implement kpsewhich --all for normal searches

git-svn-id: svn://tug.org/texlive/trunk@5825 c570f23f-e606-0410-a88d-b1316a301751

1 files changed, 35 insertions, 0 deletions

diff --git a/Build/source/texk/kpathsea/str-list.c b/Build/source/texk/kpathsea/str-list.c
index 66df7b791cc..8b4ab314dfe 100644
--- a/Build/source/texk/kpathsea/str-list.c
+++ b/Build/source/texk/kpathsea/str-list.c
@@ -118,3 +118,38 @@ str_list_free P1C(str_list_type *, l)
       STR_LIST (*l) = NULL;
     }
 }
+
+
+
+/* Remove duplicate elements from L, freeing their space.  Since our
+   lists are so short, we do a maximally inefficient bubble search.  */
+
+void
+str_list_uniqify P1C(str_list_type *, l)
+{
+  unsigned e;
+  str_list_type ret = str_list_init ();
+  
+  for (e = 0; e < STR_LIST_LENGTH (*l); e++) {
+    string elt1 = STR_LIST_ELT (*l, e);
+    unsigned f;
+    for (f = e + 1; f < STR_LIST_LENGTH (*l); f++) {
+      string elt2 = STR_LIST_ELT (*l, f);
+      /* I don't think our list should ever contain NULL's, but if
+         it does, let it stay and don't bother collapsing multiple
+         NULL's into one.  */
+      if (FILESTRCASEEQ (elt1, elt2)) {
+        break;
+      }
+    }
+    
+    if (f == STR_LIST_LENGTH (*l)) {
+      str_list_add (&ret, elt1); /* not found */
+    } else {
+      free (elt1);  /* duplicate, forget this one */
+    }
+  }
+  
+  /* Replace the passed list with what we constructed.  */
+  *l = ret;
+}