DOIService.java

1 
package edu.ucsb.cs.citelines.services;
2
3 
import java.util.regex.Matcher;
4 
import java.util.regex.Pattern;
5 
import org.springframework.stereotype.Service;
6
7 
/**
8 
 * Recognizes a DOI (Digital Object Identifier) expressed in any of several common formats and
9 
 * normalizes it to its canonical bare form (ANSI/NISO Z39.84), e.g. {@code
10 
 * 10.1038/s41586-020-2649-2}.
11 
 *
12 
 * <p>Recognized formats include:
13 
 *
14 
 * <ul>
15 
 *   <li>Actionable web URLs, e.g. {@code https://doi.org/10.1038/s41586-020-2649-2} or the
16 
 *       deprecated {@code http://dx.doi.org/10.1038/s41586-020-2649-2}
17 
 *   <li>shortDOI URLs, e.g. {@code https://doi.org/c234} or {@code https://doi.org/10/c234}
18 
 *   <li>Plain/bare DOI strings, e.g. {@code 10.1038/s41586-020-2649-2}
19 
 *   <li>The {@code doi:} prefix, e.g. {@code doi:10.1038/s41586-020-2649-2}
20 
 *   <li>The {@code urn:doi:} namespace, e.g. {@code urn:doi:10.1038/s41586-020-2649-2}
21 
 *   <li>The (deprecated) {@code info:doi/} URI scheme, e.g. {@code
22 
 *       info:doi/10.1038/s41586-020-2649-2}
23 
 * </ul>
24 
 */
25 
@Service
26 
public class DOIService {
27
28 
  // Matches a bare/canonical DOI, wherever it appears within a larger string (e.g. embedded in a
29 
  // URL, an HTML anchor, or a doi:/urn:doi:/info:doi/ prefixed identifier).
30 
  private static final Pattern DOI_PATTERN =
31 
      Pattern.compile("\\b10\\.\\d{4,}(?:\\.\\d+)*/[^\\s<>\"]+", Pattern.CASE_INSENSITIVE);
32
33 
  // Matches a shortDOI, either as a bare "10/xxxx" identifier or as a doi.org URL, with or
34 
  // without the leading "10/" segment (e.g. "10/c234", "https://doi.org/c234", or
35 
  // "https://doi.org/10/c234").
36 
  private static final Pattern SHORT_DOI_PATTERN =
37 
      Pattern.compile(
38 
          "^(?:https?://(?:dx\\.)?doi\\.org/(?:10/)?|10/)([a-zA-Z0-9]+)$",
39 
          Pattern.CASE_INSENSITIVE);
40
41 
  /**
42 
   * Normalizes a DOI given in any of the recognized formats to its canonical bare form.
43 
   *
44 
   * @param rawDoi the DOI, in any recognized format
45 
   * @return the canonical, lower-cased, bare DOI (e.g. {@code 10.1038/s41586-020-2649-2})
46 
   * @throws IllegalArgumentException if {@code rawDoi} cannot be recognized as a DOI in any of the
47 
   *     supported formats
48 
   */
49 
  public String normalizeRawDOI(String rawDoi) {
50 1 1. normalizeRawDOI : negated conditional → KILLED
 
    if (rawDoi != null) {
51 
      String trimmed = rawDoi.trim();
52
53 
      Matcher shortDoiMatcher = SHORT_DOI_PATTERN.matcher(trimmed);
54 1 1. normalizeRawDOI : negated conditional → KILLED
 
      if (shortDoiMatcher.matches()) {
55 1 1. normalizeRawDOI : replaced return value with "" for edu/ucsb/cs/citelines/services/DOIService::normalizeRawDOI → KILLED
 
        return "10/" + shortDoiMatcher.group(1).toLowerCase();
56 
      }
57
58 
      Matcher doiMatcher = DOI_PATTERN.matcher(trimmed);
59 1 1. normalizeRawDOI : negated conditional → KILLED
 
      if (doiMatcher.find()) {
60 1 1. normalizeRawDOI : replaced return value with "" for edu/ucsb/cs/citelines/services/DOIService::normalizeRawDOI → KILLED
 
        return doiMatcher.group().toLowerCase();
61 
      }
62 
    }
63
64 
    throw new IllegalArgumentException("Argument cannot be recognized as a DOI: " + rawDoi);
65 
  }
66 
}

Mutations

50

1.1
Location : normalizeRawDOI
Killed by : edu.ucsb.cs.citelines.services.DOIServiceTests.[engine:junit-jupiter]/[class:edu.ucsb.cs.citelines.services.DOIServiceTests]/[method:a_shortdoi_url_with_10_prefix_is_normalized()] negated conditional → KILLED

54

1.1
Location : normalizeRawDOI
Killed by : edu.ucsb.cs.citelines.services.DOIServiceTests.[engine:junit-jupiter]/[class:edu.ucsb.cs.citelines.services.DOIServiceTests]/[method:a_shortdoi_url_with_10_prefix_is_normalized()] negated conditional → KILLED

55

1.1
Location : normalizeRawDOI
Killed by : edu.ucsb.cs.citelines.services.DOIServiceTests.[engine:junit-jupiter]/[class:edu.ucsb.cs.citelines.services.DOIServiceTests]/[method:a_bare_shortdoi_is_normalized()] replaced return value with "" for edu/ucsb/cs/citelines/services/DOIService::normalizeRawDOI → KILLED

59

1.1
Location : normalizeRawDOI
Killed by : edu.ucsb.cs.citelines.services.DOIServiceTests.[engine:junit-jupiter]/[class:edu.ucsb.cs.citelines.services.DOIServiceTests]/[method:a_bare_doi_is_returned_unchanged()] negated conditional → KILLED

60

1.1
Location : normalizeRawDOI
Killed by : edu.ucsb.cs.citelines.services.DOIServiceTests.[engine:junit-jupiter]/[class:edu.ucsb.cs.citelines.services.DOIServiceTests]/[method:a_bare_doi_is_returned_unchanged()] replaced return value with "" for edu/ucsb/cs/citelines/services/DOIService::normalizeRawDOI → KILLED

Active mutators

Tests examined


Report generated by PIT 1.17.0