Index: src/org/openstreetmap/josm/data/validation/tests/TagChecker.java =================================================================== --- src/org/openstreetmap/josm/data/validation/tests/TagChecker.java (revision 14586) +++ src/org/openstreetmap/josm/data/validation/tests/TagChecker.java (working copy) @@ -74,6 +74,9 @@ private static final Map harmonizedKeys = new HashMap<>(); /** The spell check preset values which are not stored in TaggingPresets */ private static volatile MultiMap additionalPresetsValueData; + /** The spell check preset values which are not stored in TaggingPresets */ + private static volatile MultiMap oftenUsedValueData = new MultiMap<>(); + /** The TagChecker data */ private static final List checkerData = new ArrayList<>(); private static final List ignoreDataStartsWith = new ArrayList<>(); @@ -240,7 +243,9 @@ ignoreDataEndsWith.add(line); break; case "K:": - ignoreDataTag.add(Tag.ofString(line)); + Tag tag = Tag.ofString(line); + ignoreDataTag.add(tag); + oftenUsedValueData.put(tag.getKey(), tag.getValue()); break; default: if (!key.startsWith(";")) { @@ -531,86 +536,100 @@ withErrors.put(p, "UPK"); } } else if (!isTagInPresets(key, value)) { - // try to fix common typos and check again if value is still unknown - final String harmonizedValue = harmonizeValue(prop.getValue()); - String fixedValue = null; - Set possibleValues = getPresetValues(key); - List fixVals = new ArrayList<>(); - int maxPresetValueLen = 0; - if (possibleValues.contains(harmonizedValue)) { - fixedValue = harmonizedValue; - } else { - // use Levenshtein distance to find typical typos - int minDist = MAX_LEVENSHTEIN_DISTANCE + 1; - String closest = null; - for (String possibleVal : possibleValues) { - if (possibleVal.isEmpty()) - continue; - maxPresetValueLen = Math.max(maxPresetValueLen, possibleVal.length()); - if (harmonizedValue.length() < 3 && possibleVal.length() >= harmonizedValue.length() + MAX_LEVENSHTEIN_DISTANCE) { - // don't suggest fix value when given value is short and lengths are too different - // for example surface=u would result in surface=mud - continue; - } - int dist = Utils.getLevenshteinDistance(possibleVal, harmonizedValue); - if (dist >= harmonizedValue.length()) { - // short value, all characters are different. Don't warn, might say Value '10' for key 'fee' looks like 'no'. - continue; - } - if (dist < minDist) { - closest = possibleVal; - minDist = dist; - fixVals.clear(); - fixVals.add(possibleVal); - } else if (dist == minDist) { - fixVals.add(possibleVal); - } - } - if (minDist <= MAX_LEVENSHTEIN_DISTANCE && maxPresetValueLen > MAX_LEVENSHTEIN_DISTANCE - && (harmonizedValue.length() > 3 || minDist < MAX_LEVENSHTEIN_DISTANCE)) { - if (fixVals.size() < 2) { - fixedValue = closest; - } else { - Collections.sort(fixVals); - // misspelled preset value with multiple good alternatives - errors.add(TestError.builder(this, Severity.WARNING, MISSPELLED_VALUE_NO_FIX) - .message(tr("Misspelled property value"), - marktr("Value ''{0}'' for key ''{1}'' looks like one of {2}."), - prop.getValue(), key, fixVals) - .primitives(p).build()); - withErrors.put(p, "WPV"); - continue; - } - } + tryGuess(p, key, value, withErrors); + } + } + if (checkFixmes && key != null && value != null && !value.isEmpty() && isFixme(key, value) && !withErrors.contains(p, "FIXME")) { + errors.add(TestError.builder(this, Severity.OTHER, FIXME) + .message(tr("FIXMES")) + .primitives(p) + .build()); + withErrors.put(p, "FIXME"); + } + } + } + + private void tryGuess(OsmPrimitive p, String key, String value, MultiMap withErrors) { + // try to fix common typos and check again if value is still unknown + final String harmonizedValue = harmonizeValue(value); + String fixedValue = null; + List fixVals = new ArrayList<>(); + int maxPresetValueLen = 0; + Set presetValues = getPresetValues(key); + Set oftenUsedValues = oftenUsedValueData.get(key); + for (Set possibleValues: Arrays.asList(presetValues, oftenUsedValues)) { + if (possibleValues != null && possibleValues.contains(harmonizedValue)) { + fixedValue = harmonizedValue; + break; + } + } + if (fixedValue == null) { + // use Levenshtein distance to find typical typos + int minDist = MAX_LEVENSHTEIN_DISTANCE + 1; + String closest = null; + for (Set possibleValues: Arrays.asList(presetValues, oftenUsedValues)) { + if (possibleValues == null) + continue; + for (String possibleVal : possibleValues) { + if (possibleVal.isEmpty()) + continue; + maxPresetValueLen = Math.max(maxPresetValueLen, possibleVal.length()); + if (harmonizedValue.length() < 3 && possibleVal.length() >= harmonizedValue.length() + MAX_LEVENSHTEIN_DISTANCE) { + // don't suggest fix value when given value is short and lengths are too different + // for example surface=u would result in surface=mud + continue; } - if (fixedValue != null && possibleValues.contains(fixedValue)) { - final String newValue = fixedValue; - // misspelled preset value - errors.add(TestError.builder(this, Severity.WARNING, MISSPELLED_VALUE) - .message(tr("Misspelled property value"), - marktr("Value ''{0}'' for key ''{1}'' looks like ''{2}''."), prop.getValue(), key, newValue) - .primitives(p) - .build()); - withErrors.put(p, "WPV"); - } else { - // unknown preset value - errors.add(TestError.builder(this, Severity.OTHER, INVALID_VALUE) - .message(tr("Presets do not contain property value"), - marktr("Value ''{0}'' for key ''{1}'' not in presets."), prop.getValue(), key) - .primitives(p) - .build()); - withErrors.put(p, "UPV"); + int dist = Utils.getLevenshteinDistance(possibleVal, harmonizedValue); + if (dist >= harmonizedValue.length()) { + // short value, all characters are different. Don't warn, might say Value '10' for key 'fee' looks like 'no'. + continue; } + if (dist < minDist) { + closest = possibleVal; + minDist = dist; + fixVals.clear(); + fixVals.add(possibleVal); + } else if (dist == minDist) { + fixVals.add(possibleVal); + } } } - if (checkFixmes && key != null && value != null && !value.isEmpty() && isFixme(key, value) && !withErrors.contains(p, "FIXME")) { - errors.add(TestError.builder(this, Severity.OTHER, FIXME) - .message(tr("FIXMES")) - .primitives(p) - .build()); - withErrors.put(p, "FIXME"); + + if (minDist <= MAX_LEVENSHTEIN_DISTANCE && maxPresetValueLen > MAX_LEVENSHTEIN_DISTANCE + && (harmonizedValue.length() > 3 || minDist < MAX_LEVENSHTEIN_DISTANCE)) { + if (fixVals.size() < 2) { + fixedValue = closest; + } else { + Collections.sort(fixVals); + // misspelled preset value with multiple good alternatives + errors.add(TestError.builder(this, Severity.WARNING, MISSPELLED_VALUE_NO_FIX) + .message(tr("Unknown property value"), + marktr("Value ''{0}'' for key ''{1}'' is not in the presets, maybe one of {2} is meant?"), + value, key, fixVals) + .primitives(p).build()); + withErrors.put(p, "WPV"); + return; + } } } + if (fixedValue != null) { + final String newValue = fixedValue; + // misspelled preset value + errors.add(TestError.builder(this, Severity.WARNING, MISSPELLED_VALUE) + .message(tr("Unknown property value"), + marktr("Value ''{0}'' for key ''{1}'' is not in the presets, maybe ''{2}'' is meant?"), value, key, newValue) + .primitives(p) + .build()); + withErrors.put(p, "WPV"); + } else { + // unknown preset value + errors.add(TestError.builder(this, Severity.OTHER, INVALID_VALUE) + .message(tr("Presets do not contain property value"), + marktr("Value ''{0}'' for key ''{1}'' not in presets."), value, key) + .primitives(p) + .build()); + withErrors.put(p, "UPV"); + } } private static boolean isFixme(String key, String value) { Index: test/unit/org/openstreetmap/josm/data/validation/tests/TagCheckerTest.java =================================================================== --- test/unit/org/openstreetmap/josm/data/validation/tests/TagCheckerTest.java (revision 14586) +++ test/unit/org/openstreetmap/josm/data/validation/tests/TagCheckerTest.java (working copy) @@ -104,8 +104,8 @@ public void testMisspelledTag() throws IOException { final List errors = test(OsmUtils.createPrimitive("node landuse=forrest")); assertEquals(1, errors.size()); - assertEquals("Misspelled property value", errors.get(0).getMessage()); - assertEquals("Value 'forrest' for key 'landuse' looks like 'forest'.", errors.get(0).getDescription()); + assertEquals("Unknown property value", errors.get(0).getMessage()); + assertEquals("Value 'forrest' for key 'landuse' is not in the presets, maybe 'forest' is meant?", errors.get(0).getDescription()); assertEquals(Severity.WARNING, errors.get(0).getSeverity()); assertFalse(errors.get(0).isFixable()); } @@ -118,8 +118,10 @@ public void testMisspelledTag2() throws IOException { final List errors = test(OsmUtils.createPrimitive("node highway=servics")); assertEquals(1, errors.size()); - assertEquals("Misspelled property value", errors.get(0).getMessage()); - assertEquals("Value 'servics' for key 'highway' looks like one of [service, services].", errors.get(0).getDescription()); + assertEquals("Unknown property value", errors.get(0).getMessage()); + assertEquals( + "Value 'servics' for key 'highway' is not in the presets, maybe one of [service, services] is meant?", + errors.get(0).getDescription()); assertEquals(Severity.WARNING, errors.get(0).getSeverity()); assertFalse(errors.get(0).isFixable()); } @@ -132,8 +134,9 @@ public void testMisspelledTag3() throws IOException { final List errors = test(OsmUtils.createPrimitive("node highway=residentail")); assertEquals(1, errors.size()); - assertEquals("Misspelled property value", errors.get(0).getMessage()); - assertEquals("Value 'residentail' for key 'highway' looks like 'residential'.", errors.get(0).getDescription()); + assertEquals("Unknown property value", errors.get(0).getMessage()); + assertEquals("Value 'residentail' for key 'highway' is not in the presets, maybe 'residential' is meant?", + errors.get(0).getDescription()); assertEquals(Severity.WARNING, errors.get(0).getSeverity()); assertFalse(errors.get(0).isFixable()); } @@ -204,8 +207,9 @@ public void testValueDifferentCase() throws IOException { final List errors = test(OsmUtils.createPrimitive("node highway=Residential")); assertEquals(1, errors.size()); - assertEquals("Misspelled property value", errors.get(0).getMessage()); - assertEquals("Value 'Residential' for key 'highway' looks like 'residential'.", errors.get(0).getDescription()); + assertEquals("Unknown property value", errors.get(0).getMessage()); + assertEquals("Value 'Residential' for key 'highway' is not in the presets, maybe 'residential' is meant?", + errors.get(0).getDescription()); assertEquals(Severity.WARNING, errors.get(0).getSeverity()); assertFalse(errors.get(0).isFixable()); }