*/
package jalview.io.gff;
-import static jalview.io.FeaturesFile.MAP_ATTRIBUTE_PREFIX;
-
import jalview.analysis.SequenceIdMatcher;
import jalview.datamodel.AlignedCodonFrame;
import jalview.datamodel.AlignmentI;
import jalview.datamodel.SequenceDummy;
import jalview.datamodel.SequenceFeature;
import jalview.datamodel.SequenceI;
-import jalview.io.FeaturesFile;
import jalview.util.MapList;
import jalview.util.StringUtils;
*/
public abstract class GffHelperBase implements GffHelperI
{
- private static final String COMMA = ",";
+ private static final String INVALID_GFF_ATTRIBUTE_FORMAT = "Invalid GFF attribute format: ";
+
+ protected static final String COMMA = ",";
- private static final String NOTE = "Note";
+ protected static final String EQUALS = "=";
+
+ protected static final String NOTE = "Note";
/*
* GFF columns 1-9 (zero-indexed):
}
/**
- * Parses the input line to a map of name / value(s) pairs. For example the line
- * <br>
+ * Parses the input line to a map of name / value(s) pairs. For example the
+ * line
+ *
+ * <pre>
* Notes=Fe-S;Method=manual curation, prediction; source = Pfam; Notes = Metal
- * <br>
+ * </pre>
+ *
* if parsed with delimiter=";" and separators {' ', '='} <br>
* would return a map with { Notes={Fe=S, Metal}, Method={manual curation,
* prediction}, source={Pfam}} <br>
*
* This method supports parsing of either GFF2 format (which uses space ' ' as
- * the name/value delimiter, and allows multiple occurrences of the same name),
- * or GFF3 format (which uses '=' as the name/value delimiter, and strictly does
- * not allow repeat occurrences of the same name - but does allow a
- * comma-separated list of values).
+ * the name/value delimiter, and allows multiple occurrences of the same
+ * name), or GFF3 format (which uses '=' as the name/value delimiter, and
+ * strictly does not allow repeat occurrences of the same name - but does
+ * allow a comma-separated list of values).
* <p>
* Returns a (possibly empty) map of lists of values by attribute name.
*
* @param text
* @param namesDelimiter
- * the major delimiter between name-value pairs
+ * the major delimiter between name-value pairs
* @param nameValueSeparator
- * separator used between name and value
+ * separator used between name and value
* @param valuesDelimiter
- * delimits a list of more than one value
+ * delimits a list of more than one value
* @return
*/
public static Map<String, List<String>> parseNameValuePairs(String text,
return map;
}
- for (String pair : text.trim().split(namesDelimiter))
+ /*
+ * split by major delimiter (; for GFF3)
+ */
+ for (String nameValuePair : text.trim().split(namesDelimiter))
{
- pair = pair.trim();
- if (pair.length() == 0)
+ nameValuePair = nameValuePair.trim();
+ if (nameValuePair.length() == 0)
{
continue;
}
- int sepPos = pair.indexOf(nameValueSeparator);
+ /*
+ * find name/value separator (= for GFF3)
+ */
+ int sepPos = nameValuePair.indexOf(nameValueSeparator);
if (sepPos == -1)
{
// no name=value found
continue;
}
- String key = pair.substring(0, sepPos).trim();
- String values = pair.substring(sepPos + 1).trim();
- if (values.length() > 0)
+ String name = nameValuePair.substring(0, sepPos).trim();
+ String values = nameValuePair.substring(sepPos + 1).trim();
+ if (values.isEmpty())
{
- List<String> vals = map.get(key);
- if (vals == null)
- {
- vals = new ArrayList<>();
- map.put(key, vals);
- }
+ continue;
+ }
- /*
- * special case: formatted as jvmap_AttName={a=b,c=d,...}
- * save the value within { } for parsing at a later stage
- */
- if (key.startsWith(MAP_ATTRIBUTE_PREFIX))
- {
+ List<String> vals = map.get(name);
+ if (vals == null)
+ {
+ vals = new ArrayList<>();
+ map.put(name, vals);
+ }
- if (key.length() > MAP_ATTRIBUTE_PREFIX.length()
- && values.startsWith("{")
- && values.endsWith("}"))
- {
- vals.add(values.substring(1, values.length() - 1));
- }
- else
- {
- System.err.println("Malformed GFF data '" + values.toString()
- + "' for " + key);
- }
- }
- else
+ /*
+ * if 'values' contains more name/value separators, parse as a map
+ * (nested sub-attribute values)
+ */
+ if (values.indexOf(nameValueSeparator) != -1)
+ {
+ vals.add(values);
+ }
+ else
+ {
+ for (String val : values.split(valuesDelimiter))
{
- for (String val : values.split(valuesDelimiter))
- {
- vals.add(val);
- }
+ vals.add(val);
}
}
}
+
return map;
}
{
String key = attr.getKey();
List<String> values = attr.getValue();
- if (key.startsWith(FeaturesFile.MAP_ATTRIBUTE_PREFIX))
+ if (values.size() == 1 && values.get(0).contains(EQUALS))
{
- key = key.substring(FeaturesFile.MAP_ATTRIBUTE_PREFIX.length());
- Map<String, String> valueMap = parseAttributeMap(values);
+ /*
+ * 'value' is actually nested subattributes as x=a,y=b,z=c
+ */
+ Map<String, String> valueMap = parseAttributeMap(values.get(0));
sf.setValue(key, valueMap);
}
else
}
/**
- * Parses one or more list of comma-separated key=value pairs into a Map of
- * {key, value}
+ * Parses a (GFF3 format) list of comma-separated key=value pairs into a Map
+ * of {@code key,
+ * value} <br>
+ * An input string like {@code a=b,c,d=e,f=g,h} is parsed to
+ *
+ * <pre>
+ * a = "b,c"
+ * d = "e"
+ * f = "g,h"
+ * </pre>
+ *
+ * @param s
*
- * @param values
* @return
*/
- protected Map<String, String> parseAttributeMap(List<String> values)
+ protected static Map<String, String> parseAttributeMap(String s)
{
Map<String, String> map = new HashMap<>();
- for (String entry : values)
+ String[] fields = s.split(EQUALS);
+
+ /*
+ * format validation
+ */
+ boolean valid = true;
+ if (fields.length < 2)
+ {
+ /*
+ * need at least A=B here
+ */
+ valid = false;
+ }
+ else if (fields[0].isEmpty() || fields[0].contains(COMMA))
+ {
+ /*
+ * A,B=C is not a valid start, nor is =C
+ */
+ valid = false;
+ }
+ else
{
- String[] fields = entry.split(COMMA);
- for (String field : fields)
+ for (int i = 1; i < fields.length - 1; i++)
{
- String[] keyValue = field.split("=");
- if (keyValue.length == 2)
+ if (fields[i].isEmpty() || !fields[i].contains(COMMA))
{
- String theKey = StringUtils.urlDecode(keyValue[0],
- GFF_ENCODABLE);
- String theValue = StringUtils.urlDecode(keyValue[1],
- GFF_ENCODABLE);
- map.put(theKey, theValue);
+ /*
+ * intermediate tokens must include value,name
+ */
+ valid = false;
}
}
}
+
+ if (!valid)
+ {
+ System.err.println(INVALID_GFF_ATTRIBUTE_FORMAT + s);
+ return map;
+ }
+
+ int i = 0;
+ while (i < fields.length - 1)
+ {
+ boolean lastPair = i == fields.length - 2;
+ String before = fields[i];
+ String after = fields[i + 1];
+
+ /*
+ * if 'key' looks like a,b,c then the last token is the
+ * key
+ */
+ String theKey = before.contains(COMMA)
+ ? before.substring(before.lastIndexOf(COMMA) + 1)
+ : before;
+
+ theKey = theKey.trim();
+ if (theKey.isEmpty())
+ {
+ System.err.println(INVALID_GFF_ATTRIBUTE_FORMAT + s);
+ map.clear();
+ return map;
+ }
+
+ /*
+ * if 'value' looks like a,b,c then all but the last token is the value,
+ * unless this is the last field (no more = to follow), in which case
+ * all of it makes up the value
+ */
+ String theValue = after.contains(COMMA) && !lastPair
+ ? after.substring(0, after.lastIndexOf(COMMA))
+ : after;
+ map.put(StringUtils.urlDecode(theKey, GFF_ENCODABLE),
+ StringUtils.urlDecode(theValue, GFF_ENCODABLE));
+ i += 1;
+ }
+
return map;
}