tei2zenodo.py

#!/usr/bin/env python3

# this script adds new (old) versions to the TEI zenodo space
# it assumes the zip archives in the current folder along with 
# the readme files. Naming convention is `tei-x.x.x.zip` for the
# archives and `readme-x.x.x.html.xml`(!) for the readme files.
# NB: The readme files (which are to be found in the current release)
# need to be transformed first by tei2zenodo-readme.xsl

import requests
import json
import re
import argparse

ACCESS_TOKEN='put your token here'
INITIAL_DEPOSIT_URL="https://sandbox.zenodo.org/api/deposit/depositions/371993"

def newVersion(initial_deposit_url):
    "create a new draft version"
    r = requests.post(initial_deposit_url + "/actions/newversion", params={'access_token': ACCESS_TOKEN})
    if r.status_code != 201:
        raise Exception("Creation of new version failed: " + str(r.json()))
    return r.json()['links']['latest_draft'] 

def uploadFile(deposit_url, tei_version):
    "upload a file to the new draft, using the yet undocumented upload API, see https://github.com/zenodo/zenodo/issues/833"
    bucket_url = requests.get(deposit_url, params={'access_token': ACCESS_TOKEN}).json()['links']['bucket']
    filename = 'tei-' + tei_version + '.zip'
    r = requests.put('%s/%s' % (bucket_url,filename),
                data=open('tei-' + tei_version + '.zip', 'rb'),
                headers={"Accept":"application/json",
                "Authorization":"Bearer %s" % ACCESS_TOKEN,
                "Content-Type":"application/octet-stream"}) 
    if r.status_code != 200:
        raise Exception("Upload failed: " + str(r.json()))
    return r

def deleteAllFiles(deposit_url):
    "delete all files that are currently associated with the publication"
    files = requests.get(deposit_url + '/files', params={'access_token': ACCESS_TOKEN})
    for file in files.json():
        requests.delete(file['links']['self'], params={'access_token': ACCESS_TOKEN})
    return

def updateMetadata(deposit_url, tei_version):
    "update the metadata of the deposit"
    # first, get the readme file
    try:
        readme = open('readme-' + tei_version + '.html.xml', 'rt')
        desc = readme.read()
        readme.close()
    except IOError:
        print('Failed to open readme file for version ' + tei_version)
        desc = 'Need to add a description manually'
    # extract the date from the closing address of the readme by matching a regex
    dates = re.findall(r'Date: (\d{4}-\d{2}-\d{2})<br', desc)
    if len(dates) == 0:
        date = '1970-01-01'
    else:
        date = dates[0]
    # update the metadata fields of the new deposit
    metadata = requests.get(deposit_url, params={'access_token': ACCESS_TOKEN}).json()['metadata']
    metadata['version'] = 'v' + tei_version
    metadata['description'] = desc
    metadata['publication_date'] = date
    # finally, do the upload
    url = deposit_url + "?access_token=" + ACCESS_TOKEN
    headers = {"Content-Type": "application/json"}
    r = requests.put(url, data=json.dumps({"metadata": metadata}), headers=headers)
    if r.status_code != 200:
        raise Exception("Update failed: " + str(r.json()))
    return r

def publish(deposit_url):
    "publish the draft"
    r = requests.post(deposit_url + '/actions/publish', params={'access_token': ACCESS_TOKEN})
    if r.status_code != 202:
        raise Exception("Publish failed: " + str(r.json()))
    return r

parser = argparse.ArgumentParser()
parser.add_argument("-v", "--version", help="specify the TEI version you want to upload to Zenodo")
args = parser.parse_args()
tei_version=args.version  

# *****
# main stuff starts here
# *****

# create a new version
deposit_url = newVersion(INITIAL_DEPOSIT_URL)
# delete old uploads (which are copied to the new deposit by default)
deleteAllFiles(deposit_url)
# upload the zip archive
upload = uploadFile(deposit_url, tei_version)
# update the metadata
update = updateMetadata(deposit_url, tei_version)
# publish the deposit
publish(deposit_url)