bump pdfbox to 3.0.1

deps.edn

@@ -1,6 +1,7 @@
 {:deps {org.clojure/clojure {:mvn/version "1.11.1"},
-        org.apache.pdfbox/pdfbox {:mvn/version "2.0.29"}
-        org.apache.pdfbox/pdfbox-tools {:mvn/version "2.0.29"}}
+        org.apache.pdfbox/pdfbox {:mvn/version "3.0.1"}
+        org.apache.pdfbox/pdfbox-io {:mvn/version "3.0.1"}
+        org.apache.pdfbox/pdfbox-tools {:mvn/version "3.0.1"}}
  :aliases {:test {:extra-paths ["test"]
                   :extra-deps {org.clojure/test.check {:mvn/version "RELEASE"}}}
            :runner {:extra-deps {com.cognitect/test-runner

project.clj

@@ -4,5 +4,6 @@
   :license {:name "BSD"
             :url "https://opensource.org/license/bsd-3-clause/"}
   :dependencies [[org.clojure/clojure "1.11.1"]
-                 [org.apache.pdfbox/pdfbox "2.0.29"]
-                 [org.apache.pdfbox/pdfbox-tools "2.0.29"]])
+                 [org.apache.pdfbox/pdfbox "3.0.1"]
+                 [org.apache.pdfbox/pdfbox-io "3.0.1"]
+                 [org.apache.pdfbox/pdfbox-tools "3.0.1"]])

src/pdfboxing/common.clj

@@ -1,55 +1,44 @@
 (ns pdfboxing.common
   (:require [clojure.java.io :as io])
   (:import (java.io File)
-           (org.apache.pdfbox.pdmodel PDDocument)
-           (org.apache.pdfbox.io RandomAccessFile)
-           (org.apache.pdfbox.pdfparser PDFParser)))
+           (org.apache.pdfbox Loader)
+           (org.apache.pdfbox.pdmodel PDDocument)))
 
-(defn try-get-as-pdf
-  "Try and get the pdf-file-or-path as a PDF.
-  Returns nil if pdf-file-or-path could not be loaded as a PDF."
-  [pdf-file-or-path]
-  (let [^File pdf-file (io/as-file pdf-file-or-path)
-        random-access-file (RandomAccessFile. pdf-file "r")
-        parser (PDFParser. random-access-file)]
-    (try
-      (.parse parser)
-      (.getPDDocument parser)
-      (catch Exception _))))
+
+(defn load-pdf-from-media [pdf-file-or-path]
+  (try
+    (-> pdf-file-or-path
+        ^File (io/as-file)
+        (Loader/loadPDF))
+    (catch Exception _)))
 
 (defn is-pdf?
   "Confirm that the PDF supplied is really a PDF"
   [pdf-file-or-path]
-  (if-let [pdf (try-get-as-pdf pdf-file-or-path)]
+  (if-let [pdf (load-pdf-from-media pdf-file-or-path)]
     (try
       (not (nil? pdf))
       (finally
         (.close pdf)))
     false))
 
-(defn load-pdf
-  "Load a given PDF only after checking if it really is a PDF"
-  [pdf-file-or-path]
-  (if-let [pdf (try-get-as-pdf pdf-file-or-path)]
-    pdf
-    (throw (IllegalArgumentException. (format "%s is not a PDF file" pdf-file-or-path)))))
-
 (defprotocol PDFDocument
   "return an object from which text can be extracted"
   (obtain-document [source]))
 
 (extend-protocol PDFDocument
   (Class/forName "[B") ;; byte-array
+  #_{:clj-kondo/ignore [:function-name]}
   (obtain-document [source]
-    (PDDocument/load source))
+    (Loader/loadPDF source))
 
   String
   (obtain-document [source]
-    (load-pdf source))
+    (load-pdf-from-media source))
 
   File
   (obtain-document [source]
-    (load-pdf source))
+    (load-pdf-from-media source))
 
   PDDocument
   (obtain-document [source]

src/pdfboxing/html.clj

@@ -6,6 +6,6 @@
   "Convert a PDF document to a simple HTML file.
   `pdf-doc` - string, name of a PDF document that is in the root folder
   this is command line tool from pdfbox more info at:
-  https://pdfbox.apache.org/2.0/commandline.html#extracttext"
+  https://pdfbox.apache.org/3.0/commandline.html#extracttext"
   [pdf-doc]
-  (ExtractText/main (into-array String ["-html" pdf-doc])))
+  (ExtractText/main (into-array String ["-html" "-i" pdf-doc])))

src/pdfboxing/merge.clj

@@ -1,15 +1,15 @@
 (ns pdfboxing.merge
   (:require [pdfboxing.common :as common])
-  (:import (java.io InputStream OutputStream)
+  (:import (java.io File InputStream OutputStream)
+           (org.apache.pdfbox.io IOUtils RandomAccessRead RandomAccessReadBuffer)
            (org.apache.pdfbox.multipdf PDFMergerUtility)
            (org.apache.pdfbox.pdmodel PDDocument
-                                      PDPage
-                                      PDPageContentStream)
+                                      PDPage PDPageContentStream)
            (org.apache.pdfbox.pdmodel.common PDRectangle)
            (org.apache.pdfbox.pdmodel.graphics.image PDImageXObject)))
 
 (defn throw-exception
-  [message]
+  [^String message]
   (throw (IllegalArgumentException. message)))
 
 (defn check-if-present
@@ -43,18 +43,26 @@
   {:pre [(arg-check output input)]}
   (let [merger (PDFMergerUtility.)]
     (doseq [source input]
-      (.addSource merger source))
+      (condp instance? source
+        File (.addSource merger ^File source)
+        String (.addSource merger ^String source)
+        InputStream (.addSource merger ^RandomAccessRead
+                                (RandomAccessReadBuffer. ^InputStream source))))
     (cond
       (instance? OutputStream output)
       (.setDestinationStream merger output)
 
       :else
       (.setDestinationFileName merger output))
-    (.mergeDocuments merger)))
+    (.mergeDocuments merger (IOUtils/createMemoryOnlyStreamCache))
+    (condp instance? output
+      File (.close output)
+      OutputStream (.close output)
+      nil)))
 
 (defn- add-image-to-page
   "Adds image as a page to the document object"
-  [doc ^org.apache.pdfbox.pdmodel.graphics.image.PDImageXObject image]
+  [^PDDocument doc ^PDImageXObject image]
   (let [page-size PDRectangle/A4
         original-width (.getWidth image)
         original-height (.getHeight image)