Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Tree: 63f4e576b2
Fetching contributors…

Cannot retrieve contributors at this time

57 lines (46 sloc) 1.723 kB
;; Clojure Interface to Apache Tika library
;; TODO: add functions for class
(ns tika
(:import ( InputStream File FileInputStream))
(:import ( URL))
(:import (org.apache.tika.parser Parser AutoDetectParser ParseContext))
(:import (org.apache.tika.language LanguageIdentifier))
(:import (org.apache.tika.metadata Metadata))
(:import (org.apache.tika Tika))
(:import (org.apache.tika.sax BodyContentHandler))
(def #^{:private true} tika-class (Tika.))
(defn- conv-metadata [#^Metadata mdata]
(let [names (.names mdata)]
(zipmap (map #(keyword (.toLowerCase %1)) names)
(map #(seq (.getValues mdata %1)) names))))
(defmulti parse
"Parses given file or stream, returning map of document's metadata and text (:text key)"
(defmethod parse InputStream [ifile]
(let [parser (new AutoDetectParser)
context (new ParseContext)
metadata (new Metadata)
handler (new BodyContentHandler)
(.set context Parser parser)
(.parse parser ifile handler metadata context)
(assoc (conv-metadata metadata) :text (.toString handler))))
(defmethod parse String [filename]
(parse (FileInputStream. (File. filename))))
(defmethod parse File [file]
(parse (new FileInputStream file)))
(defmulti detect-mime-type
"Detects mime-type of given file or stream"
(defmethod detect-mime-type InputStream [ifile]
(.detect tika-class ifile))
(defmethod detect-mime-type String [filename]
(.detect tika-class filename))
(defmethod detect-mime-type File [file]
(.detect tika-class file))
(defn detect-language
"Detects language of given text"
[#^String text]
(.getLanguage (LanguageIdentifier. text)))
Jump to Line
Something went wrong with that request. Please try again.