1 | // -------------------- Macke related subroutines -------------- |
---|
2 | |
---|
3 | #include "macke.h" |
---|
4 | #include "wrap.h" |
---|
5 | #include "parser.h" |
---|
6 | #include "rdp_info.h" |
---|
7 | |
---|
8 | |
---|
9 | #define MACKELIMIT 10000 |
---|
10 | |
---|
11 | static int macke_abbrev(const char *line, char *key, int index) { |
---|
12 | // Get the key from a macke line. |
---|
13 | // returns index behind delimiting ':' |
---|
14 | index = Skip_white_space(line, index); |
---|
15 | int len = parse_key_word(line+index, key, " :\t\n"); |
---|
16 | return index+len+1; |
---|
17 | } |
---|
18 | |
---|
19 | static void macke_continue_line(const char *key, char *oldname, char*& var, Reader& reader) { |
---|
20 | // Append macke continue line. |
---|
21 | |
---|
22 | for (++reader; reader.line(); ++reader) { |
---|
23 | if (has_content(reader.line())) { |
---|
24 | char name[TOKENSIZE]; |
---|
25 | int index = macke_abbrev(reader.line(), name, 0); |
---|
26 | |
---|
27 | if (!str_equal(name, oldname)) break; |
---|
28 | |
---|
29 | char newkey[TOKENSIZE]; |
---|
30 | index = macke_abbrev(reader.line(), newkey, index); |
---|
31 | if (!str_equal(newkey, key)) break; |
---|
32 | |
---|
33 | skip_eolnl_and_append_spaced(var, reader.line() + index); |
---|
34 | } |
---|
35 | } |
---|
36 | } |
---|
37 | static void macke_one_entry_in(Reader& reader, const char *key, char *oldname, char*& var, int index) { |
---|
38 | // Get one Macke entry. |
---|
39 | if (has_content(var)) |
---|
40 | skip_eolnl_and_append_spaced(var, reader.line() + index); |
---|
41 | else |
---|
42 | freedup(var, reader.line() + index); |
---|
43 | |
---|
44 | macke_continue_line(key, oldname, var, reader); |
---|
45 | } |
---|
46 | |
---|
47 | static void macke_read_seq(Seq& seq, char*& seqabbr, Reader& reader) { |
---|
48 | ca_assert(seq.is_empty()); |
---|
49 | for (; reader.line(); ++reader) { // read in sequence data line by line |
---|
50 | if (!has_content(reader.line())) continue; |
---|
51 | |
---|
52 | char name[TOKENSIZE]; |
---|
53 | int index = macke_abbrev(reader.line(), name, 0); |
---|
54 | |
---|
55 | if (seqabbr) { |
---|
56 | if (!str_equal(seqabbr, name)) break; // stop if reached different abbrev |
---|
57 | } |
---|
58 | else { |
---|
59 | seqabbr = nulldup(name); |
---|
60 | } |
---|
61 | |
---|
62 | int seqnum; |
---|
63 | char data[LINESIZE]; |
---|
64 | int scanned = sscanf(reader.line() + index, "%d%s", &seqnum, data); |
---|
65 | if (scanned != 2) throw_errorf(80, "Failed to parse '%s'", reader.line()); |
---|
66 | |
---|
67 | for (int indj = seq.get_len(); indj < seqnum; indj++) seq.add('.'); |
---|
68 | for (int indj = 0; data[indj] != '\n' && data[indj] != '\0'; indj++) seq.add(data[indj]); |
---|
69 | } |
---|
70 | } |
---|
71 | |
---|
72 | void macke_origin(Seq& seq, char*& seqabbr, Reader& reader) { |
---|
73 | // Read in sequence data in macke file. |
---|
74 | ca_assert(seqabbr); // = macke.seqabbr |
---|
75 | macke_read_seq(seq, seqabbr, reader); |
---|
76 | } |
---|
77 | |
---|
78 | void macke_out_header(Writer& write) { |
---|
79 | // Output the Macke format header. |
---|
80 | write.out("#-\n#-\n#-\teditor\n"); |
---|
81 | const char *date = today_date(); |
---|
82 | write.outf("#-\t%s\n#-\n#-\n", date); |
---|
83 | } |
---|
84 | |
---|
85 | void macke_seq_display_out(const Macke& macke, Writer& write, Format inType, bool first_sequence) { |
---|
86 | // Output the Macke format each sequence format (wot?) |
---|
87 | if (first_sequence) { |
---|
88 | write.outf("#-\tReference sequence: %s\n", macke.seqabbr); |
---|
89 | write.out("#-\tAttributes:\n"); |
---|
90 | } |
---|
91 | |
---|
92 | write.out("#=\t\t"); |
---|
93 | if (write.out(macke.seqabbr)<8) write.out('\t'); |
---|
94 | write.out("\tin out vis prt ord "); |
---|
95 | if (inType == SWISSPROT) { |
---|
96 | write.out("pro lin n>c"); |
---|
97 | } |
---|
98 | else { |
---|
99 | write.out(macke.rna_or_dna == 'r' ? "rna" : "dna"); |
---|
100 | write.out(" lin 5>3"); |
---|
101 | } |
---|
102 | write.out(" func"); |
---|
103 | if (first_sequence) write.out(" ref"); |
---|
104 | write.out('\n'); |
---|
105 | } |
---|
106 | |
---|
107 | static void macke_print_line(Writer& write, const char *prefix, const char *content) { |
---|
108 | // print a macke line and wrap around line after MACKEMAXLINE column. |
---|
109 | WrapMode(true).print(write, prefix, prefix, content, MACKEMAXLINE); |
---|
110 | } |
---|
111 | |
---|
112 | static void macke_print_prefixed_line(const Macke& macke, Writer& write, const char *tag, const char *content) { |
---|
113 | ca_assert(has_content(content)); |
---|
114 | |
---|
115 | char prefix[LINESIZE]; |
---|
116 | sprintf(prefix, "#:%s:%s:", macke.seqabbr, tag); |
---|
117 | |
---|
118 | macke_print_line(write, prefix, content); |
---|
119 | } |
---|
120 | |
---|
121 | static bool macke_print_prefixed_line_if_content(const Macke& macke, Writer& write, const char *tag, const char *content) { |
---|
122 | if (!has_content(content)) return false; |
---|
123 | macke_print_prefixed_line(macke, write, tag, content); |
---|
124 | return true; |
---|
125 | } |
---|
126 | |
---|
127 | static const char *genbankEntryComments[] = { |
---|
128 | "KEYWORDS", |
---|
129 | "GenBank ACCESSION", |
---|
130 | "auth", |
---|
131 | "title", |
---|
132 | "jour", |
---|
133 | "standard", |
---|
134 | "Source of strain", |
---|
135 | "Former name", |
---|
136 | "Alternate name", |
---|
137 | "Common name", |
---|
138 | "Host organism", |
---|
139 | "RDP ID", |
---|
140 | "Sequencing methods", |
---|
141 | "3' end complete", |
---|
142 | "5' end complete", |
---|
143 | }; |
---|
144 | |
---|
145 | static bool macke_is_genbank_entry_comment(const char *Str) { |
---|
146 | char keyword[TOKENSIZE]; |
---|
147 | macke_key_word(Str, 0, keyword); |
---|
148 | |
---|
149 | return lookup_keyword(keyword, genbankEntryComments) >= 0; |
---|
150 | } |
---|
151 | |
---|
152 | static void macke_print_keyword_rem(const Macke& macke, int index, Writer& write) { |
---|
153 | // Print out keyworded remark line in Macke file with wrap around functionality. |
---|
154 | // (Those keywords are defined in GenBank COMMENTS by RDP group) |
---|
155 | char first[LINESIZE]; sprintf(first, "#:%s:rem:", macke.seqabbr); |
---|
156 | char other[LINESIZE]; sprintf(other, "%s:%*s", first, RDP_SUBKEY_INDENT, ""); |
---|
157 | const char *remark = macke.get_rem(index); |
---|
158 | |
---|
159 | WrapMode(true).print(write, first, other, remark, MACKEMAXLINE); |
---|
160 | } |
---|
161 | |
---|
162 | void macke_seq_info_out(const Macke& macke, Writer& write) { |
---|
163 | // Output sequence information |
---|
164 | |
---|
165 | macke_print_prefixed_line_if_content(macke, write, "name", macke.name); |
---|
166 | macke_print_prefixed_line_if_content(macke, write, "strain", macke.strain); |
---|
167 | macke_print_prefixed_line_if_content(macke, write, "subsp", macke.subspecies); |
---|
168 | macke_print_prefixed_line_if_content(macke, write, "atcc", macke.atcc); |
---|
169 | macke_print_prefixed_line_if_content(macke, write, "rna", macke.rna); // old version entry |
---|
170 | macke_print_prefixed_line_if_content(macke, write, "date", macke.date); |
---|
171 | macke_print_prefixed_line_if_content(macke, write, "acs", macke.acs) |
---|
172 | || macke_print_prefixed_line_if_content(macke, write, "acs", macke.nbk); // old version entry |
---|
173 | macke_print_prefixed_line_if_content(macke, write, "auth", macke.author); |
---|
174 | macke_print_prefixed_line_if_content(macke, write, "jour", macke.journal); |
---|
175 | macke_print_prefixed_line_if_content(macke, write, "title", macke.title); |
---|
176 | macke_print_prefixed_line_if_content(macke, write, "who", macke.who); |
---|
177 | |
---|
178 | // print out remarks, wrap around if more than MACKEMAXLINE columns |
---|
179 | for (int indi = 0; indi < macke.get_rem_count(); indi++) { |
---|
180 | if (macke_is_genbank_entry_comment(macke.get_rem(indi))) { |
---|
181 | macke_print_keyword_rem(macke, indi, write); |
---|
182 | } |
---|
183 | else { // general comment |
---|
184 | macke_print_prefixed_line(macke, write, "rem", macke.get_rem(indi)); |
---|
185 | } |
---|
186 | } |
---|
187 | } |
---|
188 | |
---|
189 | int macke_key_word(const char *line, int index, char *key) { |
---|
190 | // Find the key in Macke line. |
---|
191 | // return position behind ':' delimiter |
---|
192 | int len = parse_key_word(line+index, key, ":\n"); |
---|
193 | return len ? index+len+1 : index; |
---|
194 | } |
---|
195 | |
---|
196 | void macke_seq_data_out(const Seq& seq, const Macke& macke, Writer& write) { |
---|
197 | // Output Macke format sequence data |
---|
198 | int indj, indk; |
---|
199 | |
---|
200 | if (seq.get_len() > MACKELIMIT) { |
---|
201 | warningf(145, "Length of sequence data is %d over AE2's limit %d.", seq.get_len(), MACKELIMIT); |
---|
202 | } |
---|
203 | |
---|
204 | const char *sequence = seq.get_seq(); |
---|
205 | for (indk = indj = 0; indk < seq.get_len(); indk++) { |
---|
206 | if (indj == 0) |
---|
207 | write.outf("%s%6d ", macke.seqabbr, indk); |
---|
208 | |
---|
209 | write.out(sequence[indk]); |
---|
210 | |
---|
211 | indj++; |
---|
212 | if (indj == 50) { |
---|
213 | indj = 0; |
---|
214 | write.out('\n'); |
---|
215 | } |
---|
216 | } |
---|
217 | |
---|
218 | if (indj != 0) |
---|
219 | write.out('\n'); |
---|
220 | // every sequence |
---|
221 | } |
---|
222 | |
---|
223 | void MackeReader::read_to_start() { |
---|
224 | r1->skipOverLinesThat(Not(isMackeSeqInfo)); // skip to #:; where the sequence information is |
---|
225 | r2->skipOverLinesThat(isMackeNonSeq); // skip to where sequence data starts |
---|
226 | r3->skipOverLinesThat(Not(isMackeSeqHeader)); // skip to #=; where sequence first appears |
---|
227 | } |
---|
228 | |
---|
229 | |
---|
230 | MackeReader::MackeReader(const char *inName_) |
---|
231 | : inName(ARB_strdup(inName_)), |
---|
232 | seqabbr(dummy), |
---|
233 | dummy(NULL), |
---|
234 | r1(new Reader(inName)), |
---|
235 | r2(new Reader(inName)), |
---|
236 | r3(new Reader(inName)), |
---|
237 | using_reader(&r1) |
---|
238 | { |
---|
239 | read_to_start(); |
---|
240 | } |
---|
241 | |
---|
242 | MackeReader::~MackeReader() { |
---|
243 | char *msg = NULL; |
---|
244 | const Convaln_exception *exc = Convaln_exception::exception_thrown(); |
---|
245 | |
---|
246 | ca_assert(using_reader); |
---|
247 | delete *using_reader; *using_reader = NULL; |
---|
248 | |
---|
249 | // avoid that all 3 readers decorate the error |
---|
250 | if (exc) msg = ARB_strdup(exc->get_msg()); |
---|
251 | delete r3; r3 = NULL; |
---|
252 | delete r2; r2 = NULL; |
---|
253 | delete r1; r1 = NULL; |
---|
254 | if (exc) { exc->replace_msg(msg); free(msg); } |
---|
255 | |
---|
256 | free(inName); |
---|
257 | } |
---|
258 | |
---|
259 | class MackeParser : public Parser { |
---|
260 | Macke& macke; |
---|
261 | |
---|
262 | public: |
---|
263 | MackeParser(Macke& macke_, Seq& seq_, Reader& reader_) : Parser(seq_, reader_), macke(macke_) {} |
---|
264 | |
---|
265 | void parse_section() OVERRIDE { |
---|
266 | ca_assert(0); // @@@ unused yet |
---|
267 | } |
---|
268 | }; |
---|
269 | |
---|
270 | |
---|
271 | bool MackeReader::macke_in(Macke& macke) { |
---|
272 | // Read in one sequence data from Macke file. |
---|
273 | char oldname[TOKENSIZE], name[TOKENSIZE]; |
---|
274 | char key[TOKENSIZE]; |
---|
275 | int index; |
---|
276 | |
---|
277 | // r1 points to sequence information |
---|
278 | // r2 points to sequence data |
---|
279 | // r3 points to sequence names |
---|
280 | |
---|
281 | usingReader(r3); |
---|
282 | if (!r3->line() || !isMackeSeqHeader(r3->line())) return false; |
---|
283 | |
---|
284 | // skip to next "#:" line or end of file |
---|
285 | usingReader(r1); |
---|
286 | r1->skipOverLinesThat(Not(isMackeSeqInfo)); |
---|
287 | |
---|
288 | // read in sequence name |
---|
289 | usingReader(r3); |
---|
290 | index = macke_abbrev(r3->line(), oldname, 2); |
---|
291 | freedup(macke.seqabbr, oldname); |
---|
292 | seqabbr = macke.seqabbr; |
---|
293 | |
---|
294 | // read sequence information |
---|
295 | usingReader(r1); |
---|
296 | for (index = macke_abbrev(r1->line(), name, 2); |
---|
297 | r1->line() && isMackeSeqInfo(r1->line()) && str_equal(name, oldname); |
---|
298 | ) |
---|
299 | { |
---|
300 | index = macke_abbrev(r1->line(), key, index); |
---|
301 | if (str_equal(key, "name")) { |
---|
302 | macke_one_entry_in(*r1, "name", oldname, macke.name, index); |
---|
303 | } |
---|
304 | else if (str_equal(key, "atcc")) { |
---|
305 | macke_one_entry_in(*r1, "atcc", oldname, macke.atcc, index); |
---|
306 | } |
---|
307 | else if (str_equal(key, "rna")) { |
---|
308 | // old version entry |
---|
309 | macke_one_entry_in(*r1, "rna", oldname, macke.rna, index); |
---|
310 | } |
---|
311 | else if (str_equal(key, "date")) { |
---|
312 | macke_one_entry_in(*r1, "date", oldname, macke.date, index); |
---|
313 | } |
---|
314 | else if (str_equal(key, "nbk")) { |
---|
315 | // old version entry |
---|
316 | macke_one_entry_in(*r1, "nbk", oldname, macke.nbk, index); |
---|
317 | } |
---|
318 | else if (str_equal(key, "acs")) { |
---|
319 | macke_one_entry_in(*r1, "acs", oldname, macke.acs, index); |
---|
320 | } |
---|
321 | else if (str_equal(key, "subsp")) { |
---|
322 | macke_one_entry_in(*r1, "subsp", oldname, macke.subspecies, index); |
---|
323 | } |
---|
324 | else if (str_equal(key, "strain")) { |
---|
325 | macke_one_entry_in(*r1, "strain", oldname, macke.strain, index); |
---|
326 | } |
---|
327 | else if (str_equal(key, "auth")) { |
---|
328 | macke_one_entry_in(*r1, "auth", oldname, macke.author, index); |
---|
329 | } |
---|
330 | else if (str_equal(key, "title")) { |
---|
331 | macke_one_entry_in(*r1, "title", oldname, macke.title, index); |
---|
332 | } |
---|
333 | else if (str_equal(key, "jour")) { |
---|
334 | macke_one_entry_in(*r1, "jour", oldname, macke.journal, index); |
---|
335 | } |
---|
336 | else if (str_equal(key, "who")) { |
---|
337 | macke_one_entry_in(*r1, "who", oldname, macke.who, index); |
---|
338 | } |
---|
339 | else if (str_equal(key, "rem")) { |
---|
340 | macke.add_remark(r1->line()+index); |
---|
341 | ++(*r1); |
---|
342 | } |
---|
343 | else { |
---|
344 | warningf(144, "Unidentified AE2 key word #%s#", key); |
---|
345 | ++(*r1); |
---|
346 | } |
---|
347 | if (r1->line() && isMackeSeqInfo(r1->line())) index = macke_abbrev(r1->line(), name, 2); |
---|
348 | else index = 0; |
---|
349 | } |
---|
350 | |
---|
351 | ++(*r3); |
---|
352 | |
---|
353 | return true; |
---|
354 | } |
---|
355 | |
---|
356 | bool MackeReader::read_one_entry(Seq& seq) { |
---|
357 | data.reinit(); |
---|
358 | if (!macke_in(data) || !read_seq_data(seq)) abort(); |
---|
359 | seq.set_id(data.get_id()); |
---|
360 | return ok(); |
---|
361 | } |
---|
362 | |
---|
363 | |
---|