1 | // =============================================================== // |
---|
2 | // // |
---|
3 | // File : PT_etc.cxx // |
---|
4 | // Purpose : // |
---|
5 | // // |
---|
6 | // Institute of Microbiology (Technical University Munich) // |
---|
7 | // http://www.arb-home.de/ // |
---|
8 | // // |
---|
9 | // =============================================================== // |
---|
10 | |
---|
11 | #include "probe.h" |
---|
12 | |
---|
13 | #include <PT_server_prototypes.h> |
---|
14 | #include "pt_prototypes.h" |
---|
15 | #include <struct_man.h> |
---|
16 | #include <arb_strbuf.h> |
---|
17 | |
---|
18 | void set_table_for_PT_N_mis(int ignored_Nmismatches, int when_less_than_Nmismatches) { |
---|
19 | // calculate table for PT_N mismatches |
---|
20 | // |
---|
21 | // 'ignored_Nmismatches' specifies, how many N-mismatches will be accepted as |
---|
22 | // matches, when overall number of N-mismatches is below 'when_less_than_Nmismatches'. |
---|
23 | // |
---|
24 | // above that limit, every N-mismatch counts as mismatch |
---|
25 | |
---|
26 | if ((when_less_than_Nmismatches-1)>PT_POS_TREE_HEIGHT) when_less_than_Nmismatches = PT_POS_TREE_HEIGHT+1; |
---|
27 | if (ignored_Nmismatches >= when_less_than_Nmismatches) ignored_Nmismatches = when_less_than_Nmismatches-1; |
---|
28 | |
---|
29 | psg.w_N_mismatches[0] = 0; |
---|
30 | int mm; |
---|
31 | for (mm = 1; mm<when_less_than_Nmismatches; ++mm) { |
---|
32 | psg.w_N_mismatches[mm] = mm>ignored_Nmismatches ? mm-ignored_Nmismatches : 0; |
---|
33 | } |
---|
34 | pt_assert(mm <= (PT_POS_TREE_HEIGHT+1)); |
---|
35 | for (; mm <= PT_POS_TREE_HEIGHT; ++mm) { |
---|
36 | psg.w_N_mismatches[mm] = mm; |
---|
37 | } |
---|
38 | } |
---|
39 | |
---|
40 | void pt_export_error(PT_local *locs, const char *error) { |
---|
41 | freedup(locs->ls_error, error); |
---|
42 | } |
---|
43 | |
---|
44 | static const gene_struct *get_gene_struct_by_internal_gene_name(const char *gene_name) { |
---|
45 | gene_struct to_search(gene_name, "", ""); |
---|
46 | |
---|
47 | gene_struct_index_internal::const_iterator found = gene_struct_internal2arb.find(&to_search); |
---|
48 | return (found == gene_struct_internal2arb.end()) ? 0 : *found; |
---|
49 | } |
---|
50 | static const gene_struct *get_gene_struct_by_arb_species_gene_name(const char *species_gene_name) { |
---|
51 | const char *slash = strchr(species_gene_name, '/'); |
---|
52 | if (!slash) { |
---|
53 | fprintf(stderr, "Internal error: '%s' should be in format 'organism/gene'\n", species_gene_name); |
---|
54 | return 0; |
---|
55 | } |
---|
56 | |
---|
57 | int slashpos = slash-species_gene_name; |
---|
58 | char *organism = strdup(species_gene_name); |
---|
59 | organism[slashpos] = 0; |
---|
60 | |
---|
61 | gene_struct to_search("", organism, species_gene_name+slashpos+1); |
---|
62 | free(organism); |
---|
63 | |
---|
64 | gene_struct_index_arb::const_iterator found = gene_struct_arb2internal.find(&to_search); |
---|
65 | return (found == gene_struct_arb2internal.end()) ? 0 : *found; |
---|
66 | } |
---|
67 | |
---|
68 | static const char *arb2internal_name(const char *name) { |
---|
69 | // convert arb name ('species/gene') into internal shortname |
---|
70 | const gene_struct *found = get_gene_struct_by_arb_species_gene_name(name); |
---|
71 | return found ? found->get_internal_gene_name() : 0; |
---|
72 | } |
---|
73 | |
---|
74 | const char *virt_name(PT_probematch *ml) |
---|
75 | { |
---|
76 | // get the name with a virtual function |
---|
77 | if (gene_flag) { |
---|
78 | const gene_struct *gs = get_gene_struct_by_internal_gene_name(psg.data[ml->name].get_name()); |
---|
79 | return gs ? gs->get_arb_species_name() : "<cantResolveName>"; |
---|
80 | } |
---|
81 | else { |
---|
82 | pt_assert(psg.data[ml->name].get_name()); |
---|
83 | return psg.data[ml->name].get_name(); |
---|
84 | } |
---|
85 | } |
---|
86 | |
---|
87 | const char *virt_fullname(PT_probematch * ml) |
---|
88 | { |
---|
89 | if (gene_flag) { |
---|
90 | const gene_struct *gs = get_gene_struct_by_internal_gene_name(psg.data[ml->name].get_name()); |
---|
91 | return gs ? gs->get_arb_gene_name() : "<cantResolveGeneFullname>"; |
---|
92 | } |
---|
93 | else { |
---|
94 | return psg.data[ml->name].get_fullname() ? psg.data[ml->name].get_fullname() : "<undefinedFullname>"; |
---|
95 | } |
---|
96 | } |
---|
97 | |
---|
98 | #define MAX_LIST_PART_SIZE 50 |
---|
99 | |
---|
100 | static const char *get_list_part(const char *list, int& offset) { |
---|
101 | // scans strings with format "xxxx#yyyy#zzzz" |
---|
102 | // your first call should be with offset == 0 |
---|
103 | // |
---|
104 | // returns : static copy of each part or 0 when done |
---|
105 | // offset is incremented by this function and set to -1 when all parts were returned |
---|
106 | |
---|
107 | static char buffer[2][MAX_LIST_PART_SIZE+1]; |
---|
108 | static int curr_buff = 0; // toggle buffer to allow 2 parallel gets w/o invalidation |
---|
109 | |
---|
110 | if (offset<0) return 0; // already done |
---|
111 | curr_buff ^= 1; // toggle buffer |
---|
112 | |
---|
113 | const char *numsign = strchr(list+offset, '#'); |
---|
114 | int num_offset; |
---|
115 | if (numsign) { |
---|
116 | num_offset = numsign-list; |
---|
117 | pt_assert(list[num_offset] == '#'); |
---|
118 | } |
---|
119 | else { // last list part |
---|
120 | num_offset = offset+strlen(list+offset); |
---|
121 | pt_assert(list[num_offset] == 0); |
---|
122 | } |
---|
123 | |
---|
124 | // now num_offset points to next '#' or to end-of-string |
---|
125 | |
---|
126 | int len = num_offset-offset; |
---|
127 | pt_assert(len <= MAX_LIST_PART_SIZE); |
---|
128 | |
---|
129 | memcpy(buffer[curr_buff], list+offset, len); |
---|
130 | buffer[curr_buff][len] = 0; // EOS |
---|
131 | |
---|
132 | offset = (list[num_offset] == '#') ? num_offset+1 : -1; // set offset for next part |
---|
133 | |
---|
134 | return buffer[curr_buff]; |
---|
135 | } |
---|
136 | |
---|
137 | #undef MAX_LIST_PART_SIZE |
---|
138 | |
---|
139 | char *ptpd_read_names(PT_local *locs, const char *names_list, const char *checksums, const char*& error) { |
---|
140 | /* read the name list separated by '#' and set the flag for the group members, |
---|
141 | + returns a list of names which have not been found |
---|
142 | */ |
---|
143 | |
---|
144 | // clear 'is_group' |
---|
145 | for (int i = 0; i < psg.data_count; i++) { |
---|
146 | psg.data[i].set_group_state(0); // Note: probes are designed for species with is_group == 1 |
---|
147 | } |
---|
148 | locs->group_count = 0; |
---|
149 | error = 0; |
---|
150 | |
---|
151 | if (!names_list) { |
---|
152 | error = "Can't design probes for no species (species list is empty)"; |
---|
153 | return 0; |
---|
154 | } |
---|
155 | |
---|
156 | int noff = 0; |
---|
157 | int coff = 0; |
---|
158 | |
---|
159 | GBS_strstruct *not_found = 0; |
---|
160 | |
---|
161 | while (noff >= 0) { |
---|
162 | pt_assert(coff >= 0); // otherwise 'checksums' contains less elements than 'names_list' |
---|
163 | const char *arb_name = get_list_part(names_list, noff); |
---|
164 | const char *internal_name = arb_name; // differs only for gene pt server |
---|
165 | |
---|
166 | if (arb_name[0] == 0) { |
---|
167 | pt_assert(names_list[0] == 0); |
---|
168 | break; // nothing marked |
---|
169 | } |
---|
170 | |
---|
171 | if (gene_flag) { |
---|
172 | const char *slash = strchr(arb_name, '/'); |
---|
173 | |
---|
174 | if (!slash) { |
---|
175 | // ARB has to send 'species/gene'. |
---|
176 | // If it did not, user did not mark 'Gene probes ?' flag |
---|
177 | |
---|
178 | error = GBS_global_string("Expected '/' in '%s' (this PT-server can only design probes for genes)", arb_name); |
---|
179 | break; |
---|
180 | } |
---|
181 | |
---|
182 | internal_name = arb2internal_name(arb_name); |
---|
183 | pt_assert(internal_name); |
---|
184 | } |
---|
185 | |
---|
186 | int idx = GBS_read_hash(psg.namehash, internal_name); |
---|
187 | bool found = false; |
---|
188 | |
---|
189 | if (idx) { |
---|
190 | --idx; // because 0 means not found! |
---|
191 | |
---|
192 | if (checksums) { |
---|
193 | const char *checksum = get_list_part(checksums, coff); |
---|
194 | // if sequence checksum changed since pt server was updated -> not found |
---|
195 | found = atol(checksum) == psg.data[idx].get_checksum(); |
---|
196 | } |
---|
197 | else { |
---|
198 | found = true; |
---|
199 | } |
---|
200 | |
---|
201 | if (found) { |
---|
202 | psg.data[idx].set_group_state(1); // mark |
---|
203 | locs->group_count++; |
---|
204 | } |
---|
205 | } |
---|
206 | |
---|
207 | if (!found) { // name not found -> put into result |
---|
208 | if (not_found == 0) not_found = GBS_stropen(1000); |
---|
209 | else GBS_chrcat(not_found, '#'); |
---|
210 | GBS_strcat(not_found, arb_name); |
---|
211 | } |
---|
212 | } |
---|
213 | |
---|
214 | char *result = not_found ? GBS_strclose(not_found) : 0; |
---|
215 | if (error) freenull(result); |
---|
216 | return result; |
---|
217 | } |
---|
218 | |
---|
219 | bytestring *PT_unknown_names(PT_pdc *pdc) { |
---|
220 | static bytestring un = { 0, 0 }; |
---|
221 | PT_local *locs = (PT_local*)pdc->mh.parent->parent; |
---|
222 | delete un.data; |
---|
223 | |
---|
224 | const char *error; |
---|
225 | un.data = ptpd_read_names(locs, pdc->names.data, pdc->checksums.data, error); |
---|
226 | if (un.data) { |
---|
227 | un.size = strlen(un.data) + 1; |
---|
228 | pt_assert(!error); |
---|
229 | } |
---|
230 | else { |
---|
231 | un.data = strdup(""); |
---|
232 | un.size = 1; |
---|
233 | if (error) pt_export_error(locs, error); |
---|
234 | } |
---|
235 | return &un; |
---|
236 | } |
---|
237 | |
---|
238 | void PT_init_base_string_counter(char *str, char initval, int size) |
---|
239 | { |
---|
240 | memset(str, initval, size+1); |
---|
241 | str[size] = 0; |
---|
242 | } |
---|
243 | |
---|
244 | void PT_inc_base_string_count(char *str, char initval, char maxval, int size) |
---|
245 | { |
---|
246 | int i; |
---|
247 | if (!size) { |
---|
248 | str[0]=255; |
---|
249 | return; |
---|
250 | } |
---|
251 | for (i=size-1; i>=0; i--) { |
---|
252 | str[i]++; |
---|
253 | if (str[i] >= maxval) { |
---|
254 | str[i] = initval; |
---|
255 | if (!i) str[i]=255; // end flag |
---|
256 | } |
---|
257 | else { |
---|
258 | break; |
---|
259 | } |
---|
260 | } |
---|
261 | } |
---|