1. add strict mode

2. bump version to 0.1.3
hwding · Sep 9, 2017 · 3c30f64 · 3c30f64
1 parent 83865bf
commit 3c30f64
Show file tree

Hide file tree

Showing 9 changed files with 61 additions and 34 deletions.
diff --git a/README.md b/README.md
@@ -25,13 +25,16 @@ Remove text stamps of **any font**, **any encoding** and **any language** with p
 
 ## Usage
 ```
-Usage: 
+Usage:
    [OPTION] -i [INPUT PDF] -k [KEYWORDS...] (-o [OUTPUT PDF])
    [OPTION] -I [INPUT DIR] -k [KEYWORDS...] (-O [OUTPUT DIR])
 
 Options:
-   -d,  --directly          directly modify the input file(s), which makes option o/O unnecessary
+   -d,  --directly          directly modify the input file(s), option o/O is
+                            unnecessary when this option is on
    -r,  --recursive         process files in the given dir recursively
+   -s,  --strict            use strict mode, a text area is considered as water mark
+                            only if its content strictly equals one of the keywords
 ```
 
 ## Get it now
@@ -48,10 +51,10 @@ Make sure you have `wget` installed.
 #### Run
 ```shell
 # For single file processing
-➜ unstamp -i "C Recipes.pdf" -o "C Recipes.unstamped.pdf" -k www.allitebooks.com
+➜ unstamp -i "C Recipes.pdf" -o "C Recipes.unstamped.pdf" -k www.allitebooks.com -s
 ➜ unstamp -i RoR.pdf -o RoR.unstamped.pdf -k 图灵社区会员
 # Or
-➜ unstamp -i "C Recipes.pdf" -d -k www.allitebooks.com
+➜ unstamp -i "C Recipes.pdf" -d -k www.allitebooks.com -s
 ➜ unstamp -i RoR.pdf -d -k 图灵社区会员
 
 # For massive files processing

diff --git a/pom.xml b/pom.xml
@@ -6,7 +6,7 @@
     <modelVersion>4.0.0</modelVersion>
     <groupId>com.amastigote</groupId>
     <artifactId>unstamper</artifactId>
-    <version>0.1.2</version>
+    <version>0.1.3</version>
     <description>Text stamp remover for PDF files.</description>
     <name>pdf-unstamper</name>
     <url>https://github.com/hwding/pdf-unstamper</url>

diff --git a/script/install b/script/install
@@ -8,7 +8,7 @@ user_bin=`echo ~`"/bin/"
 locl_bin="/usr/local/bin/"
 jar_name="pdf-unstamper.jar"
 exe_name="unstamp"
-_version="0.1.2"
+_version="0.1.3"
 jar_durl="https://github.com/hwding/pdf-unstamper/releases/download/$_version/$jar_name"
 wrapper="#!/bin/bash\njava -jar ${user_bin}${jar_name} \"\$@\"\n"
 

diff --git a/src/com/amastigote/unstamper/Main.java b/src/com/amastigote/unstamper/Main.java
@@ -1,6 +1,6 @@
 /*
   AUTH | hwding
-  DATE | Sep 05 2017
+  DATE | Sep 10 2017
   DESC | text stamp remover for PDF files
   MAIL | [email protected]
   GITH | github.com/hwding
@@ -37,7 +37,8 @@ public static void main(@NotNull String[] args) {
             System.exit(0);
         } else {
             TaskRunner.init(
-                    commandLine.getOptionValues('k'));
+                    commandLine.getOptionValues('k'),
+                    commandLine.hasOption('s'));
 
             if (commandLine.hasOption('i') && (commandLine.hasOption('o') || commandLine.hasOption('d'))) {
                 if (commandLine.hasOption('d'))

diff --git a/src/com/amastigote/unstamper/core/Processor.java b/src/com/amastigote/unstamper/core/Processor.java
@@ -1,6 +1,6 @@
 /*
   AUTH | hwding
-  DATE | Sep 05 2017
+  DATE | Sep 10 2017
   DESC | text stamp remover for PDF files
   MAIL | [email protected]
   GITH | github.com/hwding
@@ -27,7 +27,8 @@
 public class Processor {
     public static void process(
             @NotNull File file,
-            @NotNull String[] strings) {
+            @NotNull String[] strings,
+            @NotNull boolean useStrict) {
         AtomicBoolean processAllOk = new AtomicBoolean(true);
         GeneralLogger.Processor.procInProgress(file.getName());
 
@@ -61,7 +62,7 @@ public static void process(
                         if (e instanceof COSString) {
                             /* Ignore Any Exception During Parallel Processing */
                             try {
-                                if (TextStampRecognizer.recognize(strings, ((COSString) e).getBytes(), pdFonts))
+                                if (TextStampRecognizer.recognize(strings, ((COSString) e).getBytes(), pdFonts, useStrict))
                                     ((COSString) e).setValue(new byte[0]);
                             } catch (Exception ignored) {
                             }

diff --git a/src/com/amastigote/unstamper/core/TextStampRecognizer.java b/src/com/amastigote/unstamper/core/TextStampRecognizer.java
@@ -1,6 +1,6 @@
 /*
   AUTH | hwding
-  DATE | Sep 05 2017
+  DATE | Sep 10 2017
   DESC | text stamp remover for PDF files
   MAIL | [email protected]
   GITH | github.com/hwding
@@ -18,15 +18,16 @@ class TextStampRecognizer {
     private static boolean recognizeWithFont(
             @NotNull String[] keywords,
             @NotNull byte[] inputText,
-            @NotNull Set<PDFont> pdFonts) {
-        String bs = generateByteString(inputText);
+            @NotNull Set<PDFont> pdFonts,
+            @NotNull boolean useStrict) {
+        String encodedInput = generateByteString(inputText);
         for (PDFont f : pdFonts) {
             if (f == null) continue;
             for (String k : keywords) {
                 try {
-                    byte[] encodedKeywords = f.encode(k);
-                    if (bs.contains(generateByteString(encodedKeywords)))
-                        return true;
+                    byte[] encodedKeywordBytes = f.encode(k);
+                    final String encodedKeyword = generateByteString(encodedKeywordBytes);
+                    if (checkDuplicate(encodedInput, encodedKeyword, useStrict)) return true;
                 } catch (IOException | IllegalArgumentException ignored) {
                 }
             }
@@ -36,26 +37,38 @@ private static boolean recognizeWithFont(
 
     private static boolean recognizePlain(
             @NotNull String[] keywords,
-            @NotNull byte[] inputText
+            @NotNull byte[] inputText,
+            @NotNull boolean useStrict
     ) {
-        for (String k : keywords) {
-            if (new String(inputText).contains(k)) return true;
+        for (String k : keywords)
+            if (checkDuplicate(new String(inputText), k, useStrict)) return true;
+        return false;
+    }
+
+    private static boolean checkDuplicate(
+            @NotNull String input,
+            @NotNull String keyword,
+            @NotNull boolean useStrict) {
+        if (useStrict) {
+            if (input.equals(keyword)) return true;
+        } else {
+            if (input.contains(keyword)) return true;
         }
         return false;
     }
 
     static boolean recognize(@NotNull String[] keywords,
                              @NotNull byte[] inputText,
-                             @NotNull Set<PDFont> pdFonts) {
-        return recognizePlain(keywords, inputText) ||
-                recognizeWithFont(keywords, inputText, pdFonts);
+                             @NotNull Set<PDFont> pdFonts,
+                             @NotNull boolean useStrict) {
+        return recognizePlain(keywords, inputText, useStrict) ||
+                recognizeWithFont(keywords, inputText, pdFonts, useStrict);
     }
 
     private static String generateByteString(@NotNull byte[] bytes) {
         StringBuilder stringBuilder = new StringBuilder();
-        for (byte b : bytes) {
+        for (byte b : bytes)
             stringBuilder.append(Byte.toString(b));
-        }
         return stringBuilder.toString();
     }
 }
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 | Sep 05 2017
+  DATE | Sep 10 2017
   DESC | text stamp remover for PDF files
   MAIL | [email protected]
   GITH | github.com/hwding
@@ -12,13 +12,16 @@
 public class GeneralLogger {
     public static class Help {
         private static final String usage =
-                "\nPDF-UnStamper ver. 0.1.2 by hwding@GitHub\n" +
+                "\nPDF-UnStamper ver. 0.1.3 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" +
                         "\nOptions:" +
-                        "\n   -d,  --directly          directly modify the input file(s), which makes option o/O unnecessary" +
-                        "\n   -r,  --recursive         process files in the given dir recursively\n";
+                        "\n   -d,  --directly          directly modify the input file(s), option o/O is\n" +
+                        "                            unnecessary when this option is on" +
+                        "\n   -r,  --recursive         process files in the given dir recursively" +
+                        "\n   -s,  --strict            use strict mode, a text area is considered as water mark\n" +
+                        "                            only if its content strictly equals one of the keywords\n";
 
         public static void print() {
             System.out.println(usage);
@@ -61,7 +64,7 @@ public static void procInProgress(@NotNull String fn) {
         }
 
         public static void procFinished() {
-            System.out.println(" GOOD");
+            System.out.println(" done");
         }
     }
 }
diff --git a/src/com/amastigote/unstamper/util/OptionManager.java b/src/com/amastigote/unstamper/util/OptionManager.java
@@ -1,6 +1,6 @@
 /*
   AUTH | hwding
-  DATE | Aug 25 2017
+  DATE | Sep 10 2017
   DESC | text stamp remover for PDF files
   MAIL | [email protected]
   GITH | github.com/hwding
@@ -18,6 +18,7 @@ public class OptionManager {
     private final static Option optionK = new Option("k", true, null);
     private final static Option optionD = new Option("d", "directly", false, null);
     private final static Option optionR = new Option("r", "recursive", false, null);
+    private final static Option optionS = new Option("s", "strict", false, null);
 
     public static Options buildOptions() {
         Options options = new Options();
@@ -34,6 +35,7 @@ public static Options buildOptions() {
         options.addOption(optionD);
         options.addOption(optionR);
         options.addOption(optionK);
+        options.addOption(optionS);
         return options;
     }
 }
diff --git a/src/com/amastigote/unstamper/util/TaskRunner.java b/src/com/amastigote/unstamper/util/TaskRunner.java
@@ -1,6 +1,6 @@
 /*
   AUTH | hwding
-  DATE | Sep 05 2017
+  DATE | Sep 10 2017
   DESC | text stamp remover for PDF files
   MAIL | [email protected]
   GITH | github.com/hwding
@@ -19,9 +19,13 @@
 
 public class TaskRunner {
     private static String[] keywords;
+    private static boolean useStrict;
 
-    public static void init(@NotNull String[] keywords) {
+    public static void init(
+            @NotNull String[] keywords,
+            @NotNull boolean useStrict) {
         TaskRunner.keywords = keywords;
+        TaskRunner.useStrict = useStrict;
     }
 
     public static void procSingleFile(
@@ -42,7 +46,7 @@ public static void procSingleFile(
     }
 
     private static void submitToProcessor(@NotNull File file) {
-        Processor.process(file, keywords);
+        Processor.process(file, keywords, useStrict);
     }
 
     public static void procSingleFileDirectly(@NotNull String ifn) {