On Wed, 07 Jun 2023 at 21:33:29 +0100, Simon McVittie wrote: > [x] attach debdiff against the package in (old)stable
That was, in fact, a lie. See attached (or the nmudiff on #1037206 if you'd prefer the unfiltered version). smcv
debdiff appstream-glib_0.7.18-1{,+deb11u1}.dsc | filterdiff -p1 -x'debian/patches/*.patch' diffstat for appstream-glib-0.7.18 appstream-glib-0.7.18 debian/.gitignore | 1 debian/changelog | 10 debian/patches/Improve-handling-of-em-and-code-tags.patch | 220 ++++++++++ debian/patches/Properly-initialize-AsNodeToXmlHelper.patch | 34 + debian/patches/Support-em-code-tags.patch | 118 +++++ debian/patches/series | 4 debian/patches/trivial-Turn-is_-em-code-_text-fields-into-bitfields.patch | 26 + libappstream-glib/as-node.c | 120 ++++- libappstream-glib/as-self-test.c | 51 ++ 9 files changed, 552 insertions(+), 32 deletions(-) diff -Nru appstream-glib-0.7.18/debian/changelog appstream-glib-0.7.18/debian/changelog --- appstream-glib-0.7.18/debian/changelog 2020-12-21 23:14:10.000000000 +0000 +++ appstream-glib-0.7.18/debian/changelog 2023-06-07 19:25:59.000000000 +0100 @@ -1,3 +1,13 @@ +appstream-glib (0.7.18-1+deb11u1) bullseye; urgency=medium + + * Add patches from upstream to cope with <em> and <code> in metadata. + Older versions of appstream-glib mis-parse upstream metadata that + contains <em> and <code>, causing flatpak 1.12.x or older to fail + to load the metadata now published by Flathub. The symptom is that + `flatpak search` fails. (Closes: #1037206, LP: #2023215) + + -- Simon McVittie <s...@debian.org> Wed, 07 Jun 2023 19:25:59 +0100 + appstream-glib (0.7.18-1) unstable; urgency=medium [ Matthias Klumpp ] diff -Nru appstream-glib-0.7.18/debian/.gitignore appstream-glib-0.7.18/debian/.gitignore --- appstream-glib-0.7.18/debian/.gitignore 1970-01-01 01:00:00.000000000 +0100 +++ appstream-glib-0.7.18/debian/.gitignore 2023-06-07 19:25:59.000000000 +0100 @@ -0,0 +1 @@ +*~ 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 1970-01-01 01:00:00.000000000 +0100 +++ appstream-glib-0.7.18/debian/patches/series 2023-06-07 19:25:59.000000000 +0100 @@ -0,0 +1,4 @@ +Support-em-code-tags.patch +Properly-initialize-AsNodeToXmlHelper.patch +trivial-Turn-is_-em-code-_text-fields-into-bitfields.patch +Improve-handling-of-em-and-code-tags.patch diff -Nru appstream-glib-0.7.18/libappstream-glib/as-node.c appstream-glib-0.7.18/libappstream-glib/as-node.c --- appstream-glib-0.7.18/libappstream-glib/as-node.c 2020-09-07 11:20:43.894573000 +0100 +++ appstream-glib-0.7.18/libappstream-glib/as-node.c 2023-06-07 20:58:11.000000000 +0100 @@ -555,6 +555,8 @@ AsNode *current; AsNodeFromXmlFlags flags; const gchar * const *locales; + guint8 is_em_text:1; + guint8 is_code_text:1; } AsNodeToXmlHelper; /** @@ -604,6 +606,16 @@ 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 +674,53 @@ 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) { + helper->is_em_text = 0; + return; + } + if (g_strcmp0 (element_name, "code") == 0) { + 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; } @@ -693,8 +752,9 @@ if (i >= text_len) return; - /* split up into lines and add each with spaces stripped */ - if (data->cdata != NULL) { + 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, @@ -703,37 +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, "<em>"); + if (helper->is_code_text) + g_string_append (str, "<code>"); + + g_string_append_len (str, text, text_len); + + if (helper->is_code_text) + g_string_append (str, "</code>"); + if (helper->is_em_text) + g_string_append (str, "</em>"); + + 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 @@ -790,7 +846,7 @@ AsNodeFromXmlFlags flags, GError **error) { - AsNodeToXmlHelper helper; + AsNodeToXmlHelper helper = {0}; AsNode *root = NULL; gboolean ret; g_autoptr(GError) error_local = NULL; @@ -927,7 +983,7 @@ GCancellable *cancellable, GError **error) { - AsNodeToXmlHelper helper; + AsNodeToXmlHelper helper = {0}; GError *error_local = NULL; AsNode *root = NULL; const gchar *content_type = NULL; diff -Nru appstream-glib-0.7.18/libappstream-glib/as-self-test.c appstream-glib-0.7.18/libappstream-glib/as-self-test.c --- appstream-glib-0.7.18/libappstream-glib/as-self-test.c 2020-09-07 11:20:43.896573000 +0100 +++ appstream-glib-0.7.18/libappstream-glib/as-self-test.c 2023-06-07 20:58:11.000000000 +0100 @@ -2861,6 +2861,20 @@ "<!-- this documents bar -->" "<bar key=\"value\">baz</bar>" "</foo>"; + const gchar *valid_em_code = "<description>" + "<p>" + "It now also supports <em>em</em> and <code>code</code> tags." + "</p>" + "</description>"; + const gchar *valid_em_code_2 = "<description>" + "<p><em>Emphasis</em> at the start of the paragraph</p>" + "</description>"; + const gchar *valid_em_code_empty = "<description>" + "<p><em></em></p>" + "</description>"; + const gchar *valid_em_code_empty_2 = "<description>" + "<p>empty <em></em> emphasis</p>" + "</description>"; GError *error = NULL; AsNode *n2; AsNode *root; @@ -2924,6 +2938,43 @@ 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); + g_assert_cmpstr (as_node_get_data (n2), ==, "It now also supports <em>em</em> and <code>code</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), ==, "<em>Emphasis</em> 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 */ root = as_node_from_xml (valid, AS_NODE_FROM_XML_FLAG_KEEP_COMMENTS,