chg: [dev] gen_defence_university.py no longer outputs empty strings, lists

This commit is contained in:
VVX7 2020-08-22 13:01:20 -04:00
parent 1cddf4b7cd
commit 5e54fc2022

View file

@ -196,22 +196,8 @@ def _gen_galaxy(scrape):
for uni in scrape: for uni in scrape:
new_template = template = { new_template = template = {
# university description
"description": "", "description": "",
"meta": { "meta": {
"supervising agency": [],
"subsidiary": [],
"category": [],
"topics": [],
# Defence labs
"Major Defence Laboratories": [],
# Defence labs
"alias": [],
"risk": "",
"address": "",
"lat": "",
"long": "",
# External link to ASPI
"refs": [] "refs": []
}, },
"uuid": "", "uuid": "",
@ -222,34 +208,57 @@ def _gen_galaxy(scrape):
new_template["meta"]["refs"].append(uni["url"]) new_template["meta"]["refs"].append(uni["url"])
for intro in uni["intro"]:
new_template["description"] += intro["clause"]
new_template["value"] = uni["name"] + f" ({uni['_name']})" new_template["value"] = uni["name"] + f" ({uni['_name']})"
new_template["meta"]["risk"] = uni["risk statement"] def _append_meta(key, meta):
if uni.get(meta):
values = []
for value in uni[meta]:
if value != "":
values.append(value)
if values:
new_template["meta"][key] = values
for alias in uni["aliases"]: if uni.get("intro"):
new_template["meta"]["alias"].append(alias) for intro in uni["intro"]:
new_template["description"] += intro["clause"]
if new_template["description"] == "":
new_template["description"] += uni["name"] + f" ({uni['_name']})"
else:
new_template["description"] += uni["name"] + f" ({uni['_name']})"
for agency in uni["supervising agencies"]: if uni.get("risk"):
new_template["meta"]["supervising agency"].append(agency) if uni.get("risk") != "":
new_template["meta"]["risk"] = uni["risk statement"]
if uni.get("subsidiaries"): _append_meta("aliases", "aliases")
for subsidiary in uni["subsidiaries"]:
new_template["meta"]["subsidiary"].append(subsidiary)
if uni.get("topics"): _append_meta("supervising agencies", "supervising agencies")
for topic in uni["topics"]:
new_template["meta"]["topics"].append(topic)
for category in uni["categories"]: _append_meta("subsidiaries", "subsidiaries")
new_template["meta"]["category"].append(category)
_append_meta("topics", "topics")
_append_meta("categories", "categories")
if uni.get("sections"):
labs = []
for section in uni["sections"]:
if section["title"] == "Major defence laboratories":
for lab in section["body"]:
if lab.get("name"):
if lab["name"] != "":
labs.append(lab["name"])
if labs:
new_template["meta"]["major defence laboratories"] = labs
if uni.get("location"): if uni.get("location"):
new_template["meta"]["address"] = uni["location"][0]["name"] if uni.get(uni["location"][0]["name"]) != "":
new_template["meta"]["lat"] = uni["location"][0]["lat"] new_template["meta"]["address"] = uni["location"][0]["name"]
new_template["meta"]["long"] = uni["location"][0]["long"] if uni.get(uni["location"][0]["lat"]) != "":
new_template["meta"]["lat"] = uni["location"][0]["lat"]
if uni.get(uni["location"][0]["long"]) != "":
new_template["meta"]["long"] = uni["location"][0]["long"]
base["values"].append(new_template) base["values"].append(new_template)
@ -276,6 +285,8 @@ def main():
galaxy = _gen_galaxy(articles) galaxy = _gen_galaxy(articles)
print(galaxy)
with open("china-defence-universities.json", "w") as g: with open("china-defence-universities.json", "w") as g:
g.write(json.dumps(galaxy, indent=4, sort_keys=True)) g.write(json.dumps(galaxy, indent=4, sort_keys=True))