Provides functions to extract content from a doc using Tika parsers in Clojure

tika-parser.clj

;; NOTE: Please add [org.apache.tika/tika-parsers "1.25] and [org.apache.tika/tika-core "1.25] in your dependencies

(import '[org.apache.tika.parser AutoDetectParser])
(import '[org.apache.tika.metadata Metadata])
(import '[org.apache.tika.sax BodyContentHandler])
(import '[java.nio.file Files])
(import '[java.io ByteArrayInputStream ByteArrayOutputStream])
(require '[clojure.java.io :as io])

(defn parse-doc
  "Parses resume content into a string when a file's byte array is passed"
  [file-byte-array]
  (let [bais (ByteArrayInputStream. file-byte-array)
        baos (ByteArrayOutputStream.)]
    (-> (AutoDetectParser.)
        (.parse bais
                (BodyContentHandler. baos)
                (Metadata.)))
    (let [parsed-output (.toString baos "UTF-8")]
      (.close baos)
      (.close bais)
      parsed-output)))

(def file (io/file "path/to/file"))
(def file-bytes (Files/readAllBytes (.toPath file)))
(def parsed-file (parse-doc file-bytes))