def parse_collection(collection):
works = []
docs = json.loads(collection)
for doc in docs:
corpus, nn = doc.split(",")
c = Corpus(corpus)
for n in nn.strip("[]").split(", "):
w = c.work_by_docix(int(n))
works.append(w)
return Collection(works=works)
def update(corpus, docix, path):
"""Reindex a document by index number. """
docix = int(docix)
with click_spinner.spinner():
c = Corpus(corpus)
c.update(docix=docix, path=Path(path))
if c.work_by_docix(docix).searchable:
click.echo(f"[+] updated document {docix} in '{corpus}'")
else:
click.echo(f"[-] failed")
def history():
db.connect()
history = []
for h in Search.select():
history.append(h)
db.close()
kwargs = request.args
id = kwargs.get("id")
db.connect()
s = Search.get_by_id(id)
db.close()
works = []
work_ids = json.loads(s.collection)
for work_id in work_ids:
corpus, docix = work_id.split(",")
c = Corpus(corpus)
w = c.work_by_docix(int(docix))
works.append(w)
results = [
r.html for r in SearchResult.select().where(SearchResult.search == s)
]
response = {
"version": __version__,
"collection": s.collection,
"works": works,
"corpora": _corpora,
"query": s.query,
"history": history,
"results": results,
"count": len(s.results),
"id": id
}
return render_template("index.html", **response)
def search():
if request.method == "POST":
form = request.form
else:
form = request.args
collection = form.get("collection")
query = form.get("query")
works = []
ids = json.loads(collection)
for id in ids:
corpus, n = id.split(",")
c = Corpus(corpus)
w = c.work_by_docix(int(n))
works.append(w)
results = search_request(works, query)
if results:
count_matches, count_documents, _ = results.count
db.connect()
try:
s = (Search.get(query=query, collection=collection), )
except Search.DoesNotExist:
names = [f"{work.author}, {work.title}" for work in works]
s = Search.create(
query=query,
collection=collection,
prettified=f"{'; '.join(names)}",
count_matches=count_matches,
count_documents=count_documents,
minscore=results.minscore,
top=results.top,
start_dt=results.start_dt,
end_dt=results.end_dt,
maxchars=results.maxchars,
surround=results.surround,
)
s.save()
finally:
for href in results.highlights:
r = SearchResult.get_or_create(search=s,
html=next(as_html([
href,
])))
if r[1]:
r[0].save()
db.close()
db.connect()
history = []
for h in Search.select():
history.append(h)
db.close()
response = {
"version": __version__,
"collection": collection,
"works": works,
"corpora": _corpora,
"query": query,
"history": history,
"results": as_html(results.highlights) if results else [],
"count": results.count if results else (0, 0),
"id": history[-1].id
}
return render_template("index.html", **response)