From 077324a16fd6c7d2307bfa52a9863ca25cfc6a52 Mon Sep 17 00:00:00 2001
From: Ludovic Courtès <ludo@gnu.org>
Date: Tue, 2 Aug 2022 15:29:55 +0200
Subject: read-print: Recognize page breaks.

* guix/read-print.scm (<page-break>, page-break?, page-break)
(char-set:whitespace-sans-page-break): New variables.
(space?): New procedure.
(read-vertical-space): Use it.
(read-until-end-of-line): New procedure.
(read-with-comments): Add #\page case.
(pretty-print-with-comments): Add 'page-break?' case.
* tests/read-print.scm ("read-with-comments: top-level page break"): New
test.
Add round-trip test with page break within an sexp.
---
 guix/read-print.scm  | 46 +++++++++++++++++++++++++++++++++++++++++++---
 tests/read-print.scm | 22 ++++++++++++++++++++++
 2 files changed, 65 insertions(+), 3 deletions(-)

diff --git a/guix/read-print.scm b/guix/read-print.scm
index 2b626ba281..33ed6e3dbe 100644
--- a/guix/read-print.scm
+++ b/guix/read-print.scm
@@ -35,6 +35,9 @@
             vertical-space-height
             canonicalize-vertical-space
 
+            page-break
+            page-break?
+
             comment
             comment?
             comment->string
@@ -83,6 +86,18 @@
       "Return a vertical space corresponding to a single blank line."
       unit)))
 
+(define <page-break>
+  (make-record-type '<page-break> '()
+                    #:parent <blank>
+                    #:extensible? #f))
+
+(define page-break?           (record-predicate <page-break>))
+(define page-break
+  (let ((break ((record-type-constructor <page-break>))))
+    (lambda ()
+      break)))
+
+
 (define <comment>
   ;; Comments.
   (make-record-type '<comment> '(str margin?)
@@ -105,12 +120,17 @@ end with newline, otherwise an error is raised."
             (&message (message "invalid comment string")))))
   (string->comment str margin?))
 
+(define char-set:whitespace-sans-page-break
+  ;; White space, excluding #\page.
+  (char-set-difference char-set:whitespace (char-set #\page)))
+
+(define (space? chr)
+  "Return true if CHR is white space, except for page breaks."
+  (char-set-contains? char-set:whitespace-sans-page-break chr))
+
 (define (read-vertical-space port)
   "Read from PORT until a non-vertical-space character is met, and return a
 single <vertical-space> record."
-  (define (space? chr)
-    (char-set-contains? char-set:whitespace chr))
-
   (let loop ((height 1))
     (match (read-char port)
       (#\newline (loop (+ 1 height)))
@@ -118,6 +138,15 @@ single <vertical-space> record."
       ((? space?) (loop height))
       (chr (unread-char chr port) (vertical-space height)))))
 
+(define (read-until-end-of-line port)
+  "Read white space from PORT until the end of line, included."
+  (let loop ()
+    (match (read-char port)
+      (#\newline #t)
+      ((? eof-object?) #t)
+      ((? space?) (loop))
+      (chr (unread-char chr port)))))
+
 (define (read-with-comments port)
   "Like 'read', but include <blank> objects when they're encountered."
   ;; Note: Instead of implementing this functionality in 'read' proper, which
@@ -148,6 +177,11 @@ single <vertical-space> record."
               (if blank-line?
                   (read-vertical-space port)
                   (loop #t return)))
+             ((eqv? chr #\page)
+              ;; Assume that a page break is on a line of its own and read
+              ;; subsequent white space and newline.
+              (read-until-end-of-line port)
+              (page-break))
              ((char-set-contains? char-set:whitespace chr)
               (loop blank-line? return))
              ((memv chr '(#\( #\[))
@@ -444,6 +478,12 @@ FORMAT-VERTICAL-SPACE; a useful value of 'canonicalize-vertical-space'."
            (loop (- i 1))))
        (display (make-string indent #\space) port)
        indent)
+      ((? page-break?)
+       (unless delimited? (newline port))
+       (display #\page port)
+       (newline port)
+       (display (make-string indent #\space) port)
+       indent)
       (('quote lst)
        (unless delimited? (display " " port))
        (display "'" port)
diff --git a/tests/read-print.scm b/tests/read-print.scm
index f915b7e2d2..70be7754f8 100644
--- a/tests/read-print.scm
+++ b/tests/read-print.scm
@@ -70,6 +70,21 @@
             (read-with-comments port)
             (read-with-comments port)))))
 
+(test-equal "read-with-comments: top-level page break"
+  (list (comment ";; Begin.\n") (vertical-space 1)
+        (page-break)
+        (comment ";; End.\n"))
+  (call-with-input-string "\
+;; Begin.
+
+
+;; End.\n"
+    (lambda (port)
+      (list (read-with-comments port)
+            (read-with-comments port)
+            (read-with-comments port)
+            (read-with-comments port)))))
+
 (test-pretty-print "(list 1 2 3 4)")
 (test-pretty-print "((a . 1) (b . 2))")
 (test-pretty-print "(a b c . boom)")
@@ -229,6 +244,13 @@ mnopqrstuvwxyz.\")"
                 ;; Comment after blank line.
                 two)")
 
+(test-pretty-print "\
+(begin
+  break
+
+  ;; page break above
+  end)")
+
 (test-equal "pretty-print-with-comments, canonicalize-comment"
   "\
 (list abc
-- 
cgit 1.4.1