From 4eabf5d72c6b01bbf11180280ef9d28d5fe587bf Mon Sep 17 00:00:00 2001
From: Nicholas Noll <nbnoll@eml.cc>
Date: Tue, 28 Apr 2020 11:59:59 -0700
Subject: feat: added number of nodes & leafs to tree data structure

---
 sys/libbio/io/newick.c | 50 ++++++++++++++++++++++++++++++++++----------------
 1 file changed, 34 insertions(+), 16 deletions(-)

(limited to 'sys/libbio/io/newick.c')

diff --git a/sys/libbio/io/newick.c b/sys/libbio/io/newick.c
index 0c9921b..81b5195 100644
--- a/sys/libbio/io/newick.c
+++ b/sys/libbio/io/newick.c
@@ -65,20 +65,22 @@ tokstr(struct Token tok)
 // TODO: Bounds checking on buffer
 static
 struct Token
-lex(io·Peeker s, void* impl)
+lex(io·Peeker s, void* fp)
 {
+#define isvalidchar(C) (' ' < (C) && (C) < ':') || (':' < (C) && (C) <= '~')
+
     byte *c;
     struct Token tok;
     static byte b[1024];
     c  = b;
-    *c = s.get(impl);
+    *c = s.get(fp);
 
     if (isspace(*c)) {
         while (isspace(*c)) {
-            *(++c) = s.get(impl);
+            *(++c) = s.get(fp);
         }
 
-        s.unget(impl, *c);
+        s.unget(fp, *c);
         Assert(c - b < 1024);
 
         *c        = 0;
@@ -101,11 +103,11 @@ lex(io·Peeker s, void* impl)
     case '0': case '1': case '2': case '3': case '4':
     case '5': case '6': case '7': case '8': case '9':
         while (isdigit(*c)) {
-    NUM:    *(++c) = s.get(impl);
+    NUM:    *(++c) = s.get(fp);
         }
         if (*c == '.') goto NUM;
 
-        s.unget(impl, *c);
+        s.unget(fp, *c);
         Assert(c - b < 1024);
 
         *c        = 0;
@@ -114,11 +116,12 @@ lex(io·Peeker s, void* impl)
         return tok;
 
     default:
-        while (isalnum(*c)) {
-            *(++c) = s.get(impl);
+        while (isvalidchar(*c)) { 
+            *(++c) = s.get(fp);
         }
+        printf("ended on character %c\n", *c);
 
-        s.unget(impl, *c);
+        s.unget(fp, *c);
         Assert(c - b < 1024);
 
         *c        = '\0';
@@ -126,6 +129,7 @@ lex(io·Peeker s, void* impl)
         tok.lit.s = b;
         return tok;
     }
+#undef isvalidchar
 }
 
 static
@@ -248,7 +252,6 @@ parse(struct Parser *p)
                     goto ERROR;
                 }
                 node->name = str·new(tok.lit.s);
-                printf("settting name %s\n", node->name);
             } else {
                 if (p->tok.kind != tok·lparen && p->tok.kind != tok·comma) {
                     errorf("format error: misplaced identifier for leaf found");
@@ -308,12 +311,21 @@ ERROR:
     return 1;
 }
 
-bio·Tree
-bio·readnewick(io·Peeker stream, void *s, mem·Allocator heap, void *h)
+error
+bio·readnewick(io·Peeker stream, void *s, mem·Allocator heap, void *h, bio·Tree *tree)
 {
     error err;
     struct Parser p;
-    bio·Tree tree;
+    enum 
+    {
+        error·nil,
+        error·notree,
+        error·parse,
+    };
+
+    if (!tree) {
+        return error·notree;
+    }
 
     p = (struct Parser){ 
         .lev   = 0,
@@ -327,11 +339,17 @@ bio·readnewick(io·Peeker stream, void *s, mem·Allocator heap, void *h)
     err = parse(&p);
     if (err) {
         errorf("parsing failed\n");
-        return tree;
+        return error·parse;
     }
-    tree.root = p.root;
 
-    return tree;
+    tree->root  = p.root;
+    tree->nleaf = 0;
+    tree->nnode = 0;
+
+    phylo·countnodes(tree->root, &tree->nnode);
+    phylo·countleafs(tree->root, &tree->nleaf);
+
+    return error·nil;
 }
 
 // -----------------------------------------------------------------------
-- 
cgit v1.2.1