feat: print reason summary in case of dryrun (#1778)

* feat: print summary of needrun reasons on dryrun

* fmt

* display more rules

* use str of reason

* fix bug

* fix adding rulename

* polishing

Co-authored-by: Felix Mölder <felix.moelder@uni-due.de>

snakemake/dag.py

@@ -2362,6 +2362,22 @@ def edge(a, b):
                 ],
             )
 
+    def print_reasons(self):
+        """Print summary of execution reasons."""
+        reasons = defaultdict(set)
+        for job in self.needrun_jobs(exclude_finished=False):
+            for reason in self.reason(job).get_names():
+                reasons[reason].add(job.rule.name)
+        if reasons:
+            msg = "Reasons:\n    (check individual jobs above for details)"
+            for reason, rules in sorted(reasons.items()):
+                rules = sorted(rules)
+                if len(rules) > 50:
+                    rules = rules[:50] + ["..."]
+                rules = ", ".join(rules)
+                msg += f"\n    {reason}:\n        {rules}"
+            logger.info(msg)
+
     def stats(self):
         from tabulate import tabulate
 

snakemake/jobs.py

@@ -1603,6 +1603,34 @@ def mark_finished(self):
         "called if the job has been run"
         self.finished = True
 
+    def get_names(self):
+        if self.forced:
+            yield "forced"
+        if self.noio:
+            yield "neither input nor output"
+        if self.nooutput:
+            yield "run or shell but no output"
+        if self._missing_output:
+            yield "missing output files"
+        if self._incomplete_output:
+            yield "incomplete output files"
+        if self._updated_input:
+            yield "updated input files"
+        if self._updated_input_run:
+            yield "input files updated by another job"
+        if self.pipe:
+            yield "pipe output needed by consuming job"
+        if self.service:
+            yield "provides service for consuming job"
+        if self.input_changed:
+            yield "set of input files has changed since last execution"
+        if self.code_changed:
+            yield "code has changed since last execution"
+        if self.params_changed:
+            yield "params have changed since last execution"
+        if self.software_stack_changed:
+            yield "software environment definition has changed since last execution"
+
     def __str__(self):
         def format_file(f):
             if is_flagged(f, "sourcecache_entry"):
@@ -1619,7 +1647,7 @@ def format_files(files):
         else:
             if self.noio:
                 s.append(
-                    "Rules with neither input nor " "output files are always executed."
+                    "Rules with neither input nor output files are always executed."
                 )
             elif self.nooutput:
                 s.append(

snakemake/workflow.py

@@ -1080,6 +1080,7 @@ def log_provenance_info():
             if dryrun:
                 if len(dag):
                     logger.run_info("\n".join(dag.stats()))
+                    dag.print_reasons()
                     log_provenance_info()
                 logger.info("")
                 logger.info(