# HG changeset patch
# User ali <ali@juiblex.co.uk>
# Date 1327883791 0
# Node ID 6a13fe0fc19e2ae100e0986865f6538c60d81787
# Parent  cf332d440466601401a5a8ce21918dee775cec27
Add a testcase for user-defined typos

diff -r cf332d440466 -r 6a13fe0fc19e bl/Makefile.am
--- a/bl/Makefile.am	Fri Jan 27 23:59:51 2012 +0000
+++ b/bl/Makefile.am	Mon Jan 30 00:36:31 2012 +0000
@@ -3,4 +3,5 @@
 LIBS=$(GLIB_LIBS)
 
 noinst_LTLIBRARIES=libbl.la
-libbl_la_SOURCES=bl.h textfileutils.c textfileutils.h spawn.c spawn.h
+libbl_la_SOURCES=bl.h textfileutils.c textfileutils.h spawn.c spawn.h \
+	path.c path.h mkdtemp.c mkdtemp.h
diff -r cf332d440466 -r 6a13fe0fc19e bl/bl.h
--- a/bl/bl.h	Fri Jan 27 23:59:51 2012 +0000
+++ b/bl/bl.h	Mon Jan 30 00:36:31 2012 +0000
@@ -1,2 +1,4 @@
 #include <bl/textfileutils.h>
 #include <bl/spawn.h>
+#include <bl/path.h>
+#include <bl/mkdtemp.h>
diff -r cf332d440466 -r 6a13fe0fc19e bl/mkdtemp.c
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/bl/mkdtemp.c	Mon Jan 30 00:36:31 2012 +0000
@@ -0,0 +1,35 @@
+#include <stdlib.h>
+#include <string.h>
+#include <errno.h>
+#include <glib.h>
+#include <glib/gstdio.h>
+#include "mkdtemp.h"
+
+#if !HAVE_G_MKDTEMP
+char *g_mkdtemp(char *template)
+{
+#if !defined(WIN32) && HAVE_MKDTEMP
+    return mkdtemp(template);
+#else
+    char *s;
+    for(;;)
+    {
+	s=g_strdup(template);
+	mktemp(s);
+	if (!*s)
+	{
+	    g_free(s);
+	    errno=EEXIST;
+	    return NULL;
+	}
+	if (g_mkdir(s,0700)>=0)
+	{
+	    strcpy(template,s);
+	    g_free(s);
+	    return template;
+	}
+	g_free(s);
+    }
+#endif	/* !defined(WIN32) && HAVE_MKDTEMP */
+}
+#endif	/* !HAVE_G_MKDTEMP */
diff -r cf332d440466 -r 6a13fe0fc19e bl/mkdtemp.h
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/bl/mkdtemp.h	Mon Jan 30 00:36:31 2012 +0000
@@ -0,0 +1,8 @@
+#ifndef BL_MKDTEMP_H
+#define BL_MKDTEMP_H
+
+#if !HAVE_G_MKDTEMP
+char *g_mkdtemp(char *template);
+#endif
+
+#endif /* BL_MKDTEMP_H */
diff -r cf332d440466 -r 6a13fe0fc19e bl/path.c
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/bl/path.c	Mon Jan 30 00:36:31 2012 +0000
@@ -0,0 +1,53 @@
+#ifdef WIN32
+#include <windows.h>
+#endif
+#include <stdlib.h>
+#include <glib.h>
+#include <bl/bl.h>
+
+/*
+ * Return an absolute path to <path>.
+ * Note that this function makes no attempt to return a unique path, or
+ * to remove "." or ".." entries. It simply returns a path which will
+ * be unaffected by subsequent calls to chdir().
+ */
+char *path_to_absolute(const char *path)
+{
+#ifdef WIN32
+    long len;
+    gunichar2 *path2;
+    gunichar2 *abs2;
+    char *abs;
+    path2=g_utf8_to_utf16(path,-1,NULL,NULL,NULL);
+    if (!path2)
+	return NULL;
+    len=GetFullPathNameW(path2,0,NULL,NULL);	/* len includes nul */
+    if (!len)
+    {
+	g_free(path2);
+	return NULL;
+    }
+    abs2=g_new(gunichar2,len);
+    len=GetFullPathNameW(path2,len,abs2,NULL);	/* len excludes nul */
+    g_free(path2);
+    if (!len)
+    {
+	g_free(abs2);
+	return NULL;
+    }
+    abs=g_utf16_to_utf8(abs2,len,NULL,NULL,NULL);
+    g_free(abs2);
+    return abs;
+#else
+    char *s,*abs;
+    if (*path=='/')
+	abs=g_strdup(path);
+    else
+    {
+	s=g_get_current_dir();
+	abs=g_build_filename(s,path,NULL);
+	g_free(s);
+    }
+    return abs;
+#endif
+}
diff -r cf332d440466 -r 6a13fe0fc19e bl/path.h
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/bl/path.h	Mon Jan 30 00:36:31 2012 +0000
@@ -0,0 +1,6 @@
+#ifndef BL_PATH_H
+#define BL_PATH_H
+
+char *path_to_absolute(const char *path);
+
+#endif	/* BL_PATH_H */
diff -r cf332d440466 -r 6a13fe0fc19e bl/spawn.c
--- a/bl/spawn.c	Fri Jan 27 23:59:51 2012 +0000
+++ b/bl/spawn.c	Mon Jan 30 00:36:31 2012 +0000
@@ -7,8 +7,8 @@
 
 #define SPAWN_BUFSIZE	128
 
-gboolean spawn_sync(char **argv,char **standard_output,int *exit_status,
-  GError **error)
+gboolean spawn_sync(const char *working_directory,char **argv,
+  char **standard_output,int *exit_status,GError **error)
 {
 /* Don't use g_spawn_sync on WIN32 for now to avoid needing the helper */
 #ifndef WIN32
@@ -17,8 +17,8 @@
     GSpawnFlags flags=G_SPAWN_SEARCH_PATH;
     if (!standard_output)
 	flags=G_SPAWN_STDOUT_TO_DEV_NULL;
-    retval=g_spawn_sync(NULL,argv,NULL,flags,NULL,NULL,standard_output,
-      &standard_error,exit_status,error);
+    retval=g_spawn_sync(working_directory,argv,NULL,flags,NULL,NULL,
+      standard_output,&standard_error,exit_status,error);
     if (standard_error)
 	fputs(standard_error,stderr);
     g_free(standard_error);
@@ -29,7 +29,21 @@
     FILE *fp;
     int i,r;
     size_t n,len;
+    char *current_dir;
     GString *command_line,*string;
+    if (working_directory)
+    {
+	current_dir=g_get_current_dir();
+	if (g_chdir(working_directory)<0)
+	{
+	    g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
+	      "%s: %s",working_directory,g_strerror(errno));
+	    g_free(current_dir);
+	    return FALSE;
+	}
+    }
+    else
+	current_dir=NULL;
     command_line=g_string_new(NULL);
     for(i=0;argv[i];i++)
     {
@@ -39,6 +53,13 @@
     }
     fp=popen(command_line->str,"r");
     g_string_free(command_line,TRUE);
+    if (current_dir)
+    {
+	if (g_chdir(current_dir)<0)
+	    g_error("Failed to restore current directory: %s",
+	      g_strerror(errno));
+	g_free(current_dir);
+    }
     if (!fp)
     {
 	g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
diff -r cf332d440466 -r 6a13fe0fc19e bl/spawn.h
--- a/bl/spawn.h	Fri Jan 27 23:59:51 2012 +0000
+++ b/bl/spawn.h	Mon Jan 30 00:36:31 2012 +0000
@@ -3,7 +3,7 @@
 
 #include <glib.h>
 
-gboolean spawn_sync(char **argv,char **standard_output,int *exit_status,
-  GError **error);
+gboolean spawn_sync(const char *working_directory,char **argv,
+  char **standard_output,int *exit_status,GError **error);
 
 #endif /* BL_SPAWN_H */
diff -r cf332d440466 -r 6a13fe0fc19e configure.ac
--- a/configure.ac	Fri Jan 27 23:59:51 2012 +0000
+++ b/configure.ac	Mon Jan 30 00:36:31 2012 +0000
@@ -45,6 +45,9 @@
 AS_IF([test "$enable_shared" = no],[
   LDFLAGS="$LDFLAGS -static-libtool-libs"
 ])
+LT_OUTPUT
+ac_compile="./libtool --mode=compile --tag=CC $ac_compile"
+ac_link="./libtool --mode=link --tag=CC $ac_link"
 PKG_PROG_PKG_CONFIG
 
 ##################################################
@@ -60,15 +63,6 @@
 ##################################################
 PKG_CHECK_MODULES([GLIB],[glib-2.0])
 
-# NOTE: If we are using a static version of glib then we
-# should define GLIB_STATIC_COMPILATION. This isn't needed
-# when glib is built only for static use (in which case
-# glibconfig.h will already define GLIB_STATIC_COMPILATION).
-# It's not easy to tell if libtool will actually link with
-# a static glib but luckily we don't currently need to;
-# this pre-processor define only affects the behaviour of
-# libraries which use glib and we don't have any.
-
 # Glib 2.30 (at least) doesn't include "-framework Carbon" in Libs.private
 # which causes link to fail for static glib on Mac (gnome bug #668152).
 # Test for this problem and implement a workaround.
@@ -97,10 +91,39 @@
   ])
 ])
 
+# If we are using a static version of glib then we should define
+# GLIB_STATIC_COMPILATION. This isn't needed when glib is built
+# only for static use (in which case glibconfig.h will already
+# define GLIB_STATIC_COMPILATION).
+AS_IF([test "$enable_shared" = no],[
+  AC_MSG_CHECKING([whether GLIB_STATIC_COMPILATION needs to be defined])
+  save_CFLAGS="$CFLAGS"
+  save_LIBS="$LIBS"
+  CFLAGS="$GLIB_CFLAGS"
+  LIBS="$GLIB_LIBS"
+  AC_TRY_LINK([
+  #include <glib.h>
+  ],[
+  return glib_major_version!=2;
+  ],[static_compilation=no],[static_compilation=yes])
+  CFLAGS="$save_CFLAGS"
+  LIBS="$save_LIBS"
+  AC_MSG_RESULT([$static_compilation])
+  AS_IF([test "$static_compilation" = yes],[
+    GLIB_CFLAGS="$GLIB_CFLAGS -DGLIB_STATIC_COMPILATION"
+  ])
+])
+
 ##################################################
 # Checks for library functions.
 ##################################################
-AC_CHECK_FUNCS_ONCE([mkstemp])
+save_CFLAGS="$CFLAGS"
+save_LIBS="$LIBS"
+CFLAGS="$GLIB_CFLAGS"
+LIBS="$GLIB_LIBS"
+AC_CHECK_FUNCS([g_mkdtemp mkdtemp],[break])
+CFLAGS="$save_CFLAGS"
+LIBS="$save_LIBS"
 
 ##################################################
 # Checks for processor independent files.
diff -r cf332d440466 -r 6a13fe0fc19e test/compatibility/Makefile.am
--- a/test/compatibility/Makefile.am	Fri Jan 27 23:59:51 2012 +0000
+++ b/test/compatibility/Makefile.am	Mon Jan 30 00:36:31 2012 +0000
@@ -2,6 +2,7 @@
 TESTS=missing-space.tst spaced-punctuation.tst html-tag.tst html-symbol.tst \
 	spaced-doublequote.tst mismatched-quotes.tst he-be.tst digits.tst \
 	extra-period.tst ellipsis.tst short-line.tst abbreviation.tst \
-	example.tst non-ascii.tst embedded-lf.tst
+	example.tst non-ascii.tst embedded-lf.tst markup.tst \
+	user-defined-typo.tst
 
 dist_pkgdata_DATA=$(TESTS)
diff -r cf332d440466 -r 6a13fe0fc19e test/compatibility/markup.tst
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/compatibility/markup.tst	Mon Jan 30 00:36:31 2012 +0000
@@ -0,0 +1,6 @@
+**************** OPTIONS ****************
+-m
+-d
+**************** INPUT ****************
+&ldquo;He went <i>thataway!</i>&rdquo;
+**************** EXPECTED ****************
diff -r cf332d440466 -r 6a13fe0fc19e test/compatibility/user-defined-typo.tst
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/compatibility/user-defined-typo.tst	Mon Jan 30 00:36:31 2012 +0000
@@ -0,0 +1,17 @@
+**************** OPTIONS ****************
+-u
+**************** INPUT(gutcheck.typ) ****************
+arid
+**************** INPUT ****************
+I am the very model of a modern Major-General,
+I've information vegetable, animal, and mineral,
+I know the kings of England, arid I quote the fights historical
+From Marathon to Waterloo, in order categorical;
+I'm very well acquainted, too, with matters mathematical,
+I understand equations, both the simple and quadratical,
+About binomial theorem I'm teeming with a lot o' news--
+With many cheerful facts about the square of the hypotenuse.
+**************** EXPECTED ****************
+
+I know the kings of England, arid I quote the fights historical
+    Line 3 column 29 - Query possible scanno arid
diff -r cf332d440466 -r 6a13fe0fc19e test/harness/Makefile.am
--- a/test/harness/Makefile.am	Fri Jan 27 23:59:51 2012 +0000
+++ b/test/harness/Makefile.am	Mon Jan 30 00:36:31 2012 +0000
@@ -4,5 +4,6 @@
 LIBS=$(GLIB_LIBS)
 
 loupe_test_SOURCES=loupe-test.c testcase.c testcase.h testcaseio.c \
-	testcaseio.h testcaseparser.c testcaseparser.h
+	testcaseio.h testcaseparser.c testcaseparser.h testcaseinput.c \
+	testcaseinput.h
 loupe_test_LDADD=../../bl/libbl.la
diff -r cf332d440466 -r 6a13fe0fc19e test/harness/testcase.c
--- a/test/harness/testcase.c	Fri Jan 27 23:59:51 2012 +0000
+++ b/test/harness/testcase.c	Mon Jan 30 00:36:31 2012 +0000
@@ -1,10 +1,12 @@
 #include <stdlib.h>
 #include <stdio.h>
 #include <string.h>
-#include <unistd.h>
 #include <errno.h>
+#include <glib.h>
+#include <glib/gstdio.h>
 #include <bl/bl.h>
 #include "testcase.h"
+#include "testcaseinput.h"
 
 GQuark testcase_error_quark(void)
 {
@@ -12,20 +14,6 @@
 }
 
 /*
- * As write(), but with error handling.
- */
-static size_t write_file(int fd,const char *buf,size_t count,GError **error)
-{
-    if (write(fd,buf,count)<count)
-    {
-	g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
-	  "Error writing bookloupe input file: %s",g_strerror(errno));
-	return -1;
-    }
-    return count;
-}
-
-/*
  * Return the length (in bytes) of any common prefix between s1 and s2.
  * The returned length will always represent an exact number of characters.
  */
@@ -79,91 +67,144 @@
 }
 
 /*
- * Replace \n with \r\n and U+240A (visible symbol for LF) with \n
+ * Create all the input files needed by a testcase and, if required,
+ * a temporary directory in which to store them.
  */
-char *unix2dos(const char *text)
+gboolean testcase_create_input_files(Testcase *testcase,GError **error)
 {
-    gunichar c;
-    const gunichar visible_lf=0x240A;
-    GString *string;
-    string=g_string_new(NULL);
-    while(*text)
+    GSList *link,*link2;
+    if (testcase->flags&TESTCASE_TMP_DIR)
     {
-	c=g_utf8_get_char(text);
-	text=g_utf8_next_char(text);
-	if (c=='\n')
-	    g_string_append(string,"\r\n");
-	else if (c==visible_lf)
-	    g_string_append_c(string,'\n');
-	else
-	    g_string_append_unichar(string,c);
+	testcase->tmpdir=g_strdup("TEST-XXXXXX");
+	if (!g_mkdtemp(testcase->tmpdir))
+	{
+	    g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
+	      "Failed to create temporary directory: %s",g_strerror(errno));
+	    g_free(testcase->tmpdir);
+	    testcase->tmpdir=NULL;
+	    return FALSE;
+	}
     }
-    return g_string_free(string,FALSE);
+    for(link=testcase->inputs;link;link=link->next)
+	if (!testcase_input_create(testcase,link->data,error))
+	{
+	    for(link2=testcase->inputs;link2!=link;link2=link2->next)
+		(void)testcase_input_remove(testcase,link2->data,NULL);
+	    if (testcase->tmpdir)
+	    {
+		(void)g_rmdir(testcase->tmpdir);
+		g_free(testcase->tmpdir);
+		testcase->tmpdir=NULL;
+	    }
+	    return FALSE;
+	}
+    return TRUE;
 }
 
-gboolean spawn_bootloupe(const char *encoding,const char *standard_input,
-  char **standard_output,char **filename,GError **error)
+/*
+ * Remove all the input files used by a testcase and, if created,
+ * the temporary directory in which they are stored.
+ */
+gboolean testcase_remove_input_files(Testcase *testcase,GError **error)
+{
+    GSList *link;
+    GError *tmp_err=NULL;
+    gboolean retval=TRUE;
+    for(link=testcase->inputs;link;link=link->next)
+	if (!testcase_input_remove(testcase,link->data,&tmp_err))
+	{
+	    if (error && !*error)
+		g_propagate_error(error,tmp_err);
+	    else
+		g_clear_error(&tmp_err);
+	    retval=FALSE;
+	}
+    if (testcase->tmpdir)
+    {
+	if (g_rmdir(testcase->tmpdir))
+	{
+	    if (error && !*error)
+		g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
+		  "Failed to remove temporary directory: %s",g_strerror(errno));
+	    retval=FALSE;
+	}
+	g_free(testcase->tmpdir);
+	testcase->tmpdir=NULL;
+    }
+    return retval;
+}
+
+/*
+ * Replace every occurance of an input file name in <str> with the
+ * filename which holds that input. For input files with fixed names,
+ * this is a noop. For input files which use the "XXXXXX" sequence
+ * to create a unique filename, the XXXXXX will be replaced with the
+ * 6 characters that were chosen to be unique.
+ */
+char *testcase_resolve_input_files(Testcase *testcase,const char *str)
+{
+    GSList *link;
+    gsize offset,pos;
+    char *s;
+    TestcaseInput *input;
+    GString *filename=g_string_new(str);
+    for(link=testcase->inputs;link;link=link->next)
+    {
+	input=link->data;
+	if (!input->name_used)
+	{
+	    g_warning("%s: Input file uninstantiated",input->name);
+	    continue;
+	}
+	offset=0;
+	do
+	{
+	    s=strstr(filename->str+offset,input->name);
+	    if (s)
+	    {
+		pos=s-filename->str;
+		g_string_overwrite(filename,pos,input->name_used);
+		offset=pos+strlen(input->name);
+	    }
+	} while(s);
+    }
+    return g_string_free(filename,FALSE);
+}
+
+gboolean testcase_spawn_bookloupe(Testcase *testcase,char **standard_output,
+  GError **error)
 {
     gboolean r;
-    int fd,exit_status;
-    size_t n,pos,offset;
-    char input[]="TEST-XXXXXX";
-    char *command[3];
-    char *output,*s,*t;
+    int i,exit_status;
+    char **argv;
+    char *output,*s;
     GError *tmp_err=NULL;
-    if (standard_input)
-    {
-	if (encoding)
-	{
-	    t=unix2dos(standard_input);
-	    s=g_convert(t,-1,encoding,"UTF-8",NULL,&n,&tmp_err);
-	    g_free(t);
-	    if (!s)
-	    {
-		g_propagate_prefixed_error(error,tmp_err,
-		  "Conversion to %s failed: ",encoding);
-		return FALSE;
-	    }
-	}
-	else
-	{
-	    s=unix2dos(standard_input);
-	    n=strlen(s);
-	}
-    }
+    if (testcase->options)
+	argv=g_new(char *,g_strv_length(testcase->options)+3);
     else
-    {
-	n=0;
-	s=NULL;
-    }
-    fd=g_mkstemp(input);
-    if (n && write_file(fd,s,n,error)!=n)
-    {
-	g_free(s);
-	close(fd);
-	(void)remove(input);
-	return FALSE;
-    }
-    g_free(s);
-    close(fd);
-    command[0]=getenv("BOOKLOUPE");
-    if (!command[0])
-	command[0]="." G_DIR_SEPARATOR_S "bookloupe";
-    command[1]=input;
-    command[2]=NULL;
+	argv=g_new(char *,3);
+    s=getenv("BOOKLOUPE");
+    if (!s)
+	s="bookloupe";
+    argv[0]=path_to_absolute(s);
+    for(i=0;testcase->options && testcase->options[i];i++)
+	argv[i+1]=testcase_resolve_input_files(testcase,testcase->options[i]);
+    argv[i+1]=testcase_resolve_input_files(testcase,"TEST-XXXXXX");
+    argv[i+2]=NULL;
     if (standard_output)
     {
-	r=spawn_sync(command,&s,&exit_status,error);
+	r=spawn_sync(testcase->tmpdir,argv,&s,&exit_status,error);
 	if (r)
 	{
-	    if (encoding)
+	    if (testcase->encoding)
 	    {
-		output=g_convert(s,-1,"UTF-8",encoding,NULL,NULL,&tmp_err);
+		output=g_convert(s,-1,"UTF-8",testcase->encoding,NULL,NULL,
+		  &tmp_err);
 		g_free(s);
 		if (!output)
 		{
 		    g_propagate_prefixed_error(error,tmp_err,
-		      "Conversion from %s failed: ",encoding);
+		      "Conversion from %s failed: ",testcase->encoding);
 		    r=FALSE;
 		}
 	    }
@@ -182,18 +223,16 @@
     }
     else
     {
-	r=spawn_sync(command,NULL,&exit_status,error);
+	r=spawn_sync(testcase->tmpdir,argv,NULL,&exit_status,error);
 	output=NULL;
     }
-    (void)remove(input);
+    g_strfreev(argv);
     if (r && exit_status)
     {
 	g_set_error(error,TESTCASE_ERROR,TESTCASE_ERROR_FAILED,
 	  "bookloupe exited with code %d",exit_status);
 	r=FALSE;
     }
-    if (r && filename)
-	*filename=g_strdup(input);
     if (r && standard_output)
 	*standard_output=output;
     return r;
@@ -211,19 +250,34 @@
     GString *header,*expected;
     char *output,*filename,*s;
     GError *error=NULL;
+    if (!testcase_create_input_files(testcase,&error))
+    {
+	fprintf(stderr,"%s: FAIL\n",testcase->basename);
+	fprintf(stderr,"%s\n",error->message);
+	g_error_free(error);
+	return FALSE;
+    }
     if (testcase->expected)
-	r=spawn_bootloupe(testcase->encoding,testcase->input,&output,&filename,
-	  &error);
+	r=testcase_spawn_bookloupe(testcase,&output,&error);
     else
     {
-	r=spawn_bootloupe(testcase->encoding,testcase->input,NULL,NULL,&error);
-        output=filename=NULL;
+	r=testcase_spawn_bookloupe(testcase,NULL,&error);
+        output=NULL;
     }
     if (!r)
     {
 	fprintf(stderr,"%s: FAIL\n",testcase->basename);
 	fprintf(stderr,"%s\n",error->message);
 	g_error_free(error);
+	(void)testcase_remove_input_files(testcase,NULL);
+	return FALSE;
+    }
+    filename=testcase_resolve_input_files(testcase,"TEST-XXXXXX");
+    if (!testcase_remove_input_files(testcase,&error))
+    {
+	fprintf(stderr,"%s: FAIL\n",testcase->basename);
+	fprintf(stderr,"%s\n",error->message);
+	g_error_free(error);
 	return FALSE;
     }
     if (testcase->expected)
@@ -272,8 +326,8 @@
 	g_string_free(header,TRUE);
 	g_string_free(expected,TRUE);
     }
+    g_free(filename);
     g_free(output);
-    g_free(filename);
     if (r)
 	fprintf(stderr,"%s: PASS\n",testcase->basename);
     return r;
@@ -285,8 +339,10 @@
 void testcase_free(Testcase *testcase)
 {
     g_free(testcase->basename);
-    g_free(testcase->input);
+    g_slist_foreach(testcase->inputs,(GFunc)testcase_input_free,NULL);
+    g_slist_free(testcase->inputs);
     g_free(testcase->expected);
     g_free(testcase->encoding);
+    g_strfreev(testcase->options);
     g_free(testcase);
 }
diff -r cf332d440466 -r 6a13fe0fc19e test/harness/testcase.h
--- a/test/harness/testcase.h	Fri Jan 27 23:59:51 2012 +0000
+++ b/test/harness/testcase.h	Mon Jan 30 00:36:31 2012 +0000
@@ -11,11 +11,14 @@
 
 typedef struct {
     char *basename;
-    char *input;
+    char *tmpdir;
+    GSList *inputs;
     char *expected;
     char *encoding;	/* The character encoding to talk to BOOKLOUPE in */
+    char **options;
     enum {
 	TESTCASE_XFAIL=1<<0,
+	TESTCASE_TMP_DIR=1<<1,
     } flags;
 } Testcase;
 
diff -r cf332d440466 -r 6a13fe0fc19e test/harness/testcaseinput.c
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/harness/testcaseinput.c	Mon Jan 30 00:36:31 2012 +0000
@@ -0,0 +1,182 @@
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <unistd.h>
+#include <errno.h>
+#include <fcntl.h>
+#ifdef WIN32
+#include <io.h>
+#endif
+#include <glib.h>
+#include <glib/gstdio.h>
+#include <bl/bl.h>
+#include "testcase.h"
+#include "testcaseinput.h"
+
+#ifndef O_BINARY
+#define O_BINARY 0
+#endif
+
+/*
+ * As write(), but with error handling.
+ */
+static size_t write_file(int fd,const char *buf,size_t count,GError **error)
+{
+    if (write(fd,buf,count)<count)
+    {
+	g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
+	  "Error writing bookloupe input file: %s",g_strerror(errno));
+	return -1;
+    }
+    return count;
+}
+
+/*
+ * Replace \n with \r\n and U+240A (visible symbol for LF) with \n
+ */
+static char *unix2dos(const char *text)
+{
+    gunichar c;
+    const gunichar visible_lf=0x240A;
+    GString *string;
+    string=g_string_new(NULL);
+    while(*text)
+    {
+	c=g_utf8_get_char(text);
+	text=g_utf8_next_char(text);
+	if (c=='\n')
+	    g_string_append(string,"\r\n");
+	else if (c==visible_lf)
+	    g_string_append_c(string,'\n');
+	else
+	    g_string_append_unichar(string,c);
+    }
+    return g_string_free(string,FALSE);
+}
+
+/*
+ * Create an input file needed for a testcase (as specified in <input>).
+ * The file is written in the encoding specified for communicating with
+ * bookloupe. The name_used field of <input> is filled in with the name
+ * of the created file (which may be different than the name specified
+ * if that contained "XXXXXX" to be replaced by a unique string).
+ */
+gboolean testcase_input_create(Testcase *testcase,TestcaseInput *input,
+  GError **error)
+{
+    int fd;
+    size_t n;
+    char *filename,*s,*t;
+    GError *tmp_err=NULL;
+    if (input->contents)
+    {
+	if (testcase->encoding)
+	{
+	    t=unix2dos(input->contents);
+	    s=g_convert(t,-1,testcase->encoding,"UTF-8",NULL,&n,&tmp_err);
+	    g_free(t);
+	    if (!s)
+	    {
+		g_propagate_prefixed_error(error,tmp_err,
+		  "Conversion to %s failed: ",testcase->encoding);
+		return FALSE;
+	    }
+	}
+	else
+	{
+	    s=unix2dos(input->contents);
+	    n=strlen(s);
+	}
+    }
+    else
+    {
+	n=0;
+	s=NULL;
+    }
+    g_free(input->name_used);
+    input->name_used=NULL;
+    if (testcase->tmpdir)
+	filename=g_build_filename(testcase->tmpdir,input->name,NULL);
+    else
+	filename=g_strdup(input->name);
+    if (strstr(input->name,"XXXXXX"))
+	fd=g_mkstemp(filename);
+    else
+	fd=g_open(filename,O_WRONLY|O_CREAT|O_EXCL|O_BINARY,0600);
+    if (fd<0)
+    {
+	g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
+	  "%s: %s",filename,g_strerror(errno));
+	g_free(s);
+	return FALSE;
+    }
+    input->name_used=g_strdup(filename+strlen(filename)-strlen(input->name));
+    if (n && write_file(fd,s,n,error)!=n)
+    {
+	g_free(s);
+	close(fd);
+	(void)g_unlink(filename);
+	g_free(filename);
+	g_free(input->name_used);
+	input->name_used=NULL;
+	return FALSE;
+    }
+    g_free(s);
+    if (close(fd)<0)
+    {
+	g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
+	  "%s: %s",filename,g_strerror(errno));
+	(void)g_unlink(filename);
+	g_free(filename);
+	g_free(input->name_used);
+	input->name_used=NULL;
+	return FALSE;
+    }
+    g_free(filename);
+    return TRUE;
+}
+
+/*
+ * Remove an input file created with testcase_input_create()
+ */
+gboolean testcase_input_remove(Testcase *testcase,TestcaseInput *input,
+  GError **error)
+{
+    char *filename;
+    if (input->name_used)
+    {
+	if (testcase->tmpdir)
+	    filename=g_build_filename(testcase->tmpdir,input->name_used,NULL);
+	else
+	    filename=g_strdup(input->name_used);
+	if (g_unlink(filename)<0)
+	{
+	    g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
+	      "%s: %s",filename,g_strerror(errno));
+	    return FALSE;
+	}
+	g_free(filename);
+	g_free(input->name_used);
+	input->name_used=NULL;
+    }
+    return TRUE;
+}
+
+/* Create a new description of an input file needed for a testcase */
+TestcaseInput *testcase_input_new(const char *name,const char *contents)
+{
+    TestcaseInput *input;
+    input=g_new0(TestcaseInput,1);
+    input->name=g_strdup(name);
+    input->contents=g_strdup(contents);
+    return input;
+}
+
+/* Free the description of a testcase input file */
+void testcase_input_free(TestcaseInput *input)
+{
+    g_free(input->name);
+    g_free(input->name_used);
+    g_free(input->contents);
+    g_free(input);
+}
diff -r cf332d440466 -r 6a13fe0fc19e test/harness/testcaseinput.h
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/harness/testcaseinput.h	Mon Jan 30 00:36:31 2012 +0000
@@ -0,0 +1,20 @@
+#ifndef TESTCASE_INPUT_H
+#define TESTCASE_INPUT_H
+
+#include <glib.h>
+#include "testcase.h"
+
+typedef struct {
+    char *name;
+    char *name_used;
+    char *contents;
+} TestcaseInput;
+
+gboolean testcase_input_create(Testcase *testcase,TestcaseInput *input,
+  GError **error);
+gboolean testcase_input_remove(Testcase *testcase,TestcaseInput *input,
+  GError **error);
+TestcaseInput *testcase_input_new(const char *name,const char *contents);
+void testcase_input_free(TestcaseInput *input);
+
+#endif	/* TESTCASE_INPUT_H */
diff -r cf332d440466 -r 6a13fe0fc19e test/harness/testcaseio.c
--- a/test/harness/testcaseio.c	Fri Jan 27 23:59:51 2012 +0000
+++ b/test/harness/testcaseio.c	Mon Jan 30 00:36:31 2012 +0000
@@ -4,6 +4,7 @@
 #include <glib.h>
 #include <bl/bl.h>
 #include "testcaseparser.h"
+#include "testcaseinput.h"
 #include "testcaseio.h"
 
 /*
@@ -15,7 +16,8 @@
 {
     Testcase *testcase;
     TestcaseParser *parser;
-    char *s;
+    TestcaseInput *input=NULL;
+    char *s,*arg;
     const char *tag,*text;
     gboolean found_tag=FALSE;
     parser=testcase_parser_new_from_file(filename);
@@ -34,12 +36,44 @@
 	*s='\0';
     while(testcase_parser_get_next_tag(parser,&tag,&text))
     {
-	if (!testcase->input && !strcmp(tag,"INPUT"))
-	    testcase->input=g_strdup(text);
+	if (!input && !strcmp(tag,"INPUT"))
+	    input=testcase_input_new("TEST-XXXXXX",text);
+	else if (g_str_has_prefix(tag,"INPUT(") && tag[strlen(tag)-1]==')')
+	{
+	    arg=g_strndup(tag+6,strlen(tag)-7);
+	    s=g_path_get_dirname(arg);
+	    if (strcmp(s,"."))
+	    {
+		/*
+		 * Otherwise it would be possible to overwrite an arbitary
+		 * file on somebody's computer by getting them to run a
+		 * testcase!
+		 */
+		fprintf(stderr,
+		  "%s: Input files may not have a directory component\n",arg);
+		g_free(s);
+		g_free(arg);
+		testcase_free(testcase);
+		testcase_parser_free(parser);
+		return NULL;
+	    }
+	    g_free(s);
+	    testcase->inputs=g_slist_prepend(testcase->inputs,
+	      testcase_input_new(arg,text));
+	    if (!strstr(arg,"XXXXXX"))
+		testcase->flags|=TESTCASE_TMP_DIR;
+	    g_free(arg);
+	}
 	else if (!testcase->expected && !strcmp(tag,"EXPECTED"))
 	    testcase->expected=g_strdup(text);
 	else if (!testcase->encoding && !strcmp(tag,"ENCODING"))
 	    testcase->encoding=g_strchomp(g_strdup(text));
+	else if (!testcase->encoding && !strcmp(tag,"OPTIONS"))
+	{
+	    testcase->options=g_strsplit(text,"\n",0);
+	    g_free(testcase->options[g_strv_length(testcase->options)-1]);
+	    testcase->options[g_strv_length(testcase->options)-1]=NULL;
+	}
 	else
 	{
 	    fprintf(stderr,"%s: Not a valid testcase (%s)\n",filename,tag);
@@ -61,6 +95,9 @@
 	testcase_parser_free(parser);
 	return NULL;
     }
+    if (!input)
+	input=testcase_input_new("TEST-XXXXXX",NULL);
+    testcase->inputs=g_slist_prepend(testcase->inputs,input);
     testcase_parser_free(parser);
     return testcase;
 }