From dd7f39cb35c4a1368982f386dc85facd18cbab4e Mon Sep 17 00:00:00 2001 From: Victor-Buendia Date: Sun, 28 Aug 2022 14:32:09 -0300 Subject: [PATCH] Fix capitalization after en-dash characters in title --- CHANGELOG.md | 1 + .../logic/formatter/casechanger/Word.java | 29 +- .../casechanger/CapitalizeFormatterTest.java | 46 +- .../importer/fileformat/IsiImporterTest.java | 641 +++++++++--------- 4 files changed, 388 insertions(+), 329 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index a869d7a09f0..e510976a435 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -31,6 +31,7 @@ Note that this project **does not** adhere to [Semantic Versioning](http://semve - We fixed a bug where spaces are trimmed when highlighting differences in the Entries merge dialog. [koppor#371](https://github.com/koppor/jabref/issues/371) - We fixed several bugs regarding the manual and the autosave of library files that sometimes lead to exceptions or data loss. [#8448](https://github.com/JabRef/jabref/issues/8484), [#8746](https://github.com/JabRef/jabref/issues/8746), [#6684](https://github.com/JabRef/jabref/issues/6684), [#6644](https://github.com/JabRef/jabref/issues/6644), [#6102](https://github.com/JabRef/jabref/issues/6102), [#6002](https://github.com/JabRef/jabref/issues/6000) - We fixed an issue where applied save actions on saving the library file would lead to the dialog "The libary has been modified by another program" popping up [#4877](https://github.com/JabRef/jabref/issues/4877) +- Title case capitalizes after en-dash in title. [#9068](https://github.com/JabRef/jabref/issues/9068) ### Removed diff --git a/src/main/java/org/jabref/logic/formatter/casechanger/Word.java b/src/main/java/org/jabref/logic/formatter/casechanger/Word.java index bf5eee51bb6..8126664796f 100644 --- a/src/main/java/org/jabref/logic/formatter/casechanger/Word.java +++ b/src/main/java/org/jabref/logic/formatter/casechanger/Word.java @@ -10,30 +10,42 @@ /** * Represents a word in a title of a bibtex entry. *

- * A word can have protected chars (enclosed in '{' '}') and may be a small (a, an, the, ...) word. + * A word can have protected chars (enclosed in '{' '}') and may be a small (a, + * an, the, ...) word. */ public final class Word { /** * Set containing common lowercase function words */ public static final Set SMALLER_WORDS; + public static final Set DASHES_N_HYPHENS; private final char[] chars; private final boolean[] protectedChars; static { Set smallerWords = new HashSet<>(); + Set dashesAndHyphens = new HashSet<>(); // Articles smallerWords.addAll(Arrays.asList("a", "an", "the")); // Prepositions - smallerWords.addAll(Arrays.asList("above", "about", "across", "against", "along", "among", "around", "at", "before", "behind", "below", "beneath", "beside", "between", "beyond", "by", "down", "during", "except", "for", "from", "in", "inside", "into", "like", "near", "of", "off", "on", "onto", "since", "to", "toward", "through", "under", "until", "up", "upon", "with", "within", "without")); + smallerWords.addAll(Arrays.asList("above", "about", "across", "against", "along", "among", "around", "at", + "before", "behind", "below", "beneath", "beside", "between", "beyond", "by", "down", "during", "except", + "for", "from", "in", "inside", "into", "like", "near", "of", "off", "on", "onto", "since", "to", + "toward", "through", "under", "until", "up", "upon", "with", "within", "without")); // Conjunctions smallerWords.addAll(Arrays.asList("and", "but", "for", "nor", "or", "so", "yet")); + // Dashes and Hyphens + dashesAndHyphens.addAll(Arrays.asList('-', '~', '֊', '־', '᐀', '‐', '‑', '‒', '–', '—', '―', '⁓', '⁻', '₋', '−', + '⸗', '⸺', '⸻', '〜', '〰', '゠', '︱', '︲', '﹘', '﹣', '－')); + // unmodifiable for thread safety SMALLER_WORDS = smallerWords.stream() - .map(word -> word.toLowerCase(Locale.ROOT)) - .collect(Collectors.toUnmodifiableSet()); + .map(word -> word.toLowerCase(Locale.ROOT)) + .collect(Collectors.toUnmodifiableSet()); + + DASHES_N_HYPHENS = dashesAndHyphens; } public Word(char[] chars, boolean[] protectedChars) { @@ -46,7 +58,8 @@ public Word(char[] chars, boolean[] protectedChars) { } /** - * Case-insensitive check against {@link Word#SMALLER_WORDS}. Checks for common function words. + * Case-insensitive check against {@link Word#SMALLER_WORDS}. Checks for common + * function words. */ public static boolean isSmallerWord(String word) { return SMALLER_WORDS.contains(word.toLowerCase(Locale.ROOT)); @@ -77,9 +90,9 @@ public void toLowerCase() { public void toUpperFirst() { for (int i = 0; i < chars.length; i++) { if (!protectedChars[i]) { - chars[i] = (i == 0) ? - Character.toUpperCase(chars[i]) : - Character.toLowerCase(chars[i]); + chars[i] = (i == 0 || DASHES_N_HYPHENS.contains(chars[i - 1])) + ? Character.toUpperCase(chars[i]) + : Character.toLowerCase(chars[i]); } } } diff --git a/src/test/java/org/jabref/logic/formatter/casechanger/CapitalizeFormatterTest.java b/src/test/java/org/jabref/logic/formatter/casechanger/CapitalizeFormatterTest.java index 6210e48b96b..8cb0aab0c60 100644 --- a/src/test/java/org/jabref/logic/formatter/casechanger/CapitalizeFormatterTest.java +++ b/src/test/java/org/jabref/logic/formatter/casechanger/CapitalizeFormatterTest.java @@ -8,7 +8,8 @@ import static org.junit.jupiter.api.Assertions.assertEquals; /** - * Tests in addition to the general tests from {@link org.jabref.logic.formatter.FormatterTest} + * Tests in addition to the general tests from + * {@link org.jabref.logic.formatter.FormatterTest} */ public class CapitalizeFormatterTest { @@ -41,8 +42,47 @@ public void formatExample() { "upper each {NOT} first, Upper Each {NOT} First", // multiple words lower case with {} "Upper {E}ach {NOT} First, Upper {E}ach {NOT} First", // multiple words correct with {} "UPPER {E}ACH {NOT} FIRST, Upper {E}ach {NOT} First", // multiple words upper case with {} - "upper each first {NOT} {this}, Upper Each First {NOT} {this}", // multiple words in lower and upper case with {} - "upper each first {N}OT {t}his, Upper Each First {N}ot {t}his", // multiple words in lower and upper case with {} part 2 + "upper each first {NOT} {this}, Upper Each First {NOT} {this}", // multiple words in lower and upper case + // with {} + "upper each first {N}OT {t}his, Upper Each First {N}ot {t}his", // multiple words in lower and upper case + // with {} part 2 + "first-second USING hypheN, First-Second Using Hyphen", // multiple words in lower and upper case using + // hyphen + "breaking Your Next TEST-cASE, Breaking Your Next Test-Case", // multiple words in lower and upper case + // using hyphen + "THIS looks〰LIKE a daSH, This Looks〰Like A Dash", // dash-like character + "one-dash, One-Dash", // testing all dash-like characters + "one~dash, One~Dash", + "one֊dash, One֊Dash", + "one־dash, One־Dash", + "one᐀dash, One᐀Dash", + "one‐dash, One‐Dash", + "one‑dash, One‑Dash", + "one‒dash, One‒Dash", + "one–dash, One–Dash", + "one—dash, One—Dash", + "one―dash, One―Dash", + "one⁓dash, One⁓Dash", + "one⁻dash, One⁻Dash", + "one₋dash, One₋Dash", + "one−dash, One−Dash", + "one⸗dash, One⸗Dash", + "one⸺dash, One⸺Dash", + "one⸻dash, One⸻Dash", + "one〜dash, One〜Dash", + "one〰dash, One〰Dash", + "one゠dash, One゠Dash", + "one︱dash, One︱Dash", + "one︲dash, One︲Dash", + "one﹘dash, One﹘Dash", + "one﹣dash, One﹣Dash", + "one－dash, One－Dash", + "--, --", // testing weird cases + "--this is one sentence, --This Is One Sentence", + "-d-o-i-t right-, -D-O-I-T Right-", + "testing hy---------------phen, Testing Hy---------------Phen", + "its getting cr--a---z-y, Its Getting Cr--A---Z-Y", + "does it wor⸻k with cra゠︱zy dashes?, Does It Wor⸻K With Cra゠︱Zy Dashes?", }) public void testInputs(String input, String expectedResult) { String formattedStr = formatter.format(input); diff --git a/src/test/java/org/jabref/logic/importer/fileformat/IsiImporterTest.java b/src/test/java/org/jabref/logic/importer/fileformat/IsiImporterTest.java index f0b385f74d4..621a080eeb7 100644 --- a/src/test/java/org/jabref/logic/importer/fileformat/IsiImporterTest.java +++ b/src/test/java/org/jabref/logic/importer/fileformat/IsiImporterTest.java @@ -25,324 +25,329 @@ public class IsiImporterTest { - private static final String FILE_ENDING = ".isi"; - private final IsiImporter importer = new IsiImporter(); - - private static Stream fileNames() throws IOException { - Predicate fileName = name -> name.startsWith("IsiImporterTest") - && !name.contains("Empty") - && name.endsWith(FILE_ENDING); - return ImporterTestEngine.getTestFiles(fileName).stream(); - } - - private static Stream invalidFileNames() throws IOException { - Predicate fileName = name -> !name.startsWith("IsiImporterTest"); - return ImporterTestEngine.getTestFiles(fileName).stream(); - } - - @Test - public void testParseMonthException() { - IsiImporter.parseMonth("20l06 06-07"); - } - - @Test - public void testGetFormatName() { - assertEquals("ISI", importer.getName()); - } - - @Test - public void testGetCLIId() { - assertEquals("isi", importer.getId()); - } - - @Test - public void testsGetExtensions() { - assertEquals(StandardFileType.ISI, importer.getFileType()); - } - - @Test - public void testGetDescription() { - assertEquals("Importer for the ISI Web of Science, INSPEC and Medline format.", - importer.getDescription()); - } - - @ParameterizedTest - @MethodSource("fileNames") - public void testIsRecognizedFormatAccepted(String fileName) throws IOException { - ImporterTestEngine.testIsRecognizedFormat(importer, fileName); - } - - @ParameterizedTest - @MethodSource("invalidFileNames") - public void testIsRecognizedFormatRejected(String fileName) throws IOException { - ImporterTestEngine.testIsNotRecognizedFormat(importer, fileName); - } - - @Test - public void testProcessSubSup() { - HashMap subs = new HashMap<>(); - - subs.put(StandardField.TITLE, "/sub 3/"); - IsiImporter.processSubSup(subs); - assertEquals("$_3$", subs.get(StandardField.TITLE)); - - subs.put(StandardField.TITLE, "/sub 3 /"); - IsiImporter.processSubSup(subs); - assertEquals("$_3$", subs.get(StandardField.TITLE)); - - subs.put(StandardField.TITLE, "/sub 31/"); - IsiImporter.processSubSup(subs); - assertEquals("$_{31}$", subs.get(StandardField.TITLE)); - - subs.put(StandardField.ABSTRACT, "/sub 3/"); - IsiImporter.processSubSup(subs); - assertEquals("$_3$", subs.get(StandardField.ABSTRACT)); - - subs.put(StandardField.REVIEW, "/sub 31/"); - IsiImporter.processSubSup(subs); - assertEquals("$_{31}$", subs.get(StandardField.REVIEW)); - - subs.put(StandardField.TITLE, "/sup 3/"); - IsiImporter.processSubSup(subs); - assertEquals("$^3$", subs.get(StandardField.TITLE)); - - subs.put(StandardField.TITLE, "/sup 31/"); - IsiImporter.processSubSup(subs); - assertEquals("$^{31}$", subs.get(StandardField.TITLE)); - - subs.put(StandardField.ABSTRACT, "/sup 3/"); - IsiImporter.processSubSup(subs); - assertEquals("$^3$", subs.get(StandardField.ABSTRACT)); - - subs.put(StandardField.REVIEW, "/sup 31/"); - IsiImporter.processSubSup(subs); - assertEquals("$^{31}$", subs.get(StandardField.REVIEW)); - - subs.put(StandardField.TITLE, "/sub $Hello/"); - IsiImporter.processSubSup(subs); - assertEquals("$_{\\$Hello}$", subs.get(StandardField.TITLE)); - } - - @Test - public void testImportEntries1() throws IOException, URISyntaxException { - Path file = Path.of(IsiImporterTest.class.getResource("IsiImporterTest1.isi").toURI()); - List entries = importer.importDatabase(file).getDatabase().getEntries(); - BibEntry entry = entries.get(0); - - assertEquals(1, entries.size()); - assertEquals(Optional.of("Optical properties of MgO doped LiNbO$_3$ single crystals"), - entry.getField(StandardField.TITLE)); - assertEquals( - Optional.of( - "James Brown and James Marc Brown and Brown, J. M. and Brown, J. and Brown, J. M. and Brown, J."), - entry.getField(StandardField.AUTHOR)); - assertEquals(StandardEntryType.Article, entry.getType()); - assertEquals(Optional.of("Optical Materials"), entry.getField(StandardField.JOURNAL)); - assertEquals(Optional.of("2006"), entry.getField(StandardField.YEAR)); - assertEquals(Optional.of("28"), entry.getField(StandardField.VOLUME)); - assertEquals(Optional.of("5"), entry.getField(StandardField.NUMBER)); - assertEquals(Optional.of("467--72"), entry.getField(StandardField.PAGES)); - } - - @Test - public void testImportEntries2() throws IOException, URISyntaxException { - Path file = Path.of(IsiImporterTest.class.getResource("IsiImporterTest2.isi").toURI()); - List entries = importer.importDatabase(file).getDatabase().getEntries(); - BibEntry entry = entries.get(0); - - assertEquals(3, entries.size()); - assertEquals(Optional.of("Optical properties of MgO doped LiNbO$_3$ single crystals"), - entry.getField(StandardField.TITLE)); - assertEquals(StandardEntryType.Misc, entry.getType()); - assertEquals(Optional.of("Optical Materials"), entry.getField(StandardField.JOURNAL)); - assertEquals(Optional.of("2006"), entry.getField(StandardField.YEAR)); - assertEquals(Optional.of("28"), entry.getField(StandardField.VOLUME)); - assertEquals(Optional.of("5"), entry.getField(StandardField.NUMBER)); - assertEquals(Optional.of("467-72"), entry.getField(StandardField.PAGES)); - } - - @Test - public void testImportEntriesINSPEC() throws IOException, URISyntaxException { - Path file = Path.of(IsiImporterTest.class.getResource("IsiImporterTestInspec.isi").toURI()); - List entries = importer.importDatabase(file).getDatabase().getEntries(); - - BibEntry first = entries.get(0); - BibEntry second = entries.get(1); - - if (first.getField(StandardField.TITLE).equals( - Optional.of("Optical and photoelectric spectroscopy of photorefractive Sn$_2$P$_2$S$_6$ crystals"))) { - BibEntry tmp = first; - first = second; - second = tmp; + private static final String FILE_ENDING = ".isi"; + private final IsiImporter importer = new IsiImporter(); + + private static Stream fileNames() throws IOException { + Predicate fileName = name -> name.startsWith("IsiImporterTest") + && !name.contains("Empty") + && name.endsWith(FILE_ENDING); + return ImporterTestEngine.getTestFiles(fileName).stream(); } - assertEquals(2, entries.size()); - assertEquals( - Optional.of( - "Second harmonic generation of continuous wave ultraviolet light and production of beta -BaB$_2$O$_4$ optical waveguides"), - first.getField(StandardField.TITLE)); - assertEquals(StandardEntryType.Article, first.getType()); - - assertEquals(Optional.of("Degl'Innocenti, R. and Guarino, A. and Poberaj, G. and Gunter, P."), - first.getField(StandardField.AUTHOR)); - assertEquals(Optional.of("Applied Physics Letters"), first.getField(StandardField.JOURNAL)); - assertEquals(Optional.of("2006"), first.getField(StandardField.YEAR)); - assertEquals(Optional.of(Month.JULY), first.getMonth()); - assertEquals(Optional.of("89"), first.getField(StandardField.VOLUME)); - assertEquals(Optional.of("4"), first.getField(StandardField.NUMBER)); - assertEquals(Optional.of("Lorem ipsum abstract"), first.getField(StandardField.ABSTRACT)); - assertEquals(Optional.of("Aip"), first.getField(StandardField.PUBLISHER)); - assertEquals( - Optional.of("Optical and photoelectric spectroscopy of photorefractive Sn$_2$P$_2$S$_6$ crystals"), - second.getField(StandardField.TITLE)); - assertEquals(StandardEntryType.Article, second.getType()); - } - - @Test - public void testImportEntriesWOS() throws IOException, URISyntaxException { - Path file = Path.of(IsiImporterTest.class.getResource("IsiImporterTestWOS.isi").toURI()); - List entries = importer.importDatabase(file).getDatabase().getEntries(); - - BibEntry first = entries.get(0); - BibEntry second = entries.get(1); - - assertEquals(2, entries.size()); - - assertEquals(Optional.of("Optical and photoelectric spectroscopy of photorefractive Sn2P2S6 crystals"), - first.getField(StandardField.TITLE)); - assertEquals(Optional.of("Optical waveguides in Sn2P2S6 by low fluence MeV He+ ion implantation"), - second.getField(StandardField.TITLE)); - - assertEquals(Optional.of("Journal of Physics-condensed Matter"), first.getField(StandardField.JOURNAL)); - } - - @Test - public void testIsiAuthorsConvert() { - assertEquals( - "James Brown and James Marc Brown and Brown, J. M. and Brown, J. and Brown, J. M. and Brown, J.", - IsiImporter.isiAuthorsConvert( - "James Brown and James Marc Brown and Brown, J.M. and Brown, J. and Brown, J.M. and Brown, J.")); - - assertEquals( - "Joffe, Hadine and Hall, Janet E. and Gruber, Staci and Sarmiento, Ingrid A. and Cohen, Lee S. and Yurgelun-Todd, Deborah and Martin, Kathryn A.", - IsiImporter.isiAuthorsConvert( - "Joffe, Hadine; Hall, Janet E; Gruber, Staci; Sarmiento, Ingrid A; Cohen, Lee S; Yurgelun-Todd, Deborah; Martin, Kathryn A")); - } - - @Test - public void testMonthConvert() { - assertEquals("#jun#", IsiImporter.parseMonth("06")); - assertEquals("#jun#", IsiImporter.parseMonth("JUN")); - assertEquals("#jun#", IsiImporter.parseMonth("jUn")); - assertEquals("#may#", IsiImporter.parseMonth("MAY-JUN")); - assertEquals("#jun#", IsiImporter.parseMonth("2006 06")); - assertEquals("#jun#", IsiImporter.parseMonth("2006 06-07")); - assertEquals("#jul#", IsiImporter.parseMonth("2006 07 03")); - assertEquals("#may#", IsiImporter.parseMonth("2006 May-Jun")); - } - - @Test - public void testIsiAuthorConvert() { - assertEquals("James Brown", IsiImporter.isiAuthorConvert("James Brown")); - assertEquals("James Marc Brown", IsiImporter.isiAuthorConvert("James Marc Brown")); - assertEquals("Brown, J. M.", IsiImporter.isiAuthorConvert("Brown, J.M.")); - assertEquals("Brown, J.", IsiImporter.isiAuthorConvert("Brown, J.")); - assertEquals("Brown, J. M.", IsiImporter.isiAuthorConvert("Brown, JM")); - assertEquals("Brown, J.", IsiImporter.isiAuthorConvert("Brown, J")); - assertEquals("Brown, James", IsiImporter.isiAuthorConvert("Brown, James")); - assertEquals("Hall, Janet E.", IsiImporter.isiAuthorConvert("Hall, Janet E")); - assertEquals("", IsiImporter.isiAuthorConvert("")); - } - - @Test - public void testImportIEEEExport() throws IOException, URISyntaxException { - Path file = Path.of(IsiImporterTest.class.getResource("IEEEImport1.txt").toURI()); - List entries = importer.importDatabase(file).getDatabase().getEntries(); - BibEntry entry = entries.get(0); - - assertEquals(1, entries.size()); - assertEquals(StandardEntryType.Article, entry.getType()); - assertEquals(Optional.of("Geoscience and Remote Sensing Letters, IEEE"), entry.getField(StandardField.JOURNAL)); - assertEquals(Optional.of("Improving Urban Road Extraction in High-Resolution " - + "Images Exploiting Directional Filtering, Perceptual " + "Grouping, and Simple Topological Concepts"), - entry.getField(StandardField.TITLE)); - assertEquals(Optional.of("4"), entry.getField(StandardField.VOLUME)); - assertEquals(Optional.of("3"), entry.getField(StandardField.NUMBER)); - assertEquals(Optional.of("1545-598X"), entry.getField(new UnknownField("SN"))); - assertEquals(Optional.of("387--391"), entry.getField(StandardField.PAGES)); - assertEquals(Optional.of("Gamba, P. and Dell'Acqua, F. and Lisini, G."), entry.getField(StandardField.AUTHOR)); - assertEquals(Optional.of("2006"), entry.getField(StandardField.YEAR)); - assertEquals(Optional.of("Perceptual grouping, street extraction, urban remote sensing"), - entry.getField(StandardField.KEYWORDS)); - assertEquals(Optional.of("Lorem ipsum abstract"), entry.getField(StandardField.ABSTRACT)); - } - - @Test - public void testIEEEImport() throws IOException, URISyntaxException { - Path file = Path.of(IsiImporterTest.class.getResource("IEEEImport1.txt").toURI()); - List entries = importer.importDatabase(file).getDatabase().getEntries(); - BibEntry entry = entries.get(0); - - assertEquals(1, entries.size()); - assertEquals(StandardEntryType.Article, entry.getType()); - assertEquals(Optional.of("Geoscience and Remote Sensing Letters, IEEE"), entry.getField(StandardField.JOURNAL)); - assertEquals( - Optional.of( - "Improving Urban Road Extraction in High-Resolution Images Exploiting Directional Filtering, Perceptual Grouping, and Simple Topological Concepts"), - entry.getField(StandardField.TITLE)); - assertEquals(Optional.of("4"), entry.getField(StandardField.VOLUME)); - assertEquals(Optional.of("3"), entry.getField(StandardField.NUMBER)); - assertEquals(Optional.of("1545-598X"), entry.getField(new UnknownField("SN"))); - assertEquals(Optional.of("387--391"), entry.getField(StandardField.PAGES)); - assertEquals(Optional.of("Gamba, P. and Dell'Acqua, F. and Lisini, G."), entry.getField(StandardField.AUTHOR)); - assertEquals(Optional.of("2006"), entry.getField(StandardField.YEAR)); - assertEquals(Optional.of("Perceptual grouping, street extraction, urban remote sensing"), - entry.getField(StandardField.KEYWORDS)); - assertEquals(Optional.of("Lorem ipsum abstract"), entry.getField(StandardField.ABSTRACT)); - } - - @Test - public void testImportEntriesMedline() throws IOException, URISyntaxException { - Path file = Path.of(IsiImporterTest.class.getResource("IsiImporterTestMedline.isi").toURI()); - List entries = importer.importDatabase(file).getDatabase().getEntries(); - - BibEntry first = entries.get(0); - BibEntry second = entries.get(1); - - assertEquals(2, entries.size()); - assertEquals( - Optional.of("Effects of modafinil on cognitive performance and alertness during sleep deprivation."), - first.getField(StandardField.TITLE)); - assertEquals(Optional.of("Wesensten, Nancy J."), first.getField(StandardField.AUTHOR)); - assertEquals(Optional.of("Curr Pharm Des"), first.getField(StandardField.JOURNAL)); - assertEquals(Optional.of("2006"), first.getField(StandardField.YEAR)); - assertEquals(Optional.empty(), first.getField(StandardField.MONTH)); - assertEquals(Optional.of("12"), first.getField(StandardField.VOLUME)); - assertEquals(Optional.of("20"), first.getField(StandardField.NUMBER)); - assertEquals(Optional.of("2457--71"), first.getField(StandardField.PAGES)); - assertEquals(StandardEntryType.Article, first.getType()); - assertEquals( - Optional.of( - "Estrogen therapy selectively enhances prefrontal cognitive processes: a randomized, double-blind, placebo-controlled study with functional magnetic resonance imaging in perimenopausal and recently postmenopausal women."), - second.getField(StandardField.TITLE)); - assertEquals( - Optional.of( - "Joffe, Hadine and Hall, Janet E. and Gruber, Staci and Sarmiento, Ingrid A. and Cohen, Lee S. and Yurgelun-Todd, Deborah and Martin, Kathryn A."), - second.getField(StandardField.AUTHOR)); - assertEquals(Optional.of("2006"), second.getField(StandardField.YEAR)); - assertEquals(Optional.of(Month.MAY), second.getMonth()); - assertEquals(Optional.of("13"), second.getField(StandardField.VOLUME)); - assertEquals(Optional.of("3"), second.getField(StandardField.NUMBER)); - assertEquals(Optional.of("411--22"), second.getField(StandardField.PAGES)); - assertEquals(StandardEntryType.Article, second.getType()); - } - - @Test - public void testImportEntriesEmpty() throws IOException, URISyntaxException { - Path file = Path.of(IsiImporterTest.class.getResource("IsiImporterTestEmpty.isi").toURI()); - - List entries = importer.importDatabase(file).getDatabase().getEntries(); - - assertEquals(1, entries.size()); - } + private static Stream invalidFileNames() throws IOException { + Predicate fileName = name -> !name.startsWith("IsiImporterTest"); + return ImporterTestEngine.getTestFiles(fileName).stream(); + } + + @Test + public void testParseMonthException() { + IsiImporter.parseMonth("20l06 06-07"); + } + + @Test + public void testGetFormatName() { + assertEquals("ISI", importer.getName()); + } + + @Test + public void testGetCLIId() { + assertEquals("isi", importer.getId()); + } + + @Test + public void testsGetExtensions() { + assertEquals(StandardFileType.ISI, importer.getFileType()); + } + + @Test + public void testGetDescription() { + assertEquals("Importer for the ISI Web of Science, INSPEC and Medline format.", + importer.getDescription()); + } + + @ParameterizedTest + @MethodSource("fileNames") + public void testIsRecognizedFormatAccepted(String fileName) throws IOException { + ImporterTestEngine.testIsRecognizedFormat(importer, fileName); + } + + @ParameterizedTest + @MethodSource("invalidFileNames") + public void testIsRecognizedFormatRejected(String fileName) throws IOException { + ImporterTestEngine.testIsNotRecognizedFormat(importer, fileName); + } + + @Test + public void testProcessSubSup() { + HashMap subs = new HashMap<>(); + + subs.put(StandardField.TITLE, "/sub 3/"); + IsiImporter.processSubSup(subs); + assertEquals("$_3$", subs.get(StandardField.TITLE)); + + subs.put(StandardField.TITLE, "/sub 3 /"); + IsiImporter.processSubSup(subs); + assertEquals("$_3$", subs.get(StandardField.TITLE)); + + subs.put(StandardField.TITLE, "/sub 31/"); + IsiImporter.processSubSup(subs); + assertEquals("$_{31}$", subs.get(StandardField.TITLE)); + + subs.put(StandardField.ABSTRACT, "/sub 3/"); + IsiImporter.processSubSup(subs); + assertEquals("$_3$", subs.get(StandardField.ABSTRACT)); + + subs.put(StandardField.REVIEW, "/sub 31/"); + IsiImporter.processSubSup(subs); + assertEquals("$_{31}$", subs.get(StandardField.REVIEW)); + + subs.put(StandardField.TITLE, "/sup 3/"); + IsiImporter.processSubSup(subs); + assertEquals("$^3$", subs.get(StandardField.TITLE)); + + subs.put(StandardField.TITLE, "/sup 31/"); + IsiImporter.processSubSup(subs); + assertEquals("$^{31}$", subs.get(StandardField.TITLE)); + + subs.put(StandardField.ABSTRACT, "/sup 3/"); + IsiImporter.processSubSup(subs); + assertEquals("$^3$", subs.get(StandardField.ABSTRACT)); + + subs.put(StandardField.REVIEW, "/sup 31/"); + IsiImporter.processSubSup(subs); + assertEquals("$^{31}$", subs.get(StandardField.REVIEW)); + + subs.put(StandardField.TITLE, "/sub $Hello/"); + IsiImporter.processSubSup(subs); + assertEquals("$_{\\$Hello}$", subs.get(StandardField.TITLE)); + } + + @Test + public void testImportEntries1() throws IOException, URISyntaxException { + Path file = Path.of(IsiImporterTest.class.getResource("IsiImporterTest1.isi").toURI()); + List entries = importer.importDatabase(file).getDatabase().getEntries(); + BibEntry entry = entries.get(0); + + assertEquals(1, entries.size()); + assertEquals(Optional.of("Optical properties of MgO doped LiNbO$_3$ single crystals"), + entry.getField(StandardField.TITLE)); + assertEquals( + Optional.of( + "James Brown and James Marc Brown and Brown, J. M. and Brown, J. and Brown, J. M. and Brown, J."), + entry.getField(StandardField.AUTHOR)); + assertEquals(StandardEntryType.Article, entry.getType()); + assertEquals(Optional.of("Optical Materials"), entry.getField(StandardField.JOURNAL)); + assertEquals(Optional.of("2006"), entry.getField(StandardField.YEAR)); + assertEquals(Optional.of("28"), entry.getField(StandardField.VOLUME)); + assertEquals(Optional.of("5"), entry.getField(StandardField.NUMBER)); + assertEquals(Optional.of("467--72"), entry.getField(StandardField.PAGES)); + } + + @Test + public void testImportEntries2() throws IOException, URISyntaxException { + Path file = Path.of(IsiImporterTest.class.getResource("IsiImporterTest2.isi").toURI()); + List entries = importer.importDatabase(file).getDatabase().getEntries(); + BibEntry entry = entries.get(0); + + assertEquals(3, entries.size()); + assertEquals(Optional.of("Optical properties of MgO doped LiNbO$_3$ single crystals"), + entry.getField(StandardField.TITLE)); + assertEquals(StandardEntryType.Misc, entry.getType()); + assertEquals(Optional.of("Optical Materials"), entry.getField(StandardField.JOURNAL)); + assertEquals(Optional.of("2006"), entry.getField(StandardField.YEAR)); + assertEquals(Optional.of("28"), entry.getField(StandardField.VOLUME)); + assertEquals(Optional.of("5"), entry.getField(StandardField.NUMBER)); + assertEquals(Optional.of("467-72"), entry.getField(StandardField.PAGES)); + } + + @Test + public void testImportEntriesINSPEC() throws IOException, URISyntaxException { + Path file = Path.of(IsiImporterTest.class.getResource("IsiImporterTestInspec.isi").toURI()); + List entries = importer.importDatabase(file).getDatabase().getEntries(); + + BibEntry first = entries.get(0); + BibEntry second = entries.get(1); + + if (first.getField(StandardField.TITLE).equals( + Optional.of("Optical and photoelectric spectroscopy of photorefractive Sn$_2$P$_2$S$_6$ crystals"))) { + BibEntry tmp = first; + first = second; + second = tmp; + } + + assertEquals(2, entries.size()); + assertEquals( + Optional.of( + "Second harmonic generation of continuous wave ultraviolet light and production of beta -BaB$_2$O$_4$ optical waveguides"), + first.getField(StandardField.TITLE)); + assertEquals(StandardEntryType.Article, first.getType()); + + assertEquals(Optional.of("Degl'Innocenti, R. and Guarino, A. and Poberaj, G. and Gunter, P."), + first.getField(StandardField.AUTHOR)); + assertEquals(Optional.of("Applied Physics Letters"), first.getField(StandardField.JOURNAL)); + assertEquals(Optional.of("2006"), first.getField(StandardField.YEAR)); + assertEquals(Optional.of(Month.JULY), first.getMonth()); + assertEquals(Optional.of("89"), first.getField(StandardField.VOLUME)); + assertEquals(Optional.of("4"), first.getField(StandardField.NUMBER)); + assertEquals(Optional.of("Lorem ipsum abstract"), first.getField(StandardField.ABSTRACT)); + assertEquals(Optional.of("Aip"), first.getField(StandardField.PUBLISHER)); + assertEquals( + Optional.of("Optical and photoelectric spectroscopy of photorefractive Sn$_2$P$_2$S$_6$ crystals"), + second.getField(StandardField.TITLE)); + assertEquals(StandardEntryType.Article, second.getType()); + } + + @Test + public void testImportEntriesWOS() throws IOException, URISyntaxException { + Path file = Path.of(IsiImporterTest.class.getResource("IsiImporterTestWOS.isi").toURI()); + List entries = importer.importDatabase(file).getDatabase().getEntries(); + + BibEntry first = entries.get(0); + BibEntry second = entries.get(1); + + assertEquals(2, entries.size()); + + assertEquals(Optional.of("Optical and photoelectric spectroscopy of photorefractive Sn2P2S6 crystals"), + first.getField(StandardField.TITLE)); + assertEquals(Optional.of("Optical waveguides in Sn2P2S6 by low fluence MeV He+ ion implantation"), + second.getField(StandardField.TITLE)); + + assertEquals(Optional.of("Journal of Physics-Condensed Matter"), first.getField(StandardField.JOURNAL)); + } + + @Test + public void testIsiAuthorsConvert() { + assertEquals( + "James Brown and James Marc Brown and Brown, J. M. and Brown, J. and Brown, J. M. and Brown, J.", + IsiImporter.isiAuthorsConvert( + "James Brown and James Marc Brown and Brown, J.M. and Brown, J. and Brown, J.M. and Brown, J.")); + + assertEquals( + "Joffe, Hadine and Hall, Janet E. and Gruber, Staci and Sarmiento, Ingrid A. and Cohen, Lee S. and Yurgelun-Todd, Deborah and Martin, Kathryn A.", + IsiImporter.isiAuthorsConvert( + "Joffe, Hadine; Hall, Janet E; Gruber, Staci; Sarmiento, Ingrid A; Cohen, Lee S; Yurgelun-Todd, Deborah; Martin, Kathryn A")); + } + + @Test + public void testMonthConvert() { + assertEquals("#jun#", IsiImporter.parseMonth("06")); + assertEquals("#jun#", IsiImporter.parseMonth("JUN")); + assertEquals("#jun#", IsiImporter.parseMonth("jUn")); + assertEquals("#may#", IsiImporter.parseMonth("MAY-JUN")); + assertEquals("#jun#", IsiImporter.parseMonth("2006 06")); + assertEquals("#jun#", IsiImporter.parseMonth("2006 06-07")); + assertEquals("#jul#", IsiImporter.parseMonth("2006 07 03")); + assertEquals("#may#", IsiImporter.parseMonth("2006 May-Jun")); + } + + @Test + public void testIsiAuthorConvert() { + assertEquals("James Brown", IsiImporter.isiAuthorConvert("James Brown")); + assertEquals("James Marc Brown", IsiImporter.isiAuthorConvert("James Marc Brown")); + assertEquals("Brown, J. M.", IsiImporter.isiAuthorConvert("Brown, J.M.")); + assertEquals("Brown, J.", IsiImporter.isiAuthorConvert("Brown, J.")); + assertEquals("Brown, J. M.", IsiImporter.isiAuthorConvert("Brown, JM")); + assertEquals("Brown, J.", IsiImporter.isiAuthorConvert("Brown, J")); + assertEquals("Brown, James", IsiImporter.isiAuthorConvert("Brown, James")); + assertEquals("Hall, Janet E.", IsiImporter.isiAuthorConvert("Hall, Janet E")); + assertEquals("", IsiImporter.isiAuthorConvert("")); + } + + @Test + public void testImportIEEEExport() throws IOException, URISyntaxException { + Path file = Path.of(IsiImporterTest.class.getResource("IEEEImport1.txt").toURI()); + List entries = importer.importDatabase(file).getDatabase().getEntries(); + BibEntry entry = entries.get(0); + + assertEquals(1, entries.size()); + assertEquals(StandardEntryType.Article, entry.getType()); + assertEquals(Optional.of("Geoscience and Remote Sensing Letters, IEEE"), + entry.getField(StandardField.JOURNAL)); + assertEquals(Optional.of("Improving Urban Road Extraction in High-Resolution " + + "Images Exploiting Directional Filtering, Perceptual " + + "Grouping, and Simple Topological Concepts"), + entry.getField(StandardField.TITLE)); + assertEquals(Optional.of("4"), entry.getField(StandardField.VOLUME)); + assertEquals(Optional.of("3"), entry.getField(StandardField.NUMBER)); + assertEquals(Optional.of("1545-598X"), entry.getField(new UnknownField("SN"))); + assertEquals(Optional.of("387--391"), entry.getField(StandardField.PAGES)); + assertEquals(Optional.of("Gamba, P. and Dell'Acqua, F. and Lisini, G."), + entry.getField(StandardField.AUTHOR)); + assertEquals(Optional.of("2006"), entry.getField(StandardField.YEAR)); + assertEquals(Optional.of("Perceptual grouping, street extraction, urban remote sensing"), + entry.getField(StandardField.KEYWORDS)); + assertEquals(Optional.of("Lorem ipsum abstract"), entry.getField(StandardField.ABSTRACT)); + } + + @Test + public void testIEEEImport() throws IOException, URISyntaxException { + Path file = Path.of(IsiImporterTest.class.getResource("IEEEImport1.txt").toURI()); + List entries = importer.importDatabase(file).getDatabase().getEntries(); + BibEntry entry = entries.get(0); + + assertEquals(1, entries.size()); + assertEquals(StandardEntryType.Article, entry.getType()); + assertEquals(Optional.of("Geoscience and Remote Sensing Letters, IEEE"), + entry.getField(StandardField.JOURNAL)); + assertEquals( + Optional.of( + "Improving Urban Road Extraction in High-Resolution Images Exploiting Directional Filtering, Perceptual Grouping, and Simple Topological Concepts"), + entry.getField(StandardField.TITLE)); + assertEquals(Optional.of("4"), entry.getField(StandardField.VOLUME)); + assertEquals(Optional.of("3"), entry.getField(StandardField.NUMBER)); + assertEquals(Optional.of("1545-598X"), entry.getField(new UnknownField("SN"))); + assertEquals(Optional.of("387--391"), entry.getField(StandardField.PAGES)); + assertEquals(Optional.of("Gamba, P. and Dell'Acqua, F. and Lisini, G."), + entry.getField(StandardField.AUTHOR)); + assertEquals(Optional.of("2006"), entry.getField(StandardField.YEAR)); + assertEquals(Optional.of("Perceptual grouping, street extraction, urban remote sensing"), + entry.getField(StandardField.KEYWORDS)); + assertEquals(Optional.of("Lorem ipsum abstract"), entry.getField(StandardField.ABSTRACT)); + } + + @Test + public void testImportEntriesMedline() throws IOException, URISyntaxException { + Path file = Path.of(IsiImporterTest.class.getResource("IsiImporterTestMedline.isi").toURI()); + List entries = importer.importDatabase(file).getDatabase().getEntries(); + + BibEntry first = entries.get(0); + BibEntry second = entries.get(1); + + assertEquals(2, entries.size()); + assertEquals( + Optional.of("Effects of modafinil on cognitive performance and alertness during sleep deprivation."), + first.getField(StandardField.TITLE)); + assertEquals(Optional.of("Wesensten, Nancy J."), first.getField(StandardField.AUTHOR)); + assertEquals(Optional.of("Curr Pharm Des"), first.getField(StandardField.JOURNAL)); + assertEquals(Optional.of("2006"), first.getField(StandardField.YEAR)); + assertEquals(Optional.empty(), first.getField(StandardField.MONTH)); + assertEquals(Optional.of("12"), first.getField(StandardField.VOLUME)); + assertEquals(Optional.of("20"), first.getField(StandardField.NUMBER)); + assertEquals(Optional.of("2457--71"), first.getField(StandardField.PAGES)); + assertEquals(StandardEntryType.Article, first.getType()); + assertEquals( + Optional.of( + "Estrogen therapy selectively enhances prefrontal cognitive processes: a randomized, double-blind, placebo-controlled study with functional magnetic resonance imaging in perimenopausal and recently postmenopausal women."), + second.getField(StandardField.TITLE)); + assertEquals( + Optional.of( + "Joffe, Hadine and Hall, Janet E. and Gruber, Staci and Sarmiento, Ingrid A. and Cohen, Lee S. and Yurgelun-Todd, Deborah and Martin, Kathryn A."), + second.getField(StandardField.AUTHOR)); + assertEquals(Optional.of("2006"), second.getField(StandardField.YEAR)); + assertEquals(Optional.of(Month.MAY), second.getMonth()); + assertEquals(Optional.of("13"), second.getField(StandardField.VOLUME)); + assertEquals(Optional.of("3"), second.getField(StandardField.NUMBER)); + assertEquals(Optional.of("411--22"), second.getField(StandardField.PAGES)); + assertEquals(StandardEntryType.Article, second.getType()); + } + + @Test + public void testImportEntriesEmpty() throws IOException, URISyntaxException { + Path file = Path.of(IsiImporterTest.class.getResource("IsiImporterTestEmpty.isi").toURI()); + + List entries = importer.importDatabase(file).getDatabase().getEntries(); + + assertEquals(1, entries.size()); + } }