def processListIndent(curNote):
"""process list (ul/ol/...) indent
Args:
curNote (Note): evernote Note
Returns:
Note
Raises:
"""
# soup = utils.htmlToSoup(curNote.content)
# enNoteSoup = soup.find("en-note")
# soup = crifanEvernote.noteContentToSoup(curNote)
# Note: makesure removed top html, then following can search out ul and ul list with recursive=False
soup = crifanEvernote.noteContentToSoup(curNote, isKeepTopHtml=False)
# allSubUlSoupList = soup.find_all("ul")
# allSubUlSoupNum = len(allSubUlSoupList)
# logging.info("Found %d all sub level ul list", allSubUlSoupNum)
topUlSoupList = soup.find_all("ul", recursive=False)
topOlSoupList = soup.find_all("ol", recursive=False)
# for debug: need debug to make sure support ol
if topOlSoupList:
logging.info("Found ol list")
topListSoupList = []
topListSoupList.extend(topUlSoupList)
topListSoupList.extend(topOlSoupList)
if topListSoupList:
directUlSoupNum = len(topListSoupList)
logging.info("Found %d top level ul/ol list", directUlSoupNum)
for curUlIdx, eachListSoup in enumerate(topListSoupList):
logging.info("%s %s %s", "-" * 20, curUlIdx, "-" * 20)
logging.debug("before eachListSoup=%s", eachListSoup)
crifanEvernoteToWordpress.removeDivInUlOl(eachListSoup)
crifanEvernoteToWordpress.processSpanInUlOl(eachListSoup)
logging.debug("after eachListSoup=%s", eachListSoup)
# soup changed, write back to note content
updatedNoteHtml = crifanEvernote.soupToNoteContent(soup)
curNote.content = updatedNoteHtml
else:
logging.info("No ul/ol list for %s", curNote.title)
return curNote
def syncNoteImage(self,
curNoteDetail,
curResource,
uploadedImgUrl,
curResList=None):
"""Sync uploaded image url into Evernote Note content, replace en-media to img
Args:
curNoteDetail (Note): evernote Note
curResource (Resource): evernote Note Resource
uploadedImgUrl (str): uploaded imge url, previously is Evernote Resource
curResList (list): evernote Note Resource list
Returns:
updated note detail
Raises:
"""
if not curResList:
curResList = curNoteDetail.resources
soup = crifanEvernote.noteContentToSoup(curNoteDetail)
curEnMediaSoup = crifanEvernote.findResourceSoup(curResource,
soup=soup)
logging.debug("curEnMediaSoup=%s", curEnMediaSoup)
# curEnMediaSoup=<en-media hash="0bbf1712d4e9afe725dd51e701c7fae6" style="width: 788px; height: auto;" type="image/jpeg"></en-media>
if curEnMediaSoup:
curImgSoup = curEnMediaSoup
curImgSoup.name = "img"
curImgSoup.attrs = {"src": uploadedImgUrl}
logging.debug("curImgSoup=%s", curImgSoup)
# curImgSoup=<img src="https://www.crifan.com/files/pic/uploads/2020/11/c8b16cafe6484131943d80267d390485.jpg"></img>
# new content string
updatedContent = crifanEvernote.soupToNoteContent(soup)
logging.debug("updatedContent=%s", updatedContent)
curNoteDetail.content = updatedContent
else:
logging.warning(
"Not found en-media node for guid=%s, mime=%s, fileName=%s",
curResource.guid, curResource.mime,
curResource.attributes.fileName)
# here even not found, still consider as processed, later will remove it
# remove resource from resource list
# oldResList = curNoteDetail.resources
# Note: avoid side-effect: alter pass in curNoteDetail object's resources list
# which will cause caller curNoteDetail.resources loop terminated earlier than expected !
# oldResList = copy.deepcopy(curNoteDetail.resources)
# oldResList.remove(curResource) # workable
# newResList = oldResList
# Note 20201206: has update above loop, so should directly update curNoteDetail.resources
# curNoteDetail.resources.remove(curResource)
# newResList = curNoteDetail.resources
curResList.remove(curResource)
newResList = curResList
# # for debug
# if not newResList:
# logging.info("empty resources list")
syncParamDict = {
# mandatory
"noteGuid": curNoteDetail.guid,
"noteTitle": curNoteDetail.title,
# optional
"newContent": curNoteDetail.content,
"newResList": newResList,
}
respNote = self.evernote.syncNote(**syncParamDict)
# logging.info("Completed sync image %s to evernote note %s", uploadedImgUrl, curNoteDetail.title)
logging.info("Completed sync image %s", uploadedImgUrl)
return respNote
def batchUploadAndSyncImage(self, curNoteDetail, isCheckExisted):
"""Batch upload image to wordpress and sync to note (replace en-media to img)
Args:
curNoteDetail (Note): evernote Note
isCheckExisted (bool): whether check image is uploaded or not
Returns:
resp Note
Raises:
"""
originResList = copy.deepcopy(curNoteDetail.resources)
totalResNum = len(originResList)
logging.info("Total resources: %d", totalResNum)
latestResList = copy.deepcopy(curNoteDetail.resources)
soup = crifanEvernote.noteContentToSoup(curNoteDetail)
for curResIdx, eachResource in enumerate(originResList):
curResNum = curResIdx + 1
logging.info("%s resource %d/%d %s", "-" * 20, curResNum,
totalResNum, "-" * 20)
curResInfoStr = crifanEvernote.genResourceInfoStr(eachResource)
if self.evernote.isValidImageResource(eachResource):
curEnMediaSoup = crifanEvernote.findResourceSoup(eachResource,
soup=soup)
if not curEnMediaSoup:
logging.warning(
"NOT upload for not found related <en-media> node for %s",
curResInfoStr)
continue
isUploadOk, respInfo = self.uploadImageToWordpress(
eachResource, isCheckExisted)
if isUploadOk:
uploadedImgUrl = respInfo["url"]
logging.info("Uploaded image url %s", uploadedImgUrl)
curImgSoup = curEnMediaSoup
curImgSoup.name = "img"
curImgSoup.attrs = {"src": uploadedImgUrl}
latestResList.remove(eachResource)
else:
logging.error(
"Failed to upload image resource=%s, respInfo=%s",
curResInfoStr, respInfo)
else:
logging.warning("NOT upload for non-image resource: %s",
curResInfoStr)
updatedContent = crifanEvernote.soupToNoteContent(soup)
newResList = latestResList
syncParamDict = {
# mandatory
"noteGuid": curNoteDetail.guid,
"noteTitle": curNoteDetail.title,
# optional
"newContent": updatedContent,
"newResList": newResList,
}
respNote = self.evernote.syncNote(**syncParamDict)
return respNote