def extract_revision_text(remark_el):
remark_children = minidom_children(remark_el)
if len(remark_children) == 1 and remark_children[0].tagName == "p":
return minidom_get_text(remark_children[0])
if len(remark_children) == 0:
return minidom_get_text(remark_el) or None
if minidom_get_text(remark_el).strip():
return None
lines = []
for child in remark_children:
if child.tagName == "p":
lines.append(minidom_get_text(child))
return "\n".join(lines)
def extract_revision_text(remark_el):
remark_children = minidom_children(remark_el)
if len(remark_children) == 1 and remark_children[0].tagName == "p":
return minidom_get_text(remark_children[0])
if len(remark_children) == 0:
return minidom_get_text(remark_el) or None
if minidom_get_text(remark_el).strip():
return None
lines = []
for child in remark_children:
if child.tagName == "p":
lines.append(minidom_get_text(child))
elif child.tagName == "ul":
for ul_child in minidom_children(child):
if ul_child.tagName == "li":
lines.append("* {}".format(minidom_get_text(ul_child)))
return "\n".join(lines)
def extract_xep_metadata(document):
header = minidom_find_header(document)
latest_revision = minidom_find_child(header, "revision")
if latest_revision is not None:
last_revision_version = minidom_get_text(
minidom_find_child(latest_revision, "version"))
last_revision_date = minidom_get_text(
minidom_find_child(latest_revision, "date"))
remark_el = minidom_find_child(latest_revision, "remark")
last_revision_remark = None
if remark_el is not None:
last_revision_remark = extract_revision_text(remark_el)
if last_revision_remark is not None:
initials_el = minidom_find_child(latest_revision, "initials")
last_revision_initials = initials_el and minidom_get_text(
initials_el)
else:
last_revision_initials = None
else:
last_revision_version = None
last_revision_date = None
last_revision_remark = None
last_revision_initials = None
status = minidom_get_text(minidom_find_child(header, "status"))
type_ = minidom_get_text(minidom_find_child(header, "type"))
abstract = " ".join(
minidom_get_text(minidom_find_child(header, "abstract")).split())
sig_el = minidom_find_child(header, "sig")
if sig_el is None:
sig = None
else:
sig = minidom_get_text(sig_el)
shortname = minidom_get_text(minidom_find_child(header, "shortname"))
if shortname.replace("-", " ").replace("_", " ").lower() in [
"not yet assigned", "n/a", "none", "to be assigned", "to be issued"
]:
shortname = None
title = minidom_get_text(minidom_find_child(header, "title"))
approver_el = minidom_find_child(header, "approver")
if approver_el is not None:
approver = minidom_get_text(approver_el)
else:
approver = "Board" if type_ == "Procedural" else "Council"
last_call_el = minidom_find_child(header, "lastcall")
if last_call_el is not None:
last_call = minidom_get_text(last_call_el)
else:
last_call = None
return {
"last_revision": {
"version": last_revision_version,
"date": last_revision_date,
"initials": last_revision_initials,
"remark": last_revision_remark,
},
"status": status,
"type": type_,
"sig": sig,
"abstract": abstract,
"shortname": shortname,
"title": title,
"approver": approver,
"last_call": last_call,
}
def extract_xep_metadata(document):
header = minidom_find_header(document)
latest_revision = minidom_find_child(header, "revision")
if latest_revision is not None:
last_revision_version = minidom_get_text(
minidom_find_child(latest_revision, "version")
)
last_revision_date = minidom_get_text(
minidom_find_child(latest_revision, "date")
)
remark_el = minidom_find_child(latest_revision, "remark")
last_revision_remark = None
if remark_el is not None:
last_revision_remark = extract_revision_text(remark_el)
if last_revision_remark is not None:
initials_el = minidom_find_child(latest_revision, "initials")
last_revision_initials = initials_el and minidom_get_text(
initials_el
)
else:
last_revision_initials = None
else:
last_revision_version = None
last_revision_date = None
last_revision_remark = None
last_revision_initials = None
status = minidom_get_text(minidom_find_child(header, "status"))
type_ = minidom_get_text(minidom_find_child(header, "type"))
abstract = " ".join(minidom_get_text(
minidom_find_child(header, "abstract")
).split())
sig_el = minidom_find_child(header, "sig")
if sig_el is None:
sig = None
else:
sig = minidom_get_text(sig_el)
shortname = minidom_get_text(minidom_find_child(header, "shortname"))
if shortname.replace("-", " ").replace("_", " ").lower() in [
"not yet assigned", "n/a", "none", "to be assigned",
"to be issued"]:
shortname = None
title = minidom_get_text(minidom_find_child(header, "title"))
approver_el = minidom_find_child(header, "approver")
if approver_el is not None:
approver = minidom_get_text(approver_el)
else:
approver = "Board" if type_ == "Procedural" else "Council"
last_call_el = minidom_find_child(header, "lastcall")
if last_call_el is not None:
last_call = minidom_get_text(last_call_el)
else:
last_call = None
return {
"last_revision": {
"version": last_revision_version,
"date": last_revision_date,
"initials": last_revision_initials,
"remark": last_revision_remark,
},
"status": status,
"type": type_,
"sig": sig,
"abstract": abstract,
"shortname": shortname,
"title": title,
"approver": approver,
"last_call": last_call,
}