Merge pull request #3864 from obsidiansystems/more-topo-sort

Abstract out topo sorting logic

src/libstore/misc.cc

@@ -4,6 +4,7 @@
 #include "local-store.hh"
 #include "store-api.hh"
 #include "thread-pool.hh"
+#include "topo-sort.hh"
 
 
 namespace nix {
@@ -256,41 +257,21 @@ void Store::queryMissing(const std::vector<StorePathWithOutputs> & targets,
 
 StorePaths Store::topoSortPaths(const StorePathSet & paths)
 {
-    StorePaths sorted;
+    return topoSort(paths,
-    StorePathSet visited, parents;
+        {[&](const StorePath & path) {
-
-    std::function<void(const StorePath & path, const StorePath * parent)> dfsVisit;
-
-    dfsVisit = [&](const StorePath & path, const StorePath * parent) {
-        if (parents.count(path))
-            throw BuildError("cycle detected in the references of '%s' from '%s'",
-                printStorePath(path), printStorePath(*parent));
-
-        if (!visited.insert(path).second) return;
-        parents.insert(path);
-
             StorePathSet references;
             try {
                 references = queryPathInfo(path)->references;
             } catch (InvalidPath &) {
             }
-
+            return references;
-        for (auto & i : references)
+        }},
-            /* Don't traverse into paths that don't exist.  That can
+        {[&](const StorePath & path, const StorePath & parent) {
-               happen due to substitutes for non-existent paths. */
+            return BuildError(
-            if (i != path && paths.count(i))
+                "cycle detected in the references of '%s' from '%s'",
-                dfsVisit(i, &path);
+                printStorePath(path),
-
+                printStorePath(parent));
-        sorted.push_back(path);
+        }});
-        parents.erase(path);
-    };
-
-    for (auto & i : paths)
-        dfsVisit(i, nullptr);
-
-    std::reverse(sorted.begin(), sorted.end());
-
-    return sorted;
 }
 
 

src/libutil/topo-sort.hh

@@ -0,0 +1,40 @@
+#include "error.hh"
+
+namespace nix {
+
+template<typename T>
+std::vector<T> topoSort(std::set<T> items,
+        std::function<std::set<T>(const T &)> getChildren,
+        std::function<Error(const T &, const T &)> makeCycleError)
+{
+    std::vector<T> sorted;
+    std::set<T> visited, parents;
+
+    std::function<void(const T & path, const T * parent)> dfsVisit;
+
+    dfsVisit = [&](const T & path, const T * parent) {
+        if (parents.count(path)) throw makeCycleError(path, *parent);
+
+        if (!visited.insert(path).second) return;
+        parents.insert(path);
+
+        std::set<T> references = getChildren(path);
+
+        for (auto & i : references)
+            /* Don't traverse into items that don't exist in our starting set. */
+            if (i != path && items.count(i))
+                dfsVisit(i, &path);
+
+        sorted.push_back(path);
+        parents.erase(path);
+    };
+
+    for (auto & i : items)
+        dfsVisit(i, nullptr);
+
+    std::reverse(sorted.begin(), sorted.end());
+
+    return sorted;
+}
+
+}