feat: added prototype for plotting log file data

4 files changed, 50 insertions, 5 deletions

diff --git a/Makefile b/Makefile
index 0fe7c0a..b3f1d20 100644
--- a/Makefile
+++ b/Makefile
@@ -55,7 +55,7 @@ staph:
 	@echo "cluster    staph"; \
 	pangraph cluster -d data/staph data/staph/assemblies/*.fna.gz
 	@echo "build      staph"; \
-	pangraph build -d data/staph -m 500 -b 0 -e 2500 -w 100 data/staph/guide.json
+	pangraph build -d data/staph -m 500 -b 0 -e 2500 -w 1000 data/staph/guide.json 2>/dev/null 1>staph-e2500-w1000.log
 
 # figures
 
diff --git a/pangraph/graph.py b/pangraph/graph.py
index f4afa97..235ef31 100644
--- a/pangraph/graph.py
+++ b/pangraph/graph.py
@@ -495,7 +495,7 @@ class Graph(object):
 
         delta = len(blk_list)-len(shared_blks)
         if delta > 0 and num_seqs > 1:
-            print(f"LEN: {delta}", end="\t")
+            print(f">LEN: {delta}", end=';')
             fd   = [None, None]
             path = [None, None]
             try:
@@ -524,15 +524,16 @@ class Graph(object):
                                     shell=True)
                         out, err = proc.communicate()
                         tree = Phylo.read(io.StringIO(out.decode('utf-8')), format='newick')
-                        print(f"-> {n} SCORE: {tree.total_branch_length()/(2*num_seqs)}")
+                        print(f"-> {n} SCORE={tree.total_branch_length()/(2*num_seqs)}", end=";")
 
                     make_tree(0)
                     make_tree(1)
+                    print("\n", end="")
             finally:
                 os.remove(path[0])
                 os.remove(path[1])
         else:
-            print(f"NO MATCH")
+            print(f">NO MATCH")
 
         # NOTE: debugging code
         # if len(blk_list) < len(shared_blks):
diff --git a/pangraph/tree.py b/pangraph/tree.py
index eeb52c8..d36b77d 100644
--- a/pangraph/tree.py
+++ b/pangraph/tree.py
@@ -383,7 +383,7 @@ class Tree(object):
         for n in self.postorder():
             if n.is_leaf():
                 continue
-            print(f"---NODE LEVEL {n.level}---")
+            print(f"+++LEVEL={n.level}+++")
             n.fapath = f"{tmpdir}/{n.name}"
             log(f"fusing {n.child[0].name} with {n.child[1].name} @ {n.name}")
             n.graph = merge(*n.child)
diff --git a/scripts/parse_log.py b/scripts/parse_log.py
new file mode 100644
index 0000000..1caae07
--- /dev/null
+++ b/scripts/parse_log.py
@@ -0,0 +1,44 @@
+#!/usr/bin/env python3
+"""
+script to process our end repair log files for plotting
+"""
+import argparse
+from collections import defaultdict
+
+level_preset = "+++LEVEL="
+level_offset = len(level_preset)
+
+score_preset = "SCORE="
+score_offset = len(score_preset)
+
+def main(args):
+    for log_path in args:
+        with open(log_path) as log:
+            level = -1
+            stats = defaultdict(lambda: {'hits':[], 'miss': 0})
+            for line in log:
+                line.rstrip('\n')
+                if line[0] == "+":
+                    assert line.startwith(level_preset), "check syntax in log file"
+                    level = int(line[level_offset:line.find("+++", level_offset)])
+                    continue
+                if line[0] == ">":
+                    if line[1:] == "NO MATCH":
+                        stats[level]['miss'] += 1
+                        continue
+                    if line[1:].startswith("LEN="):
+                        offset = [line.find(";")]
+                        offset.append(line.find(";", offset[0]))
+                        score[0] = int(line[line.find(score_preset, offset[0])+1+score_offset:offset[1])
+                        score[1] = int(line[line.find(score_preset, offset[1])+1+score_offset:)
+                        stats[level]['hits'].extend(score)
+
+                raise ValueError(f"invalid syntax: {line}")
+        print(stats)
+
+parser = argparse.ArgumentParser(description='process our data log files on end repair')
+parser.add_argument('files', type=str, nargs='+')
+
+if __name__ == "__main__":
+    args = parser.parse_args()
+    main(args.files)