|
3 | 3 | import java.io.IOException; |
4 | 4 | import java.nio.file.Files; |
5 | 5 | import java.nio.file.Path; |
6 | | -import java.util.ArrayList; |
7 | 6 | import java.util.List; |
8 | 7 | import java.util.Optional; |
9 | 8 | import java.util.regex.Matcher; |
|
19 | 18 | * Parses the contents of a .blg (BibTeX log) file to extract warning messages. |
20 | 19 | */ |
21 | 20 | public class BibtexLogParser { |
22 | | - private static final Pattern WARNING_PATTERN = Pattern.compile("^Warning--(?<message>[a-zA-Z ]+) in (?<entryKey>[^\\s]+)$"); |
| 21 | + private static final Pattern BIBTEX_WARNING_PATTERN = Pattern.compile("^Warning--(?<message>[a-zA-Z ]+) in (?<entryKey>[^\\s]+)$"); |
| 22 | + private static final Pattern BIBLATEX_WARNING_PATTERN = Pattern.compile( |
| 23 | + "(?:(?:\\[\\d+\\] )?Biber\\.pm:\\d+> )?WARN - Datamodel: [a-z]+ entry '(?<entryKey>[^']+)' \\((?<fileName>[^)]+)\\): (?<message>.+)"); |
| 24 | + |
23 | 25 | private static final String EMPTY_FIELD_PREFIX = "empty"; |
| 26 | + private static final String INVALID_FIELD_PREFIX = "field '"; |
| 27 | + private static final String MULTI_INVALID_FIELD_PREFIX = "field - one of '"; |
24 | 28 |
|
25 | 29 | public List<BibWarning> parseBiblog(@NonNull Path blgFilePath) throws IOException { |
26 | | - List<BibWarning> warnings = new ArrayList<>(); |
27 | | - List<String> lines = Files.readAllLines(blgFilePath); |
28 | | - for (String line : lines) { |
29 | | - Optional<BibWarning> potentialWarning = parseWarningLine(line); |
30 | | - potentialWarning.ifPresent(warnings::add); |
31 | | - } |
32 | | - return warnings; |
| 30 | + return Files.lines(blgFilePath) |
| 31 | + .map(this::parseWarningLine) |
| 32 | + .flatMap(Optional::stream) |
| 33 | + .toList(); |
33 | 34 | } |
34 | 35 |
|
35 | | - /** |
36 | | - * Parses a single line from the .blg file to identify a warning. |
37 | | - * <p> |
38 | | - * Currently supports parsing warnings of the format: |
39 | | - * <pre> |
40 | | - * Warning--[message] in [entryKey] |
41 | | - * </pre> |
42 | | - * For example: {@code Warning--empty journal in Scholey_2013} |
43 | | - * |
44 | | - * @param line a single line from the .blg file |
45 | | - * @return an Optional containing a {@link BibWarning} if a match is found, or empty otherwise |
46 | | - */ |
47 | | - private Optional<BibWarning> parseWarningLine(String line) { |
48 | | - // TODO: Support additional warning formats |
49 | | - Matcher matcher = WARNING_PATTERN.matcher(line); |
50 | | - if (!matcher.find()) { |
51 | | - return Optional.empty(); |
| 36 | + /// Parses a single line from a .blg file to identify a warning. |
| 37 | + /// |
| 38 | + /// This method supports two warning formats: |
| 39 | + /// |
| 40 | + /// 1. **BibTeX Warnings:** Simple warnings from the legacy BibTeX backend. |
| 41 | + /// `Warning--[message] in [entryKey]` |
| 42 | + /// For example: `Warning--empty journal in Scholey_2013` |
| 43 | + /// |
| 44 | + /// 2. **BibLaTeX Datamodel Warnings:** Detailed warnings from the Biber backend, including datamodel validation issues. |
| 45 | + /// `[Log line] > WARN - Datamodel: [entry type] entry '[entryKey]' ([fileName]): [message]` |
| 46 | + /// For example: `Biber.pm:123> WARN - Datamodel: article entry 'Scholey_2013' (file.bib): Invalid field 'journal'` |
| 47 | + /// |
| 48 | + /// @param line The single line from the .blg file to parse. |
| 49 | + /// |
| 50 | + /// @returns An `Optional` containing a `BibWarning` if a match is found, or an empty `Optional` otherwise. |
| 51 | + Optional<BibWarning> parseWarningLine(String line) { |
| 52 | + Matcher bibtexMatcher = BIBTEX_WARNING_PATTERN.matcher(line); |
| 53 | + if (bibtexMatcher.find()) { |
| 54 | + String message = bibtexMatcher.group("message").trim(); |
| 55 | + String entryKey = bibtexMatcher.group("entryKey"); |
| 56 | + // Extract field name for warnings related to empty fields (e.g., "empty journal" -> fieldName = "journal") |
| 57 | + String fieldName = null; |
| 58 | + if (message.startsWith(EMPTY_FIELD_PREFIX)) { |
| 59 | + fieldName = message.substring(EMPTY_FIELD_PREFIX.length()).trim(); |
| 60 | + fieldName = FieldFactory.parseField(fieldName).getName(); |
| 61 | + } |
| 62 | + |
| 63 | + return Optional.of(new BibWarning( |
| 64 | + SeverityType.WARNING, |
| 65 | + message, |
| 66 | + fieldName, |
| 67 | + entryKey |
| 68 | + )); |
52 | 69 | } |
53 | 70 |
|
54 | | - String message = matcher.group("message").trim(); |
55 | | - String entryKey = matcher.group("entryKey"); |
56 | | - // Extract field name for warnings related to empty fields (e.g., "empty journal" -> fieldName = "journal") |
57 | | - String fieldName = null; |
58 | | - if (message.startsWith(EMPTY_FIELD_PREFIX)) { |
59 | | - fieldName = message.substring(EMPTY_FIELD_PREFIX.length()).trim(); |
60 | | - fieldName = FieldFactory.parseField(fieldName).getName(); |
| 71 | + Matcher biblatexMatcher = BIBLATEX_WARNING_PATTERN.matcher(line); |
| 72 | + if (biblatexMatcher.find()) { |
| 73 | + String message = biblatexMatcher.group("message").trim(); |
| 74 | + String entryKey = biblatexMatcher.group("entryKey"); |
| 75 | + String fieldName = null; |
| 76 | + |
| 77 | + // Extract field name for warnings related to invalid fields (e.g., "Invalid field 'publisher' for entrytype 'article'" -> fieldName = "publisher") |
| 78 | + String lowerCaseMessage = message.toLowerCase(); |
| 79 | + if (lowerCaseMessage.contains(INVALID_FIELD_PREFIX)) { |
| 80 | + int startIndex = lowerCaseMessage.indexOf(INVALID_FIELD_PREFIX) + INVALID_FIELD_PREFIX.length(); |
| 81 | + int endIndex = lowerCaseMessage.indexOf('\'', startIndex); |
| 82 | + if (endIndex != -1) { |
| 83 | + fieldName = lowerCaseMessage.substring(startIndex, endIndex).trim(); |
| 84 | + fieldName = FieldFactory.parseField(fieldName).getName(); |
| 85 | + } |
| 86 | + } else if (lowerCaseMessage.contains(MULTI_INVALID_FIELD_PREFIX)) { |
| 87 | + int startIndex = lowerCaseMessage.indexOf(MULTI_INVALID_FIELD_PREFIX) + MULTI_INVALID_FIELD_PREFIX.length(); |
| 88 | + int endIndex = lowerCaseMessage.indexOf('\'', startIndex); |
| 89 | + if (endIndex != -1) { |
| 90 | + fieldName = lowerCaseMessage.substring(startIndex, endIndex).trim().split(",")[0].trim(); |
| 91 | + fieldName = FieldFactory.parseField(fieldName).getName(); |
| 92 | + } |
| 93 | + } |
| 94 | + |
| 95 | + return Optional.of(new BibWarning( |
| 96 | + SeverityType.WARNING, |
| 97 | + message, |
| 98 | + fieldName, |
| 99 | + entryKey |
| 100 | + )); |
61 | 101 | } |
62 | 102 |
|
63 | | - return Optional.of(new BibWarning( |
64 | | - SeverityType.WARNING, |
65 | | - message, |
66 | | - fieldName, |
67 | | - entryKey |
68 | | - )); |
| 103 | + return Optional.empty(); |
69 | 104 | } |
70 | 105 | } |
0 commit comments