Method parse

tools/src/main/java/be/gov/data/tools/LinkChecker.java:114–155

Parse CSV file with URLs. Last column of the row should be the URL.

()

     * Last column of the row should be the URL.
     */
    private static void parse() {
        String file = getProperty("urlfile");
        String outfile = getProperty("outfile");
        
        int skiplines = Integer.parseInt(getProperty("skiplines"));
        List<String> okcodes = Arrays.asList(getProperty("httpok").split(","));
       
        try (   BufferedReader r = Files.newBufferedReader(Paths.get(file));
                BufferedWriter w = Files.newBufferedWriter(Paths.get(outfile));) {
           
            logger.info("Loading urls from {}, skipping {} lines", file, skiplines);
            
            String line = r.readLine();
            
            // skip header line(s)
            for(int i = 0; (i < skiplines) && (line != null); i++) {
                line = r.readLine();
            }
         
            int urlOK = 0;
            int urlBad = 0;
            int code = -1;
         
            while(line != null) {
                // SPARQL CSV writer always uses "," as separator
                String s[] = line.split(",");
                code = checkURL(s[s.length-1]);
                if (okcodes.contains(Integer.toString(code))) {
                    urlOK++;
                } else {
                    urlBad++;
                    w.write(code + ";" + line);
                    w.newLine();
                }
                line = r.readLine();
            }
            logger.info("Checked {} urls: {} ok, {} not ok", 
                                urlOK + urlBad, urlOK, urlBad);
        } catch (IOException ex) {
            logger.error("Error loading file", ex);
        }
    }
    
    /**
     * Main program