courier-unicode: unicode_bidi_cleaned_size, unicode::bidi_override

10 files changed, 211 insertions, 19 deletions

diff --git a/unicode/ChangeLog b/unicode/ChangeLog
index 1995736..fcb1c10 100644
--- a/unicode/ChangeLog
+++ b/unicode/ChangeLog
@@ -1,3 +1,5 @@
+2.2.1
+
 2021-02-14  Sam Varshavchik  <mrsam@courier-mta.com>
 
 	* unicode_bidi_calc and unicode_bidi_calc_levels return a
diff --git a/unicode/Makefile.am b/unicode/Makefile.am
index 32380d3..5877d22 100644
--- a/unicode/Makefile.am
+++ b/unicode/Makefile.am
@@ -97,6 +97,7 @@ man_MANS= \
         $(srcdir)/man/unicode[\:][\:]bidi_embed_paragraph_level.3 \
         $(srcdir)/man/unicode[\:][\:]bidi_get_direction.3 \
         $(srcdir)/man/unicode[\:][\:]bidi_logical_order.3 \
+        $(srcdir)/man/unicode[\:][\:]bidi_override.3 \
         $(srcdir)/man/unicode[\:][\:]bidi_reorder.3 \
         $(srcdir)/man/unicode[\:][\:]iconvert[\:][\:]convert.3 \
         $(srcdir)/man/unicode[\:][\:]iconvert[\:][\:]convert_tocase.3 \
@@ -119,6 +120,7 @@ man_MANS= \
         $(srcdir)/man/unicode_bidi_calc.3 \
         $(srcdir)/man/unicode_bidi_calc_levels.3 \
         $(srcdir)/man/unicode_bidi_calc_types.3 \
+        $(srcdir)/man/unicode_bidi_cleaned_size.3 \
         $(srcdir)/man/unicode_bidi_cleanup.3 \
         $(srcdir)/man/unicode_bidi_direction.3 \
         $(srcdir)/man/unicode_bidi_embed.3 \
@@ -219,7 +221,7 @@ libcourier_unicode_la_SOURCES=\
 			bidi_mirroring.h \
 			unicode_categories.c
 
-libcourier_unicode_la_LDFLAGS=-version-info 6:0:2
+libcourier_unicode_la_LDFLAGS=-version-info 7:0:0
 
 EXTRA_DIST=$(noinst_SCRIPTS) $(man_MANS) $(PACKAGE).spec \
 	m4/courier-unicode.m4 \
diff --git a/unicode/biditest2.C b/unicode/biditest2.C
index 7787a33..6ab347b 100644
--- a/unicode/biditest2.C
+++ b/unicode/biditest2.C
@@ -274,6 +274,9 @@ void character_test()
 				 std::reverse(b+index, b+index+n);
 			 });
 
+		size_t cleaned_size=unicode_bidi_cleaned_size(s.c_str(),
+							      s.size(), 0);
+
 		n=0;
 		unicode::bidi_cleanup
 			(s, levels,
@@ -285,6 +288,17 @@ void character_test()
 				 ++n;
 			 });
 
+		if (cleaned_size != s.size())
+		{
+			std::cerr << "Regression, line "
+				  << linenum
+				  << ": default cleaned size"
+				  << std::endl
+				  << "   Expected size: " << cleaned_size
+				  << ", actual size: " << s.size()
+				  << std::endl;
+			exit(1);
+		}
 		if (render_order != actual_render_order)
 		{
 			std::cerr << "Regression, line "
@@ -408,6 +422,12 @@ void character_test()
 			}
 
 			unicode::bidi_reorder(new_string, std::get<0>(ret));
+
+			cleaned_size=unicode_bidi_cleaned_size
+				(new_string.c_str(),
+				 new_string.size(),
+				 UNICODE_BIDI_CLEANUP_CANONICAL);
+
 			unicode::bidi_cleanup(new_string,
 					      std::get<0>(ret),
 					      []
@@ -416,6 +436,20 @@ void character_test()
 					      },
 					      UNICODE_BIDI_CLEANUP_CANONICAL);
 
+			if (cleaned_size != new_string.size())
+			{
+				std::cerr << "Regression, line "
+					  << linenum
+					  << ": canonoical cleaned size"
+					  << std::endl
+					  << "   Expected size: "
+					  << cleaned_size
+					  << ", actual size: "
+					  << new_string.size()
+					  << std::endl;
+				exit(1);
+			}
+
 			/* New string is now back in logical order */
 
 			if (new_string == s && std::get<0>(ret) == levels)
diff --git a/unicode/book.xml b/unicode/book.xml
index 2a83033..0b45433 100644
--- a/unicode/book.xml
+++ b/unicode/book.xml
@@ -334,6 +334,7 @@ make install DESTDIR=/tmp/courier-unicode-instimage # For example.</programlisti
 	  <refname>unicode_bidi_calc</refname>
 	  <refname>unicode_bidi_reorder</refname>
 	  <refname>unicode_bidi_cleanup</refname>
+	  <refname>unicode_bidi_cleaned_size</refname>
 	  <refname>unicode_bidi_logical_order</refname>
 	  <refname>unicode_bidi_embed</refname>
 	  <refname>unicode_bidi_embed_paragraph_level</refname>
@@ -395,6 +396,13 @@ make install DESTDIR=/tmp/courier-unicode-instimage # For example.</programlisti
 	    </funcprototype>
 
 	    <funcprototype>
+	      <funcdef>size_t <function>unicode_bidi_cleaned_size</function></funcdef>
+              <paramdef>const char32_t *<parameter>string</parameter></paramdef>
+              <paramdef>size_t <parameter>n</parameter></paramdef>
+	      <paramdef>int <parameter>options</parameter></paramdef>
+	    </funcprototype>
+
+	    <funcprototype>
 	      <funcdef>size_t <function>unicode_bidi_logical_order</function></funcdef>
               <paramdef>char32_t *<parameter>string</parameter></paramdef>
               <paramdef>unicode_bidi_level_t *<parameter>levels</parameter></paramdef>
@@ -538,6 +546,9 @@ make install DESTDIR=/tmp/courier-unicode-instimage # For example.</programlisti
 		  to remove the characters from the string which are used
 		  by the bi-directional algorithm, and are not needed for
 		  rendering the text.
+		  <function>unicode_bidi_cleaned_size</function>() is
+		  available to determine, in advance, how many characters
+		  will remain.
 		</para>
 	      </listitem>
 	    </orderedlist>
@@ -865,6 +876,11 @@ make install DESTDIR=/tmp/courier-unicode-instimage # For example.</programlisti
 	      canonical rendering order for their string and embedding level
 	      values.
             </para>
+	    <para>
+	      The parameters to <function>unicode_bidi_cleaned_size</function>()
+	      are a pointer to the unicode string, its size, and
+	      the bitmask option to <function>unicode_bidi_cleanup</function>().
+            </para>
 	  </refsect2>
 
 	  <refsect2 id="unicode_bidi_embed">
@@ -2906,6 +2922,7 @@ make install DESTDIR=/tmp/courier-unicode-instimage # For example.</programlisti
 	  <refname>unicode::bidi_embed</refname>
 	  <refname>unicode::bidi_embed_paragraph_level</refname>
 	  <refname>unicode::bidi_get_direction</refname>
+	  <refname>unicode::bidi_override</refname>
 	  <refpurpose>unicode bi-directional algorithm</refpurpose>
 	</refnamediv>
 
@@ -3035,6 +3052,13 @@ make install DESTDIR=/tmp/courier-unicode-instimage # For example.</programlisti
 	      <paramdef>size_t <parameter>starting_pos</parameter>=0</paramdef>
 	      <paramdef>size_t <parameter>n</parameter>=(size_t)-1</paramdef>
 	    </funcprototype>
+
+	    <funcprototype>
+	      <funcdef>std::u32string <function>bidi_override</function></funcdef>
+              <paramdef>const std::u32string &amp;<parameter>string</parameter></paramdef>
+              <paramdef>unicode_bidi_level_t <parameter>direction</parameter></paramdef>
+	      <paramdef>int <parameter>cleanup_options</parameter>=0</paramdef>
+	    </funcprototype>
           </funcsynopsis>
 	</refsynopsisdiv>
 
@@ -3190,6 +3214,41 @@ auto [levels, direction]=unicode::bidi_calc(types);
 	      </para>
 	    </listitem>
 
+	    <listitem>
+	      <para>
+		<function>unicode::bidi_override</function>
+		modifies the passed-in <parameter>string</parameter> as
+		follows:
+	      </para>
+
+	      <itemizedlist>
+		<listitem>
+		  <para>
+		    <function>unicode::bidi_cleanup</function>() is applied
+		    with the specified, or defaulted,
+		    <replaceable>cleanup_options</replaceable>
+		  </para>
+		</listitem>
+
+		<listitem>
+		  <para>
+		    Either the <literal>LRO</literal> or an
+		    <literal>RLO</literal> override marker gets prepended
+		    to the Unicode string, forcing the entire string to
+		    be interpreted in a single rendering direction, when
+		    processed by the Unicode bi-directional algorithm.
+		  </para>
+		</listitem>
+	      </itemizedlist>
+
+	      <para>
+		<function>unicode::bidi_override</function> makes it
+		possible to use a Unicode-aware application or algorithm
+		in a context that only works with text that's always
+		displayed in a fixed direction, allowing graceful handling
+		of input containing bi-directional text.
+	      </para>
+	    </listitem>
           </itemizedlist>
 
 	  <refsect2 id="unicode_cpp_bidi_literals">
diff --git a/unicode/configure.ac b/unicode/configure.ac
index 66c1b97..1cc3b76 100644
--- a/unicode/configure.ac
+++ b/unicode/configure.ac
@@ -117,7 +117,19 @@ fi
 CFLAGS="-I.. -I$srcdir/.. $CFLAGS"
 CXXFLAGS="-I.. -I$srcdir/.. $CXXFLAGS"
 
-HVERSION="`echo $VERSION | tr -d '.'`"
+set -- `echo "$VERSION" | tr '.' ' '`
+
+v=$1
+r=`echo "00"$2 | sed 's/.*(...)$/$1/'`
+p=$3
+
+if test "$p" = ""
+   then p="0"
+fi
+
+p=`echo "00"$p | sed 's/.*(...)$/$[]1/'`
+
+HVERSION="$v$r$p"
 AC_SUBST(HVERSION)
 
 AM_CONDITIONAL(HAVE_DOCS,[test -f $srcdir/docbook/icon.gif])
diff --git a/unicode/courier-unicode-version.m4.in b/unicode/courier-unicode-version.m4.in
index d40b5b7..94b0c04 100644
--- a/unicode/courier-unicode-version.m4.in
+++ b/unicode/courier-unicode-version.m4.in
@@ -14,13 +14,16 @@ fi
 set -- `echo "$vers" | tr '.' ' '`
 
 v=$[]1
-r=$[]2
+r=`echo "00"$[]2 | sed 's/.*(...)$/$[]1/'`
+
 p=$[]3
 
 if test "$p" = ""
    then p="0"
 fi
 
+p=`echo "00"$p | sed 's/.*(...)$/$[]1/'`
+
 AC_TRY_COMPILE([
 #include <courier-unicode.h>
 #ifndef COURIER_UNICODE_VERSION
diff --git a/unicode/courier-unicode.h.in b/unicode/courier-unicode.h.in
index e0a5b99..a1a502c 100644
--- a/unicode/courier-unicode.h.in
+++ b/unicode/courier-unicode.h.in
@@ -705,6 +705,10 @@ extern size_t unicode_bidi_cleanup(char32_t *string,
 				   void (*removed_callback)(size_t, void *),
 				   void *);
 
+extern size_t unicode_bidi_cleaned_size(const char32_t *string,
+					size_t n,
+					int options);
+
 extern void unicode_bidi_logical_order(char32_t *string,
 				       unicode_bidi_level_t *levels,
 				       size_t n,
@@ -2354,14 +2358,19 @@ std::u32string bidi_embed(const std::u32string &string,
 //! In order for the unicode string to have the specified default
 //! paragraph embedding level.
 
-extern char32_t bidi_embed_paragraph_level(const std::u32string &string,
-					   unicode_bidi_level_t level);
+char32_t bidi_embed_paragraph_level(const std::u32string &string,
+				    unicode_bidi_level_t level);
 
 //! Compute default direction of text
 
-extern unicode_bidi_direction bidi_get_direction(const std::u32string &string,
-						 size_t starting_pos=0,
-						 size_t n=(size_t)-1);
+unicode_bidi_direction bidi_get_direction(const std::u32string &string,
+					  size_t starting_pos=0,
+					  size_t n=(size_t)-1);
+
+//! Override bidi direction.
+std::u32string bidi_override(const std::u32string &s,
+			     unicode_bidi_level_t direction,
+			     int cleanup_options=0);
 
 #if 0
 {
diff --git a/unicode/courier-unicode.spec.in b/unicode/courier-unicode.spec.in
index f7d1eb6..440d6f1 100644
--- a/unicode/courier-unicode.spec.in
+++ b/unicode/courier-unicode.spec.in
@@ -1,5 +1,12 @@
 Summary: Courier Unicode Library
+%if 0%{?compat:1}
+Name: courier-unicode%(echo @VERSION@ | tr -d '.')
+
+%define __brp_ldconfig %{nil}
+
+%else
 Name: courier-unicode
+%endif
 Version: @VERSION@
 Release: 1%{?dist}%{?courier_release}
 License: GPLv3
@@ -11,10 +18,14 @@ BuildRequires: perl
 BuildRequires: gcc-c++
 BuildRequires: %{__make}
 
+%if 0%{?compat:1}
+
+%else
 %package devel
 Summary: Courier Unicode Library development files
 Group: Development/Libraries
 Requires: %{name} = 0:%{version}-%{release}
+%endif
 
 %description
 This library implements several algorithms related to the Unicode
@@ -24,13 +35,17 @@ This package installs only the run-time libraries needed by applications that
 use this library. Install the "courier-unicode-devel" package if you want
 to develop new applications using this library.
 
+%if 0%{?compat:1}
+
+%else
 %description devel
 This package contains development files for the Courier Unicode Library.
 Install this package if you want to develop applications that uses this
 unicode library.
+%endif
 
 %prep
-%setup -q
+%setup -q -n courier-unicode-@VERSION@
 %configure
 %build
 %{__make} -s %{?_smp_mflags}
@@ -39,12 +54,27 @@ unicode library.
 rm -rf $RPM_BUILD_ROOT
 %{__make} install DESTDIR=$RPM_BUILD_ROOT
 
-%post -p /sbin/ldconfig
-%postun -p /sbin/ldconfig
+%if 0%{?compat:1}
+find $RPM_BUILD_ROOT%{_libdir} -type l -print | xargs rm -f
+rm -rf $RPM_BUILD_ROOT%{_includedir}
+rm -f $RPM_BUILD_ROOT%{_libdir}/*.a
+rm -f $RPM_BUILD_ROOT%{_libdir}/*.la
+rm -rf $RPM_BUILD_ROOT%{_datadir}/aclocal
+rm -rf $RPM_BUILD_ROOT%{_mandir}
+%endif
 
 %clean
 rm -rf $RPM_BUILD_ROOT
 
+%post -p /sbin/ldconfig
+%postun -p /sbin/ldconfig
+
+%if 0%{?compat:1}
+%files
+%defattr(-,root,root,-)
+%{_libdir}/*.so.*
+
+%else
 
 %files
 %defattr(-,root,root,-)
@@ -58,6 +88,7 @@ rm -rf $RPM_BUILD_ROOT
 %{_libdir}/*.la
 %{_libdir}/*.a
 %{_datadir}/aclocal/*.m4
+%endif
 
 %changelog
 * Sun Jan 12 2014 Sam Varshavchik <mrsam@octopus.email-scan.com> - 1.0
diff --git a/unicode/unicode_bidi.c b/unicode/unicode_bidi.c
index b97ec25..1aa4a88 100644
--- a/unicode/unicode_bidi.c
+++ b/unicode/unicode_bidi.c
@@ -2063,12 +2063,14 @@ void unicode_bidi_reorder(char32_t *p,
 	level_run_layers_deinit(&layers);
 }
 
-size_t unicode_bidi_cleanup(char32_t *string,
-			    unicode_bidi_level_t *levels,
-			    size_t n,
-			    int cleanup_options,
-			    void (*removed_callback)(size_t, void *),
-			    void *arg)
+static size_t unicode_bidi_count_or_cleanup(const char32_t *string,
+					    char32_t *dest,
+					    unicode_bidi_level_t *levels,
+					    size_t n,
+					    int cleanup_options,
+					    void (*removed_callback)(size_t,
+								     void *),
+					    void *arg)
 {
 	size_t i=0;
 	for (size_t j=0; j<n; ++j)
@@ -2090,13 +2092,34 @@ size_t unicode_bidi_cleanup(char32_t *string,
 		if (levels)
 			levels[i]=levels[j] & 1;
 
-		string[i]=(cleanup_options & UNICODE_BIDI_CLEANUP_BNL)
-			&& cl == UNICODE_BIDI_TYPE_B ? '\n' : string[j];
+		if (dest)
+			dest[i]=(cleanup_options & UNICODE_BIDI_CLEANUP_BNL)
+				&& cl == UNICODE_BIDI_TYPE_B ? '\n' : string[j];
 		++i;
 	}
 	return i;
 }
 
+size_t unicode_bidi_cleanup(char32_t *string,
+			    unicode_bidi_level_t *levels,
+			    size_t n,
+			    int cleanup_options,
+			    void (*removed_callback)(size_t, void *),
+			    void *arg)
+{
+	return unicode_bidi_count_or_cleanup(string, string, levels, n,
+					     cleanup_options, removed_callback,
+					     arg);
+}
+
+size_t unicode_bidi_cleaned_size(const char32_t *string,
+				 size_t n,
+				 int cleanup_options)
+{
+	return unicode_bidi_count_or_cleanup(string, NULL, NULL, n,
+					     cleanup_options, NULL, NULL);
+}
+
 void unicode_bidi_logical_order(char32_t *string,
 				unicode_bidi_level_t *levels,
 				size_t n,
diff --git a/unicode/unicodecpp.C b/unicode/unicodecpp.C
index e91f756..04d9879 100644
--- a/unicode/unicodecpp.C
+++ b/unicode/unicodecpp.C
@@ -949,3 +949,20 @@ unicode_bidi_direction unicode::bidi_get_direction(const std::u32string &string,
 
 	return unicode_bidi_get_direction(string.c_str()+starting_pos, n);
 }
+
+std::u32string unicode::bidi_override(const std::u32string &s,
+				      unicode_bidi_level_t direction,
+				      int cleanup_options)
+{
+	std::u32string ret;
+
+	ret.reserve(s.size()+1);
+
+	ret.push_back(' ');
+	ret.insert(ret.end(), s.begin(), s.end());
+
+	bidi_cleanup(ret, [](size_t) {}, cleanup_options);
+	ret.at(0)=direction & 1 ? UNICODE_RLO : UNICODE_LRO;
+
+	return ret;
+}