[bittorrent] content search

larytet.8708132 at bloglines.com larytet.8708132 at bloglines.com
Wed May 18 13:40:07 EDT 2005


i added context sensitive search to the client this is part of the initiial
functional requirements and major piece of core functionality. the solution
is not scalable for the rodi crawler (dictionary is a hashtable), but it's
good enough for many users. in the current version only ACII (7 bits characters)
files are parsed. i am working on parsing of PDF files (any links to small-extract-text-only
Java implemetation of PDF parser ?)

let's say you are looking for words
"Los Angeles"
let' say that file A contains both words in the filename -
"Map of Los Angeles"
and file B ("History of the US") contains those same
words in the text, but not in the filename
rank of A is higher than rank
of B.
A will be returned by search engine first.
if file C filename is "Map
of Los Gatos" than rankA > rankC > rankB


alternative to the search is
RodiHash file - works exactly as torrent file, but XML based. Default block
size in Rodi is 4MB



More information about the BitTorrent mailing list