* The new normaliser now passes the unit tests.

7 files changed, 179 insertions, 34 deletions

diff --git a/src/fstate.cc b/src/fstate.cc
index 596d63a5f406..aad961ef58e8 100644
--- a/src/fstate.cc
+++ b/src/fstate.cc
@@ -459,6 +459,37 @@ static Slice parseSlice(FState fs)
 }
 
 
+static ATermList unparseIds(const FSIds & ids)
+{
+    ATermList l = ATempty;
+    for (FSIds::const_iterator i = ids.begin();
+         i != ids.end(); i++)
+        l = ATinsert(l,
+            ATmake("<str>", ((string) *i).c_str()));
+    return ATreverse(l);
+}
+
+
+static FState unparseSlice(const Slice & slice)
+{
+    ATermList roots = unparseIds(slice.roots);
+    
+    ATermList elems = ATempty;
+    for (SliceElems::const_iterator i = slice.elems.begin();
+         i != slice.elems.end(); i++)
+        elems = ATinsert(elems,
+            ATmake("(<str>, <str>, <term>)",
+                i->path.c_str(),
+                ((string) i->id).c_str(),
+                unparseIds(i->refs)));
+
+    return ATmake("Slice(<term>, <term>)", roots, elems);
+}
+
+
+typedef set<FSId> FSIdSet;
+
+
 Slice normaliseFState(FSId id)
 {
     debug(format("normalising fstate"));
@@ -494,24 +525,23 @@ Slice normaliseFState(FSId id)
     /* Realise inputs (and remember all input paths). */
     FSIds inIds;
     parseIds(ins, inIds);
-    
-    SliceElems inElems; /* !!! duplicates */
-    StringSet inPathsSet;
+
+    typedef map<string, SliceElem> ElemMap;
+
+    ElemMap inMap;
+
     for (FSIds::iterator i = inIds.begin(); i != inIds.end(); i++) {
         Slice slice = normaliseFState(*i);
         realiseSlice(slice);
 
         for (SliceElems::iterator j = slice.elems.begin();
              j != slice.elems.end(); j++)
-        {
-            inElems.push_back(*j);
-            inPathsSet.insert(j->path);
-        }
+            inMap[j->path] = *j;
     }
 
     Strings inPaths;
-    copy(inPathsSet.begin(), inPathsSet.end(),
-        inserter(inPaths, inPaths.begin()));
+    for (ElemMap::iterator i = inMap.begin(); i != inMap.end(); i++)
+        inPaths.push_back(i->second.path);
 
     /* Build the environment. */
     Environment env;
@@ -533,6 +563,7 @@ Slice normaliseFState(FSId id)
         if (!ATmatch(t, "(<str>, <str>)", &s1, &s2))
             throw badTerm("string expected", t);
         outPaths.push_back(OutPath(s1, parseHash(s2)));
+        inPaths.push_back(s1);
         outs = ATgetNext(outs);
     }
 
@@ -548,6 +579,7 @@ Slice normaliseFState(FSId id)
 
     /* Check whether the output paths were created, and register each
        one. */
+    FSIdSet used;
     for (list<OutPath>::iterator i = outPaths.begin(); 
          i != outPaths.end(); i++)
     {
@@ -557,25 +589,81 @@ Slice normaliseFState(FSId id)
         registerPath(path, i->second);
         slice.roots.push_back(i->second);
 
-        Strings outPaths = filterReferences(path, inPaths);
+        Strings refs = filterReferences(path, inPaths);
+
+        SliceElem elem;
+        elem.path = path;
+        elem.id = i->second;
+
+        for (Strings::iterator j = refs.begin(); j != refs.end(); j++) {
+            ElemMap::iterator k;
+            if ((k = inMap.find(*j)) != inMap.end()) {
+                elem.refs.push_back(k->second.id);
+                used.insert(k->second.id);
+            } else abort(); /* fix! check in created paths */
+        }
+
+        slice.elems.push_back(elem);
+    }
+
+    for (ElemMap::iterator i = inMap.begin();
+         i != inMap.end(); i++)
+    {
+        FSIdSet::iterator j = used.find(i->second.id);
+        if (j == used.end())
+            debug(format("NOT referenced: `%1%'") % i->second.path);
+        else {
+            debug(format("referenced: `%1%'") % i->second.path);
+            slice.elems.push_back(i->second);
+        }
     }
 
+    FState nf = unparseSlice(slice);
+    debug(printTerm(nf));
+    storeSuccessor(id, nf);
+
     return slice;
 }
 
 
-void realiseSlice(Slice slice)
+static void checkSlice(const Slice & slice)
+{
+    if (slice.elems.size() == 0)
+        throw Error("empty slice");
+
+    FSIdSet decl;
+    for (SliceElems::const_iterator i = slice.elems.begin();
+         i != slice.elems.end(); i++)
+    {
+        debug((string) i->id);
+        decl.insert(i->id);
+    }
+    
+    for (FSIds::const_iterator i = slice.roots.begin();
+         i != slice.roots.end(); i++)
+        if (decl.find(*i) == decl.end())
+            throw Error(format("undefined id: %1%") % (string) *i);
+    
+    for (SliceElems::const_iterator i = slice.elems.begin();
+         i != slice.elems.end(); i++)
+        for (FSIds::const_iterator j = i->refs.begin();
+             j != i->refs.end(); j++)
+            if (decl.find(*j) == decl.end())
+                throw Error(format("undefined id: %1%") % (string) *j);
+}
+
+
+void realiseSlice(const Slice & slice)
 {
     debug(format("realising slice"));
     Nest nest(true);
 
-    if (slice.elems.size() == 0)
-        throw Error("empty slice");
+    checkSlice(slice);
 
     /* Perhaps all paths already contain the right id? */
 
     bool missing = false;
-    for (SliceElems::iterator i = slice.elems.begin();
+    for (SliceElems::const_iterator i = slice.elems.begin();
          i != slice.elems.end(); i++)
     {
         SliceElem elem = *i;
@@ -596,7 +684,7 @@ void realiseSlice(Slice slice)
     }
 
     /* For each element, expand its id at its path. */
-    for (SliceElems::iterator i = slice.elems.begin();
+    for (SliceElems::const_iterator i = slice.elems.begin();
          i != slice.elems.end(); i++)
     {
         SliceElem elem = *i;
diff --git a/src/fstate.hh b/src/fstate.hh
index f06a4807ef5d..a195281646b6 100644
--- a/src/fstate.hh
+++ b/src/fstate.hh
@@ -113,7 +113,7 @@ void registerSuccessor(const FSId & id1, const FSId & id2);
 Slice normaliseFState(FSId id);
 
 /* Realise a Slice in the file system. */
-void realiseSlice(Slice slice);
+void realiseSlice(const Slice & slice);
 
 
 #endif /* !__FSTATE_H */
diff --git a/src/hash.cc b/src/hash.cc
index 72dcd790c7dc..97bf8f7856f0 100644
--- a/src/hash.cc
+++ b/src/hash.cc
@@ -28,6 +28,16 @@ bool Hash::operator != (Hash h2)
 }
 
 
+bool Hash::operator < (const Hash & h) const
+{
+    for (unsigned int i = 0; i < hashSize; i++) {
+        if (hash[i] < h.hash[i]) return true;
+        if (hash[i] > h.hash[i]) return false;
+    }
+    return false;
+}
+
+
 Hash::operator string() const
 {
     ostringstream str;
diff --git a/src/hash.hh b/src/hash.hh
index 509a27912a72..5ae5dc22aed8 100644
--- a/src/hash.hh
+++ b/src/hash.hh
@@ -22,6 +22,9 @@ struct Hash
     /* Check whether two hash are not equal. */
     bool operator != (Hash h2);
 
+    /* For sorting. */
+    bool operator < (const Hash & h) const;
+
     /* Convert a hash code into a hexadecimal representation. */
     operator string() const;
 };
diff --git a/src/references.cc b/src/references.cc
index de7a4b339c68..a42c1aed02c1 100644
--- a/src/references.cc
+++ b/src/references.cc
@@ -1,3 +1,5 @@
+#include <map>
+
 #include <sys/types.h>
 #include <sys/stat.h>
 #include <unistd.h>
@@ -9,24 +11,24 @@
 
 
 static void search(const string & s,
-    Strings & refs, Strings & seen)
+    Strings & ids, Strings & seen)
 {
-    for (Strings::iterator i = refs.begin();
-         i != refs.end(); )
+    for (Strings::iterator i = ids.begin();
+         i != ids.end(); )
     {
         if (s.find(*i) == string::npos)
             i++;
         else {
             debug(format("found reference to `%1%'") % *i);
             seen.push_back(*i);
-            i = refs.erase(i);
+            i = ids.erase(i);
         }
     }
 }
 
 
 void checkPath(const string & path,
-    Strings & refs, Strings & seen)
+    Strings & ids, Strings & seen)
 {
     struct stat st;
     if (lstat(path.c_str(), &st))
@@ -39,8 +41,8 @@ void checkPath(const string & path,
         while (errno = 0, dirent = readdir(dir)) {
             string name = dirent->d_name;
             if (name == "." || name == "..") continue;
-            search(name, refs, seen);
-            checkPath(path + "/" + name, refs, seen);
+            search(name, ids, seen);
+            checkPath(path + "/" + name, ids, seen);
         }
 
         closedir(dir); /* !!! close on exception */
@@ -58,7 +60,7 @@ void checkPath(const string & path,
         if (read(fd, buf, st.st_size) != st.st_size)
             throw SysError(format("reading file %1%") % path);
 
-        search(string(buf, st.st_size), refs, seen);
+        search(string(buf, st.st_size), ids, seen);
         
         delete buf; /* !!! autodelete */
 
@@ -69,30 +71,40 @@ void checkPath(const string & path,
         char buf[st.st_size];
         if (readlink(path.c_str(), buf, st.st_size) != st.st_size)
             throw SysError(format("reading symbolic link `%1%'") % path);
-        search(string(buf, st.st_size), refs, seen);
+        search(string(buf, st.st_size), ids, seen);
     }
     
     else throw Error(format("unknown file type: %1%") % path);
 }
 
 
-Strings filterReferences(const string & path, const Strings & _refs)
+Strings filterReferences(const string & path, const Strings & paths)
 {
-    Strings refs;
+    map<string, string> backMap;
+    Strings ids;
     Strings seen;
 
     /* For efficiency (and a higher hit rate), just search for the
        hash part of the file name.  (This assumes that all references
        have the form `HASH-bla'). */
-    for (Strings::const_iterator i = _refs.begin();
-         i != _refs.end(); i++)
+    for (Strings::const_iterator i = paths.begin();
+         i != paths.end(); i++)
     {
         string s = string(baseNameOf(*i), 0, 32);
         parseHash(s);
-        refs.push_back(s);
+        ids.push_back(s);
+        backMap[s] = *i;
     }
 
-    checkPath(path, refs, seen);
+    checkPath(path, ids, seen);
+
+    Strings found;
+    for (Strings::iterator i = seen.begin(); i != seen.end(); i++)
+    {
+        map<string, string>::iterator j;
+        if ((j = backMap.find(*i)) == backMap.end()) abort();
+        found.push_back(j->second);
+    }
 
-    return seen;
+    return found;
 }
diff --git a/src/test-builder-2.sh b/src/test-builder-2.sh
index 010e1c80576b..a12fa27a67fd 100644
--- a/src/test-builder-2.sh
+++ b/src/test-builder-2.sh
@@ -5,4 +5,4 @@ echo "builder 2"
 mkdir $out || exit 1
 cd $out || exit 1
 echo "Hallo Wereld" > bla
-cat $src >> bla
\ No newline at end of file
+echo $builder >> bla
diff --git a/src/test.cc b/src/test.cc
index 3437650ac217..219281c8befc 100644
--- a/src/test.cc
+++ b/src/test.cc
@@ -137,8 +137,8 @@ void runTests()
     realise(fs2id);
     realise(fs2id);
 
-    string out1fn = nixStore + "/hello.txt";
     string out1id = hashString("foo"); /* !!! bad */
+    string out1fn = nixStore + "/" + (string) out1id + "-hello.txt";
     FState fs3 = ATmake(
         "Derive([(<str>, <str>)], [<str>], <str>, <str>, [(\"out\", <str>)])",
         out1fn.c_str(),
@@ -153,6 +153,38 @@ void runTests()
     realise(fs3id);
     realise(fs3id);
 
+
+    FSId builder4id;
+    string builder4fn;
+    addToStore("./test-builder-2.sh", builder4fn, builder4id);
+
+    FState fs4 = ATmake(
+        "Slice([<str>], [(<str>, <str>, [])])",
+        ((string) builder4id).c_str(),
+        builder4fn.c_str(),
+        ((string) builder4id).c_str());
+    FSId fs4id = writeTerm(fs4, "", 0);
+
+    realise(fs4id);
+
+    string out5id = hashString("bar"); /* !!! bad */
+    string out5fn = nixStore + "/" + (string) out5id + "-hello2";
+    FState fs5 = ATmake(
+        "Derive([(<str>, <str>)], [<str>], <str>, <str>, [(\"out\", <str>), (\"builder\", <str>)])",
+        out5fn.c_str(),
+        ((string) out5id).c_str(),
+        ((string) fs4id).c_str(),
+        ((string) builder4fn).c_str(),
+        thisSystem.c_str(),
+        out5fn.c_str(),
+        ((string) builder4fn).c_str());
+    debug(printTerm(fs5));
+    FSId fs5id = writeTerm(fs5, "", 0);
+
+    realise(fs5id);
+    realise(fs5id);
+
+
 #if 0
     FState fs2 = ATmake(
         "Path(<str>, Hash(<str>), [])",