| 1 | // ================================================================ // |
|---|
| 2 | // // |
|---|
| 3 | // File : Importer.h // |
|---|
| 4 | // Purpose : // |
|---|
| 5 | // // |
|---|
| 6 | // Coded by Ralf Westram (coder@reallysoft.de) in November 2006 // |
|---|
| 7 | // Institute of Microbiology (Technical University Munich) // |
|---|
| 8 | // http://www.arb-home.de/ // |
|---|
| 9 | // // |
|---|
| 10 | // ================================================================ // |
|---|
| 11 | #ifndef IMPORTER_H |
|---|
| 12 | #define IMPORTER_H |
|---|
| 13 | |
|---|
| 14 | #ifndef BUFFEREDFILEREADER_H |
|---|
| 15 | #include <BufferedFileReader.h> |
|---|
| 16 | #endif |
|---|
| 17 | #ifndef SMARTPTR_H |
|---|
| 18 | #include <smartptr.h> |
|---|
| 19 | #endif |
|---|
| 20 | #ifndef METATAG_H |
|---|
| 21 | #include "MetaTag.h" |
|---|
| 22 | #endif |
|---|
| 23 | |
|---|
| 24 | class DBwriter; |
|---|
| 25 | class Feature; |
|---|
| 26 | |
|---|
| 27 | enum FeatureLineType { |
|---|
| 28 | FL_START = 1, // start of feature (e.g. 'CDS 352120..353193'). starts at offset 5 |
|---|
| 29 | // all types below start at offset 21 (or higher): |
|---|
| 30 | FL_QUALIFIER = 2, // start of qualifier (e.g. '/codon_start=1') |
|---|
| 31 | FL_QUALIFIER_NODATA = 4, // start of qualifier w/o data (e.g. '/pseudo') |
|---|
| 32 | FL_QUALIFIER_QUOTED = 8, // start of qualifier with quoted data (e.g. '/product="phosphate"') |
|---|
| 33 | FL_QUALIFIER_QUOTE_OPENED = 16, // start of qualifier with quoted data (e.g. '/product="phosphate') |
|---|
| 34 | FL_CONTINUED_QUOTE_CLOSED = 32, // something terminated by a quote ('"') |
|---|
| 35 | FL_CONTINUED = 64, // other |
|---|
| 36 | |
|---|
| 37 | // meta types: |
|---|
| 38 | FL_META_QUALIFIER = (FL_QUALIFIER|FL_QUALIFIER_NODATA|FL_QUALIFIER_QUOTED|FL_QUALIFIER_QUOTE_OPENED), |
|---|
| 39 | FL_META_CONTINUED = (FL_CONTINUED_QUOTE_CLOSED|FL_CONTINUED), |
|---|
| 40 | }; |
|---|
| 41 | |
|---|
| 42 | class FeatureLine { |
|---|
| 43 | void interpret_as_continued_line(); |
|---|
| 44 | public: |
|---|
| 45 | string name; // feature or qualifier name (only valid for FL_START, FL_QUALIFIER...) |
|---|
| 46 | string rest; // rest of line (behind '=' for FL_QUALIFIER..., not for FL_QUALIFIER_NODATA) |
|---|
| 47 | string orgLine; |
|---|
| 48 | FeatureLineType type; |
|---|
| 49 | |
|---|
| 50 | FeatureLine(const string& line); |
|---|
| 51 | bool reinterpret_as_continued_line(); |
|---|
| 52 | }; |
|---|
| 53 | |
|---|
| 54 | typedef SmartPtr<Feature> FeaturePtr; |
|---|
| 55 | typedef SmartPtr<FeatureLine> FeatureLinePtr; |
|---|
| 56 | typedef std::vector<FeatureLinePtr> FeatureLines; |
|---|
| 57 | |
|---|
| 58 | class Importer : virtual Noncopyable { |
|---|
| 59 | protected: |
|---|
| 60 | DBwriter& db_writer; |
|---|
| 61 | LineReader& flatfile; |
|---|
| 62 | MetaTagTranslator tagTranslator; |
|---|
| 63 | FeatureLines pushedFeatureLines; // pushed back feature lines |
|---|
| 64 | stringVector warnings; |
|---|
| 65 | long expectedSeqLength; // length read from LOCUS or ID line ( = 0 -> no length info found) |
|---|
| 66 | |
|---|
| 67 | void expectLine(string& line) { if (!flatfile.getLine(line)) throw flatfile.lineError("Unexpected EOF"); } |
|---|
| 68 | const MetaTag *findTag(const string& tag) { return tagTranslator.get(tag); } |
|---|
| 69 | |
|---|
| 70 | virtual bool readFeatureTableLine(string& line) = 0; |
|---|
| 71 | |
|---|
| 72 | FeatureLinePtr getFeatureTableLine(); |
|---|
| 73 | void backFeatureTableLine(FeatureLinePtr& fline) { pushedFeatureLines.push_back(fline); } |
|---|
| 74 | |
|---|
| 75 | FeatureLinePtr getUnwrappedFeatureTableLine(); |
|---|
| 76 | |
|---|
| 77 | FeaturePtr parseFeature(); |
|---|
| 78 | void parseFeatureTable(); |
|---|
| 79 | |
|---|
| 80 | virtual void import_section() = 0; |
|---|
| 81 | |
|---|
| 82 | void show_warnings(const string& import_of_what); |
|---|
| 83 | |
|---|
| 84 | public: |
|---|
| 85 | Importer(LineReader& Flatfile, DBwriter& DB_writer, const MetaTag *meta_description); |
|---|
| 86 | virtual ~Importer() {} |
|---|
| 87 | |
|---|
| 88 | void import(); |
|---|
| 89 | void warning(const char *msg); // add a warning |
|---|
| 90 | }; |
|---|
| 91 | |
|---|
| 92 | |
|---|
| 93 | class GenebankImporter : public Importer { |
|---|
| 94 | void import_section() OVERRIDE; |
|---|
| 95 | bool readFeatureTableLine(string& line) OVERRIDE; |
|---|
| 96 | void parseSequence(const string& tag, const string& headerline); |
|---|
| 97 | |
|---|
| 98 | public: |
|---|
| 99 | GenebankImporter(LineReader& Flatfile, DBwriter& DB_writer); |
|---|
| 100 | ~GenebankImporter() OVERRIDE {} |
|---|
| 101 | |
|---|
| 102 | }; |
|---|
| 103 | |
|---|
| 104 | |
|---|
| 105 | class EmblImporter : public Importer { |
|---|
| 106 | void import_section() OVERRIDE; |
|---|
| 107 | bool readFeatureTableLine(string& line) OVERRIDE; |
|---|
| 108 | void parseSequence(const string& headerline); |
|---|
| 109 | |
|---|
| 110 | public: |
|---|
| 111 | EmblImporter(LineReader& Flatfile, DBwriter& DB_writer); |
|---|
| 112 | ~EmblImporter() OVERRIDE {} |
|---|
| 113 | }; |
|---|
| 114 | |
|---|
| 115 | |
|---|
| 116 | #else |
|---|
| 117 | #error Importer.h included twice |
|---|
| 118 | #endif // IMPORTER_H |
|---|
| 119 | |
|---|
