123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344 |
- <?php
- function tripal_phylogeny_parse_newick_file($file_name) {
-
- global $tripal_phylogeny_bootstrap;
- $tripal_phylogeny_bootstrap = 1;
- $tripal_phylogeny_index = 1;
- $tree = [];
- $fp = fopen($file_name, 'r');
- if ($fp) {
- $tree = tripal_phylogeny_parse_newick_tree($fp);
- }
- else {
-
- }
- return $tree;
- }
- function tripal_phylogeny_parse_newick_tree($fp, $depth = 0) {
- $subtree = tripal_phylogeny_parse_newick_subtree($fp, $depth);
- $subtree['is_root'] = 1;
-
-
- $token = tripal_phylogeny_parse_newick_get_token($fp);
- if ($token == ";") {
-
- return $subtree;
- }
- tripal_phylogeny_parse_newick_replace_token($fp);
-
- $length = tripal_phylogeny_parse_newick_length($fp, $depth);
- $subtree['length'] = $length;
-
- $token = tripal_phylogeny_parse_newick_get_token($fp);
- if ($token != ';') {
- print "Syntax Error: missing trailing semicolon.\n";
- exit;
- }
- return $subtree;
- }
- function tripal_phylogeny_parse_newick_subtree($fp, $depth) {
- $internal = tripal_phylogeny_parse_newick_internal($fp, $depth + 1);
- if (!is_array($internal)) {
- $leaf_node = tripal_phylogeny_parse_newick_leaf($fp, $depth);
- return [
- 'name' => $leaf_node,
- 'depth' => $depth,
- 'is_leaf' => TRUE,
- 'descendents' => 0,
- ];
- }
- else {
- $internal['depth'] = $depth;
- }
- return $internal;
- }
- function tripal_phylogeny_parse_newick_branch($fp, $depth) {
- $subtree = tripal_phylogeny_parse_newick_subtree($fp, $depth);
- $length = tripal_phylogeny_parse_newick_length($fp, $depth);
- $subtree['length'] = $length;
- return $subtree;
- }
- function tripal_phylogeny_parse_newick_internal($fp, $depth) {
-
- if (tripal_phylogeny_parse_newick_get_token($fp) != '(') {
- tripal_phylogeny_parse_newick_replace_token($fp);
- return FALSE;
- }
- $branches = tripal_phylogeny_parse_newick_branchset($fp, $depth);
- if (!is_array($branches)) {
- return FALSE;
- }
-
- if (tripal_phylogeny_parse_newick_get_token($fp) != ')') {
- tripal_phylogeny_parse_newick_replace_token($fp);
- return FALSE;
- }
- $internal_node = tripal_phylogeny_parse_newick_name($fp, $depth);
- $descendent_count = 0;
- for ($i = 0; $i < count($branches); $i++) {
- $branches[$i]['parent'] = $internal_node;
- $descendent_count += 1 + $branches[$i]['descendents'];
- }
- return [
- 'name' => $internal_node,
- 'depth' => $depth,
- 'branch_set' => $branches,
- 'is_internal' => TRUE,
- 'descendents' => $descendent_count,
- ];
- }
- function tripal_phylogeny_parse_newick_branchset($fp, $depth) {
- $branches = [];
- $num_read = 0;
- $branch = tripal_phylogeny_parse_newick_branch($fp, $depth);
- $branches[] = $branch;
-
-
- if (!is_array($branch)) {
- return FALSE;
- }
-
-
- $token = tripal_phylogeny_parse_newick_get_token($fp);
- if ($token == ',') {
- $rbranches = tripal_phylogeny_parse_newick_branchset($fp, $depth);
- foreach ($rbranches as $branch) {
- $branches[] = $branch;
- }
- }
- else {
- tripal_phylogeny_parse_newick_replace_token($fp);
- }
- return $branches;
- }
- function tripal_phylogeny_parse_newick_leaf($fp, $depth) {
- return tripal_phylogeny_parse_newick_name($fp, $depth);
- }
- function tripal_phylogeny_parse_newick_name($fp, $depth) {
- global $tripal_phylogeny_bootstrap;
- $token = tripal_phylogeny_parse_newick_get_token($fp);
-
-
- if ($token == ':' or $token == ',' or $token == ';' or $token == ')') {
- tripal_phylogeny_parse_newick_replace_token($fp);
-
- return $tripal_phylogeny_bootstrap++;
- }
-
- if ($token == '(') {
- tripal_phylogeny_parse_newick_replace_token($fp);
- return FALSE;
- }
- return $token;
- }
- function tripal_phylogeny_parse_newick_length($fp, $depth) {
- $length = '';
- $token = tripal_phylogeny_parse_newick_get_token($fp);
-
-
- if ($token == ',' or $token == ';' or $token == ')') {
- tripal_phylogeny_parse_newick_replace_token($fp);
- return '';
- }
-
-
- if ($token != ':') {
- tripal_phylogeny_parse_newick_replace_token($fp);
- return FALSE;
- }
-
- $token = tripal_phylogeny_parse_newick_get_token($fp);
-
- if ($token == '(') {
- exit();
- }
- return $token;
- }
- function tripal_phylogeny_parse_newick_get_token($fp) {
-
- global $tripal_phylogeny_fp_pos;
- $tripal_phylogeny_fp_pos = ftell($fp);
- $token = '';
- $in_quote = FALSE;
- $num_read = 0;
- $c = fgetc($fp);
- while (!feof($fp)) {
- $num_read++;
- switch ($c) {
-
-
-
- case ';':
- case '(':
- case ')':
- case ',':
- case ':':
- if (!$token) {
- return $c;
- }
- else {
-
- fseek($fp, $tripal_phylogeny_fp_pos + $num_read - 1);
- return $token;
- }
- break;
-
-
- case '\'':
- case '"':
- if (!$in_quote) {
- $in_quote = TRUE;
- }
- else {
- $in_quote = FALSE;
- }
- break;
- case " ":
- case "\t":
- case "\r":
- case "\n":
- if ($in_quote) {
- $token .= $c;
- }
- break;
-
- default:
- $token .= $c;
- }
- $c = fgetc($fp);
- }
- return $token;
- }
- function tripal_phylogeny_parse_newick_replace_token($fp) {
- global $tripal_phylogeny_fp_pos;
- fseek($fp, $tripal_phylogeny_fp_pos);
- $tripal_phylogeny_fp_pos = ftell($fp);
- }
|