fix for TIKA-3134 contributed by tothd

tika-parsers/src/main/java/org/apache/tika/parser/pdf/AbstractPDF2XHTML.java

@@ -539,7 +539,7 @@ protected void endPage(PDPage page) throws IOException {
                 doOCROnCurrentPage();
             } else if (config.getOcrStrategy().equals(PDFParserConfig.OCR_STRATEGY.AUTO)) {
                 //TODO add more sophistication
-                if (totalCharsPerPage < 10 || unmappedUnicodeCharsPerPage > 10) {
+                if (totalCharsPerPage < config.getTotalCharsPerPage() || unmappedUnicodeCharsPerPage > config.getUnmappedUnicodeCharsPerPage()) {
                     doOCROnCurrentPage();
                 }
             }

tika-parsers/src/main/java/org/apache/tika/parser/pdf/PDFParserConfig.java

@@ -158,6 +158,10 @@ private static OCR_STRATEGY parse(String s) {
 
     private boolean detectAngles = false;
 
+	private int totalCharsPerPage = 10;
+
+    private int unmappedUnicodeCharsPerPage = 10;
+
     public PDFParserConfig() {
         init(this.getClass().getResourceAsStream("PDFParser.properties"));
     }
@@ -263,6 +267,14 @@ private void init(InputStream is) {
 
         maxMainMemoryBytes = getLongProp(props.getProperty("maxMainMemoryBytes"), -1);
         detectAngles = getBooleanProp(props.getProperty("detectAngles"), false);
+
+		setTotalCharsPerPage(
+			getIntProp(props.getProperty("totalCharsPerPage"),
+				getTotalCharsPerPage()));
+
+		setUnmappedUnicodeCharsPerPage(
+			getIntProp(props.getProperty("unmappedUnicodeCharsPerPage"), 
+				getUnmappedUnicodeCharsPerPage()));
     }
 
     /**
@@ -838,6 +850,22 @@ public boolean getDetectAngles() {
         return detectAngles;
     }
 
+	public void setTotalCharsPerPage(int totalCharsPerPage) {
+        this.totalCharsPerPage = totalCharsPerPage;
+    }
+
+    public int getTotalCharsPerPage() {
+        return totalCharsPerPage;
+    }
+
+    public void setUnmappedUnicodeCharsPerPage(int unmappedUnicodeCharsPerPage) {
+        this.unmappedUnicodeCharsPerPage = unmappedUnicodeCharsPerPage;
+    }
+
+    public int getUnmappedUnicodeCharsPerPage() {
+        return unmappedUnicodeCharsPerPage;
+    }
+
     @Override
     public boolean equals(Object o) {
         if (this == o) return true;
@@ -864,6 +892,8 @@ public boolean equals(Object o) {
         if (!getOcrImageFormatName().equals(config.getOcrImageFormatName())) return false;
         if (getExtractActions() != config.getExtractActions()) return false;
         if (!getAccessChecker().equals(config.getAccessChecker())) return false;
+		if (getTotalCharsPerPage() != config.getTotalCharsPerPage()) return false;
+        if (getUnmappedUnicodeCharsPerPage() != config.getUnmappedUnicodeCharsPerPage()) return false;
         return getMaxMainMemoryBytes() == config.getMaxMainMemoryBytes();
     }
 
@@ -889,6 +919,8 @@ public int hashCode() {
         result = 31 * result + (getCatchIntermediateIOExceptions() ? 1 : 0);
         result = 31 * result + (getExtractActions() ? 1 : 0);
         result = 31 * result + Long.valueOf(getMaxMainMemoryBytes()).hashCode();
+		result = 31 * result + getTotalCharsPerPage();
+        result = 31 * result + getUnmappedUnicodeCharsPerPage();
         return result;
     }
 
@@ -915,6 +947,8 @@ public String toString() {
                 ", extractActions=" + extractActions +
                 ", catchIntermediateIOExceptions=" + catchIntermediateIOExceptions +
                 ", maxMainMemoryBytes=" + maxMainMemoryBytes +
+				", totalCharsPerPage=" + totalCharsPerPage +
+                ", unmappedUnicodeCharsPerPage=" + unmappedUnicodeCharsPerPage +
                 '}';
     }
 }