X-Git-Url: http://git.annexia.org/?a=blobdiff_plain;f=csv.ml;h=6ca424b09ee1faa34556da43a40389dbc156712f;hb=680fc80e414200998978f834a83bb8200b2a6609;hp=2a22bf983eef6e3e372b825de554e0e08eee65fe;hpb=ae6f56ba8e80a1c05aa076404dd14145b306780a;p=ocaml-csv.git

diff --git a/csv.ml b/csv.ml
index 2a22bf9..6ca424b 100644
--- a/csv.ml
+++ b/csv.ml
@@ -1,6 +1,6 @@
 (* csv.ml - comma separated values parser
  *
- * $Id: csv.ml,v 1.8 2006-02-15 13:25:58 rich Exp $
+ * $Id: csv.ml,v 1.11 2006-07-19 09:41:58 rich Exp $
  *)
 
 (* The format of CSV files:
@@ -40,6 +40,8 @@
  * subset of it into a matrix.
  *)
 
+open Printf
+
 type t = string list list
 
 exception Bad_CSV_file of string
@@ -265,6 +267,38 @@ let sub r c rows cols csv =
   let csv = set_columns cols csv in
   csv
 
+(* Compare two rows for semantic equality - ignoring any blank cells
+ * at the end of each row.
+ *)
+let rec compare_row (row1 : string list) row2 =
+  match row1, row2 with
+  | [], [] -> 0
+  | x :: xs, y :: ys ->
+      let c = compare x y in
+      if c <> 0 then c else compare_row xs ys
+  | "" :: xs , [] ->
+      compare_row xs []
+  | x :: xs, [] ->
+      1
+  | [], "" :: ys ->
+      compare_row [] ys
+  | [], y :: ys ->
+      -1
+
+(* Semantic equality for CSV files. *)
+let rec compare (csv1 : t) csv2 =
+  match csv1, csv2 with
+  | [], [] -> 0
+  | x :: xs, y :: ys ->
+      let c = compare_row x y in
+      if c <> 0 then c else compare xs ys
+  | x :: xs, [] ->
+      let c = compare_row x [] in
+      if c <> 0 then c else compare xs []
+  | [], y :: ys ->
+      let c = compare_row [] y in
+      if c <> 0 then c else compare [] ys
+
 let to_array csv =
   Array.of_list (List.map Array.of_list csv)
 
@@ -337,18 +371,18 @@ let save_out_readable chan csv =
 
     match csv with
       | [] -> []
-      | r :: _ ->
-	  let n = List.length r in
-	  let lengths = List.map (List.map String.length) csv in
+      | row1 :: rest ->
+	  let lengths_row1 = List.map String.length row1 in
+	  let lengths_rest = List.map (List.map String.length) rest in
 	  let max2rows r1 r2 =
-	    let rp = List.combine r1 r2 in
+	    let rp =
+	      try List.combine r1 r2
+	      with
+		Invalid_argument "List.combine" ->
+		  failwith (sprintf "Csv.save_out_readable: internal error: length r1 = %d, length r2 = %d" (List.length r1) (List.length r2)) in
 	    List.map (fun ((a : int), (b : int)) -> max a b) rp
 	  in
-	  let rec repeat x = function
-	    | 0 -> []
-	    | i -> x :: repeat x (i-1)
-	  in
-	  List.fold_left max2rows (repeat 0 n) lengths in
+	  List.fold_left max2rows lengths_row1 lengths_rest in
 
   (* Print out each cell at the correct width. *)
   let rec repeat f = function
@@ -361,9 +395,17 @@ let save_out_readable chan csv =
 	output_string chan cell;
 	output_char chan '\n'
     | row ->				(* Other. *)
-	let row = List.combine widths row in
+	(* Pair up each cell with its max width. *)
+	let row =
+	  let rec loop = function
+	    | ([], _) -> []
+	    | (_, []) -> failwith "Csv.save_out_readable: internal error"
+	    | (cell :: cells, width :: widths) ->
+		(cell, width) :: loop (cells, widths)
+	  in
+	  loop (row, widths) in
 	List.iter (
-	  fun (width, cell) ->
+	  fun (cell, width) ->
 	    output_string chan cell;
 	    let n = String.length cell in
 	    repeat (fun () -> output_char chan ' ') (width - n + 1)