(* csv.ml - comma separated values parser
*
- * $Id: csv.ml,v 1.9 2006-02-23 15:24:25 rich Exp $
+ * $Id: csv.ml,v 1.13 2006-11-24 09:43:15 rich Exp $
*)
(* The format of CSV files:
* subset of it into a matrix.
*)
+open Printf
+
+(* Uncomment the next line to enable Extlib's List function. These
+ * avoid stack overflows on really huge CSV files.
+ *)
+(*open ExtList*)
+
type t = string list list
exception Bad_CSV_file of string
let c = compare_row [] y in
if c <> 0 then c else compare [] ys
+(* Concatenate - arrange left to right. *)
+let rec concat = function
+ | [] -> []
+ | [csv] -> csv
+ | left_csv :: csvs ->
+ (* Concatenate the remaining CSV files. *)
+ let right_csv = concat csvs in
+
+ (* Set the height of the left and right CSVs to the same. *)
+ let nr_rows = max (lines left_csv) (lines right_csv) in
+ let left_csv = set_rows nr_rows left_csv in
+ let right_csv = set_rows nr_rows right_csv in
+
+ (* Square off the left CSV. *)
+ let left_csv = square left_csv in
+
+ (* Prepend the right CSV rows with the left CSV rows. *)
+ List.map (
+ fun (left_row, right_row) -> List.append left_row right_row
+ ) (List.combine left_csv right_csv)
+
let to_array csv =
Array.of_list (List.map Array.of_list 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
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)