modules/dsv/fsm/dsv-context: Validate row lengths

* modules/dsv/fsm/dsv-context.scm (add-row): When "validate?" option is set,
  check if rows have consistent lengths.
* modules/dsv/rfc4180.scm (dsv->scm): Add "validate?" option.
* modules/dsv/unix.scm (dsv->scm): Add "validate?" option.
* modules/dsv/cli/common.scm (print-file): Validate input data.
* modules/dsv.scm (dsv->scm): Handle "validate?" option.
* NEWS: Update.

NEWS

@@ -7,6 +7,9 @@ Copyright (C) Artyom V. Poptsov <[email protected]>
   are permitted in any medium without royalty provided the copyright
   notice and this notice are preserved.
 
+* Unreleased
+** Parsers now can check if the table rows have consistent lengths
+
 * Changes in version 0.7.1 (2023-10-22)
 ** =dsv= now handles =--number= option
 This option allows to add numbers to rows and columns of a table.

modules/dsv.scm

@@ -84,6 +84,7 @@
                    #:key
                    (format         'unix)
                    (comment-prefix 'default)
+                   (validate?      #f)
                    (debug-mode?    #f)
                    (log-driver     #f)
                    (log-opt        '()))
@@ -100,10 +101,12 @@ values, or throw 'dsv-parser-error' on an error."
      (unix:dsv->scm port
                     #:delimiter      delimiter
                     #:comment-prefix comment-prefix
+                    #:validate?      validate?
                     #:debug-mode?    debug-mode?))
     ((rfc4180)
      (rfc4180:dsv->scm port
                        #:delimiter      delimiter
+                       #:validate?      validate?
                        #:debug-mode?    debug-mode?))
     (else
      (dsv-error "Unknown format" format))))

modules/dsv/cli/common.scm

@@ -142,6 +142,7 @@ of numbers, or #f if an error occurred."
       (error "Could not determine a file delimiter" input-port))
 
     (let* ((table (remove-empty-rows (dsv->scm input-port delim #:format fmt
+                                               #:validate? #t
                                                #:debug-mode? debug-mode?)))
            (table (if filter-col-proc
                       (table-filter-column filter-col-proc table)

modules/dsv/fsm/dsv-context.scm

@@ -91,9 +91,41 @@
       context
       (add-field context)))
 
+(define (throw-row-length-error context row row-length expected-row-length)
+  (context-log-error context
+                     "Inconsistent length mismatch on line ~a: expected ~a, got ~a"
+                     (context-row-number context)
+                     expected-row-length
+                     row-length)
+  (error (format #f
+                 "Inconsistent row length on line ~a: expected ~a, got ~a; "
+                 (context-row-number context)
+                 expected-row-length
+                 row-length)
+         (context-port context)
+         (context-row-number context)
+         (context-col-number context)
+         row
+         context))
+
 (define* (add-row context #:optional char)
-  (clear-stanza
-   (push-event-to-result context (context-stanza/reversed context))))
+  (let ((result (context-result context))
+        (stanza (context-stanza/reversed context)))
+    (if (or (not (hash-ref (context-custom-data context) 'validate?))
+            (null? result))
+        (clear-stanza
+         (push-event-to-result context (context-stanza/reversed context)))
+        (let* ((stanza-length   (length stanza))
+               (last-row        (car result))
+               (last-row-length (length last-row)))
+          (if (not (equal? stanza-length last-row-length))
+              (throw-row-length-error context
+                                      stanza
+                                      stanza-length
+                                      last-row-length)
+              (clear-stanza
+               (push-event-to-result context
+                                     (context-stanza/reversed context))))))))
 
 (define* (prepare-result context #:optional char)
   (reverse-result context))

modules/dsv/rfc4180.scm

@@ -107,6 +107,7 @@
 (define* (dsv->scm port
                    #:key
                    (debug-mode? #f)
+                   (validate? #f)
                    (delimiter %default-delimiter))
   (let* ((fsm (make <rfc4180-fsm>
                 #:pre-action pre-action
@@ -119,6 +120,7 @@
                                             `((delimiter . ,(if (equal? delimiter 'default)
                                                                 %default-delimiter
                                                                 delimiter))
+                                              (validate?      . ,validate?)
                                               (comment-prefix . #f)))))))
     (context-result context)))
 

modules/dsv/unix.scm

@@ -73,6 +73,7 @@
                    #:key
                    (debug-mode? #f)
                    (delimiter %default-delimiter)
+                   (validate? #f)
                    (comment-prefix %default-comment-prefix))
   (let* ((fsm (make <unix-fsm>
                 #:pre-action pre-action
@@ -85,6 +86,7 @@
                                             `((delimiter . ,(if (equal? delimiter 'default)
                                                                 %default-delimiter
                                                                 delimiter))
+                                              (validate? . ,validate?)
                                               (comment-prefix . ,(if (equal? comment-prefix 'default)
                                                                      %default-comment-prefix
                                                                      comment-prefix))))))))