- provide a more elegant and efficient way to load fonts in each page

- bump version & head comment date
hwding · Sep 4, 2017 · 08ac529 · 08ac529
1 parent 1615901
commit 08ac529
Show file tree

Hide file tree

Showing 4 changed files with 30 additions and 44 deletions.
diff --git a/pom.xml b/pom.xml
@@ -6,8 +6,8 @@
     <modelVersion>4.0.0</modelVersion>
     <groupId>com.amastigote</groupId>
     <artifactId>unstamper</artifactId>
-    <version>0.1.1</version>
-    <description>text stamp remover for PDF files</description>
+    <version>0.1.2</version>
+    <description>Text stamp remover for PDF files.</description>
     <name>pdf-unstamper</name>
     <url>https://github.com/hwding/pdf-unstamper</url>
     <build>

diff --git a/script/install b/script/install
@@ -1,13 +1,13 @@
 #!/bin/bash
 
 #AUTH hwding
-#DATE AUG/25/2017
+#DATE SEP/04/2017
 #DESC install unstamp as a command
 
 user_bin=`echo ~`"/bin/"
 jar_name="pdf-unstamper.jar"
 exe_name="unstamp"
-_version="0.1.1"
+_version="0.1.2"
 jar_durl="https://github.com/hwding/pdf-unstamper/releases/download/$_version/$jar_name"
 
 function chk_f() {

diff --git a/src/com/amastigote/unstamper/core/Processor.java b/src/com/amastigote/unstamper/core/Processor.java
@@ -1,14 +1,13 @@
 /*
   AUTH | hwding
-  DATE | Aug 27 2017
+  DATE | Sep 04 2017
   DESC | text stamp remover for PDF files
   MAIL | [email protected]
   GITH | github.com/hwding
  */
 package com.amastigote.unstamper.core;
 
 import com.amastigote.unstamper.log.GeneralLogger;
-import org.apache.pdfbox.cos.COSName;
 import org.apache.pdfbox.cos.COSString;
 import org.apache.pdfbox.pdfparser.PDFStreamParser;
 import org.apache.pdfbox.pdfwriter.ContentStreamWriter;
@@ -19,11 +18,9 @@
 import java.io.File;
 import java.io.IOException;
 import java.io.OutputStream;
-import java.util.Collections;
 import java.util.HashSet;
 import java.util.List;
 import java.util.Set;
-import java.util.stream.Collectors;
 
 public class Processor {
     public static void process(File file, String[] strings) {
@@ -35,45 +32,34 @@ public static void process(File file, String[] strings) {
             PDDocument pdDocument = PDDocument.load(file);
             pdDocument.getPages().forEach(pdPage -> {
                 try {
-                    /* START: loading font resources for further parsing */
+                    /* START: loading font resources from current page */
                     PDFStreamParser pdfStreamParser = new PDFStreamParser(pdPage);
                     pdfStreamParser.parse();
 
-                    List<Object> objects =
-                            Collections.synchronizedList(pdfStreamParser.getTokens());
+                    List<Object> objects = pdfStreamParser.getTokens();
+                    Set<PDFont> pdFonts = new HashSet<>();
 
-                    List<Object> cosNames =
-                            objects.parallelStream()
-                                    .filter(e -> e instanceof COSName)
-                                    .collect(Collectors.toList());
-
-                    Set<PDFont> pdFonts =
-                            Collections.synchronizedSet(new HashSet<>());
-
-                    cosNames.parallelStream()
-                            .forEach(e -> {
-                                /* Ignore Any Exception During Parallel Processing */
-                                try {
-                                    PDFont pdFont = pdPage.getResources().getFont(((COSName) e));
-                                    if (pdFont != null)
-                                        pdFonts.add(pdFont);
-                                } catch (Exception ignored) {
-                                }
-                            });
+                    pdPage.getResources().getFontNames().forEach(e -> {
+                        /* Ignore Any Exception During Parallel Processing */
+                        try {
+                            PDFont pdFont = pdPage.getResources().getFont(e);
+                            if (pdFont != null)
+                                pdFonts.add(pdFont);
+                        } catch (Exception ignored) {
+                        }
+                    });
                     /* END */
-                    objects
-                            .parallelStream()
-                            .forEach(e -> {
-                                        if (e instanceof COSString) {
-                                    /* Ignore Any Exception During Parallel Processing */
-                                            try {
-                                                if (TextStampRecognizer.recognize(strings, ((COSString) e).getBytes(), pdFonts))
-                                                    ((COSString) e).setValue(new byte[0]);
-                                            } catch (Exception ignored) {
-                                            }
-                                        }
-                                    }
-                            );
+
+                    objects.parallelStream().forEach(e -> {
+                        if (e instanceof COSString) {
+                            /* Ignore Any Exception During Parallel Processing */
+                            try {
+                                if (TextStampRecognizer.recognize(strings, ((COSString) e).getBytes(), pdFonts))
+                                    ((COSString) e).setValue(new byte[0]);
+                            } catch (Exception ignored) {
+                            }
+                        }
+                    });
 
                     PDStream newContents = new PDStream(pdDocument);
                     OutputStream out = newContents.createOutputStream();

diff --git a/src/com/amastigote/unstamper/log/GeneralLogger.java b/src/com/amastigote/unstamper/log/GeneralLogger.java
@@ -1,6 +1,6 @@
 /*
   AUTH | hwding
-  DATE | Aug 27 2017
+  DATE | Sep 04 2017
   DESC | text stamp remover for PDF files
   MAIL | [email protected]
   GITH | github.com/hwding
@@ -10,7 +10,7 @@
 public class GeneralLogger {
     public static class Help {
         private static final String usage =
-                "\nPDF-UnStamper ver. 0.1.1 by hwding@GitHub\n" +
+                "\nPDF-UnStamper ver. 0.1.2 by hwding@GitHub\n" +
                         "\nUsage: " +
                         "\n   [OPTION] -i [INPUT PDF] -k [KEYWORDS...] (-o [OUTPUT PDF])" +
                         "\n   [OPTION] -I [INPUT DIR] -k [KEYWORDS...] (-O [OUTPUT DIR])\n" +