def uploadImageToWordpress(self, imgResource, isCheckExisted=False):
"""Upload image resource to wordpress
Args:
imgResource (Resouce): evernote image Resouce
isCheckExisted (bool): whether check image is uploaded or not
Returns:
(bool, dict)
Raises:
"""
respInfo = {}
imgData = imgResource.data
imgBytes = imgData.body
imgDataSize = imgData.size
# guid:'f6956c30-ef0b-475f-a2b9-9c2f49622e35'
imgGuid = imgResource.guid
logging.debug("imgGuid=%s, imgDataSize=%s", imgGuid, imgDataSize)
curImg = utils.bytesToImage(imgBytes)
logging.debug("curImg=%s", curImg)
# # for debug
# curImg.show()
imgFormat = curImg.format # 'PNG'
imgSuffix = utils.ImageFormatToSuffix[imgFormat] # 'png'
imgMime = utils.ImageSuffixToMime[imgSuffix] # 'image/png'
# curDatetimeStr = utils.getCurDatetimeStr() # '20200307_173141'
processedGuid = imgGuid.replace(
"-", "") # 'f6956c30ef0b475fa2b99c2f49622e35'
# imgeFilename = "%s.%s" % (curDatetimeStr, imgSuffix) # '20200307_173141.png'
imgeFilename = "%s.%s" % (processedGuid, imgSuffix
) # 'f6956c30ef0b475fa2b99c2f49622e35.png'
isNeedUpload = True
if isCheckExisted:
generatedImgeUrl = self.wordpress.generateUploadedImageUrl(
imgeFilename)
# https://www.crifan.org/files/pic/uploads/2021/03/f60ea32cf4664b41922431f4ea015621.jpg
# 'url':'https://www.crifan.org/files/pic/uploads/2021/03/f60ea32cf4664b41922431f4ea015621-1.jpg'
isValid = utils.isValidImageUrl(
generatedImgeUrl, proxies=self.wordpress.requestsProxies)
if isValid:
logging.info("Found existed image %s", generatedImgeUrl)
isUploadImgOk = True
respInfo["url"] = generatedImgeUrl
isNeedUpload = False
if isNeedUpload:
isUploadImgOk, respInfo = self.wordpress.createMedia(
imgMime, imgeFilename, imgBytes)
logging.debug("%s to upload resource %s to wordpress",
isUploadImgOk, imgGuid)
return isUploadImgOk, respInfo
def uploadImageToWordpress(self, imgResource):
"""Upload image resource to wordpress
Args:
imgResource (Resouce): evernote image Resouce
Returns:
(bool, dict)
Raises:
"""
imgData = imgResource.data
imgBytes = imgData.body
imgDataSize = imgData.size
# guid:'f6956c30-ef0b-475f-a2b9-9c2f49622e35'
imgGuid = imgResource.guid
logging.debug("imgGuid=%s, imgDataSize=%s", imgGuid, imgDataSize)
curImg = utils.bytesToImage(imgBytes)
logging.debug("curImg=%s", curImg)
# # for debug
# curImg.show()
imgFormat = curImg.format # 'PNG'
imgSuffix = utils.ImageFormatToSuffix[imgFormat] # 'png'
imgMime = utils.ImageSuffixToMime[imgSuffix] # 'image/png'
# curDatetimeStr = utils.getCurDatetimeStr() # '20200307_173141'
processedGuid = imgGuid.replace(
"-", "") # 'f6956c30ef0b475fa2b99c2f49622e35'
# imgeFilename = "%s.%s" % (curDatetimeStr, imgSuffix) # '20200307_173141.png'
imgeFilename = "%s.%s" % (processedGuid, imgSuffix
) # 'f6956c30ef0b475fa2b99c2f49622e35.png'
isUploadImgOk, respInfo = self.wordpress.createMedia(
imgMime, imgeFilename, imgBytes)
logging.debug("%s to upload resource %s to wordpress", isUploadImgOk,
imgGuid)
return isUploadImgOk, respInfo
def updateNoteImageResouces(noteDetail, newResList):
"""Update note resources and content, with new resource and updated <en-media> content
Args:
noteDetail (Note): Evernote note with details
Returns:
updated note detail
Raises:
"""
validNewResList = []
originResList = noteDetail.resources
# originContent = noteDetail.content
# soup = BeautifulSoup(originContent, 'html.parser')
soup = crifanEvernote.noteContentToSoup(noteDetail)
# # for debug: try restore not show image
# resSoupList = soup.find_all("en-media")
for curIdx, curRes in enumerate(originResList):
foundMediaNode = crifanEvernote.findResourceSoup(curRes, soup=soup)
# # for debug: try restore not show image
# if not foundMediaNode:
# foundMediaNode = resSoupList[curIdx]
if foundMediaNode:
newRes = newResList[curIdx] # 'image/jpeg'
validNewResList.append(newRes)
newMime = newRes.mime
# newHashBytes = newRes.data.bodyHash # b'\xb8\xe8\xbb\xcc\xca\xc1\xdf&J\xbeV\xe2`\xa6K\xb7'
newResBytes = newRes.data.body
newHashStr = utils.calcMd5(
newResBytes) # '3da990710b87fcd56a84d644202b12c2'
foundMediaNode["type"] = newMime
foundMediaNode["hash"] = newHashStr
noteAttrs = foundMediaNode.attrs
# {'hash': '3da990710b87fcd56a84...44202b12c2', 'type': 'image/jpeg'}
# {'hash': 'f9fc018211046ccc0eb6...5f67d0c8d6', 'type': 'image/jpeg', 'width': '385'}
# special:
# <en-media hash="0bbf1712d4e9afe725dd51e701c7fae6" style="width: 788px; height: auto;" type="image/jpeg"></en-media>
# ->
# {'hash': '0bbf1712d4e9afe725dd51e701c7fae6', 'type': 'image/jpeg', 'width': '788', 'height': 'auto'} ?
hasWidthAttr = "width" in noteAttrs
hasHeightAttr = "height" in noteAttrs
if hasWidthAttr or hasHeightAttr:
newImg = utils.bytesToImage(newResBytes)
# newImgWidth = newImg.width
# newImgHeight = newImg.height
newImgWidth, newImgHeight = newImg.size
if hasWidthAttr:
attrWidthStr = noteAttrs["width"]
if attrWidthStr.isdigit():
curWidth = int(attrWidthStr)
if curWidth >= newImgWidth:
# old size is bigger than resized image size, so need remove old size
del foundMediaNode["width"]
else:
# is 'auto' ? -> keep not changed ?
logging.warning(
"not support ? img width value: %s",
attrWidthStr)
if hasHeightAttr:
attrHeightStr = noteAttrs["height"]
if attrHeightStr.isdigit():
curHeight = int(attrHeightStr)
if curHeight >= newImgHeight:
# old size is bigger than resized image size, so need remove old size
del foundMediaNode["height"]
else:
# is 'auto' ? -> keep not changed ?
logging.warning(
"not support ? img height value: %s",
attrHeightStr)
else:
# logging.warning("Not found resource: type=%s, hash=%s", curMime, curHashStr)
logging.warning("Not found resource: guid=%s, mime=%s",
curRes.guid, curRes.mime)
# not add to validNewResList
# newContent = soup.prettify()
# newContent = str(soup)
newContent = crifanEvernote.soupToNoteContent(soup)
noteDetail.content = newContent
# noteDetail.resources = newResList
noteDetail.resources = validNewResList
return noteDetail
def blurImageResource(self, imgRes, sensitiveInfoList):
"""Blur image resource
Note: NOT process gif image
Args:
imgRes (Resource): Note image resource
Returns:
new image resouce
Raises:
"""
isBlurred = False
# isGif = "gif" in imgRes.mime
# if isGif:
# logging.info("Omit process gif image: %s", imgRes.attributes.fileName)
# return imgRes
imgBytes = imgRes.data.body
matchResultDict = self.baiduOcr.isStrInImage(sensitiveInfoList,
imgBytes=imgBytes,
isMatchMultiple=True,
isRespShortInfo=False)
curImg = utils.bytesToImage(imgBytes)
for eachSensitiveInfo, eachMatchResult in matchResultDict.items():
isMatch, matchResultList = eachMatchResult
if isMatch:
for eachMatchResultDict in matchResultList:
matchStr, matchLocation = eachMatchResultDict # ' limao@xx1 ~/dev/crifan/gitbookgitbook_template/books/gitbook_demo master .make install', (49, 0, 72, 16)
logging.debug("matchStr=%s, matchLocation=%s", matchStr,
matchLocation)
posX, posY, posW, posH = matchLocation # (49, 0, 72, 16)
# matchBox = [posX, posY, posX + posW, posY + posH] # [49, 0, 121, 16]
# Note: add extrac width (and height) to makesure all sensitve info blured
extraWidth = 5
extraHeight = 2
matchBox = [
posX, posY, posX + posW + extraWidth,
posY + posH + extraHeight
]
cropBoxImg = curImg.crop(
matchBox
) # <PIL.Image.Image image mode=RGBA size=72x16 at 0x101B9CA90>
# Use GaussianBlur directly to blur the image 10 times
blurImg = cropBoxImg.filter(
ImageFilter.GaussianBlur(radius=10)
) # <PIL.Image.Image image mode=RGBA size=72x16 at 0x103530DF0>
curImg.paste(blurImg, matchBox)
# # for debug
# curImg.show()
isBlurred = True
if isBlurred:
logging.info("Blurred image: %s", imgRes.attributes.fileName)
# # for debug
# curImg.show()
newImgBytes = utils.imageToBytes(curImg)
newImgRes = crifanEvernote.genNewImgRes(imgRes, newImgBytes)
return newImgRes