| 1 | // ================================================================ |
|---|
| 2 | // |
|---|
| 3 | // File : convert.cxx |
|---|
| 4 | // Purpose : some helpers for global data handling |
|---|
| 5 | // |
|---|
| 6 | // ================================================================ |
|---|
| 7 | |
|---|
| 8 | #include "embl.h" |
|---|
| 9 | #include "genbank.h" |
|---|
| 10 | #include "macke.h" |
|---|
| 11 | |
|---|
| 12 | FormatReaderPtr FormatReader::create(const FormattedFile& in) { |
|---|
| 13 | switch (in.type()) { |
|---|
| 14 | case GENBANK: return new GenbankReader(in.name()); |
|---|
| 15 | case SWISSPROT: |
|---|
| 16 | case EMBL: return new EmblSwissprotReader(in.name()); |
|---|
| 17 | case MACKE: return new MackeReader(in.name()); |
|---|
| 18 | default: throw_unsupported_input_format(in.type()); |
|---|
| 19 | } |
|---|
| 20 | return NULL; |
|---|
| 21 | } |
|---|
| 22 | |
|---|
| 23 | class ConvertibleData { |
|---|
| 24 | const InputFormat& in; |
|---|
| 25 | |
|---|
| 26 | mutable OutputFormatPtr embl; |
|---|
| 27 | mutable OutputFormatPtr gbk; |
|---|
| 28 | mutable OutputFormatPtr macke; |
|---|
| 29 | |
|---|
| 30 | public: |
|---|
| 31 | ConvertibleData(const InputFormat& in_) : in(in_) {} |
|---|
| 32 | |
|---|
| 33 | Format format() const { return in.format(); } |
|---|
| 34 | |
|---|
| 35 | const Embl& to_embl(const Seq& seq) const; |
|---|
| 36 | const GenBank& to_genbank(const Seq& seq) const; |
|---|
| 37 | const Macke& to_macke(const Seq& seq) const; |
|---|
| 38 | }; |
|---|
| 39 | |
|---|
| 40 | #if defined(WARN_TODO) |
|---|
| 41 | #warning refactor below |
|---|
| 42 | #endif |
|---|
| 43 | |
|---|
| 44 | const Embl& ConvertibleData::to_embl(const Seq& seq) const { |
|---|
| 45 | Format inType = in.format(); |
|---|
| 46 | |
|---|
| 47 | if (inType == EMBL) return dynamic_cast<const Embl&>(in); |
|---|
| 48 | |
|---|
| 49 | embl = new Embl; |
|---|
| 50 | |
|---|
| 51 | bool ok = true; |
|---|
| 52 | switch (inType) { |
|---|
| 53 | case GENBANK: ok = gtoe(dynamic_cast<const GenBank&>(in), dynamic_cast<Embl&>(*embl), seq); break; |
|---|
| 54 | case MACKE: ok = mtoe(dynamic_cast<const Macke&>(in), dynamic_cast<Embl&>(*embl), seq); break; |
|---|
| 55 | default: throw_conversion_not_implemented(inType, EMBL); |
|---|
| 56 | } |
|---|
| 57 | if (!ok) throw_conversion_failure(inType, EMBL); |
|---|
| 58 | |
|---|
| 59 | ca_assert(!embl.isNull()); |
|---|
| 60 | return dynamic_cast<const Embl&>(*embl); |
|---|
| 61 | } |
|---|
| 62 | |
|---|
| 63 | const GenBank& ConvertibleData::to_genbank(const Seq& seq) const { |
|---|
| 64 | Format inType = in.format(); |
|---|
| 65 | |
|---|
| 66 | if (inType == GENBANK) return dynamic_cast<const GenBank&>(in); |
|---|
| 67 | |
|---|
| 68 | gbk = new GenBank; |
|---|
| 69 | |
|---|
| 70 | bool ok = true; |
|---|
| 71 | switch (inType) { |
|---|
| 72 | case EMBL: ok = etog(dynamic_cast<const Embl&>(in), dynamic_cast<GenBank&>(*gbk), seq); break; |
|---|
| 73 | case MACKE: ok = mtog(dynamic_cast<const Macke&>(in), dynamic_cast<GenBank&>(*gbk), seq); break; |
|---|
| 74 | default: throw_conversion_not_implemented(inType, GENBANK); |
|---|
| 75 | } |
|---|
| 76 | if (!ok) throw_conversion_failure(inType, GENBANK); |
|---|
| 77 | |
|---|
| 78 | ca_assert(!gbk.isNull()); |
|---|
| 79 | return dynamic_cast<const GenBank&>(*gbk); |
|---|
| 80 | } |
|---|
| 81 | |
|---|
| 82 | const Macke& ConvertibleData::to_macke(const Seq& seq) const { |
|---|
| 83 | Format inType = in.format(); |
|---|
| 84 | |
|---|
| 85 | if (inType == MACKE) return dynamic_cast<const Macke&>(in); |
|---|
| 86 | |
|---|
| 87 | macke = new Macke; |
|---|
| 88 | |
|---|
| 89 | bool ok = true; |
|---|
| 90 | switch (inType) { |
|---|
| 91 | case EMBL: ok = etom(dynamic_cast<const Embl&>(in), dynamic_cast<Macke&>(*macke), seq); break; |
|---|
| 92 | case GENBANK: ok = gtom(dynamic_cast<const GenBank&>(in), dynamic_cast<Macke&>(*macke)); break; |
|---|
| 93 | default: throw_conversion_not_implemented(inType, MACKE); |
|---|
| 94 | } |
|---|
| 95 | if (!ok) throw_conversion_failure(inType, MACKE); |
|---|
| 96 | |
|---|
| 97 | ca_assert(!macke.isNull()); |
|---|
| 98 | return dynamic_cast<const Macke&>(*macke); |
|---|
| 99 | } |
|---|
| 100 | |
|---|
| 101 | static void write_to_embl(FileWriter& write, const ConvertibleData& data, const Seq& seq) { |
|---|
| 102 | const Embl& embl = data.to_embl(seq); |
|---|
| 103 | embl_out(embl, seq, write); |
|---|
| 104 | write.seq_done(); |
|---|
| 105 | } |
|---|
| 106 | static void write_to_genbank(FileWriter& write, const ConvertibleData& data, const Seq& seq) { |
|---|
| 107 | const GenBank& gbk = data.to_genbank(seq); |
|---|
| 108 | genbank_out(gbk, seq, write); |
|---|
| 109 | write.seq_done(); |
|---|
| 110 | } |
|---|
| 111 | static void write_to_macke(FileWriter& write, const ConvertibleData& data, const Seq& seq, int phase, bool first) { |
|---|
| 112 | Format inType = data.format(); |
|---|
| 113 | const Macke& macke = data.to_macke(seq); |
|---|
| 114 | switch (phase) { |
|---|
| 115 | case 0: |
|---|
| 116 | macke_seq_display_out(macke, write, inType, first); |
|---|
| 117 | break; |
|---|
| 118 | case 1: |
|---|
| 119 | macke_seq_info_out(macke, write); |
|---|
| 120 | break; |
|---|
| 121 | case 2: |
|---|
| 122 | macke_seq_data_out(seq, macke, write); |
|---|
| 123 | write.seq_done(); // count only in one and last phase! |
|---|
| 124 | break; |
|---|
| 125 | } |
|---|
| 126 | } |
|---|
| 127 | |
|---|
| 128 | static void to_macke(const FormattedFile& in, const char *outf) { |
|---|
| 129 | FormatReaderPtr reader = FormatReader::create(in); |
|---|
| 130 | FileWriter write(outf); |
|---|
| 131 | SmartPtr<Warnings> suppress; |
|---|
| 132 | |
|---|
| 133 | macke_out_header(write); |
|---|
| 134 | |
|---|
| 135 | for (int phase = 0; phase<3; ++phase) { |
|---|
| 136 | bool first = true; |
|---|
| 137 | while (1) { |
|---|
| 138 | Seq seq; |
|---|
| 139 | if (!reader->read_one_entry(seq)) break; |
|---|
| 140 | write_to_macke(write, reader->get_data(), seq, phase, first); |
|---|
| 141 | |
|---|
| 142 | first = false; |
|---|
| 143 | } |
|---|
| 144 | if (phase<2) { |
|---|
| 145 | if (phase == 0) { |
|---|
| 146 | write.out("#-\n"); |
|---|
| 147 | suppress = new Warnings; // no warning messages for phase 1+2 |
|---|
| 148 | } |
|---|
| 149 | reader->rewind(); |
|---|
| 150 | } |
|---|
| 151 | } |
|---|
| 152 | } |
|---|
| 153 | |
|---|
| 154 | void convert(const FormattedFile& in, const FormattedFile& out) { |
|---|
| 155 | if (str_equal(in.name(), out.name())) |
|---|
| 156 | throw_error(30, "Input file and output file must be different file"); |
|---|
| 157 | |
|---|
| 158 | bool converted = true; |
|---|
| 159 | switch (out.type()) { |
|---|
| 160 | case EMBL: |
|---|
| 161 | case SWISSPROT: |
|---|
| 162 | case GENBANK: { |
|---|
| 163 | Format inType = in.type(); |
|---|
| 164 | Format ouType = out.type(); |
|---|
| 165 | if ((inType == GENBANK && ouType == SWISSPROT) || |
|---|
| 166 | (inType == SWISSPROT && ouType == GENBANK) || |
|---|
| 167 | (inType == EMBL && ouType == SWISSPROT) || |
|---|
| 168 | (inType == SWISSPROT && ouType == EMBL)) |
|---|
| 169 | { |
|---|
| 170 | converted = false; |
|---|
| 171 | break; |
|---|
| 172 | } |
|---|
| 173 | |
|---|
| 174 | FormatReaderPtr reader = FormatReader::create(in); |
|---|
| 175 | FileWriter write(out.name()); |
|---|
| 176 | |
|---|
| 177 | while (1) { |
|---|
| 178 | Seq seq; |
|---|
| 179 | if (!reader->read_one_entry(seq)) break; |
|---|
| 180 | if (ouType == GENBANK) { |
|---|
| 181 | write_to_genbank(write, reader->get_data(), seq); |
|---|
| 182 | } |
|---|
| 183 | else { |
|---|
| 184 | write_to_embl(write, reader->get_data(), seq); |
|---|
| 185 | } |
|---|
| 186 | } |
|---|
| 187 | break; |
|---|
| 188 | } |
|---|
| 189 | case MACKE: to_macke(in, out.name()); break; |
|---|
| 190 | case GCG: to_gcg(in, out.name()); break; |
|---|
| 191 | case NEXUS: to_paup(in, out.name()); break; |
|---|
| 192 | case PHYLIP: to_phylip(in, out.name(), false); break; |
|---|
| 193 | case FASTDNAML: to_phylip(in, out.name(), true); break; |
|---|
| 194 | case PRINTABLE: to_printable(in, out.name()); break; |
|---|
| 195 | |
|---|
| 196 | default: converted = false; break; |
|---|
| 197 | } |
|---|
| 198 | if (!converted) { |
|---|
| 199 | throw_conversion_not_supported(in.type(), out.type()); |
|---|
| 200 | } |
|---|
| 201 | } |
|---|