From 0da8c3c7c76ab06d5ccfc6ae52488d7549735059 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Kim=20Nguy=E1=BB=85n?= <kn@lri.fr>
Date: Fri, 4 May 2012 16:00:11 +0200
Subject: [PATCH] Add -doc-stats options to print document statistics.

---
 src/main.ml     |  1 +
 src/options.ml  |  3 +++
 src/options.mli |  1 +
 src/tree.ml     | 66 ++++++++++++++++++++++++++++++++++++++-----------
 4 files changed, 57 insertions(+), 14 deletions(-)

diff --git a/src/main.ml b/src/main.ml
index 48f2e46..8e0e130 100644
--- a/src/main.ml
+++ b/src/main.ml
@@ -33,6 +33,7 @@ let mk_runtime run auto doc arg count print outfile =
 
 let main v query_string output =
   Tag.init (Tree.tag_operations v);
+  if !Options.docstats then Tree.stats v;
   let query =
     time ~msg:"Parsing query" XPath.parse query_string
   in
diff --git a/src/options.ml b/src/options.ml
index 0c7c92d..1e349a6 100644
--- a/src/options.ml
+++ b/src/options.ml
@@ -19,6 +19,7 @@ let text_index_type = ref 0
 let do_perf = ref false
 let twopass = ref false
 let repeat = ref 1
+let docstats = ref false
 
 let set_index_type = function
   | "default" -> text_index_type := 0
@@ -89,6 +90,8 @@ let spec = Arg.align
     "-r", Arg.Set_int(repeat),
     " repeat query execution n time (benchmarking only, default 1)";
 
+    "-doc-stats", Arg.Set(docstats),
+    " Compute document statistics (performs full traversal)";
 
 
     "-v", Arg.Set(verbose), " verbose mode"; ] @
diff --git a/src/options.mli b/src/options.mli
index 8393ead..cfb229d 100644
--- a/src/options.mli
+++ b/src/options.mli
@@ -16,3 +16,4 @@ val text_index_type : int ref
 val do_perf : bool ref
 val twopass : bool ref
 val repeat : int ref
+val docstats : bool ref
diff --git a/src/tree.ml b/src/tree.ml
index ae256dd..ec253bc 100644
--- a/src/tree.ml
+++ b/src/tree.ml
@@ -84,7 +84,6 @@ struct
       if (!Options.index_empty_texts) || not (is_whitespace s) then
 	begin
 	  open_tag b "<$>";
-	  Printf.eprintf "Inserting >>%s<<\n" s;
 	  text b s;
 	  close_tag b "<$>";
 	end;
@@ -372,19 +371,7 @@ external tree_size : tree -> int = "caml_xml_tree_size" "noalloc"
 let size t = tree_size t.doc
 
 
-let stats t =
-  let tree = t.doc in
-  let rec loop left node acc_d total_d num_leaves =
-    if node == nil then
-      (acc_d+total_d,if left then num_leaves+1 else num_leaves)
-    else
-      let d,td = loop true (tree_first_child tree node) (acc_d+1) total_d num_leaves in
-      loop false (tree_next_sibling tree  node) (acc_d)  d td
-  in
-  let a,b = loop true root 0 0 0
-  in
-  Logger.print err_formatter "Average depth: %f, number of leaves %i@\n@?" ((float_of_int a)/. (float_of_int b)) b
-;;
+
 
 module TagS =
 struct
@@ -769,3 +756,54 @@ let full_text_query q t s =
   let res = (query_fun q) t s true in
   Hashtbl.replace _pred_cache (q,s) res;
   res.pos
+
+let stats tree =
+      let h = Hashtbl.create 1024 in
+      let depth = ref 0 in
+      let numleaves = ref 0 in
+      let numtexts = ref 0 in
+    let rec traverse tree t p d =
+      if is_nil t then
+	let oldc =
+	  try
+	    Hashtbl.find h p
+	  with Not_found -> 0
+	in
+	Hashtbl.replace h p (oldc + 1);
+	if d > !depth then depth := d;
+	incr numleaves
+      else
+	let label = tree_tag tree t in
+	if label == Tag.pcdata || label == Tag.attribute_data then incr numtexts;
+	iter_siblings tree t (label::p) (d+1)
+    and iter_siblings tree t p d =
+      if is_nil t then () else
+	let fs = tree_first_child tree t in
+	traverse tree fs p d;
+	let ns = tree_next_sibling tree t in
+	iter_siblings tree ns p d
+    in
+    traverse tree.doc root [] 0;
+    let sumdepth = Hashtbl.fold (fun p c acc -> (List.length p) * c + acc) h 0 in
+    
+    Logger.print err_formatter "Statistics :@\n\
+Average depth: %f@\n\
+Longest path: %i@\n\
+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 distinct tags: %i@\n@?"
+      (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))
+
+(*
+  Logger.print err_formatter "Average depth: %f, number of leaves %i@\n@?" ((float_of_int a)/. (float_of_int b)) b
+;;
+
+*)
-- 
2.17.1