Attachment #25597 for bug #49446

View | Details | Raw Unified | Return to bug 49446
Collapse All | Expand All

Lines 54-59 Link Here

(-)src/ooxml/java/org/apache/poi/xwpf/usermodel/XWPFParagraph.java (-1 / +8 lines)
54	import org.openxmlformats.schemas.wordprocessingml.x2006.main.STLineSpacingRule;	54	import org.openxmlformats.schemas.wordprocessingml.x2006.main.STLineSpacingRule;
55	import org.openxmlformats.schemas.wordprocessingml.x2006.main.STOnOff;	55	import org.openxmlformats.schemas.wordprocessingml.x2006.main.STOnOff;
56	import org.openxmlformats.schemas.wordprocessingml.x2006.main.STTextAlignment;	56	import org.openxmlformats.schemas.wordprocessingml.x2006.main.STTextAlignment;
		57	import org.w3c.dom.Node;
57	import org.w3c.dom.NodeList;	58	import org.w3c.dom.NodeList;
58	import org.w3c.dom.Text;	59	import org.w3c.dom.Text;
59		60
Lines 133-139 Link Here
133	while (c.toNextSelection()) {	134	while (c.toNextSelection()) {
134	XmlObject o = c.getObject();	135	XmlObject o = c.getObject();
135	if (o instanceof CTText) {	136	if (o instanceof CTText) {
136	text.append(((CTText) o).getStringValue());	137	String tagName = o.getDomNode().getNodeName();
		138	// field codes (w:instrText, defined in spec sec. 17.16.23)
		139	// come up as instances of CTText, but they only
		140	// pollute the output
		141	if (!tagName.equals("w:instrText")) {
		142	text.append(((CTText) o).getStringValue());
		143	}
137	}	144	}
138	if (o instanceof CTPTab) {	145	if (o instanceof CTPTab) {
139	text.append("\t");	146	text.append("\t");




       // Now check the first paragraph in total
       assertTrue(extractor.getText().contains("a\tb\n"));
    }
    
    /**
     * The output should not contain field codes, e.g. those specified in the
     * w:instrText tag (spec sec. 17.16.23)
     */
    public void testNoFieldCodes() {
        XWPFDocument doc = XWPFTestDataSamples.openSampleDocument("FieldCodes.docx");
        XWPFWordExtractor extractor = new XWPFWordExtractor(doc);
        String text = extractor.getText();
        assertTrue(text.length() > 0);
        assertFalse(text.contains("AUTHOR"));
        assertFalse(text.contains("CREATEDATE"));
    }
}

Return to bug 49446

Lines 237-240 Link Here

(-)src/ooxml/testcases/org/apache/poi/xwpf/extractor/TestXWPFWordExtractor.java (+13 lines)
237	// Now check the first paragraph in total	237	// Now check the first paragraph in total
238	assertTrue(extractor.getText().contains("a\tb\n"));	238	assertTrue(extractor.getText().contains("a\tb\n"));
239	}	239	}
		240
		241	/**
		242	* The output should not contain field codes, e.g. those specified in the
		243	* w:instrText tag (spec sec. 17.16.23)
		244	*/
		245	public void testNoFieldCodes() {
		246	XWPFDocument doc = XWPFTestDataSamples.openSampleDocument("FieldCodes.docx");
		247	XWPFWordExtractor extractor = new XWPFWordExtractor(doc);
		248	String text = extractor.getText();
		249	assertTrue(text.length() > 0);
		250	assertFalse(text.contains("AUTHOR"));
		251	assertFalse(text.contains("CREATEDATE"));
		252	}
240	}	253	}