From 306cca2ae3b20a975ea03e6d3f0fb219fb9d005c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Dominik=20Pant=C5=AF=C4=8Dek?=
 <dominik.pantucek@trustica.cz>
Date: Fri, 31 Mar 2023 19:20:49 +0200
Subject: [PATCH] Heavily optimize CSV reader.

---
 csv-simple.scm | 33 ++++++++++++++++++---------------
 1 file changed, 18 insertions(+), 15 deletions(-)

diff --git a/csv-simple.scm b/csv-simple.scm
index 60a3331..e20689a 100644
--- a/csv-simple.scm
+++ b/csv-simple.scm
@@ -41,49 +41,52 @@
 	 testing
 	 progress)
 
- ;; Curry version of line parser with configurable cell separator and
+  ;; Curry version of line parser with configurable cell separator and
  ;; string delimiter. Returns a list of lists of strings.
  (define ((make-csv-line-parser separator string-delimiter) line)
-   (let loop ((tokens (irregex-extract (irregex "." 'u) line))
+   (let loop ((tokens (string->list line))
 	      (res '())
 	      (state 1))
      (if (null? tokens)
-	 (reverse res)
+	 (reverse
+	  (map
+	   (lambda (cell)
+	     (list->string (reverse cell)))
+	   res))
 	 (let ((token (car tokens)))
 	   (case state
 	     ((0) ; Parsing regular unquoted cell data - separator creates new cell
-	      (if (equal? token separator)
+	      (if (eq? token separator)
 		  (loop (cdr tokens)
 			res
 			1) ; Start a new cell
 		  (loop (cdr tokens)
-			(cons (string-append (car res) token) (cdr res))
+			(cons (cons token (car res)) (cdr res))
 			0)))
 	     ((1) ; Starting a new cell - check for string delimiter
-	      (if (equal? token string-delimiter)
+	      (if (eq? token string-delimiter)
 		  (loop (cdr tokens)
-			(cons "" res) ; If it is quoted, keep even empty strings there
+			(cons '() res) ; If it is quoted, keep even empty strings there
 			2)
-		  (if (equal? token separator)
+		  (if (eq? token separator)
 		      (loop (cdr tokens)
-			    (cons "" res) ; This was an empty cell
+			    (cons '() res) ; This was an empty cell
 			    1) ; Another new cell awaiting
 		      (loop (cdr tokens)
-			    (cons token res) ; first token of regular new cell
+			    (cons (list token) res) ; first token of regular new cell
 			    0))))
 	     ((2) ; Parsing quoted cell data - no support for escaping string delimiter!
-	      (if (equal? token string-delimiter)
+	      (if (eq? token string-delimiter)
 		  (loop (cdr tokens)
 			res
 			0) ; There shouldn't be anything more, but it is safe to append the rest as normal unquoted data
 		  (loop (cdr tokens)
-			(cons (string-append (car res) token) (cdr res))
+			(cons (cons token (car res)) (cdr res))
 			2)))))))) ; Continue inside quoted data
-
  ;; Parses given CSV lines list
  (define (csv-parse-lines lines . args)
-   (let* ((separator (get-keyword #:separator args (lambda () ";")))
-	  (string-delimiter (get-keyword #:string-delimiter args (lambda () "\"")))
+   (let* ((separator (get-keyword #:separator args (lambda () #\;)))
+	  (string-delimiter (get-keyword #:string-delimiter args (lambda () #\")))
 	  (csv-parse-line (make-csv-line-parser separator string-delimiter))
 	  (total (max (sub1 (length lines)) 1)))
      (let loop ((lines lines)