v1.0.6: parser.cfm — PDFBox Loader + BufferedInputStream

index.cfm

@@ -1 +1 @@
-<cfsetting showdebugoutput=no enablecfoutputonly=yes><cfoutput>OK v1.0.5</cfoutput>
+<cfsetting showdebugoutput=no enablecfoutputonly=yes><cfoutput>OK v1.0.6</cfoutput>

parser.cfm

@@ -29,8 +29,22 @@
                     <cfset fileBytes = binaryDecode(doc.original_bytes, "base64")>
                     <!--- Байты → InputStream --->
                     <cfset bais = createObject("java", "java.io.ByteArrayInputStream").init(fileBytes)>
-                    <!--- Открываем PDF --->
+                    <cfset bis = createObject("java", "java.io.BufferedInputStream").init(bais)>
-                    <cfset pdfDoc = createObject("java", "org.apache.pdfbox.pdmodel.PDDocument").load(bais)>
+
+                    <!--- Пробуем PDFBox 3.x Loader, иначе 2.x PDDocument.load --->
+                    <cfset pdfDoc = 0>
+                    <cftry>
+                        <cfset pdfDoc = createObject("java", "org.apache.pdfbox.Loader").loadPDF(bis)>
+                        <cfcatch>
+                            <cftry>
+                                <cfset pdfDoc = createObject("java", "org.apache.pdfbox.pdmodel.PDDocument").load(bis)>
+                                <cfcatch>
+                                    <cfset arrayAppend(errors, "PDFBox: cannot open PDF — " & cfcatch.message)>
+                                </cfcatch>
+                            </cftry>
+                        </cfcatch>
+                    </cftry>
+                    <cfif NOT isNumeric(pdfDoc)>
                         <cfset stripper = createObject("java", "org.apache.pdfbox.text.PDFTextStripper")>
 
                         <cfset pageCount = pdfDoc.getNumberOfPages()>
@@ -49,6 +63,8 @@
                         </cfloop>
 
                         <cfset pdfDoc.close()>
+                    </cfif>
+                    <cfset bis.close()>
                     <cfset bais.close()>
 
                     <cfcatch>