From 47c1c8bc125d78d72b644331f4607442ad2bd627 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Linus=20Bj=C3=B6rnstam?= Date: Thu, 16 Jan 2020 12:26:32 +0100 Subject: [PATCH] * module/sxml/upstream/SSAX.scm: Fix improper handling of > in CDATA and correct tests. --- configure.ac | 4 ++-- module/sxml/upstream/SSAX.scm | 30 ++++++++---------------------- 2 files changed, 10 insertions(+), 24 deletions(-) diff --git a/configure.ac b/configure.ac index bb9a9281f..6198c7e6e 100644 --- a/configure.ac +++ b/configure.ac @@ -3,7 +3,7 @@ dnl Process this file with autoconf to produce configure. dnl define(GUILE_CONFIGURE_COPYRIGHT,[[ -Copyright 1998-2019 Free Software Foundation, Inc. +Copyright 1998-2020 Free Software Foundation, Inc. This file is part of Guile. @@ -25,7 +25,7 @@ License along with Guile. If not, see AC_PREREQ(2.61) AC_INIT([GNU Guile], - m4_esyscmd([build-aux/git-version-gen --match v2.9.\* .tarball-version]), + m4_esyscmd([build-aux/git-version-gen --match v3.0.\* .tarball-version]), [bug-guile@gnu.org]) AC_CONFIG_AUX_DIR([build-aux]) AC_CONFIG_MACRO_DIR([m4]) diff --git a/module/sxml/upstream/SSAX.scm b/module/sxml/upstream/SSAX.scm index d2b8fd925..4bfaa9bcb 100644 --- a/module/sxml/upstream/SSAX.scm +++ b/module/sxml/upstream/SSAX.scm @@ -950,11 +950,10 @@ ; CR, LF, and CRLF are treated as line delimiters, and passed ; as a single #\newline to the STR-HANDLER ; "]]>" combination is the end of the CDATA section. -; > is treated as an embedded #\> character -; Note, < and & are not specially recognized (and are not expanded)! +; Note, <, > and & are not specially recognized (and are not expanded)! (define ssax:read-cdata-body - (let ((cdata-delimiters (list char-return #\newline #\] #\&))) + (let ((cdata-delimiters (list char-return #\newline #\]))) (lambda (port str-handler seed) (let loop ((seed seed)) @@ -974,18 +973,6 @@ ((#\]) (check-after-second-braket (str-handler "]" "" seed))) (else (loop (str-handler "]]" "" seed))))))) - ((#\&) ; Note that #\& within CDATA may stand for itself - (let ((ent-ref ; it does not have to start an entity ref - (next-token-of (lambda (c) - (and (not (eof-object? c)) (char-alphabetic? c) c)) port))) - (cond ; ">" is to be replaced with #\> - ((and (string=? "gt" ent-ref) (eqv? (peek-char port) #\;)) - (read-char port) - (loop (str-handler fragment ">" seed))) - (else - (loop - (str-handler ent-ref "" - (str-handler fragment "&" seed))))))) (else ; Must be CR: if the next char is #\newline, skip it (if (eqv? (peek-char port) #\newline) (read-char port)) (loop (str-handler fragment nl seed))) @@ -1004,7 +991,7 @@ (call-with-input-string (unesc-string str) (lambda (port) (ssax:read-cdata-body port consumer '())) )))) - (write result) + (write result) (assert (equal? result expected-result))))) ) (test "]]>" '()) @@ -1017,10 +1004,9 @@ (test "%r%n%r%n]]>" '("" " NL" "" " NL")) (test "%r%n%r%na]]>" '("" " NL" "" " NL" "a" "")) (test "%r%r%r%na]]>" '("" " NL" "" " NL" "" " NL" "a" "")) - (test "abc&!!!]]>" '("abc" "&" "" "" "!!!" "")) + (test "abc&!!!]]>" '("abc&!!!" "")) (test "abc]]>>&]]]>and]]>" - '("abc" "" "]]" "" "" ">" "" "&" "gt" "" "" "&" "amp" "" ";" "" "]" "" - "]]" "" "" ">" "and" "")) + '("abc" "" "]]" "" ">>&" "" "]" "" "]]" "" ">and" "")) )) @@ -2596,10 +2582,10 @@ `(('"itemize" ('"item" "This is item 1 ") ,(unesc-string "%n") ('"item" "Item 2") ,(unesc-string "%n ")))) (test "

%n]]>]]>

" - dummy-doctype-fn `(('"P" "
" ,nl "" "]]" "" ">"))) + dummy-doctype-fn `(('"P" "
" ,nl "" "]]" ">"))) (test "

%r]]>]]>

" - dummy-doctype-fn `(('"P" "
" ,nl "" "]]" "" ">"))) + dummy-doctype-fn `(('"P" "
" ,nl "" "]]" ">"))) (test "%n%n" dummy-doctype-fn '(('"Reports" (@ ('"TStamp" "1"))))) @@ -2964,7 +2950,7 @@ `(*TOP* (P ,(unesc-string "some text <1%n\"") (B "strong") ,(unesc-string "\"%n")))) (test "

%n]]>]]>

" '() - `(*TOP* (P ,(unesc-string "
%n]]>")))) + `(*TOP* (P ,(unesc-string "
%n]]>")))) ; (test "it's%r%nand that%n%r%n%r%n%n" '() ; '(*TOP* (T1 (T2 "it's%nand that%n") "%n%n%n"))) (test "it's%r%nand that%n%r%n%r%n%n" '() -- 2.24.1