Add timeline with deciminutes

make_profiler/dot_export.py

@@ -3,6 +3,7 @@
 import collections
 import datetime
 import os
+import math
 from subprocess import Popen, PIPE
 
 def critical_path(influences, dependencies, inputs, timing):
@@ -21,6 +22,9 @@ def critical_path(influences, dependencies, inputs, timing):
             duration = 1
         targets[t]["duration"] = duration
         targets[t]["early_end"] = targets[t]["early_start"] + duration
+        # add timing tag as hour number from very start
+        targets[t]["timing_tag"] = math.ceil(targets[t]["early_end"]/60/10)
+        targets[t]["pin_timing_tag"] = True
         for z in influences[t]:
             update_queue.append(z)
             if z not in targets:
@@ -39,6 +43,9 @@ def critical_path(influences, dependencies, inputs, timing):
         targets[t]["late_start"] = targets[t]["late_end"] - targets[t]["duration"]
         for d in dependencies.get(t, []):
             for z in d:
+                # don't pin timing tags for dependencies if they're at the same tag
+                if targets[z]["timing_tag"] == targets[t]["timing_tag"]:
+                    targets[z]["pin_timing_tag"] = False
                 if z not in update_queue:
                     update_queue.append(z)
                 if "late_end" not in targets[z]:
@@ -47,10 +54,17 @@ def critical_path(influences, dependencies, inputs, timing):
                     targets[z]["late_end"] = min(targets[t]["late_start"], targets[z]["late_end"])
 
     cp = set()
+    timing_tags = {}
     for t, z in targets.items():
         if z["early_start"] == z["late_start"]:
             cp.add(t)
-    return cp
+        if z["pin_timing_tag"]:
+            if z["timing_tag"] not in timing_tags:
+                timing_tags[z["timing_tag"]] = [t]
+            else:
+                timing_tags[z["timing_tag"]].append(t)
+
+    return cp, timing_tags
 
 
 def classify_target(name, influences, dependencies, inputs, order_only):
@@ -86,7 +100,7 @@ def dot_node(name, performance, docstring, cp):
             node['URL'] = target_performance['log']
         if timing != '0:00:00':
             node['label'] += '\\n%s\\r' % timing
-            node['fontsize'] = min(max(timing_sec ** .5, node['fontsize']), 100)
+            node['fontsize'] = min(max(timing_sec ** .5, node['fontsize']), 150)
     if name in cp:
         node['color'] = '#cc0000'
     node['group'] = '/'.join(name.split('/')[:2])
@@ -107,6 +121,7 @@ def export_dot(f, influences, dependencies, order_only, performance, indirect_in
     rankdir="BT"
     ratio=0.5625
     node [shape="box"]
+    newrank=true
 """)
     groups = collections.defaultdict(set)
 
@@ -116,7 +131,7 @@ def export_dot(f, influences, dependencies, order_only, performance, indirect_in
         for t in v:
             inputs.discard(t)
 
-    cp = critical_path(influences, dependencies, inputs, performance)
+    cp, timing_tags = critical_path(influences, dependencies, inputs, performance)
 
     # cluster labels
     labels = {
@@ -127,6 +142,8 @@ def export_dot(f, influences, dependencies, order_only, performance, indirect_in
         'cluster_tools': 'Tools'
     }
 
+    hidden_nodes = []
+
     for target, infls in influences.items():
         group = classify_target(target, infls, dependencies, inputs, order_only)
         groups[group].add(target)
@@ -135,24 +152,50 @@ def export_dot(f, influences, dependencies, order_only, performance, indirect_in
         label = ''
         if k in labels:
             label = 'label="%s"' % labels[k]
+
         nodes = [dot_node(t, performance, docs.get(t, t), cp) for t in v]
         nodes.append('\"%s_DUMMY\" [shape=point style=invis]' % k)
+        if k == 'cluster_result':
+            nodes.append('rank=sink;')
+        #if k == 'cluster_inputs':
+            #nodes.append('rank=source;')
+        if k == 'cluster_tools':
+            nodes.append('rank=source;')
+        if k == 'cluster_order_only':
+            hidden_nodes = v
+            continue
+
+            nodes.append('style=invis;')
+
+
         f.write('subgraph "%s" { %s graph[style=dotted] %s }\n' % (k, label, ';\n'.join(nodes)))
 
     for k, v in influences.items():
         for t in sorted(v):
             if t in indirect_influences[k]:
                 f.write('"%s" -> "%s" [color="#00000033",weight="0",style="dashed"];\n' % (k, t))
             elif k in cp and t in cp:
-                f.write('"%s" -> "%s" [color="#cc0000",weight="3",penwidth="3",headclip="true"];\n' % (k, t))
+                f.write('"%s" -> "%s" [color="#cc0000",weight="10",penwidth="3",headclip="true"];\n' % (k, t))
             else:
                 f.write('"%s" -> "%s";\n' % (k, t))
 
     f.write('cluster_inputs_DUMMY -> cluster_tools_DUMMY -> cluster_result_DUMMY [ style=invis ];')
 
     if 'cluster_not_implemented' in groups:
         f.write('cluster_inputs_DUMMY -> cluster_not_implemented_DUMMY -> cluster_tools_DUMMY [ style=invis ];')
-        f.write('cluster_result_DUMMY -> cluster_not_implemented_DUMMY -> cluster_order_only_DUMMY [ style=invis ];')
+        f.write('cluster_not_implemented_DUMMY -> cluster_order_only_DUMMY [ style=invis ];')
+
+    def format_deciminutes(k):
+        hrs = math.floor(k / 6)
+        min = (k %6)*10
+        return '%s:%02d'%(hrs,min)
+
+    for k,v in timing_tags.items():
+        f.write('{ rank=same; ' + '%s [label="%s"]'%(k,format_deciminutes(k)) + ' [fontsize=50];' + ';'.join(['"%s"' % t for t in v if t not in hidden_nodes]) + '}')
+    tags = sorted(timing_tags.keys())
+
+    f.write('->'.join([ '%s' % k  for k in tags]))
+
     f.write('}')