Skip to content

Broken tokenization of boost #418

Description

@duckduckgrayduck
Image

Discovered today during BwM session.

If I run parse on the same query in the shell it looks like it converts the boost into scientific notation.

In [7]: from documentcloud.users.models import User
...: u = User.objects.get(username="sanjin")
...: qd = QueryDict('hl=true&version=2.0&per_page=25')
...: result = s._parse('"death" AND "taser"^50', qd, user=u)
...: print(result)
('"death" AND "taser"^5E+1', <QueryDict: {}>, None, False, False)

I think what's happening here is

User types "taser"^50
luqum parses it, Decimal(50).normalize() → 5E+1
Solr gets "taser"^5 (boost of 5 on taser) plus a mystery term E+1 which is passed to the highlighter

Metadata

Metadata

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions