X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fio%2FIdentifyFile.java;fp=src%2Fjalview%2Fio%2FIdentifyFile.java;h=0556e76a20248be0a0ad19fc9e92eaeb4a255689;hb=2595e9d4ee0dbbd3406a98c4e49a61ccde806479;hp=deae9ae7ece309e8f40515ff38306621b1acb42e;hpb=e20075ba805d744d7cc4976e2b8d5e5840fb0a8d;p=jalview.git diff --git a/src/jalview/io/IdentifyFile.java b/src/jalview/io/IdentifyFile.java index deae9ae..0556e76 100755 --- a/src/jalview/io/IdentifyFile.java +++ b/src/jalview/io/IdentifyFile.java @@ -30,8 +30,6 @@ import java.io.IOException; */ public class IdentifyFile { - public static final String FeaturesFile = "GFF or Jalview features"; - /** * Identify a datasource's file content. * @@ -39,18 +37,18 @@ public class IdentifyFile * instead. * * @param file - * DOCUMENT ME! - * @param protocol - * DOCUMENT ME! - * @return ID String + * @param sourceType + * @return + * @throws FileFormatException */ - public String identify(String file, String protocol) + public FileFormatI identify(String file, DataSourceType sourceType) + throws FileFormatException { String emessage = "UNIDENTIFIED FILE PARSING ERROR"; FileParse parser = null; try { - parser = new FileParse(file, protocol); + parser = new FileParse(file, sourceType); if (parser.isValid()) { return identify(parser); @@ -63,15 +61,22 @@ public class IdentifyFile } if (parser != null) { - return parser.errormessage; + throw new FileFormatException(parser.errormessage); } - return emessage; + throw new FileFormatException(emessage); + } + + public FileFormatI identify(FileParse source) throws FileFormatException + { + return identify(source, true); + // preserves original behaviour prior to version 2.3 } - public String identify(FileParse source) + public FileFormatI identify(AlignmentFileReaderI file, boolean closeSource) + throws IOException { - return identify(source, true); // preserves original behaviour prior to - // version 2.3 + FileParse fp = new FileParse(file.getInFile(), file.getDataSourceType()); + return identify(fp, closeSource); } /** @@ -80,11 +85,13 @@ public class IdentifyFile * * @param source * @param closeSource - * @return filetype string + * @return (best guess at) file format + * @throws FileFormatException */ - public String identify(FileParse source, boolean closeSource) + public FileFormatI identify(FileParse source, boolean closeSource) + throws FileFormatException { - String reply = "PFAM"; + FileFormatI reply = FileFormat.Pfam; String data; int bytesRead = 0; int trimmedLength = 0; @@ -123,12 +130,12 @@ public class IdentifyFile if (fileStr.lastIndexOf(".jar") > -1 || fileStr.lastIndexOf(".zip") > -1) { - reply = "Jalview"; + reply = FileFormat.Jalview; } } if (!lineswereskipped && data.startsWith("PK")) { - reply = "Jalview"; // archive. + reply = FileFormat.Jalview; // archive. break; } } @@ -137,24 +144,24 @@ public class IdentifyFile if (data.startsWith("##GFF-VERSION")) { // GFF - possibly embedded in a Jalview features file! - reply = FeaturesFile; + reply = FileFormat.Features; break; } if (looksLikeFeatureData(data)) { - reply = FeaturesFile; + reply = FileFormat.Features; break; } if (data.indexOf("# STOCKHOLM") > -1) { - reply = "STH"; + reply = FileFormat.Stockholm; break; } if (data.indexOf("_ENTRY.ID") > -1 || data.indexOf("_AUDIT_AUTHOR.NAME") > -1 || data.indexOf("_ATOM_SITE.") > -1) { - reply = "mmCIF"; + reply = FileFormat.MMCif; break; } // if (data.indexOf(">") > -1) @@ -166,14 +173,14 @@ public class IdentifyFile { // watch for PIR file attributes checkPIR = true; - reply = "PIR"; + reply = FileFormat.PIR; } // could also be BLC file, read next line to confirm data = source.nextLine(); if (data.indexOf(">") > -1) { - reply = "BLC"; + reply = FileFormat.BLC; } else { @@ -190,18 +197,19 @@ public class IdentifyFile { if (c1 == 0 && c1 == data2.indexOf("*")) { - reply = "BLC"; + reply = FileFormat.BLC; } else { - reply = "FASTA"; // possibly a bad choice - may be recognised as + reply = FileFormat.Fasta; // possibly a bad choice - may be + // recognised as // PIR } // otherwise can still possibly be a PIR file } else { - reply = "FASTA"; + reply = FileFormat.Fasta; // TODO : AMSA File is indicated if there is annotation in the // FASTA file - but FASTA will automatically generate this at the // mo. @@ -238,12 +246,12 @@ public class IdentifyFile } if (starterm) { - reply = "PIR"; + reply = FileFormat.PIR; break; } else { - reply = "FASTA"; // probably a bad choice! + reply = FileFormat.Fasta; // probably a bad choice! } } // read as a FASTA (probably) @@ -256,19 +264,19 @@ public class IdentifyFile String upper = data.toUpperCase(); if (upper.substring(lessThan).startsWith(" -1) { - reply = JSONFile.FILE_DESC; + reply = FileFormat.Json; break; } if ((data.length() < 1) || (data.indexOf("#") == 0)) @@ -279,7 +287,7 @@ public class IdentifyFile if (data.indexOf("PILEUP") > -1) { - reply = "PileUp"; + reply = FileFormat.Pileup; break; } @@ -288,32 +296,32 @@ public class IdentifyFile || ((data.indexOf("!!") > -1) && (data.indexOf("!!") < data .indexOf("_MULTIPLE_ALIGNMENT ")))) { - reply = "MSF"; + reply = FileFormat.MSF; break; } else if (data.indexOf("CLUSTAL") > -1) { - reply = "CLUSTAL"; + reply = FileFormat.Clustal; break; } else if (data.indexOf("HEADER") == 0 || data.indexOf("ATOM") == 0) { - reply = "PDB"; + reply = FileFormat.PDB; break; } else if (data.matches("\\s*\\d+\\s+\\d+\\s*")) { - reply = PhylipFile.FILE_DESC; + reply = FileFormat.Phylip; break; } else { if (!lineswereskipped && looksLikeJnetData(data)) { - reply = "JnetFile"; + reply = FileFormat.Jnet; break; } } @@ -332,14 +340,15 @@ public class IdentifyFile } catch (Exception ex) { System.err.println("File Identification failed!\n" + ex); - return source.errormessage; + throw new FileFormatException(source.errormessage); } if (trimmedLength == 0) { System.err .println("File Identification failed! - Empty file was read."); - return "EMPTY DATA FILE"; + throw new FileFormatException("EMPTY DATA FILE"); } + System.out.println("File format identified as " + reply.toString()); return reply; } @@ -393,11 +402,19 @@ public class IdentifyFile public static void main(String[] args) { - for (int i = 0; args != null && i < args.length; i++) { IdentifyFile ider = new IdentifyFile(); - String type = ider.identify(args[i], AppletFormatAdapter.FILE); + FileFormatI type = null; + try + { + type = ider.identify(args[i], DataSourceType.FILE); + } catch (FileFormatException e) + { + System.err.println(String.format( + "Error '%s' identifying file type for %s", args[i], + e.getMessage())); + } System.out.println("Type of " + args[i] + " is " + type); } if (args == null || args.length == 0)