[a-z\s/]+>", '',content)
if len(x['chapterName'].split(' '))>1:
chapter=x['chapterName'][x['chapterName'].index(' ')+1:]
else:
chapter=x['chapterName']
if len(x['volName'].split(' '))==2:
volname=x['volName'][x['volName'].index(' ')+1:]
else:
volname=x['volName']
#递归判断去除章节名前的数字
try:
int(chapter[0:4])
chapter=chapter[4:]
except:
try:
int(chapter[0:3])
chapter=chapter[3:]
except:
try:
int(chapter[0:2])
chapter=chapter[2:]
except:
try:
int(chapter[0:1])
chapter=chapter[1:]
except:
chapter=chapter
chapter=' 第'+str(chapterCount).zfill(3)+'章 '+chapter
filename=self.data['newChapterTxtDir']+'/第'+str(i).zfill(2)+'卷 '+volname+chapter+'.txt'
#self.alert(filename)
#return
filename=filename.strip()
try:
f=open(unicode(filename),'wb')
except:
self.alert(unicode('写入文件“%s”错误'%(filename)))
#self.alert('写入文件“%s”错误'%(filename))
'''try:
f.write(content.decode('utf-8').encode(self.data['outputCharset']))
except:
self.alert(unicode('写入文件(%s)失败!'%(filename)))
f.close()'''
f.write(content.decode('utf-8').encode(self.data['outputCharset']))
f.close()
#del content
#self.window.umd_progressBar.setValue(currentNum)
self.emit(QtCore.SIGNAL("updateProcegressBar(int)"),currentNum)
#self.window.umd_textEditMessageOutPut.append(unicode('处理(%s)完毕!'%(str(i).zfill(2)+'卷 '+volname+chapter)))
self.alert(unicode('处理文件(%s)完毕!'%(str(i).zfill(2)+'卷 '+volname+chapter)))
else:
self.alert(unicode('处理失败: 章节文件不存在'))
#self.alert('章节列表文件处理完毕')
#import ctypes
#libc = ctypes.CDLL('libc.so.6')
#libc.printf('Hello world!')
def alert(self, txt):
#self.window.updateStatuBar(txt)
self.emit(QtCore.SIGNAL("msg(QString)"),QtCore.QString(unicode(txt)))
return
class umd_MainWindow(QtGui.QMainWindow, Ui_umd_MainWindow):
"""
Class documentation goes here.
"""
def __init__(self, parent = None):
"""
Constructor
"""
QtGui.QMainWindow.__init__(self, parent)
try:
#某些版本必须使用reload(sys)来重新载入sys模块才包含有setdefaultencoding方法
reload(sys)
sys.setdefaultencoding('utf-8')
except:
sys.setappdefaultencoding('utf-8')
#self.bmpdir=os.path.join(os.path.dirname(os.path.abspath(sys.argv[0])), 'res')
self.setupUi(self)
#以下为窗口自动居中
screen = QtGui.QDesktopWidget().screenGeometry()
size = self.geometry()
self.move((screen.width()-size.width())/2, (screen.height()-size.height())/2)
#设置按钮为隐藏
#self.umd_pushButton_pause.setHidden(True)
self.umd_pushButton_pause.hide()
self.umd_pushButton_Resume.hide()
self.umd_pushButton_Stop.hide()
#载入上一次的配置文件
self.cfg=DictIni(os.getcwd()+'/ebook.ini',encoding= 'utf-8')
if len(self.cfg.recentSetting)>0:
self.loadRecentSetting()
#以下开始设置进度条
self.umd_progressBar.setRange(0, 100)
self.umd_progressBar.setValue(0)
#self.showFullScreen()
#self.showNormal()
#self.isMinimized()
#创建一个文件处理线程
self.thread = ebookParseWorker(parent)
#线程退出
self.connect(self.thread, QtCore.SIGNAL("finished()"), self.finished)
self.connect(self.thread, QtCore.SIGNAL("terminated()"), self.finished)
#线程输出
self.connect(self.thread, QtCore.SIGNAL("msg(QString)"), self.message)
self.connect(self.thread, QtCore.SIGNAL("updateProcegressBar(int)"), self.updateProcegressBar)
self.connect(self.thread, QtCore.SIGNAL("setProcegressBar(int)"), self.setProcegressBar)
#open('./log.txt','wb').write(self.bmpdir)
#线程定义结束
pass
def loadRecentSetting(self):
self.umd_lineEditChapterTxtNew.setText(self.cfg.recentSetting.newTxtDir.decode('utf-8'))
self.umd_textEditChapterContentRegx.setPlainText(self.cfg.recentSetting.chapterContentRegx.decode('utf-8'))
self.umd_lineEditChapterJs.setText(self.cfg.recentSetting.chapterJs.decode('utf-8'))
self.umd_lineEditChapterRegx.setText(self.cfg.recentSetting.chapterRegx.decode('utf-8'))
self.umd_lineEditChapterTxtExt.setText(self.cfg.recentSetting.txtExt.decode('utf-8'))
self.umd_lineEditChapterTxtOld.setText(self.cfg.recentSetting.oldTxtDir.decode('utf-8'))
inputCharsetIndex=self.umd_comboBoxInputCharset.findText(self.cfg.recentSetting.inputCharset.decode('utf-8'))
self.umd_comboBoxInputCharset.setCurrentIndex(inputCharsetIndex)
outputCharsetIndex=self.umd_comboBoxOutputCharset.findText(self.cfg.recentSetting.outputCharset.decode('utf-8'))
self.umd_comboBoxOutputCharset.setCurrentIndex(outputCharsetIndex)
umdChapterContentRegx=str(self.umd_textEditChapterContentRegx.toPlainText().toUtf8())
#self.umd_lineEditChapterJs.setText(umd_textEditChapterContentRegx)
umdChapterRegx=str(self.umd_lineEditChapterRegx.text().toUtf8())
umdChapterTxtExt=str(self.umd_lineEditChapterTxtExt.text().toUtf8())
umdChapterTxtOld=str(self.umd_lineEditChapterTxtOld.text().toUtf8())
umdChapterTxtNew=str(self.umd_lineEditChapterTxtNew.text().toUtf8())
umdChapterJs=str(self.umd_lineEditChapterJs.text().toUtf8())
#原始文件编码
umdInputCharset=str(self.umd_comboBoxInputCharset.currentText().toUtf8())
#新文件编码
umdOutputCharset=self.umd_comboBoxOutputCharset.currentText().toUtf8().__str__()
pass
def closeEvent(self, event):
reply = QtGui.QMessageBox.question(self, QtGui.QApplication.translate("umd_MainWindow", "消息", None, QtGui.QApplication.UnicodeUTF8),
QtGui.QApplication.translate("umd_MainWindow", "确认关闭窗口?", None, QtGui.QApplication.UnicodeUTF8), QtGui.QMessageBox.Yes, QtGui.QMessageBox.No)
if reply == QtGui.QMessageBox.Yes:
event.accept()
else:
event.ignore()
def alert(self, txt):
response = False
# buttons texts
Close = QtGui.QApplication.translate("umd_MainWindow", "关闭", None, QtGui.QApplication.UnicodeUTF8)
#RELOAD='reload'
#CANCEL='cancel'
message = QtGui.QMessageBox(self)
message.setText(QtGui.QApplication.translate("umd_MainWindow",txt, None, QtGui.QApplication.UnicodeUTF8),)
message.setWindowTitle(QtGui.QApplication.translate("umd_MainWindow", "消息", None, QtGui.QApplication.UnicodeUTF8),)
message.setIcon(QtGui.QMessageBox.Warning)
message.addButton(Close, QtGui.QMessageBox.AcceptRole)
#message.addButton(RELOAD, QtGui.QMessageBox.DestructiveRole)
#message.addButton(CANCEL, QtGui.QMessageBox.RejectRole)
#message.setDetailedText(txt)
message.exec_()
response = message.clickedButton().text()
'''
if response == SAVE:
fd = QtGui.QFileDialog(self)
newfile = fd.getSaveFileName()
if newfile:
s = codecs.open(newfile,'w','utf-8')
s.write(unicode(self.ui.editor_window.toPlainText()))
s.close()
self.ui.button_save.setEnabled(False)
# new file, remove old and add the new one to the watcher
if self.filename and str(newfile) != str(self.filename):
self.watcher.removePath(self.filename)
self.watcher.addPath(newfile)
self.filename = newfile
# reload the text in the editor
elif response == RELOAD:
s = codecs.open(self.filename,'r','utf-8').read()
self.ui.editor_window.setPlainText(s)
self.ui.button_save.setEnabled(False)
'''
@QtCore.pyqtSignature("")
def on_umd_toolButtonChapterTxtNew_clicked(self):
"""
Slot documentation goes here.
"""
#raise NotImplementedError
dlg=QtGui.QFileDialog(self)
oldDir=unicode(self.umd_lineEditChapterTxtNew.text())
if oldDir.strip()=='':
oldDir=os.getcwd()
self.chapterTxtDirOld = dlg.getExistingDirectory(self, QtGui.QApplication.translate("umd_MainWindow", "请选择处理后的TXT文件的位置", None, QtGui.QApplication.UnicodeUTF8), oldDir)
if os.path.isdir(self.chapterTxtDirOld):
self.umd_lineEditChapterTxtNew.setText(self.chapterTxtDirOld)
def setProcegressBar(self, i):
self.umd_progressBar.setRange(0, i)
pass
def updateProcegressBar(self, i):
self.umd_progressBar.setValue(i)
pass
def message(self, txt):
self.umd_textEditMessageOutPut.append(txt)
def finished(self):
#print '处理完毕'
self.umd_textEditMessageOutPut.append(unicode('所有内容处理完毕'))
#处理完毕,隐藏暂停和停止按钮
self.umd_pushButton_pause.hide()
self.umd_pushButton_Resume.hide()
self.umd_pushButton_Stop.hide()
self.umd_pushButton_submit.show()
self.umd_pushButton_close.show()
#self.umd_plainTextEditMessageOutPut.appendPlainText(unicode('所有内容处理完毕'))
@QtCore.pyqtSignature("")
def on_umd_pushButton_Resume_clicked(self):
self.thread.isWait=False
self.umd_pushButton_pause.show()
self.umd_pushButton_Resume.hide()
self.umd_pushButton_Stop.show()
pass
@QtCore.pyqtSignature("")
def on_umd_pushButton_Stop_clicked(self):
self.thread.terminate()
self.umd_pushButton_submit.show()
self.umd_pushButton_close.show()
self.umd_pushButton_pause.hide()
self.umd_pushButton_Resume.hide()
self.umd_pushButton_Stop.hide()
@QtCore.pyqtSignature("")
def on_umd_pushButton_pause_clicked(self):
self.thread.isWait=True
#隐藏按钮
self.umd_pushButton_pause.hide()
self.umd_pushButton_Resume.show()
self.umd_pushButton_Stop.show()
pass
@QtCore.pyqtSignature("")
def on_umd_toolButtonChapterTxtOld_clicked(self):
"""
Slot documentation goes here.
"""
dlg=QtGui.QFileDialog(self)
oldDir=unicode(self.umd_lineEditChapterTxtOld.text())
if oldDir.strip()=='':
oldDir=os.getcwd()
self.chapterTxtDirOld = dlg.getExistingDirectory(self, QtGui.QApplication.translate("umd_MainWindow", "请选择CHM导出的TXT文件的位置", None, QtGui.QApplication.UnicodeUTF8), oldDir)
if os.path.isdir(self.chapterTxtDirOld):
self.umd_lineEditChapterTxtOld.setText(self.chapterTxtDirOld)
@QtCore.pyqtSignature("")
def on_umd_toolButtonChapterJs_clicked(self):
"""
Slot documentation goes here.
"""
dlg=QtGui.QFileDialog(self)
oldDir=unicode(self.umd_lineEditChapterJs.text())
if oldDir.strip()=='':
oldDir=os.getcwd()
self.chapterJsFilename = dlg.getOpenFileName(self, QtGui.QApplication.translate("umd_MainWindow", "请选择目录列表JS位置", None, QtGui.QApplication.UnicodeUTF8), oldDir,QtGui.QApplication.translate("umd_MainWindow", "js文件(*.js);;文本文件(*.txt);;所有文件(*)", None, QtGui.QApplication.UnicodeUTF8))
if os.path.isfile(self.chapterJsFilename):
dir=os.path.dirname(unicode(self.chapterJsFilename))
#对于从CHM中展开的电子书,一般保持这样的结构
if dir[-2:]=='js':
dir=dir[0:len(dir)-2]+'txt'
#self.alert(dir[-2:])
self.umd_lineEditChapterJs.setText(self.chapterJsFilename)
self.umd_lineEditChapterTxtOld.setText(dir)
@QtCore.pyqtSignature("")
def on_umd_pushButton_submit_clicked(self):
"""
Slot documentation goes here.
"""
#raise NotImplementedError
umdChapterContentRegx=str(self.umd_textEditChapterContentRegx.toPlainText().toUtf8())
#print umdChapterContentRegx
#self.umd_lineEditChapterJs.setText(umd_textEditChapterContentRegx)
umdChapterRegx=str(self.umd_lineEditChapterRegx.text().toUtf8())
umdChapterTxtExt=str(self.umd_lineEditChapterTxtExt.text().toUtf8())
umdChapterTxtOld=str(self.umd_lineEditChapterTxtOld.text().toUtf8())
umdChapterTxtNew=str(self.umd_lineEditChapterTxtNew.text().toUtf8())
umdChapterJs=str(self.umd_lineEditChapterJs.text().toUtf8())
#原始文件编码
umdInputCharset=str(self.umd_comboBoxInputCharset.currentText().toUtf8())
#新文件编码
umdOutputCharset=self.umd_comboBoxOutputCharset.currentText().toUtf8().__str__()
'''
>>>print type(self.umd_comboBoxInputCharset.currentText())
>>>
>>>print type(self.umd_comboBoxInputCharset.currentText().toUtf8())
>>>
>>>print str(self.umd_comboBoxInputCharset.currentText()).encode('utf-8')
>>>选择编码
>>>dir(self.umd_comboBoxInputCharset.currentText())
>>>['KeepEmptyParts', 'NormalizationForm', 'NormalizationForm_C', 'NormalizationForm_D', 'NormalizationForm_KC', 'NormalizationForm_KD', 'SectionCaseInsensitiveSeps', 'SectionDefault', 'SectionFlag', 'SectionFlags', 'SectionIncludeLeadingSep', 'SectionIncludeTrailingSep', 'SectionSkipEmpty', 'SkipEmptyParts', 'SplitBehavior', '__add__', '__class__', '__contains__', '__delattr__', '__dict__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__getitem__', '__gt__', '__hash__', '__iadd__', '__imul__', '__init__', '__le__', '__len__', '__lt__', '__module__', '__mul__', '__ne__', '__new__', '__radd__', '__reduce__', '__reduce_ex__', '__repr__', '__rmul__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', '__weakref__', 'append', 'arg', 'at', 'capacity', 'chop', 'clear', 'compare', 'contains', 'count', 'endsWith', 'fill', 'fromAscii', 'fromLatin1', 'fromLocal8Bit', 'fromUtf8', 'indexOf', 'insert', 'isEmpty', 'isNull', 'isRightToLeft', 'isSimpleText', 'lastIndexOf', 'left', 'leftJustified', 'length', 'localeAwareCompare', 'mid', 'normalized', 'number', 'prepend', 'push_back', 'push_front', 'remove', 'repeated', 'replace', 'reserve', 'resize', 'right', 'rightJustified', 'section', 'setNum', 'simplified', 'size', 'split', 'squeeze', 'startsWith', 'toAscii', 'toCaseFolded', 'toDouble', 'toFloat', 'toInt', 'toLatin1', 'toLocal8Bit', 'toLong', 'toLongLong', 'toLower', 'toShort', 'toUInt', 'toULong', 'toULongLong', 'toUShort', 'toUpper', 'toUtf8', 'trimmed', 'truncate']
>>>self.umd_comboBoxInputCharset.currentText().__str__()
>>>选择编码
'''
#保存设置
self.cfg.recentSetting.oldTxtDir=umdChapterTxtOld
self.cfg.recentSetting.newTxtDir=umdChapterTxtNew
self.cfg.recentSetting.txtExt=umdChapterTxtExt
self.cfg.recentSetting.chapterJs=umdChapterJs
self.cfg.recentSetting.chapterRegx=umdChapterRegx
self.cfg.recentSetting.chapterContentRegx=umdChapterContentRegx
self.cfg.recentSetting.inputCharset=umdInputCharset
self.cfg.recentSetting.outputCharset=umdOutputCharset
self.cfg.save()
#开始操作
#self.alert(umdChapterContentRegx)
#重置进度条
self.umd_progressBar.setValue(0)
#重置状态窗口
self.umd_textEditMessageOutPut.clear()
#self.umd_plainTextEditMessageOutPut.clear()
'''self.thread.setVar('chmChapterRegx', unicode(umdChapterRegx))
self.thread.setVar('chmChapterJs', unicode(umdChapterJs))
self.thread.setVar('oldChapterTxtDir', unicode(umdChapterTxtOld))
self.thread.setVar('inputCharset', unicode(umdInputCharset))
self.thread.setVar('oldChapterTxtExt', unicode(umdChapterTxtExt))
self.thread.setVar('chapterTxtRegx', unicode(umdChapterContentRegx))
self.thread.setVar('newChapterTxtDir', unicode(umdChapterTxtNew))
self.thread.setVar('outputCharset', unicode(umdOutputCharset))'''
self.thread.setVar('chmChapterRegx', umdChapterRegx)
self.thread.setVar('chmChapterJs', umdChapterJs)
self.thread.setVar('oldChapterTxtDir', umdChapterTxtOld)
self.thread.setVar('inputCharset', umdInputCharset)
self.thread.setVar('oldChapterTxtExt', umdChapterTxtExt)
self.thread.setVar('chapterTxtRegx', umdChapterContentRegx)
self.thread.setVar('newChapterTxtDir', umdChapterTxtNew)
self.thread.setVar('outputCharset', umdOutputCharset)
#print self.thread.data
#开始执行
self.umd_pushButton_submit.setHidden(True)
self.umd_pushButton_close.hide()
#显示暂停和停止窗口
self.umd_pushButton_pause.show()
self.umd_pushButton_Stop.show()
#
self.thread.buildRegx()
self.thread.getChapterList()
self.thread.start()
@QtCore.pyqtSignature("")
def on_umd_pushButton_close_clicked(self):
self.close()
if __name__ == "__main__