def _check_hashes (hashes, remote_repo, remote_password):
host, port, path = parse_URL(remote_repo)
hashtext = string.join(hashes, "\n")
# create zipfile of the folder
errcode, errmsg, headers, text = https_post_multipart(host, port, remote_password,
"/action/SynchronizeRepositories/check_hashes",
[("hashes", hashtext),],
())
if errcode != 200:
note("Errcode %s received. Error text from repository was\n%s.\n", errcode, errmsg)
return None
else:
results = []
resultsA = text.split("\n")
for result in resultsA:
rr = string.strip(result)
if rr:
results.append(string.split(rr))
return results
def _push_doc (doc, remote_repo_url, remote_password):
note(3, "pushing %s (%s)...", doc.id, doc.get_metadata("title"))
tdir = tempfile.mktemp()
os.mkdir(tdir)
try:
if os.path.isdir(os.path.join(doc.folder(), "page-images")):
shutil.copytree(os.path.join(doc.folder(), "page-images"), os.path.join(tdir, "page-images"))
else:
shutil.copyfile(os.path.join(doc.folder(), "document.tiff"), os.path.join(tdir, "document.tiff"))
if os.path.exists(doc.text_path()):
shutil.copyfile(os.path.join(doc.folder(), "contents.txt"), os.path.join(tdir, "contents.txt"))
if os.path.exists(os.path.join(doc.folder(), "paragraphs.txt")):
shutil.copyfile(os.path.join(doc.folder(), "paragraphs.txt"), os.path.join(tdir, "paragraphs.txt"))
if os.path.exists(os.path.join(doc.folder(), "wordbboxes")):
shutil.copyfile(os.path.join(doc.folder(), "wordbboxes"), os.path.join(tdir, "wordbboxes"))
if os.path.exists(os.path.join(doc.folder(), "images")):
shutil.copytree(os.path.join(doc.folder(), "images"), os.path.join(tdir, "images"))
if os.path.exists(os.path.join(doc.folder(), "originals")):
shutil.copytree(os.path.join(doc.folder(), "originals"), os.path.join(tdir, "originals"))
if os.path.exists(os.path.join(doc.folder(), "links")):
shutil.copytree(os.path.join(doc.folder(), "links"), os.path.join(tdir, "links"))
# filter the metadata
conf = configurator()
md = doc.get_metadata()
newmd = {}
properties = string.split(conf.get('synchronizing-properties') or
conf.get('metadata-sharing-properties') or
conf.get('metadata-sharing-default-properties'), ':')
for property in ["title", "date", "authors", "citation"]:
if not property in properties:
properties.append(property)
for prop in properties:
if md.has_key(prop):
newmd[prop] = md[prop]
update_metadata(os.path.join(tdir, "metadata.txt"), newmd)
host, port, path = parse_URL(remote_repo_url)
# create zipfile of the folder
tmpfilename = zipup(tdir)
try:
# send it to the repository
errcode, errmsg, headers, text = https_post_multipart(host, port, remote_password,
"/action/basic/repo_add",
(("password", remote_password,),
("filetype", "zipped-folder",)),
(("newfile", tmpfilename),))
if errcode != 200:
raise ValueError("Posting of data to repository %s resulted in error %d (%s).\nReturned text was:\n%s" % (remote_repo_url, errcode, errmsg, text))
note(3, "text from remote repository was\n%s.\n", text)
note(3, "%s successfully posted to %s.", tdir, remote_repo_url)
finally:
if os.path.exists(tmpfilename):
os.unlink(tmpfilename)
return text.strip()
finally:
if os.path.exists(tdir):
shutil.rmtree(tdir)
def _pull_doc (local_repo, doc_id, remote_repo_url, remote_password):
host, port, path = parse_URL(remote_repo_url)
return _get_doc_to_file(local_repo, host, port, remote_password, '/action/externalAPI/fetch_folder?doc_id=%s' % doc_id)