Merge pull request #68 from sampoto/pdfa-perf-params

Pdf/a conversion performance parameters
solita · Aug 15, 2023 · a34addb · a34addb
2 parents c0fc5df + 8719896
commit a34addb
Show file tree

Hide file tree

Showing 4 changed files with 94 additions and 9 deletions.
diff --git a/programs/pdf2pdfa b/programs/pdf2pdfa
@@ -4,6 +4,44 @@ set -u
 
 INPUT=$1
 OUTPUT=$2
+DPI=$3
+MAXBITMAP=$4
+PDFSETTINGS=$5
+
+if [ -z "$DPI" ]; then
+  DPI=720
+fi
+re='^[0-9]+$'
+if ! [[ $DPI =~ $re ]] ; then
+  echo "error: DPI is not a number"
+  exit 2
+fi
+
+
+if [ -n "$PDFSETTINGS" ]; then
+  valid_settings=("/screen" "/ebook" "/printer" "/prepress" "/default")
+  for i in "${valid_settings[@]}"
+  do
+    if [ "$i" = "$PDFSETTINGS" ] ; then
+      found=true
+    fi
+  done
+  if [ "$found" != true ]; then
+    echo "error: Invalid PDFSETTINGS value"
+    exit 2
+  fi
+  PDFSETTINGS="-dPDFSETTINGS=$PDFSETTINGS"
+else
+  PDFSETTINGS="-dPDFSETTINGS=/default"
+fi
+
+if [ -z "$MAXBITMAP" ]; then
+  MAXBITMAP=0
+fi
+if ! [[ $MAXBITMAP =~ $re ]] ; then
+  echo "error: MAXBITMAP is not a number"
+  exit 2
+fi
 
 docker run \
   --runtime="${LAUNDRY_DOCKER_RUNTIME:-runsc}" \
@@ -15,5 +53,5 @@ docker run \
   --rm \
   laundry-programs \
     /bin/bash -c 'cat > /home/docconv/document.pdf && \
-    gs -q -dPDFA -dBATCH -dNOPAUSE -sProcessColorModel=DeviceCMYK -sDEVICE=pdfwrite -dPDFACompatibilityPolicy=1 -sOutputFile=- /home/docconv/document.pdf' \
+    gs -q -dPDFA -dBATCH -dNOPAUSE -r'$DPI' -dMaxBitmap='$MAXBITMAP' '$PDFSETTINGS' -sProcessColorModel=DeviceCMYK -sDEVICE=pdfwrite -dPDFACompatibilityPolicy=1 -sOutputFile=- /home/docconv/document.pdf' \
     < "$INPUT" > "$OUTPUT"
diff --git a/resources/public/index.html b/resources/public/index.html
@@ -6,10 +6,28 @@
 a:hover { color: #098809; text-decoration: underline }
 </style>
 <script>
+  var pdfParameters = [
+    {"type": "pdf/pdf2pdfa", key: "dpi", dtype: "number", wrapper: "pdf_dpi_elem", elem: "pdf_dpi" },
+    {"type": "pdf/pdf2pdfa", key: "maxbitmap", dtype: "number", wrapper: "pdf_maxbitmap_elem", elem: "pdf_maxbitmap" },
+    {"type": "pdf/pdf2pdfa", key: "pdfsettings", wrapper: "pdf_pdfsettings_elem", elem: "pdf_pdfsettings" }
+  ]
+
+  function updateConversionParameters() {
+    var seln = document.getElementById("pdfOperation");
+    for (const param of pdfParameters) {
+      document.getElementById(param.wrapper).hidden = (param.type !== seln.value);
+    }
+  }
+
   function pdfConverter() {
     var form = document.getElementById("pdfop");
     var seln = document.getElementById("pdfOperation");
-    converter(form, seln)
+    var args = pdfParameters.filter(item => item.type == seln.value)
+      .map(arg => ({
+        key: arg.key,
+        value: arg.dtype === "number" ? parseInt(document.getElementById(arg.elem).value).toString() : document.getElementById(arg.elem).value
+      }));
+    converter(form, seln, args)
   }
 
   function imageConverter() {
@@ -44,14 +62,15 @@
     xhr.send(formData);
   }
 
-  function converter(form, seln) {
+  function converter(form, seln, args) {
     var formData = new FormData(form);
     var xhr = new XMLHttpRequest();
     if (form[0].files.length != 1) {
       alert("Select a file first");
       return;
     }
-    xhr.open("POST", seln.value, true);
+    var argsString = args && args.length ? "?" + args.map(function(arg) { return arg.key + "=" + encodeURIComponent(arg.value) }).join("&") : ""
+    xhr.open("POST", seln.value + argsString, true);
     xhr.responseType = "blob";
     xhr.onload = function(e) {
       var urlsource= window.URL || window.webkitURL;
@@ -63,19 +82,28 @@
 </script>
 <title>Laundry</title>
 </head>
-<body>
+<body onload="updateConversionParameters()">
    <h1>Laundry</h1>
    <a href="api-docs">Swagger API</a>
 
    <hr>
 
    <p>PDF
-     <select id="pdfOperation">
+     <select id="pdfOperation" onchange="updateConversionParameters()">
        <option value="pdf/pdf-preview">preview</option>
        <option value="pdf/pdf2pdfa">to PDF/A</option>
        <option value="pdf/pdf2txt">to text
      </select>
    </p>
+   <div id="pdf_dpi_elem">
+     <input type="number" id="pdf_dpi" name="dpi" value="720"> DPI
+   </div>
+   <div id="pdf_maxbitmap_elem">
+     <input type="number" id="pdf_maxbitmap" name="maxbitmap" value="0"> Maxbitmap
+   </div>
+   <div id="pdf_pdfsettings_elem">
+     <input type="text" id="pdf_pdfsettings" name="pdfsettings" value="/default"> Pdfsettings
+   </div>
    <form id="pdfop">
    <input type="file" name="file">
    <input type="button" value="Convert" onclick="pdfConverter()">

diff --git a/src/laundry/pdf.clj b/src/laundry/pdf.clj
@@ -18,11 +18,14 @@
         (io/delete-file path)))))
 
 ;; pdf/a converter
-(s/defn api-pdf2pdfa [env, tempfile :- java.io.File]
+(s/defn api-pdf2pdfa [env, tempfile :- java.io.File, dpinum :- s/Int, maxbitmapnum :- s/Int, pdfsettings :- s/Str]
   (let [in-path (.getAbsolutePath tempfile)
         out-path  (str (.getAbsolutePath tempfile) ".pdf")
+        dpi (str dpinum)
+        maxbitmap (str maxbitmapnum)
+        pdfsettings (str pdfsettings)
         res (shell-out! (str (:tools env) "/pdf2pdfa")
-                        in-path out-path)]
+                        in-path out-path dpi maxbitmap pdfsettings)]
     (.delete tempfile)
     (if (= (:exit res) 0)
       (htresp/content-type
@@ -77,10 +80,11 @@
          (api-pdf2txt env tempfile)))
      (POST "/pdf2pdfa" []
        :summary "attempt to convert a PDF file to PDF/A"
+       :query-params [{dpi :- s/Int 720} {maxbitmap :- s/Int 0} {pdfsettings :- s/Str "/default"}]
        :multipart-params [file :- upload/TempFileUpload]
        :middleware [wrap-multipart-params]
        (let [tempfile (:tempfile file)
              filename (:filename file)]
          (info "PDF converter received " filename "(" (:size file) "b)")
          (.deleteOnExit tempfile) ;; cleanup if VM is terminated
-         (api-pdf2pdfa env tempfile))))))
+         (api-pdf2pdfa env tempfile dpi maxbitmap pdfsettings))))))
diff --git a/test/laundry/pdf_test.clj b/test/laundry/pdf_test.clj
@@ -39,6 +39,21 @@
     (is (= "application/pdf" (get-in response [:headers "Content-Type"])))
     (is (clojure.string/starts-with? body "%PDF-"))))
 
+(deftest ^:integration api-pdf-pdf2pdfa-with-opt-args
+  (let [app (fixture/get-app)
+        file (io/file (io/resource "hypno.pdf"))
+        _ (assert file)
+        request (-> (mock/request :post "/pdf/pdf2pdfa")
+                    (assoc-in [:query-params :dpi] 720)
+                    (assoc-in [:query-params :maxbitmap] 0)
+                    (assoc-in [:query-params :pdfsettings] "/default")
+                    (merge (peridot.multipart/build {:file file})))
+        response (app request)
+        body (ring.util.request/body-string response)]
+    (is (= 200 (:status response)))
+    (is (= "application/pdf" (get-in response [:headers "Content-Type"])))
+    (is (clojure.string/starts-with? body "%PDF-"))))
+
 (deftest ^:integration api-pdf-pdf2txt
   (let [app (fixture/get-app)
         file (io/file (io/resource "hypno.pdf"))