Garmelon/wikilyze
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Make json format more consistent

Browse source
This commit is contained in:
Joscha 2022-10-03 15:00:23 +02:00
parent 51096c99e1
commit ecdeb4086a
1 changed files with 18 additions and 12 deletions
Download patch file Download diff file Expand all files Collapse all files

26
sift/sift.py
View file

@ -12,22 +12,28 @@ def eprint(*args, **kwargs):
def process_page(page): def process_page(page):
# https://pythonhosted.org/mwxml/iteration.html#mwxml.Page # https://pythonhosted.org/mwxml/iteration.html#mwxml.Page
eprint(f"{page.namespace:4}: {page.id:8} - {page.title}") print(f"{page.namespace:4}: {page.id:8} - {page.title}")
info = {"ns": page.namespace, "id": page.id, "title": page.title}
if page.redirect:
info["redirect"] = page.redirect
else:
[revision] = list(page) # Every page has exactly one revision [revision] = list(page) # Every page has exactly one revision
if revision.text is not None: text = revision.text or ""
info["length"] = len(revision.text)
links = [] links = []
for link in wtp.parse(revision.text).wikilinks: for link in wtp.parse(text).wikilinks:
start, end = link.span start, end = link.span
links.append((link.title, start, end)) links.append((link.title, start, end))
info["links"] = links
info = {
"ns": page.namespace,
"id": page.id,
"title": page.title,
"length": len(text),
"links": links,
}
if page.redirect:
assert len(links) == 1
assert links[0] == page.redirect
info["redirect"] = page.redirect
print(json.dumps(info, check_circular=False, separators=(",", ":"))) print(json.dumps(info, check_circular=False, separators=(",", ":")))