Add as_utils_license_to_spdx()

This moves the Fedora-Legal->SPDX conversion code from libappstream-builder to
libappstream-glib, which allows front ends like GNOME Software to display a
valid SPDX ID when no AppStream or AppData information is available.

It also fixes the multi-token search which was not working before.

4 files changed, 114 insertions, 85 deletions

diff --git a/libappstream-builder/asb-package-rpm.c b/libappstream-builder/asb-package-rpm.c
index 9c5fe0d..42b15cd 100644
--- a/libappstream-builder/asb-package-rpm.c
+++ b/libappstream-builder/asb-package-rpm.c
@@ -78,72 +78,8 @@ static void
 asb_package_rpm_set_license (AsbPackage *pkg, const gchar *license)
 {
 	guint i;
-	guint j;
 	g_autofree gchar *new = NULL;
 	g_auto(GStrv) tokens = NULL;
-	struct {
-		const gchar	*fedora;
-		const gchar	*spdx;
-	} convert[] =  {
-		{ "AGPLv3",			"AGPL-3.0" },
-		{ "AGPLv3+",			"AGPL-3.0" },
-		{ "AGPLv3 with exceptions",	"AGPL-3.0" },
-		{ "AGPLv3+ with exceptions",	"AGPL-3.0" },
-		{ "Artistic 2.0",		"Artistic-2.0" },
-		{ "Artistic",			"Artistic-1.0" },
-		{ "Artistic clarified",		"Artistic-2.0" },
-		{ "ASL 1.1",			"Apache-1.1" },
-		{ "ASL 2.0",			"Apache-2.0" },
-		{ "Boost",			"BSL-1.0" },
-		{ "BSD",			"BSD-3-Clause" },
-		{ "BSD with advertising",	"BSD-3-Clause" },
-		{ "CC0",			"CC0-1.0" },
-		{ "CC-BY",			"CC-BY-3.0" },
-		{ "CC-BY-SA",			"CC-BY-SA-3.0" },
-		{ "CDDL",			"CDDL-1.0" },
-		{ "CeCILL-C",			"CECILL-C" },
-		{ "CeCILL",			"CECILL-2.0" },
-		{ "CPAL",			"CPAL-1.0" },
-		{ "CPL",			"CPL-1.0" },
-		{ "EPL",			"EPL-1.0" },
-		{ "Free Art",			"ClArtistic" },
-		{ "GFDL",			"GFDL-1.3" },
-		{ "GPL+",			"GPL-1.0+" },
-		{ "GPLv2",			"GPL-2.0" },
-		{ "GPLv2+",			"GPL-2.0+" },
-		{ "GPLV2",			"GPL-2.0" },
-		{ "GPLv2 with exceptions",	"GPL-2.0-with-font-exception" },
-		{ "GPLv2+ with exceptions",	"GPL-2.0-with-font-exception" },
-		{ "GPLv3",			"GPL-3.0" },
-		{ "GPLv3+",			"GPL-3.0+" },
-		{ "GPLV3+",			"GPL-3.0+" },
-		{ "GPLv3+ with exceptions",	"GPL-3.0+" },
-		{ "GPLv3 with exceptions",	"GPL-3.0-with-GCC-exception" },
-		{ "GPL+ with exceptions",	"GPL-2.0-with-font-exception" },
-		{ "IBM",			"IPL-1.0" },
-		{ "LGPL+",			"LGPL-2.1+" },
-		{ "LGPLv2.1",			"LGPL-2.1" },
-		{ "LGPLv2",			"LGPL-2.1" },
-		{ "LGPLv2+",			"LGPL-2.1+" },
-		{ "LGPLv2 with exceptions",	"LGPL-2.0" },
-		{ "LGPLv2+ with exceptions",	"LGPL-2.0+" },
-		{ "LGPLv3",			"LGPL-3.0" },
-		{ "LGPLv3+",			"LGPL-3.0+" },
-		{ "LPPL",			"LPPL-1.3c" },
-		{ "MIT with advertising",	"MIT" },
-		{ "MPLv1.0",			"MPL-1.0" },
-		{ "MPLv1.1",			"MPL-1.1" },
-		{ "MPLv2.0",			"MPL-2.0" },
-		{ "Netscape",			"NPL-1.1" },
-		{ "OFL",			"OFL-1.1" },
-		{ "Python",			"Python-2.0" },
-		{ "QPL",			"QPL-1.0" },
-		{ "QPL with exceptions",	"QPL-1.0" },
-		{ "SPL",			"SPL-1.0" },
-		{ "zlib",			"Zlib" },
-		{ "ZPLv2.0",			"ZPL-2.0" },
-		{ "Unlicense",			"CC0-1.0" },
-		{ NULL, NULL } };
 
 	/* this isn't supposed to happen */
 	if (license == NULL) {
@@ -152,9 +88,9 @@ asb_package_rpm_set_license (AsbPackage *pkg, const gchar *license)
 		return;
 	}
 
-	/* tokenize the license string and try to convert the Fedora license
-	 * string to a SPDX license the best we can */
-	tokens = as_utils_spdx_license_tokenize (license);
+	/* tokenize the license string and log non SPDX licenses */
+	new = as_utils_license_to_spdx (license);
+	tokens = as_utils_spdx_license_tokenize (new);
 	for (i = 0; tokens[i] != NULL; i++) {
 
 		/* ignore */
@@ -168,30 +104,12 @@ asb_package_rpm_set_license (AsbPackage *pkg, const gchar *license)
 		if (tokens[i][0] == '@')
 			continue;
 
-		/* convert */
-		for (j = 0; convert[j].fedora != NULL; j++) {
-			if (g_strcmp0 (tokens[i], convert[j].fedora) == 0) {
-				g_free (tokens[i]);
-				tokens[i] = g_strdup_printf ("@%s", convert[j].spdx);
-				asb_package_log (pkg,
-						 ASB_PACKAGE_LOG_LEVEL_DEBUG,
-						 "Converting Fedora license "
-						 "'%s' to SPDX '%s'",
-						 convert[j].fedora,
-						 convert[j].spdx);
-				break;
-			}
-		}
-		if (convert[j].fedora != NULL)
-			continue;
-
 		/* no matching SPDX entry */
 		asb_package_log (pkg,
 				 ASB_PACKAGE_LOG_LEVEL_WARNING,
 				 "Unable to currently map Fedora "
 				 "license '%s' to SPDX", tokens[i]);
 	}
-	new = as_utils_spdx_license_detokenize (tokens);
 	asb_package_set_license (pkg, new);
 }
 
diff --git a/libappstream-glib/as-self-test.c b/libappstream-glib/as-self-test.c
index 5a2a14d..0730676 100644
--- a/libappstream-glib/as-self-test.c
+++ b/libappstream-glib/as-self-test.c
@@ -3508,6 +3508,11 @@ as_test_utils_spdx_token_func (void)
 	g_assert (as_utils_is_spdx_license ("CC0 AND GFDL-1.3"));
 	g_assert (as_utils_is_spdx_license ("NOASSERTION"));
 	g_assert (!as_utils_is_spdx_license ("CC0 dave"));
+
+	/* importing non-SPDX formats */
+	tmp = as_utils_license_to_spdx ("CC0 and (Public Domain and GPLv3+ with exceptions)");
+	g_assert_cmpstr (tmp, ==, "CC0-1.0 AND (LicenseRef-public-domain AND GPL-3.0+)");
+	g_free (tmp);
 }
 
 static void
diff --git a/libappstream-glib/as-utils.c b/libappstream-glib/as-utils.c
index 2ed13cf..7e8d71c 100644
--- a/libappstream-glib/as-utils.c
+++ b/libappstream-glib/as-utils.c
@@ -791,6 +791,111 @@ as_utils_is_spdx_license (const gchar *license)
 }
 
 /**
+ * as_utils_license_to_spdx:
+ * @license: a not-quite SPDX license string, e.g. "GPLv3+"
+ *
+ * Converts a non-SPDX license into an SPDX format string where possible.
+ *
+ * Returns: the best-effort SPDX license string
+ *
+ * Since: 0.2.5
+ **/
+gchar *
+as_utils_license_to_spdx (const gchar *license)
+{
+	GString *str;
+	guint i;
+	guint j;
+	guint license_len;
+	struct {
+		const gchar	*old;
+		const gchar	*new;
+	} convert[] =  {
+		{ " with exceptions",		NULL },
+		{ " with advertising",		NULL },
+		{ " and ",			" AND " },
+		{ " or ",			" OR " },
+		{ "AGPLv3+",			"AGPL-3.0" },
+		{ "AGPLv3",			"AGPL-3.0" },
+		{ "Artistic 2.0",		"Artistic-2.0" },
+		{ "Artistic clarified",		"Artistic-2.0" },
+		{ "Artistic",			"Artistic-1.0" },
+		{ "ASL 1.1",			"Apache-1.1" },
+		{ "ASL 2.0",			"Apache-2.0" },
+		{ "Boost",			"BSL-1.0" },
+		{ "BSD",			"BSD-3-Clause" },
+		{ "CC0",			"CC0-1.0" },
+		{ "CC-BY-SA",			"CC-BY-SA-3.0" },
+		{ "CC-BY",			"CC-BY-3.0" },
+		{ "CDDL",			"CDDL-1.0" },
+		{ "CeCILL-C",			"CECILL-C" },
+		{ "CeCILL",			"CECILL-2.0" },
+		{ "CPAL",			"CPAL-1.0" },
+		{ "CPL",			"CPL-1.0" },
+		{ "EPL",			"EPL-1.0" },
+		{ "Free Art",			"ClArtistic" },
+		{ "GFDL",			"GFDL-1.3" },
+		{ "GPL+",			"GPL-1.0+" },
+		{ "GPLv2+",			"GPL-2.0+" },
+		{ "GPLv2",			"GPL-2.0" },
+		{ "GPLv3+",			"GPL-3.0+" },
+		{ "GPLv3",			"GPL-3.0" },
+		{ "IBM",			"IPL-1.0" },
+		{ "LGPL+",			"LGPL-2.1+" },
+		{ "LGPLv2.1",			"LGPL-2.1" },
+		{ "LGPLv2+",			"LGPL-2.1+" },
+		{ "LGPLv2",			"LGPL-2.1" },
+		{ "LGPLv3+",			"LGPL-3.0+" },
+		{ "LGPLv3",			"LGPL-3.0" },
+		{ "LPPL",			"LPPL-1.3c" },
+		{ "MPLv1.0",			"MPL-1.0" },
+		{ "MPLv1.1",			"MPL-1.1" },
+		{ "MPLv2.0",			"MPL-2.0" },
+		{ "Netscape",			"NPL-1.1" },
+		{ "OFL",			"OFL-1.1" },
+		{ "Python",			"Python-2.0" },
+		{ "QPL",			"QPL-1.0" },
+		{ "SPL",			"SPL-1.0" },
+		{ "zlib",			"Zlib" },
+		{ "ZPLv2.0",			"ZPL-2.0" },
+		{ "Unlicense",			"CC0-1.0" },
+		{ "Public Domain",		"LicenseRef-public-domain" },
+		{ "Copyright only",		"LicenseRef-public-domain" },
+		{ "Proprietary",		"LicenseRef-proprietary" },
+		{ "Commercial",			"LicenseRef-proprietary" },
+		{ NULL, NULL } };
+
+	/* nothing set */
+	if (license == NULL)
+		return NULL;
+
+	/* already in SPDX format */
+	if (as_utils_is_spdx_license (license))
+		return g_strdup (license);
+
+	/* go through the string looking for case-insensitive matches */
+	str = g_string_new ("");
+	license_len = strlen (license);
+	for (i = 0; i < license_len; i++) {
+		gboolean found = FALSE;
+		for (j = 0; convert[j].old != NULL; j++) {
+			guint old_len = strlen (convert[j].old);
+			if (g_ascii_strncasecmp (license + i,
+						 convert[j].old,
+						 old_len) != 0)
+				continue;
+			if (convert[j].new != NULL)
+				g_string_append (str, convert[j].new);
+			i += old_len - 1;
+			found = TRUE;
+		}
+		if (!found)
+			g_string_append_c (str, license[i]);
+	}
+	return g_string_free (str, FALSE);
+}
+
+/**
  * as_pixbuf_blur_private:
  **/
 static void
diff --git a/libappstream-glib/as-utils.h b/libappstream-glib/as-utils.h
index 44bfa39..93e4db4 100644
--- a/libappstream-glib/as-utils.h
+++ b/libappstream-glib/as-utils.h
@@ -145,6 +145,7 @@ gboolean	 as_utils_is_category_id	(const gchar	*category_id);
 gboolean	 as_utils_is_blacklisted_id	(const gchar	*desktop_id);
 gchar		**as_utils_spdx_license_tokenize (const gchar	*license);
 gchar		*as_utils_spdx_license_detokenize (gchar	**license_tokens);
+gchar		*as_utils_license_to_spdx	(const gchar	*license);
 gchar		*as_utils_find_icon_filename	(const gchar	*destdir,
 						 const gchar	*search,
 						 GError		**error);