#include #include #include #include #include #include "alphabet.h" #include "dreme-sax.h" #include "linked-list.h" #include "red-black-tree.h" #include "sax-parser-utils.h" #include "utils.h" /***************************************************************************** * Parser States ****************************************************************************/ enum parser_state { PS_ERROR, PS_START, PS_IN_DREME, PS_IN_MODEL, PS_IN_MOTIFS, PS_IN_RUN_TIME, PS_IN_COMMAND_LINE, PS_IN_POSITIVES, PS_IN_NEGATIVES, PS_IN_ALPHABET, PS_IN_ALPHABET_LETTER, PS_IN_STRANDS, PS_IN_BACKGROUND, PS_IN_STOP, PS_IN_NORC, PS_IN_NGEN, PS_IN_ADD_PV_THRESH, PS_IN_SEED, PS_IN_HOST, PS_IN_WHEN, PS_IN_DESCRIPTION, PS_IN_MOTIF, PS_IN_POS, PS_IN_MATCH, PS_END }; typedef enum parser_state PS_EN; /***************************************************************************** * Parser State Names (for error messages) ****************************************************************************/ static char const * const state_names[] = { "ERROR", "START", "IN_DREME", "IN_MODEL", "IN_MOTIFS", "IN_RUN_TIME", "IN_COMMAND_LINE", "IN_POSITIVES", "IN_NEGATIVES", "IN_BACKGROUND", "IN_STOP", "IN_NGEN", "IN_ADD_PV_THRESH", "IN_SEED", "IN_HOST", "IN_WHEN", "IN_DESCRIPTION", "IN_MOTIF", "IN_POS", "IN_MATCH", "END" }; /***************************************************************************** * Alphabet enumeration (for old background::type attribute) ****************************************************************************/ enum dreme_alph { DREME_ALPH_DNA, // DNA alphabet DREME_ALPH_RNA // RNA alphabet }; /***************************************************************************** * Expected State Repeats ****************************************************************************/ enum expected_state { ES_ZERO_OR_ONE, ES_ONCE, ES_ONE_OR_MORE, ES_ANY }; typedef enum expected_state ES_EN; /***************************************************************************** * Used to keep track of what elements are expected and enforces limits on the * number of times a valid element can be repeated. Also avoids the * possibility that elements could be left out. ****************************************************************************/ typedef struct es ES_T; struct es { PS_EN state; // the element that is expected ES_EN expected; // the number of times the element is expected (see ES codes) int found; // the number of times the element has been found already }; /***************************************************************************** * Stores the parser state. ****************************************************************************/ typedef struct ps PS_T; struct ps { int state; // the state code (see PS codes) int udepth; // the depth of unknown tags (disabled) ATTRBUF_T attrbuf; CHARBUF_T characters; // a buffer to store characters between elements LINKLST_T *expected_stack;// a queue of expected states DREME_IO_XML_CALLBACKS_T *callbacks; // callbacks to the user struct prog_version ver; bool seen_alphabet; // distinguish between old and new bool seen_ambig; // ambiguous symbols must only appear after core RBTREE_T *alph_ids; // list of alphabet ids and their positions double *freqs; // buffer of frequencies char *motif_id; // motif id int motif_len; // motif length int last_pos; // last seen motif pos void *user_data; // the user data }; /***************************************************************************** * Report an error to the user. ****************************************************************************/ static void error(PS_T *ps, char *fmt, ...) { va_list argp; if (ps->callbacks->error) { va_start(argp, fmt); ps->callbacks->error(ps->user_data, fmt, argp); va_end(argp); } ps->state = PS_ERROR; } /***************************************************************************** * Report an attribute parse error to the user. ****************************************************************************/ void dreme_attr_parse_error(void *state, int errcode, const char *tag, const char *attr, const char *value) { PS_T *ps = (PS_T*)state; if (errcode == PARSE_ATTR_DUPLICATE) { error(ps, "Duplicate attribute %s::%s.\n", tag, attr); } else if (errcode == PARSE_ATTR_BAD_VALUE) { error(ps, "Bad value \"%s\" for attribute %s::%s.\n", value, tag, attr); } else if (errcode == PARSE_ATTR_MISSING) { error(ps, "Missing required attribute %s::%s.\n", tag, attr); } } /***************************************************************************** * Push an expected state on the stack ****************************************************************************/ void dreme_push_es(PS_T *ps, PS_EN expected_state, ES_EN expected_occurances) { ES_T *es; if (expected_state < PS_ERROR || expected_state > PS_END) { die("Bad state code!\n"); } es = mm_malloc(sizeof(ES_T)); es->state = expected_state; es->expected = expected_occurances; es->found = 0; linklst_push(es, ps->expected_stack); } /***************************************************************************** * At the start of a new element check that all previous expected elements * have been found and that the current element hasn't been repeated too * many times. ****************************************************************************/ int dreme_update_es(PS_T *ps, PS_EN next_state) { ES_T *es, old_es; if (next_state < PS_ERROR || next_state > PS_END) { die("Bad state code!\n"); } while ((es = (ES_T*)linklst_pop(ps->expected_stack)) != NULL && es->state != next_state) { old_es = *es; free(es); es = NULL; switch (old_es.expected) { case ES_ONCE: case ES_ONE_OR_MORE: if (old_es.found < 1){ error(ps, "Expected state %s not found!\n", state_names[old_es.state]); return FALSE; } default: break; } } if (es != NULL) { es->found += 1; linklst_push(es, ps->expected_stack); switch (es->expected) { case ES_ONCE: case ES_ZERO_OR_ONE: if (es->found > 1) { error(ps, "Expected state %s only once!\n", state_names[es->state]); return FALSE; } default: break; } } else { error(ps, "The state %s was not expected!\n", state_names[next_state]); return FALSE; } return TRUE; } /***************************************************************************** * dreme * contains: model, motifs, run_time * * release the release date. * version the program version. ****************************************************************************/ static void start_ele_dreme(PS_T *ps, const xmlChar **attrs) { char *release; char* names[2] = {"release", "version"}; int (*parsers[2])(char*, void*) = {ld_str, ld_version}; void *data[2] = {&release, &(ps->ver)}; BOOLEAN_T required[2] = {TRUE, TRUE}; BOOLEAN_T done[2]; parse_attributes(dreme_attr_parse_error, ps, "dreme", attrs, 2, names, parsers, data, required, done); if (ps->callbacks->start_dreme && ps->state != PS_ERROR) { ps->callbacks->start_dreme(ps->user_data, ps->ver.major, ps->ver.minor, ps->ver.patch, release); } dreme_push_es(ps, PS_IN_RUN_TIME, ES_ONCE); dreme_push_es(ps, PS_IN_MOTIFS, ES_ONCE); dreme_push_es(ps, PS_IN_MODEL, ES_ONCE); } /***************************************************************************** * /dreme * contains: model, motifs, run_time ****************************************************************************/ static void end_ele_dreme(PS_T *ps) { if (ps->callbacks->end_dreme) { ps->callbacks->end_dreme(ps->user_data); } } /***************************************************************************** * dreme > model * contains: command_line, positives, negatives, background, stop, ngen, * add_pv_thresh, seed, host, when, description ****************************************************************************/ static void start_ele_model(PS_T *ps, const xmlChar **attrs) { if (ps->callbacks->start_model && ps->state != PS_ERROR) { ps->callbacks->start_model(ps->user_data); } dreme_push_es(ps, PS_IN_DESCRIPTION, ES_ZERO_OR_ONE); dreme_push_es(ps, PS_IN_WHEN, ES_ONCE); dreme_push_es(ps, PS_IN_HOST, ES_ONCE); dreme_push_es(ps, PS_IN_SEED, ES_ONCE); dreme_push_es(ps, PS_IN_ADD_PV_THRESH, ES_ONCE); dreme_push_es(ps, PS_IN_NGEN, ES_ONCE); dreme_push_es(ps, PS_IN_NORC, ES_ZERO_OR_ONE); // removed in 4.11 dreme_push_es(ps, PS_IN_STOP, ES_ONCE); dreme_push_es(ps, PS_IN_BACKGROUND, ES_ONCE); dreme_push_es(ps, PS_IN_STRANDS, ES_ZERO_OR_ONE); // added in 4.11 dreme_push_es(ps, PS_IN_ALPHABET, ES_ZERO_OR_ONE); // added in 4.11 dreme_push_es(ps, PS_IN_NEGATIVES, ES_ONCE); dreme_push_es(ps, PS_IN_POSITIVES, ES_ONCE); dreme_push_es(ps, PS_IN_COMMAND_LINE, ES_ONCE); } /***************************************************************************** * dreme > /model * contains: command_line, positives, negatives, background, stop, ngen, * add_pv_thresh, seed, host, when, description? ****************************************************************************/ static void end_ele_model(PS_T *ps) { if (ps->callbacks->end_model && ps->state != PS_ERROR) { ps->callbacks->end_model(ps->user_data); } } /***************************************************************************** * dreme > motifs * contains: motif* ****************************************************************************/ static void start_ele_motifs(PS_T *ps, const xmlChar **attrs) { if (ps->callbacks->start_motifs && ps->state != PS_ERROR) { ps->callbacks->start_motifs(ps->user_data); } dreme_push_es(ps, PS_IN_MOTIF, ES_ANY); } /***************************************************************************** * dreme > /motifs * contains: motif* ****************************************************************************/ static void end_ele_motifs(PS_T *ps) { if (ps->callbacks->end_motifs && ps->state != PS_ERROR) { ps->callbacks->end_motifs(ps->user_data); } } /***************************************************************************** * dreme > run_time * * cpu the time the program was running on a cpu. * real the real world time the program was running. * stop the reason dreme stopped. ****************************************************************************/ static void start_ele_run_time(PS_T *ps, const xmlChar **attrs) { double cpu_time, real_time; int stop; char* stop_options[3] = {"count", "evalue", "time"}; int stop_values[3] = {DREME_STOP_COUNT, DREME_STOP_EVALUE, DREME_STOP_TIME}; MULTI_T stop_multi = {.count = 3, .options = stop_options, .outputs = stop_values, .target = &(stop)}; char* names[3] = {"cpu", "real", "stop"}; int (*parsers[3])(char*, void*) = {ld_double, ld_double, ld_multi}; void *data[3] = {&cpu_time, &real_time, &stop_multi}; BOOLEAN_T required[3] = {TRUE, TRUE, TRUE}; BOOLEAN_T done[3]; parse_attributes(dreme_attr_parse_error, ps, "run_time", attrs, 3, names, parsers, data, required, done); if (ps->callbacks->handle_run_time && ps->state != PS_ERROR) { ps->callbacks->handle_run_time(ps->user_data, cpu_time, real_time, (DREME_STOP_EN)stop); } } /***************************************************************************** * dreme > model > /command_line * the command-line used to run the program. ****************************************************************************/ static void end_ele_command_line(PS_T *ps) { if (ps->callbacks->handle_command_line && ps->state != PS_ERROR) { ps->callbacks->handle_command_line(ps->user_data, ps->characters.buffer); } } /***************************************************************************** * dreme > model > positives * * name the name of the positive input set * count the number of sequences in the positive input set * file the file containing the positive input set * last_mod_date the time the positive input set was last modified ****************************************************************************/ static void start_ele_positives(PS_T *ps, const xmlChar **attrs) { char *name, *file, *lastmod; long count; char* names[4] = {"count", "file", "last_mod_date", "name"}; int (*parsers[4])(char*, void*) = {ld_long, ld_str, ld_str, ld_str}; void *data[4] = {&count, &file, &lastmod, &name}; BOOLEAN_T required[4] = {TRUE, TRUE, TRUE, TRUE}; BOOLEAN_T done[4]; parse_attributes(dreme_attr_parse_error, ps, "positives", attrs, 4, names, parsers, data, required, done); if (ps->callbacks->handle_positives && ps->state != PS_ERROR) { ps->callbacks->handle_positives(ps->user_data, name, count, file, lastmod); } } /***************************************************************************** * dreme > model > negatives * * name the name of the negative dataset * count the number of sequences in the negative dataset * from the source of the negative dataset (eg shuffled positives) * file the file containing the negative dataset (optional) * last_mod_date the last modified date of the file (optional) ****************************************************************************/ static void start_ele_negatives(PS_T *ps, const xmlChar **attrs) { char *name, *file, *lastmod; long count; int from; file = NULL; lastmod = NULL; char* from_options[2] = {"file", "shuffled"}; int from_values[2] = {DREME_NEG_FILE, DREME_NEG_SHUFFLED}; MULTI_T from_multi = {.count = 2, .options = from_options, .outputs = from_values, .target = &(from)}; char* names[5] = {"count", "file", "from", "last_mod_date", "name"}; int (*parsers[5])(char*, void*) = {ld_long, ld_str, ld_multi, ld_str, ld_str}; void *data[5] = {&count, &file, &from_multi, &lastmod, &name}; BOOLEAN_T required[5] = {TRUE, FALSE, TRUE, FALSE, TRUE}; BOOLEAN_T done[5]; parse_attributes(dreme_attr_parse_error, ps, "negatives", attrs, 5, names, parsers, data, required, done); if (ps->state != PS_ERROR && from == DREME_NEG_FILE) { if (file == NULL) { dreme_attr_parse_error(ps, PARSE_ATTR_MISSING, "negatives", "file", NULL); } if (lastmod == NULL) { dreme_attr_parse_error(ps, PARSE_ATTR_MISSING, "negatives", "last_mod_date", NULL); } } if (ps->callbacks->handle_negatives && ps->state != PS_ERROR) { ps->callbacks->handle_negatives(ps->user_data, name, count, (DREME_NEG_EN)from, file, lastmod); } } /***************************************************************************** * DREME > model > alphabet ****************************************************************************/ static void start_ele_alphabet(PS_T *ps, const xmlChar **attrs) { char *name; int extends; char* extends_options[3] = {"dna", "protein", "rna"}; int extends_values[3] = {ALPH_FLAG_EXTENDS_DNA, ALPH_FLAG_EXTENDS_PROTEIN, ALPH_FLAG_EXTENDS_RNA}; MULTI_T extends_multi = {.count = 3, .options = extends_options, .outputs = extends_values, .target = &(extends)}; char* names[2] = {"like", "name"}; int (*parsers[2])(char*, void*) = {ld_multi, ld_str}; void *data[2] = {&extends_multi, &name}; BOOLEAN_T required[2] = {FALSE, FALSE}; BOOLEAN_T done[2]; // just so we know later on when reading the background which used to set the alphabet ps->seen_alphabet = true; // defaults name = NULL; extends = 0; parse_attributes(dreme_attr_parse_error, ps, "alphabet", attrs, 2, names, parsers, data, required, done); if (ps->callbacks->start_alphabet && ps->state != PS_ERROR) { ps->callbacks->start_alphabet(ps->user_data, name, extends); } dreme_push_es(ps, PS_IN_ALPHABET_LETTER, ES_ONE_OR_MORE); } /***************************************************************************** * DREME > model > /alphabet ****************************************************************************/ static void end_ele_alphabet(PS_T *ps) { if (ps->callbacks->end_alphabet && ps->state != PS_ERROR) { ps->callbacks->end_alphabet(ps->user_data); } } /***************************************************************************** * DREME > model > alphabet > letter ****************************************************************************/ static void start_ele_alphabet_letter(PS_T *ps, const xmlChar **attrs) { char *aliases, *id, *name, *equals, symbol, complement; int colour, idx; char* names[7] = {"aliases", "colour", "complement", "equals", "id", "name", "symbol"}; int (*parsers[7])(char*, void*) = {ld_str, ld_hex, ld_char, ld_str, ld_str, ld_str, ld_char}; void *data[7] = {&aliases, &colour, &complement, &equals, &id, &name, &symbol}; BOOLEAN_T required[7] = {FALSE, FALSE, FALSE, FALSE, TRUE, FALSE, TRUE}; BOOLEAN_T done[7]; aliases = NULL; name = NULL; equals = NULL; complement = '\0'; colour = -1; parse_attributes(dreme_attr_parse_error, ps, "letter", attrs, 7, names, parsers, data, required, done); if (ps->seen_ambig) { if (equals == NULL) { error(ps, "All core symbols must appear before any ambigous symbols.\n"); } } else if (equals == NULL) { idx = rbtree_size(ps->alph_ids); rbtree_make(ps->alph_ids, id, &idx); } else { ps->seen_ambig = true; } if (ps->callbacks->handle_alphabet_letter && ps->state != PS_ERROR) { ps->callbacks->handle_alphabet_letter(ps->user_data, id, symbol, aliases, complement, equals, name, colour); } } /***************************************************************************** * DREME > model > strands\ ****************************************************************************/ void end_ele_strands(PS_T *ps) { DREME_STRANDS_EN strands = DREME_STRANDS_GIVEN; // stop compiler complaining if (strcmp("both", ps->characters.buffer) == 0) { strands = DREME_STRANDS_BOTH; } else if (strcmp("given", ps->characters.buffer) == 0) { strands = DREME_STRANDS_GIVEN; } else if (strcmp("none", ps->characters.buffer) == 0) { strands = DREME_STRANDS_NONE; } else { error(ps, "Strands value must be both, given or none.\n"); } if (ps->callbacks->handle_strands && ps->state != PS_ERROR) { ps->callbacks->handle_strands(ps->user_data, strands); } } /***************************************************************************** * DREME > model > norc\ ****************************************************************************/ void end_ele_norc(PS_T *ps) { DREME_STRANDS_EN strands = DREME_STRANDS_GIVEN; // stop compiler complaining if (strcmp("FALSE", ps->characters.buffer) == 0) { strands = DREME_STRANDS_BOTH; } else if (strcmp("TRUE", ps->characters.buffer) == 0) { strands = DREME_STRANDS_GIVEN; } if (ps->callbacks->handle_strands && ps->state != PS_ERROR) { ps->callbacks->handle_strands(ps->user_data, strands); } } /***************************************************************************** * Reads frequency attributes into the pre-allocated freqs array. ****************************************************************************/ static void parse_freq_attrs(PS_T *ps, const char* tag, const xmlChar **attrs) { int i, ncore, seen, *idx; char *end_ptr; double value, sum; RBNODE_T *node; bool seen_bad; ncore = rbtree_size(ps->alph_ids); // initilize the freqs array if (ps->freqs == NULL) ps->freqs = mm_malloc(sizeof(double) * ncore); // reset freqs array; for (i = 0; i < ncore; i++) ps->freqs[i] = -1; seen = 0; seen_bad = false; sum = 0.0; // iterate over attributes for (i = 0; attrs[i] != NULL; i += 2) { idx = (int*)rbtree_get(ps->alph_ids, attrs[i]); if (idx != NULL) { assert(*idx < ncore); if (ps->freqs[*idx] != -1) { dreme_attr_parse_error(ps, PARSE_ATTR_DUPLICATE, tag, (const char*)attrs[i], NULL); continue; } seen++; errno = 0; // reset because we're about to check it value = strtod((const char*)attrs[i+1], &end_ptr); // allow out of range values, mainly because freqs can be very close to zero if (end_ptr == (const char*)attrs[i+1] || (errno && errno != ERANGE) || value < 0 || value > 1) { dreme_attr_parse_error(ps, PARSE_ATTR_BAD_VALUE, tag, (const char*)attrs[i], (const char*)attrs[i+1]); ps->freqs[*idx] = 0; // mark frequence as seen, even though it's bad seen_bad = true; continue; } ps->freqs[*idx] = value; sum += value; } } // check we got everthing if (seen < ncore) { // identify what we're missing for (node = rbtree_first(ps->alph_ids); node != NULL; node = rbtree_next(node)) { idx = (int*)rbtree_value(node); if (ps->freqs[*idx] == -1) { dreme_attr_parse_error(ps, PARSE_ATTR_MISSING, tag, (char*)rbtree_key(node), NULL); } } } else if (!seen_bad) { // check the frequencies sum to 1 double delta = sum - 1; delta = (delta < 0 ? -delta : delta); if (delta > (0.001 * ncore)) { // dreme writes background probabilities to 3 decimal places so assuming // the error on each is at maximum 0.001 then the total error for the // sum must be less than or equal to 0.004 error(ps, "Probabilities of %s do not sum to 1, got %g .\n", tag, sum); } } } /***************************************************************************** * dreme > model > background * * type is the alphabet DNA or RNA (optional when custom alphabet specified) * frequency of from core alphabet * from from the negative dataset or a background file * file the background file (optional) * last_mod_date the last modified date of the background file (optional) ****************************************************************************/ static void start_ele_background(PS_T *ps, const xmlChar **attrs) { int type, from; char *file, *lastmod; // set reasonable defaults type = DREME_BG_FROM_DATASET; from = DREME_ALPH_DNA; file = NULL; lastmod = NULL; char* type_options[2] = {"dna", "rna"}; int type_values[2] = {DREME_ALPH_DNA, DREME_ALPH_RNA}; MULTI_T type_multi = {.count = 2, .options = type_options, .outputs = type_values, .target = &(type)}; char* from_options[2] = {"dataset", "file"}; int from_values[2] = {DREME_BG_FROM_DATASET, DREME_BG_FROM_FILE}; MULTI_T from_multi = {.count = 2, .options = from_options, .outputs = from_values, .target = &(from)}; char* names[4] = {"file", "from", "last_mod_date", "type"}; int (*parsers[4])(char*, void*) = {ld_str, ld_multi, ld_str, ld_multi}; void *data[4] = {&file, &from_multi, &lastmod, &type_multi}; BOOLEAN_T required[4] = {FALSE, TRUE, FALSE, FALSE}; BOOLEAN_T done[4]; required[3] = !ps->seen_alphabet; parse_attributes(dreme_attr_parse_error, ps, "background", attrs, 4, names, parsers, data, required, done); if (from == DREME_BG_FROM_FILE) { if (!done[0]) dreme_attr_parse_error(ps, PARSE_ATTR_MISSING, "background", "file", NULL); if (!done[2]) dreme_attr_parse_error(ps, PARSE_ATTR_MISSING, "background", "last_mod_date", NULL); } // if we haven't seen the alphabet then we must define it from the type if (!ps->seen_alphabet) { int idx = 0; rbtree_make(ps->alph_ids, "A", &idx); idx++; rbtree_make(ps->alph_ids, "C", &idx); idx++; rbtree_make(ps->alph_ids, "G", &idx); idx++; rbtree_make(ps->alph_ids, (type == DREME_ALPH_DNA ? "T" : "U"), &idx); } parse_freq_attrs(ps, "background", attrs); if (ps->callbacks->handle_background && ps->state != PS_ERROR) { ps->callbacks->handle_background(ps->user_data, rbtree_size(ps->alph_ids), ps->freqs, from, file, lastmod); } } /***************************************************************************** * dreme > model > stop * * evalue the stopping evalue (returned as log10). * count the stopping count. * time the stopping time. ****************************************************************************/ static void start_ele_stop(PS_T *ps, const xmlChar **attrs) { int count, time; double log10evalue; char* names[3] = {"count", "evalue", "time"}; int (*parsers[3])(char*, void*) = {ld_int, ld_log10_ev, ld_int}; void *data[3] = {&count, &log10evalue, &time}; BOOLEAN_T required[3] = {FALSE, FALSE, FALSE}; BOOLEAN_T done[3]; parse_attributes(dreme_attr_parse_error, ps, "stop", attrs, 3, names, parsers, data, required, done); if (ps->callbacks->handle_stop && ps->state != PS_ERROR) { ps->callbacks->handle_stop(ps->user_data, &log10evalue, &count, &time); } } /***************************************************************************** * dreme > model > /ngen * the number of generations to check (or something like that). ****************************************************************************/ static void end_ele_ngen(PS_T *ps) { int ngen; if (ld_int(ps->characters.buffer, &ngen)) { error(ps, "Bad value \"%s\" for ngen.\n", ps->characters.buffer); } if (ps->callbacks->handle_ngen && ps->state != PS_ERROR) { ps->callbacks->handle_ngen(ps->user_data, ngen); } } /***************************************************************************** * dreme > model > /add_pv_thresh * the p-value threshold (returned as log10) for adding a word to the list of * possible motif seed values (this description is probably wrong). ****************************************************************************/ static void end_ele_add_pv_thresh(PS_T *ps) { double log10_add_pv_thresh; if (ld_log10_pv(ps->characters.buffer, &log10_add_pv_thresh)) { error(ps, "Bad value \"%s\" for add_pv_thresh.\n", ps->characters.buffer); } if (ps->callbacks->handle_add_pv_thresh && ps->state != PS_ERROR) { ps->callbacks->handle_add_pv_thresh(ps->user_data, log10_add_pv_thresh); } } /***************************************************************************** * dreme > model > /seed * the seed value used by the pseudo-random number generator. ****************************************************************************/ static void end_ele_seed(PS_T *ps) { long seed; if (ld_long(ps->characters.buffer, &seed)) { error(ps, "Bad value \"%s\" for seed.\n", ps->characters.buffer); } if (ps->callbacks->handle_seed && ps->state != PS_ERROR) { ps->callbacks->handle_seed(ps->user_data, seed); } } /***************************************************************************** * dreme > model > /host * the name of the computer which ran DREME. ****************************************************************************/ static void end_ele_host(PS_T *ps) { if (ps->callbacks->handle_host && ps->state != PS_ERROR) { ps->callbacks->handle_host(ps->user_data, ps->characters.buffer); } } /***************************************************************************** * dreme > model > /when * the time DREME was run. ****************************************************************************/ static void end_ele_when(PS_T *ps) { if (ps->callbacks->handle_when && ps->state != PS_ERROR) { ps->callbacks->handle_when(ps->user_data, ps->characters.buffer); } } /***************************************************************************** * dreme > model > /description * an optional description of the experiment. ****************************************************************************/ static void end_ele_description(PS_T *ps) { if (ps->callbacks->handle_description && ps->state != PS_ERROR) { ps->callbacks->handle_description(ps->user_data, ps->characters.buffer); } } /***************************************************************************** * dreme > motifs > motif * contains: pos+, match* * * id the identifier used by DREME * seq the DNA iupac sequence representing the motif. * length the length of the motif * nsites the number of sites used to create the motif * p the number of sequences in the positive set with the motif * n the number of sequences in the negative set with the motif * pvalue the pvalue of the motif after erasing (returned as log10) * evalue the evalue of the motif after erasing (returned as log10) * unerased_evalue the evalue of the motif without erasing (returned as log10) ****************************************************************************/ static void start_ele_motif(PS_T *ps, const xmlChar **attrs) { char *id, *seq; int length; long nsites, p_hits, n_hits; double log10pvalue, log10evalue, log10uevalue; char* names[9] = {"evalue", "id", "length", "n", "nsites", "p", "pvalue", "seq", "unerased_evalue"}; int (*parsers[9])(char*, void*) = {ld_log10_ev, ld_str, ld_int, ld_long, ld_long, ld_long, ld_log10_pv, ld_str, ld_log10_ev}; void *data[9] = {&log10evalue, &id, &length, &n_hits, &nsites, &p_hits, &log10pvalue, &seq, &log10uevalue}; BOOLEAN_T required[9] = {TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE}; BOOLEAN_T done[9]; parse_attributes(dreme_attr_parse_error, ps, "motif", attrs, 9, names, parsers, data, required, done); // copy the motif id so we can use it in any error messages if (ps->state != PS_ERROR) { int len = strlen(id); ps->motif_id = mm_malloc(sizeof(char) * (len + 1)); strcpy(ps->motif_id, id); ps->last_pos = 0; ps->motif_len = length; } if (ps->callbacks->start_motif && ps->state != PS_ERROR) { ps->callbacks->start_motif(ps->user_data, id, seq, length, nsites, p_hits, n_hits, log10pvalue, log10evalue, log10uevalue); } dreme_push_es(ps, PS_IN_MATCH, ES_ANY); dreme_push_es(ps, PS_IN_POS, ES_ONE_OR_MORE); } /***************************************************************************** * dreme > motifs > /motif * contains: pos+, match* ****************************************************************************/ static void end_ele_motif(PS_T *ps) { if (ps->state != PS_ERROR && ps->last_pos != ps->motif_len) { error(ps, "Motif %s has length %d which does not match the " "number of pos elements %d.", ps->motif_len, ps->last_pos); } if (ps->callbacks->end_motif && ps->state != PS_ERROR) { ps->callbacks->end_motif(ps->user_data); } // free stored motif id if (ps->motif_id) free(ps->motif_id); ps->motif_id = NULL; } /***************************************************************************** * dreme > motifs > motif > pos * * i index of the motif position (optional) * frequency of ****************************************************************************/ static void start_ele_pos(PS_T *ps, const xmlChar **attrs) { int pos = ps->last_pos + 1; if (!ps->seen_alphabet) { // attribute "i" only exists in the older specification without custom alphabets char* names[1] = {"i"}; int (*parsers[1])(char*, void*) = {ld_int}; void *data[1] = {&pos}; BOOLEAN_T required[1] = {TRUE}; BOOLEAN_T done[1]; parse_attributes(dreme_attr_parse_error, ps, "pos", attrs, 1, names, parsers, data, required, done); if (pos != (ps->last_pos + 1)) { error(ps, "Motif %s did not have pos %d but instead " "has pos %d in its place.\n", ps->motif_id, ps->last_pos + 1, pos); } } ps->last_pos = pos; parse_freq_attrs(ps, "pos", attrs); if (ps->callbacks->handle_pos && ps->state != PS_ERROR) { ps->callbacks->handle_pos(ps->user_data, pos, rbtree_size(ps->alph_ids), ps->freqs); } } /***************************************************************************** * dreme > motifs > motif > match * * seq word which matches motif * p number of positive sequences which contain this word * n number of negative sequences which contain this word * pvalue the pvalue of this word (returned as log10) * evalue the evalue of this word (returned as log10) ****************************************************************************/ static void start_ele_match(PS_T *ps, const xmlChar **attrs) { char *seq; long p_hits, n_hits; double log10pvalue, log10evalue; char* names[5] = {"evalue", "n", "p", "pvalue", "seq"}; int (*parsers[5])(char*, void*) = {ld_log10_ev, ld_long, ld_long, ld_log10_pv, ld_str}; void *data[5] = {&log10evalue, &n_hits, &p_hits, &log10pvalue, &seq}; BOOLEAN_T required[5] = {TRUE, TRUE, TRUE, TRUE, TRUE}; BOOLEAN_T done[5]; parse_attributes(dreme_attr_parse_error, ps, "match", attrs, 5, names, parsers, data, required, done); if (ps->callbacks->handle_match && ps->state != PS_ERROR) { ps->callbacks->handle_match(ps->user_data, seq, p_hits, n_hits, log10pvalue, log10evalue); } } /***************************************************************************** * Handle the document start ****************************************************************************/ void handle_dreme_start_doc(void *ctx) { PS_T *ps = (PS_T*)ctx; ps->state = PS_START; dreme_push_es(ps, PS_END, ES_ONCE); dreme_push_es(ps, PS_IN_DREME, ES_ONCE); } /***************************************************************************** * Handle the document end ****************************************************************************/ void handle_dreme_end_doc(void *ctx) { PS_T *ps = (PS_T*)ctx; ES_T *es; if (ps->state != PS_ERROR) dreme_update_es(ps, PS_END); while ((es = (ES_T*)linklst_pop(ps->expected_stack)) != NULL) { free(es); } } /***************************************************************************** * Handle the characters within elements. Only accepts characters * that pass the character filter. If a filter is not set then it doesn't * store anything. ****************************************************************************/ void handle_dreme_characters(void *ctx, const xmlChar *ch, int len) { PS_T *ps = (PS_T*)ctx; if (ps->state == PS_ERROR) return; if (ps->udepth) {//we don't know where we are! } else { store_xml_characters(&(ps->characters), (const char*)ch, len); } } /***************************************************************************** * Macros to greatly reduce the boilerplate code needed to keep track of the * parser state. * * DO_START_ELE compares the tag name to a possible option and if it matches * then validates the state stack to ensure that it really is a possible * option. If everything passes then it changes the state to the specified * transition state and calls the callback. It also changes the character * accept function. * CHECK_START_ELE does the same except it does not have a callback function. ****************************************************************************/ #define DO_START_ELE(_expected_,_call_,_transition_,_char_accept_) \ if (strcasecmp((char*)name, #_expected_ ) == 0) { \ if (dreme_update_es(ps, _transition_ )) { \ ps->state = _transition_; \ _call_ (ps, translate_attributes(&(ps->attrbuf), attrs)); \ ps->characters.accept = _char_accept_; \ } \ break; \ } #define CHECK_START_ELE(_expected_,_transition_,_char_accept_) \ if (strcasecmp((char*)name, #_expected_ ) == 0) { \ if (dreme_update_es(ps, _transition_ )) { \ ps->state = _transition_; \ ps->characters.accept = _char_accept_; \ } \ break; \ } /***************************************************************************** * Handle a starting element. ****************************************************************************/ void handle_dreme_start_ele(void *ctx, const xmlChar *name, const xmlChar **attrs) { PS_T *ps = (PS_T*)ctx; int known, allow_unknown; if (ps->state == PS_ERROR) return; if (ps->udepth) {//we don't know where we are! ps->udepth += 1; } else { //reset the character buffer to the begining ps->characters.pos = 0; ps->characters.buffer[0] = '\0'; known = 1; //assume we can find it allow_unknown = 1; //are unknowns allowed in this state? switch (ps->state) { case PS_START: DO_START_ELE(dreme, start_ele_dreme, PS_IN_DREME, IGNORE); known = 0; break; case PS_IN_DREME: DO_START_ELE(model, start_ele_model, PS_IN_MODEL, IGNORE); DO_START_ELE(motifs, start_ele_motifs, PS_IN_MOTIFS, IGNORE); DO_START_ELE(run_time, start_ele_run_time, PS_IN_RUN_TIME, IGNORE); known = 0; break; case PS_IN_MODEL: CHECK_START_ELE(command_line, PS_IN_COMMAND_LINE, ALL_CHARS); DO_START_ELE(positives, start_ele_positives, PS_IN_POSITIVES, IGNORE); DO_START_ELE(negatives, start_ele_negatives, PS_IN_NEGATIVES, IGNORE); DO_START_ELE(alphabet, start_ele_alphabet, PS_IN_ALPHABET, IGNORE); CHECK_START_ELE(strands, PS_IN_STRANDS, ALL_BUT_SPACE); DO_START_ELE(background, start_ele_background, PS_IN_BACKGROUND, IGNORE); DO_START_ELE(stop, start_ele_stop, PS_IN_STOP, IGNORE); CHECK_START_ELE(norc, PS_IN_NORC, ALL_BUT_SPACE); // replaced by strands in 4.11 CHECK_START_ELE(ngen, PS_IN_NGEN, ALL_BUT_SPACE); CHECK_START_ELE(add_pv_thresh, PS_IN_ADD_PV_THRESH, ALL_BUT_SPACE); CHECK_START_ELE(seed, PS_IN_SEED, ALL_BUT_SPACE); CHECK_START_ELE(host, PS_IN_HOST, ALL_BUT_SPACE); CHECK_START_ELE(when, PS_IN_WHEN, ALL_CHARS); CHECK_START_ELE(description, PS_IN_DESCRIPTION, ALL_CHARS); break; case PS_IN_ALPHABET: DO_START_ELE(letter, start_ele_alphabet_letter, PS_IN_ALPHABET_LETTER, IGNORE); break; case PS_IN_MOTIFS: DO_START_ELE(motif, start_ele_motif, PS_IN_MOTIF, IGNORE); known = 0; break; case PS_IN_MOTIF: DO_START_ELE(pos, start_ele_pos, PS_IN_POS, IGNORE); DO_START_ELE(match, start_ele_match, PS_IN_MATCH, IGNORE); known = 0; break; default: known = 0; break; } if (!known) { if (allow_unknown) { ps->udepth = 1; } else { error(ps, "Encountered illegal tag %s while in state %s\n",(char*)name, state_names[ps->state]); } } } } /***************************************************************************** * Macros to greatly reduce the boilerplate code needed to keep track of the * parser state. * * DO_END_ELE compares the state and if it matches then tries to match the * element name. If that passes then it calls the callback and if that * succeeds it transitions to the parent state. * CHECK_END_ELE does the same except it does not have a callback function. ****************************************************************************/ #define DO_END_ELE(_expected_,_call_,_state_,_transition_) \ case _state_: \ if (strcasecmp((char*)name, #_expected_) == 0) { \ known = 1; \ _call_ (ps); \ if (ps->state == _state_) ps->state = _transition_; \ } \ break #define CHECK_END_ELE(_expected_,_state_,_transition_) \ case _state_: \ if (strcasecmp((char*)name, #_expected_) == 0) { \ known = 1; \ ps->state = _transition_; \ } \ break /***************************************************************************** * Handle a closing element ****************************************************************************/ void handle_dreme_end_ele(void *ctx, const xmlChar *name) { PS_T *ps = (PS_T*)ctx; int known = 0; if (ps->state == PS_ERROR) return; if (ps->udepth) { ps->udepth -= 1; } else { switch (ps->state) { DO_END_ELE(dreme, end_ele_dreme, PS_IN_DREME, PS_END); DO_END_ELE(model, end_ele_model, PS_IN_MODEL, PS_IN_DREME); DO_END_ELE(motifs, end_ele_motifs, PS_IN_MOTIFS, PS_IN_DREME); CHECK_END_ELE(run_time, PS_IN_RUN_TIME, PS_IN_DREME); DO_END_ELE(command_line, end_ele_command_line, PS_IN_COMMAND_LINE, PS_IN_MODEL); CHECK_END_ELE(positives, PS_IN_POSITIVES, PS_IN_MODEL); CHECK_END_ELE(negatives, PS_IN_NEGATIVES, PS_IN_MODEL); DO_END_ELE(alphabet, end_ele_alphabet, PS_IN_ALPHABET, PS_IN_MODEL); CHECK_END_ELE(letter, PS_IN_ALPHABET_LETTER, PS_IN_ALPHABET); CHECK_END_ELE(strands, PS_IN_STRANDS, PS_IN_MODEL); CHECK_END_ELE(background, PS_IN_BACKGROUND, PS_IN_MODEL); CHECK_END_ELE(stop, PS_IN_STOP, PS_IN_MODEL); DO_END_ELE(norc, end_ele_norc, PS_IN_NORC, PS_IN_MODEL); DO_END_ELE(ngen, end_ele_ngen, PS_IN_NGEN, PS_IN_MODEL); DO_END_ELE(add_pv_thresh, end_ele_add_pv_thresh, PS_IN_ADD_PV_THRESH, PS_IN_MODEL); DO_END_ELE(seed, end_ele_seed, PS_IN_SEED, PS_IN_MODEL); DO_END_ELE(host, end_ele_host, PS_IN_HOST, PS_IN_MODEL); DO_END_ELE(when, end_ele_when, PS_IN_WHEN, PS_IN_MODEL); DO_END_ELE(description, end_ele_description, PS_IN_DESCRIPTION, PS_IN_MODEL); DO_END_ELE(motif, end_ele_motif, PS_IN_MOTIF, PS_IN_MOTIFS); CHECK_END_ELE(pos, PS_IN_POS, PS_IN_MOTIF); CHECK_END_ELE(match, PS_IN_MATCH, PS_IN_MOTIF); } if (!known) { error(ps, "Unexpected end tag %s in state %s\n", (char*)name, state_names[ps->state]); } } } /***************************************************************************** * Register handlers on the xmlSAXHandler structure ****************************************************************************/ void register_dreme_io_xml_sax_handlers(xmlSAXHandler *handler) { memset(handler, 0, sizeof(xmlSAXHandler)); handler->startDocument = handle_dreme_start_doc; handler->endDocument = handle_dreme_end_doc; handler->characters = handle_dreme_characters; handler->startElement = handle_dreme_start_ele; handler->endElement = handle_dreme_end_ele; } /***************************************************************************** * Creates the data to be passed to the SAX parser ****************************************************************************/ void* create_dreme_io_xml_sax_context(void *user_data, DREME_IO_XML_CALLBACKS_T *callbacks) { PS_T *ps; CHARBUF_T *buf; ps = (PS_T*)mm_malloc(sizeof(PS_T)); memset(ps, 0, sizeof(PS_T)); ps->state = PS_START; ps->udepth = 0; ps->callbacks = callbacks; ps->user_data = user_data; ps->seen_alphabet = false; ps->seen_ambig = false; ps->alph_ids = rbtree_create(rbtree_strcmp, rbtree_strcpy, free, rbtree_intcpy, free); ps->freqs = NULL; ps->motif_id = NULL; ps->last_pos = 0; //set up character buffer buf = &(ps->characters); buf->buffer = mm_malloc(sizeof(char)*10); buf->buffer[0] = '\0'; buf->size = 10; buf->pos = 0; attrbuf_init(&(ps->attrbuf)); //set up expected queue ps->expected_stack = linklst_create(); return ps; } /***************************************************************************** * Destroys the data used by the SAX parser ****************************************************************************/ void destroy_dreme_io_xml_sax_context(void *ctx) { PS_T *ps = (PS_T*)ctx; if (ps->motif_id) free(ps->motif_id); if (ps->freqs) free(ps->freqs); rbtree_destroy(ps->alph_ids); free(ps->characters.buffer); attrbuf_free(&(ps->attrbuf)); linklst_destroy_all(ps->expected_stack, free); free(ps); }