projects
/
SXSI
/
xpathcomp.git
/ commitdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
| commitdiff |
tree
raw
|
patch
|
inline
| side by side (parent:
0da8c3c
)
Bump file format magic number, invalidate all previously generated indexes.
author
Kim Nguyễn
<kn@lri.fr>
Tue, 29 May 2012 05:49:14 +0000
(07:49 +0200)
committer
Kim Nguyễn
<kn@lri.fr>
Tue, 29 May 2012 05:49:14 +0000
(07:49 +0200)
src/tree.ml
patch
|
blob
|
history
diff --git
a/src/tree.ml
b/src/tree.ml
index
ec253bc
..
e729034
100644
(file)
--- a/
src/tree.ml
+++ b/
src/tree.ml
@@
-572,7
+572,7
@@
let parse_xml_string str = node_of_t (TreeBuilder.parse_string str)
let size t = tree_size t.doc;;
let magic_string = "SXSI_INDEX"
let size t = tree_size t.doc;;
let magic_string = "SXSI_INDEX"
-let version_string = "
3
"
+let version_string = "
4
"
let pos fd =
Unix.lseek fd 0 Unix.SEEK_CUR
let pos fd =
Unix.lseek fd 0 Unix.SEEK_CUR
@@
-785,7
+785,7
@@
let stats tree =
in
traverse tree.doc root [] 0;
let sumdepth = Hashtbl.fold (fun p c acc -> (List.length p) * c + acc) h 0 in
in
traverse tree.doc root [] 0;
let sumdepth = Hashtbl.fold (fun p c acc -> (List.length p) * c + acc) h 0 in
-
+ let alltags = Ptset.Int.union tree.elements tree.attributes in
Logger.print err_formatter "Statistics :@\n\
Average depth: %f@\n\
Longest path: %i@\n\
Logger.print err_formatter "Statistics :@\n\
Average depth: %f@\n\
Longest path: %i@\n\
@@
-793,14
+793,16
@@
Number of distinct paths: %i@\n\
Number of nodes: %i@\n\
Number of leaves: %i@\n\
Number of pcdata/cdata nodes: %i@\n\
Number of nodes: %i@\n\
Number of leaves: %i@\n\
Number of pcdata/cdata nodes: %i@\n\
-Number of distinct tags: %i@\n@?"
+Number of distinct tags: %i@\n\
+Largest tag id: %i@\n@?"
(float_of_int sumdepth /. float_of_int !numleaves)
!depth
(Hashtbl.length h)
(tree_subtree_size tree.doc root)
!numleaves
!numtexts
(float_of_int sumdepth /. float_of_int !numleaves)
!depth
(Hashtbl.length h)
(tree_subtree_size tree.doc root)
!numleaves
!numtexts
- (Ptset.Int.cardinal (Ptset.Int.union tree.elements tree.attributes))
+ (Ptset.Int.cardinal alltags)
+ (Ptset.Int.max_elt alltags)
(*
Logger.print err_formatter "Average depth: %f, number of leaves %i@\n@?" ((float_of_int a)/. (float_of_int b)) b
(*
Logger.print err_formatter "Average depth: %f, number of leaves %i@\n@?" ((float_of_int a)/. (float_of_int b)) b