diff -Nru appstream-glib-0.7.18/debian/changelog appstream-glib-0.7.18/debian/changelog --- appstream-glib-0.7.18/debian/changelog 2021-11-07 19:38:32.000000000 +0100 +++ appstream-glib-0.7.18/debian/changelog 2023-06-07 17:42:20.000000000 +0200 @@ -1,3 +1,11 @@ +appstream-glib (0.7.18-2ubuntu0.1) jammy; urgency=medium + + * Add support-em-code.patch, improve-em-code-handling.patch: + - Do not fail parsing when encountering valid or + elements in AppStream XML. + + -- Matthias Klumpp Wed, 07 Jun 2023 17:42:20 +0200 + appstream-glib (0.7.18-2) unstable; urgency=medium [ Sebastian Krzyszkowiak ] diff -Nru appstream-glib-0.7.18/debian/patches/03_support-em-code.patch appstream-glib-0.7.18/debian/patches/03_support-em-code.patch --- appstream-glib-0.7.18/debian/patches/03_support-em-code.patch 1970-01-01 01:00:00.000000000 +0100 +++ appstream-glib-0.7.18/debian/patches/03_support-em-code.patch 2023-06-07 17:35:34.000000000 +0200 @@ -0,0 +1,114 @@ +From f939f14774618fd07d7019e9d0c86e1e1ae5642a Mon Sep 17 00:00:00 2001 +From: =?UTF-8?q?Philip=20M=C3=BCller?= +Date: Mon, 14 Jun 2021 22:06:01 +0200 +Subject: [PATCH] Support em/code tags Some appstream-data packages add + and or and to the files. Not all package manager can + handle that. An example would be pamac from Manjaro + +--- + libappstream-glib/as-node.c | 36 ++++++++++++++++++++++++++++++++ + libappstream-glib/as-self-test.c | 16 ++++++++++++++ + 2 files changed, 52 insertions(+) + +diff --git a/libappstream-glib/as-node.c b/libappstream-glib/as-node.c +index aafb16aa..b4159ea9 100644 +--- a/libappstream-glib/as-node.c ++++ b/libappstream-glib/as-node.c +@@ -555,6 +555,8 @@ typedef struct { + AsNode *current; + AsNodeFromXmlFlags flags; + const gchar * const *locales; ++ guint8 is_em_text; ++ guint8 is_code_text; + } AsNodeToXmlHelper; + + /** +@@ -604,6 +606,16 @@ as_node_start_element_cb (GMarkupParseContext *context, + AsNode *current; + guint i; + ++ /* do not create a child node for em and code tags */ ++ if (g_strcmp0 (element_name, "em") == 0) { ++ helper->is_em_text = 1; ++ return; ++ } ++ if (g_strcmp0 (element_name, "code") == 0) { ++ helper->is_code_text = 1; ++ return; ++ } ++ + /* check if we should ignore the locale */ + data = g_slice_new0 (AsNodeData); + +@@ -662,6 +674,16 @@ as_node_end_element_cb (GMarkupParseContext *context, + GError **error) + { + AsNodeToXmlHelper *helper = (AsNodeToXmlHelper *) user_data; ++ ++ /* do not create a child node for em and code tags */ ++ if (g_strcmp0 (element_name, "em") == 0) { ++ helper->is_em_text = 0; ++ return; ++ } ++ if (g_strcmp0 (element_name, "code") == 0) { ++ helper->is_code_text = 0; ++ return; ++ } + helper->current = helper->current->parent; + } + +@@ -695,6 +717,20 @@ as_node_text_cb (GMarkupParseContext *context, + + /* split up into lines and add each with spaces stripped */ + if (data->cdata != NULL) { ++ /* support em and code tags */ ++ if (g_strcmp0 (as_tag_data_get_name (data), "p") == 0 || ++ g_strcmp0 (as_tag_data_get_name (data), "li") == 0) { ++ g_autoptr(GString) str = g_string_new (data->cdata); ++ as_ref_string_unref (data->cdata); ++ if (helper->is_em_text) ++ g_string_append_printf (str, "%s", text); ++ else if (helper->is_code_text) ++ g_string_append_printf (str, "%s", text); ++ else ++ g_string_append (str, text); ++ data->cdata = as_ref_string_new_with_length (str->str, str->len); ++ return; ++ } + g_set_error (error, + AS_NODE_ERROR, + AS_NODE_ERROR_INVALID_MARKUP, +diff --git a/libappstream-glib/as-self-test.c b/libappstream-glib/as-self-test.c +index 2f0820fc..eafb15e1 100644 +--- a/libappstream-glib/as-self-test.c ++++ b/libappstream-glib/as-self-test.c +@@ -2865,6 +2865,11 @@ as_test_node_xml_func (void) + "" + "baz" + ""; ++ const gchar *valid_em_code = "" ++ "" ++ "It now also supports em and code tags." ++ "" ++ ""; + GError *error = NULL; + AsNode *n2; + AsNode *root; +@@ -2928,6 +2933,17 @@ as_test_node_xml_func (void) + g_string_free (xml, TRUE); + as_node_unref (root); + ++ /* support em and code tags */ ++ root = as_node_from_xml (valid_em_code, 0, &error); ++ g_assert_no_error (error); ++ g_assert (root != NULL); ++ ++ n2 = as_node_find (root, "description/p"); ++ g_assert (n2 != NULL); ++ printf ("<%s>\n", as_node_get_data (n2)); ++ g_assert_cmpstr (as_node_get_data (n2), ==, "It now also supportsem and code tags."); ++ as_node_unref (root); ++ + /* keep comments */ + root = as_node_from_xml (valid, + AS_NODE_FROM_XML_FLAG_KEEP_COMMENTS, diff -Nru appstream-glib-0.7.18/debian/patches/04_improve-em-code-handling.patch appstream-glib-0.7.18/debian/patches/04_improve-em-code-handling.patch --- appstream-glib-0.7.18/debian/patches/04_improve-em-code-handling.patch 1970-01-01 01:00:00.000000000 +0100 +++ appstream-glib-0.7.18/debian/patches/04_improve-em-code-handling.patch 2023-06-07 17:41:01.000000000 +0200 @@ -0,0 +1,242 @@ +From 674490bd54ff206f213ca4547db7fdb591a0fb3d Mon Sep 17 00:00:00 2001 +From: "Jan Alexander Steffens (heftig)" +Date: Fri, 15 Jul 2022 21:18:47 +0200 +Subject: [PATCH] Improve handling of and tags + +This is still not great code but at least somewhat an improvement. Tests +were expanded to showcase the new behavior. + +I think, ideally, we would append opening/closing tags to the ancestor +`p` or `li` node's cdata as soon as we encounter the start/end of an +`em` or `code` element. This would then also handle empty elements +correctly. +--- + libappstream-glib/as-node.c | 108 ++++++++++++++++++------------- + libappstream-glib/as-self-test.c | 39 ++++++++++- + 2 files changed, 101 insertions(+), 46 deletions(-) + +--- a/libappstream-glib/as-node.c ++++ b/libappstream-glib/as-node.c +@@ -555,8 +555,8 @@ + AsNode *current; + AsNodeFromXmlFlags flags; + const gchar * const *locales; +- guint8 is_em_text; +- guint8 is_code_text; ++ guint8 is_em_text:1; ++ guint8 is_code_text:1; + } AsNodeToXmlHelper; + + /** +@@ -674,6 +674,7 @@ + GError **error) + { + AsNodeToXmlHelper *helper = (AsNodeToXmlHelper *) user_data; ++ AsNodeData *data = helper->current->data; + + /* do not create a child node for em and code tags */ + if (g_strcmp0 (element_name, "em") == 0) { +@@ -684,6 +685,42 @@ + helper->is_code_text = 0; + return; + } ++ ++ if (data->cdata != NULL) { ++ /* split up into lines and add each with spaces stripped */ ++ if ((helper->flags & AS_NODE_FROM_XML_FLAG_LITERAL_TEXT) == 0) { ++ AsRefString *cdata = data->cdata; ++ data->cdata = as_node_reflow_text (cdata, strlen (cdata)); ++ as_ref_string_unref (cdata); ++ } ++ ++ /* intern commonly duplicated tag values and save a bit of memory */ ++ if (data->is_tag_valid) { ++ AsNode *root = g_node_get_root (helper->current); ++ switch (data->tag) { ++ case AS_TAG_CATEGORY: ++ case AS_TAG_COMPULSORY_FOR_DESKTOP: ++ case AS_TAG_CONTENT_ATTRIBUTE: ++ case AS_TAG_DEVELOPER_NAME: ++ case AS_TAG_EXTENDS: ++ case AS_TAG_ICON: ++ case AS_TAG_ID: ++ case AS_TAG_KUDO: ++ case AS_TAG_LANG: ++ case AS_TAG_METADATA_LICENSE: ++ case AS_TAG_MIMETYPE: ++ case AS_TAG_PROJECT_GROUP: ++ case AS_TAG_PROJECT_LICENSE: ++ case AS_TAG_SOURCE_PKGNAME: ++ case AS_TAG_URL: ++ as_node_cdata_to_intern (root, data); ++ break; ++ default: ++ break; ++ } ++ } ++ } ++ + helper->current = helper->current->parent; + } + +@@ -715,22 +752,9 @@ + if (i >= text_len) + return; + +- /* split up into lines and add each with spaces stripped */ +- if (data->cdata != NULL) { +- /* support em and code tags */ +- if (g_strcmp0 (as_tag_data_get_name (data), "p") == 0 || +- g_strcmp0 (as_tag_data_get_name (data), "li") == 0) { +- g_autoptr(GString) str = g_string_new (data->cdata); +- as_ref_string_unref (data->cdata); +- if (helper->is_em_text) +- g_string_append_printf (str, "%s", text); +- else if (helper->is_code_text) +- g_string_append_printf (str, "%s", text); +- else +- g_string_append (str, text); +- data->cdata = as_ref_string_new_with_length (str->str, str->len); +- return; +- } ++ if (data->cdata != NULL && ++ g_strcmp0 (as_tag_data_get_name (data), "p") != 0 && ++ g_strcmp0 (as_tag_data_get_name (data), "li") != 0) { + g_set_error (error, + AS_NODE_ERROR, + AS_NODE_ERROR_INVALID_MARKUP, +@@ -739,37 +763,33 @@ + data->cdata, text); + return; + } +- if ((helper->flags & AS_NODE_FROM_XML_FLAG_LITERAL_TEXT) > 0) { +- data->cdata = as_ref_string_new_with_length (text, text_len + 1); +- } else { +- data->cdata = as_node_reflow_text (text, (gssize) text_len); +- } + +- /* intern commonly duplicated tag values and save a bit of memory */ +- if (data->is_tag_valid && data->cdata != NULL) { +- AsNode *root = g_node_get_root (helper->current); +- switch (data->tag) { +- case AS_TAG_CATEGORY: +- case AS_TAG_COMPULSORY_FOR_DESKTOP: +- case AS_TAG_CONTENT_ATTRIBUTE: +- case AS_TAG_DEVELOPER_NAME: +- case AS_TAG_EXTENDS: +- case AS_TAG_ICON: +- case AS_TAG_ID: +- case AS_TAG_KUDO: +- case AS_TAG_LANG: +- case AS_TAG_METADATA_LICENSE: +- case AS_TAG_MIMETYPE: +- case AS_TAG_PROJECT_GROUP: +- case AS_TAG_PROJECT_LICENSE: +- case AS_TAG_SOURCE_PKGNAME: +- case AS_TAG_URL: +- as_node_cdata_to_intern (root, data); +- break; +- default: +- break; ++ /* support em and code tags */ ++ if (helper->is_em_text || helper->is_code_text || data->cdata != NULL) { ++ g_autoptr(GString) str = g_string_new (NULL); ++ ++ if (data->cdata != NULL) { ++ g_string_append (str, data->cdata); ++ as_ref_string_unref (data->cdata); + } ++ ++ if (helper->is_em_text) ++ g_string_append (str, ""); ++ if (helper->is_code_text) ++ g_string_append (str, ""); ++ ++ g_string_append_len (str, text, text_len); ++ ++ if (helper->is_code_text) ++ g_string_append (str, ""); ++ if (helper->is_em_text) ++ g_string_append (str, ""); ++ ++ data->cdata = as_ref_string_new_with_length (str->str, str->len); ++ return; + } ++ ++ data->cdata = as_ref_string_new_with_length (text, text_len); + } + + static void +@@ -826,7 +846,7 @@ + AsNodeFromXmlFlags flags, + GError **error) + { +- AsNodeToXmlHelper helper; ++ AsNodeToXmlHelper helper = {0}; + AsNode *root = NULL; + gboolean ret; + g_autoptr(GError) error_local = NULL; +@@ -963,7 +983,7 @@ + GCancellable *cancellable, + GError **error) + { +- AsNodeToXmlHelper helper; ++ AsNodeToXmlHelper helper = {0}; + GError *error_local = NULL; + AsNode *root = NULL; + const gchar *content_type = NULL; +--- a/libappstream-glib/as-self-test.c ++++ b/libappstream-glib/as-self-test.c +@@ -2866,6 +2866,15 @@ + "It now also supports em and code tags." + "" + ""; ++ const gchar *valid_em_code_2 = "" ++ "Emphasis at the start of the paragraph" ++ ""; ++ const gchar *valid_em_code_empty = "" ++ "" ++ ""; ++ const gchar *valid_em_code_empty_2 = "" ++ "empty emphasis" ++ ""; + GError *error = NULL; + AsNode *n2; + AsNode *root; +@@ -2936,8 +2945,34 @@ + + n2 = as_node_find (root, "description/p"); + g_assert (n2 != NULL); +- printf ("<%s>\n", as_node_get_data (n2)); +- g_assert_cmpstr (as_node_get_data (n2), ==, "It now also supportsem and code tags."); ++ g_assert_cmpstr (as_node_get_data (n2), ==, "It now also supports em and code tags."); ++ as_node_unref (root); ++ ++ root = as_node_from_xml (valid_em_code_2, 0, &error); ++ g_assert_no_error (error); ++ g_assert (root != NULL); ++ ++ n2 = as_node_find (root, "description/p"); ++ g_assert (n2 != NULL); ++ g_assert_cmpstr (as_node_get_data (n2), ==, "Emphasis at the start of the paragraph"); ++ as_node_unref (root); ++ ++ root = as_node_from_xml (valid_em_code_empty, 0, &error); ++ g_assert_no_error (error); ++ g_assert (root != NULL); ++ ++ n2 = as_node_find (root, "description/p"); ++ g_assert (n2 != NULL); ++ g_assert_cmpstr (as_node_get_data (n2), ==, NULL); ++ as_node_unref (root); ++ ++ root = as_node_from_xml (valid_em_code_empty_2, 0, &error); ++ g_assert_no_error (error); ++ g_assert (root != NULL); ++ ++ n2 = as_node_find (root, "description/p"); ++ g_assert (n2 != NULL); ++ g_assert_cmpstr (as_node_get_data (n2), ==, "empty emphasis"); + as_node_unref (root); + + /* keep comments */ diff -Nru appstream-glib-0.7.18/debian/patches/series appstream-glib-0.7.18/debian/patches/series --- appstream-glib-0.7.18/debian/patches/series 2021-11-07 19:19:16.000000000 +0100 +++ appstream-glib-0.7.18/debian/patches/series 2023-06-07 17:39:52.000000000 +0200 @@ -1,2 +1,4 @@ 01_as-app-Parse-custom-tags-from-DEP-11.patch 02_as-app-Parse-launchables-from-DEP-11.patch +03_support-em-code.patch +04_improve-em-code-handling.patch