libreoffice python_libreoffice python 操作word及excel文档的方法

1、开始、关闭libreoffice服务；

开始之前同步字体文件时间，是因为创建soffice服务时，服务会检查所需加载的文件的时间，如果其认为时间不符，则其可能会重新加载，耗时较长，因此需事先统一时间。

使用时如果需要多次调用，最后每次调用均开启后关闭，否则libreoffice会创建一个缓存文档并越用越大，处理时间会增加。

class OfficeProcess(object):

def __init__(self):

self.p = 0

subprocess.Popen('find /usr/share/fonts | xargs touch -m -t 201801010000.00', shell=True)

def start_office(self):

self.p = subprocess.Popen('soffice --pidfile=sof.pid --invisible --accept="socket,host=localhost,port=2002;urp;"', shell=True)

while True:

try:

local_context = uno.getComponentContext()

resolver = local_context.getServiceManager().createInstanceWithContext('com.sun.star.bridge.UnoUrlResolver', local_context)

resolver.resolve('uno:socket,host=localhost,port=2002;urp;StarOffice.ComponentContext')

return

except:

print(ts(), "wait for connecting soffice...")

time.sleep(1)

continue

def stop_office(self):

with open("sof.pid", "rb") as f:

try:

os.kill(int(f.read()), signal.SIGTERM)

self.p.wait()

except:

pass

2、init service manager

local_context = uno.getComponentContext()

service_manager = local_context.getServiceManager()

resolver = service_manager.createInstanceWithContext('com.sun.star.bridge.UnoUrlResolver', local_context)

self.ctx = resolver.resolve('uno:socket,host=localhost,port=2002;urp;StarOffice.ComponentContext')

self.smgr = self.ctx.ServiceManager

self.desktop = self.smgr.createInstanceWithContext('com.sun.star.frame.Desktop', self.ctx)

3、从二进制数据中读取doc文档

def ImportFromMemory(self, data):

istream = self.smgr.createInstanceWithContext('com.sun.star.io.SequenceInputStream', self.ctx)

istream.initialize((uno.ByteSequence(data), ))

pv = PropertyValue()

pv.Name = 'InputStream'

pv.Value = istream

self.doc = {'doc': []}

try:

self.document = self.desktop.loadComponentFromURL('private:stream/swriter', '_blank', 0, (pv, ))

self.text = self.document.getText()

except:

self.text = None

4、读取doc文档中的数据

def ExportToJson(self):

try:

l = self.__ParseText(self.text, self.__Callback(self.doc['doc']))

self.doc['length'] = l

except:

self.doc = {'doc': [], 'length': 0}

return json.dumps(self.doc)

@staticmethod

def __Callback(alist):

def Append(sth):

alist.append(sth)

return Append

def __ParseText(self, text, func):

l = 0

text_it = text.createEnumeration()

while text_it.hasMoreElements():

element = text_it.nextElement()

if element.supportsService('com.sun.star.text.Paragraph'):

l += self.__ParseParagraph(element, func)

elif element.supportsService('com.sun.star.text.TextTable'):

l += self.__ParseTable(element, func)

else:

pass

return l

def __ParseParagraph(self, paragraph, func):

p = {'paragraph': []}

l = 0

paragraph_it = paragraph.createEnumeration()

while paragraph_it.hasMoreElements():

portion = paragraph_it.nextElement()

if portion.TextPortionType == 'Text':

l += self.__ParsePortionText(portion, self.__Callback(p['paragraph']))

elif portion.TextPortionType == 'SoftPageBreak':

pass

elif portion.TextPortionType == 'TextField':

l += self.__ParsePortionText(portion, self.__Callback(p['paragraph']))

else:

l += self.__ParseTextContent(portion, self.__Callback(p['paragraph']))

if hasattr(paragraph, 'createContentEnumeration'):

l +=