Examples with InsertDelta difflib.InsertDelta used on opensource projects

Example 1 with InsertDelta

use of difflib.InsertDelta in project pdfbox by apache.

the class TestTextStripper method doTestFile.

/**
 * Validate text extraction on a single file.
 *
 * @param inFile The PDF file to validate
 * @param outDir The directory to store the output in
 * @param bLogResult Whether to log the extracted text
 * @param bSort Whether or not the extracted text is sorted
 * @throws Exception when there is an exception
 */
public void doTestFile(File inFile, File outDir, boolean bLogResult, boolean bSort) throws Exception {
    if (bSort) {
        log.info("Preparing to parse " + inFile.getName() + " for sorted test");
    } else {
        log.info("Preparing to parse " + inFile.getName() + " for standard test");
    }
    if (!outDir.exists()) {
        if (!outDir.mkdirs()) {
            throw (new Exception("Error creating " + outDir.getAbsolutePath() + " directory"));
        }
    }
    // System.out.println("  " + inFile + (bSort ? " (sorted)" : ""));
    try (PDDocument document = PDDocument.load(inFile)) {
        File outFile;
        File diffFile;
        File expectedFile;
        if (bSort) {
            outFile = new File(outDir, inFile.getName() + "-sorted.txt");
            diffFile = new File(outDir, inFile.getName() + "-sorted-diff.txt");
            expectedFile = new File(inFile.getParentFile(), inFile.getName() + "-sorted.txt");
        } else {
            outFile = new File(outDir, inFile.getName() + ".txt");
            diffFile = new File(outDir, inFile.getName() + "-diff.txt");
            expectedFile = new File(inFile.getParentFile(), inFile.getName() + ".txt");
        }
        // delete possible leftover
        diffFile.delete();
        try (OutputStream os = new FileOutputStream(outFile)) {
            os.write(0xEF);
            os.write(0xBB);
            os.write(0xBF);
            try (Writer writer = new BufferedWriter(new OutputStreamWriter(os, ENCODING))) {
                // Allows for sorted tests
                stripper.setSortByPosition(bSort);
                stripper.writeText(document, writer);
            // close the written file before reading it again
            }
        }
        if (bLogResult) {
            log.info("Text for " + inFile.getName() + ":");
            log.info(stripper.getText(document));
        }
        if (!expectedFile.exists()) {
            this.bFail = true;
            log.error("FAILURE: Input verification file: " + expectedFile.getAbsolutePath() + " did not exist");
            return;
        }
        boolean localFail = false;
        try (LineNumberReader expectedReader = new LineNumberReader(new InputStreamReader(new FileInputStream(expectedFile), ENCODING));
            LineNumberReader actualReader = new LineNumberReader(new InputStreamReader(new FileInputStream(outFile), ENCODING))) {
            while (true) {
                String expectedLine = expectedReader.readLine();
                while (expectedLine != null && expectedLine.trim().length() == 0) {
                    expectedLine = expectedReader.readLine();
                }
                String actualLine = actualReader.readLine();
                while (actualLine != null && actualLine.trim().length() == 0) {
                    actualLine = actualReader.readLine();
                }
                if (!stringsEqual(expectedLine, actualLine)) {
                    this.bFail = true;
                    localFail = true;
                    log.error("FAILURE: Line mismatch for file " + inFile.getName() + " (sort = " + bSort + ")" + " at expected line: " + expectedReader.getLineNumber() + " at actual line: " + actualReader.getLineNumber() + "\nexpected line was: \"" + expectedLine + "\"" + "\nactual line was:   \"" + actualLine + "\"" + "\n");
                // lets report all lines, even though this might produce some verbose logging
                // break;
                }
                if (expectedLine == null || actualLine == null) {
                    break;
                }
            }
        }
        if (!localFail) {
            outFile.delete();
        } else {
            // https://code.google.com/p/java-diff-utils/wiki/SampleUsage
            List<String> original = fileToLines(expectedFile);
            List<String> revised = fileToLines(outFile);
            // Compute diff. Get the Patch object. Patch is the container for computed deltas.
            Patch patch = DiffUtils.diff(original, revised);
            try (PrintStream diffPS = new PrintStream(diffFile, ENCODING)) {
                for (Object delta : patch.getDeltas()) {
                    if (delta instanceof ChangeDelta) {
                        ChangeDelta cdelta = (ChangeDelta) delta;
                        diffPS.println("Org: " + cdelta.getOriginal());
                        diffPS.println("New: " + cdelta.getRevised());
                        diffPS.println();
                    } else if (delta instanceof DeleteDelta) {
                        DeleteDelta ddelta = (DeleteDelta) delta;
                        diffPS.println("Org: " + ddelta.getOriginal());
                        diffPS.println("New: " + ddelta.getRevised());
                        diffPS.println();
                    } else if (delta instanceof InsertDelta) {
                        InsertDelta idelta = (InsertDelta) delta;
                        diffPS.println("Org: " + idelta.getOriginal());
                        diffPS.println("New: " + idelta.getRevised());
                        diffPS.println();
                    } else {
                        diffPS.println(delta);
                    }
                }
            }
        }
    }
}