enum_worksheet.clj

;; This was generated with chat.deepseeker.com in order to solve the problem that I'm reading excel tables
;; in clojure with duckdb jdbc driver and it does not have a way to enumerate excel worksheets
;; Plus side of this code is no dependencies
(ns util.enum-worksheet
  (:import
    (java.util Enumeration)
    (java.util.zip ZipFile ZipEntry)
    (java.io InputStream)
    (javax.xml.parsers DocumentBuilderFactory DocumentBuilder)
    (org.w3c.dom Document NodeList Node NamedNodeMap)))

(set! *warn-on-reflection* true)

(defn ^:private find-entry [^Enumeration entries name]
  (some #(when (= (.getName ^ZipEntry %) name) %) (enumeration-seq entries)))

(defn get-sheet-names
  "Returns a list of the sheet names in the Excel spreadsheet specified by the file path."
  [^String file-path]
  (with-open [^ZipFile zip-file (ZipFile. file-path)]
    (let [ entries (.entries zip-file)
          ^ZipEntry workbook-entry (find-entry entries "xl/workbook.xml")]
      (when workbook-entry
        (with-open [^InputStream input-stream (.getInputStream zip-file workbook-entry)]
          (let [^DocumentBuilderFactory dbf (DocumentBuilderFactory/newInstance)
                _ (.setNamespaceAware dbf true)
                ^DocumentBuilder db (.newDocumentBuilder dbf)
                ^Document doc (.parse db input-stream)
                ^NodeList sheet-nodes (.getElementsByTagNameNS doc "http://schemas.openxmlformats.org/spreadsheetml/2006/main" "sheet")
                sheet-names (keep (fn [i]
                                    (let [^Node sheet-node (.item sheet-nodes i)
                                          ^NamedNodeMap attributes (.getAttributes sheet-node)
                                          ^Node name-attr (.getNamedItem attributes "name")]
                                      (when name-attr (.getNodeValue name-attr))))
                                  (range (.getLength sheet-nodes)))]
            sheet-names)))))) 

(comment
  (prn *e)
  (get-sheet-names "file.xlsx")
  )