update mupdf

sumatrapdfreader · Nov 3, 2023 · f7f3f39 · f7f3f39
1 parent b4c744b
commit f7f3f39
Show file tree

Hide file tree

Showing 4 changed files with 80 additions and 5 deletions.
diff --git a/mupdf/include/mupdf/fitz/archive.h b/mupdf/include/mupdf/fitz/archive.h
@@ -381,7 +381,7 @@ typedef struct
 }
 fz_archive_handler;
 
-extern const fz_archive_handler fz_libarchive_archive_handler;
+FZ_DATA extern const fz_archive_handler fz_libarchive_archive_handler;
 
 void fz_register_archive_handler(fz_context *ctx, const fz_archive_handler *handler);
 

diff --git a/mupdf/source/html/html-doc.c b/mupdf/source/html/html-doc.c
@@ -220,6 +220,26 @@ mobi_lookup_metadata(fz_context *ctx, fz_document *doc_, const char *key, char *
 	return -1;
 }
 
+static int
+txt_lookup_metadata(fz_context *ctx, fz_document *doc_, const char *key, char *buf, size_t size)
+{
+	html_document *doc = (html_document*)doc_;
+	if (!strcmp(key, FZ_META_FORMAT))
+		return (int)fz_strlcpy(buf, "Text", size);
+	return -1;
+}
+
+static int
+office_lookup_metadata(fz_context *ctx, fz_document *doc_, const char *key, char *buf, size_t size)
+{
+	html_document *doc = (html_document*)doc_;
+	if (!strcmp(key, FZ_META_FORMAT))
+		return (int)fz_strlcpy(buf, "Office document", size);
+	if (!strcmp(key, FZ_META_INFO_TITLE) && doc->html->title)
+		return 1 + (int)fz_strlcpy(buf, doc->html->title, size);
+	return -1;
+}
+
 static fz_document *
 htdoc_open_document_with_buffer(fz_context *ctx, fz_archive *zip, fz_buffer *buf, int format)
 {
@@ -244,7 +264,8 @@ htdoc_open_document_with_buffer(fz_context *ctx, fz_archive *zip, fz_buffer *buf
 		case FORMAT_HTML5: doc->super.lookup_metadata = htdoc_lookup_metadata; break;
 		case FORMAT_XHTML: doc->super.lookup_metadata = xhtdoc_lookup_metadata; break;
 		case FORMAT_MOBI: doc->super.lookup_metadata = mobi_lookup_metadata; break;
-		case FORMAT_OFFICE: doc->super.lookup_metadata = NULL; break;
+		case FORMAT_TXT: doc->super.lookup_metadata = txt_lookup_metadata; break;
+		case FORMAT_OFFICE: doc->super.lookup_metadata = office_lookup_metadata; break;
 		}
 		doc->super.is_reflowable = 1;
 

diff --git a/mupdf/source/html/office.c b/mupdf/source/html/office.c
@@ -31,6 +31,8 @@ typedef struct
 
 	int footnotes_max;
 	char **footnotes;
+
+	char *title;
 } doc_info;
 
 static void
@@ -1037,6 +1039,35 @@ process_office_document(fz_context *ctx, fz_archive *arch, const char *file, doc
 		fz_rethrow(ctx);
 }
 
+static void
+process_office_document_properties(fz_context *ctx, fz_archive *arch, const char *file, doc_info *info)
+{
+	fz_xml *xml = NULL;
+	char *title;
+
+	fz_try(ctx)
+	{
+		fz_xml *pos;
+
+		xml = fz_parse_xml_archive_entry(ctx, arch, file, 1);
+
+		pos = fz_xml_find_dfs(xml, "title", NULL, NULL);
+		title = fz_xml_text(fz_xml_down(pos));
+		if (title)
+		{
+			fz_write_string(ctx, info->out, "<title>");
+			doc_escape(ctx, info->out, title);
+			fz_write_string(ctx, info->out, "</title>");
+		}
+	}
+	fz_always(ctx)
+	{
+		fz_drop_xml(ctx, xml);
+	}
+	fz_catch(ctx)
+		fz_rethrow(ctx);
+}
+
 fz_buffer *
 fz_office_to_html(fz_context *ctx, fz_html_font_set *set, fz_buffer *buffer_in, const char *user_css, fz_office_to_html_opts *opts)
 {
@@ -1047,6 +1078,7 @@ fz_office_to_html(fz_context *ctx, fz_html_font_set *set, fz_buffer *buffer_in,
 	fz_xml *pos = NULL;
 	fz_xml *rels = NULL;
 	const char *schema = "http://schemas.openxmlformats.org/officeDocument/2006/relationships/officeDocument";
+	const char *schema_props = "http://schemas.openxmlformats.org/package/2006/relationships/metadata/core-properties";
 	doc_info info = { 0 };
 
 	stream = fz_open_buffer(ctx, buffer_in);
@@ -1085,6 +1117,19 @@ fz_office_to_html(fz_context *ctx, fz_html_font_set *set, fz_buffer *buffer_in,
 		/* Try other types */
 		{
 			xml = try_parse_xml_archive_entry(ctx, archive, "_rels/.rels", 0);
+
+			fz_write_string(ctx, info.out, "<html>\n");
+
+			pos = fz_xml_find_dfs(xml, "Relationship", "Type", schema_props);
+			if (pos)
+			{
+				const char *file = fz_xml_att(pos, "Target");
+				fz_write_string(ctx, info.out, "<head>\n");
+				process_office_document_properties(ctx, archive, file, &info);
+				fz_write_string(ctx, info.out, "</head>\n");
+			}
+
+			fz_write_string(ctx, info.out, "<body>\n");
 			pos = fz_xml_find_dfs(xml, "Relationship", "Type", schema);
 			if (!pos)
 				fz_throw(ctx, FZ_ERROR_GENERIC, "Archive not docx.");

diff --git a/mupdf/source/pdf/pdf-op-run.c b/mupdf/source/pdf/pdf-op-run.c
@@ -1840,6 +1840,7 @@ push_marked_content(fz_context *ctx, pdf_run_processor *proc, const char *tagstr
 	int drop_tag = 1;
 	fz_structure standard;
 	pdf_obj *mc_dict = NULL;
+	int fallback = 0;
 
 	/* Flush any pending text so it's not in the wrong layer. */
 	pdf_flush_text(ctx, proc);
@@ -1873,10 +1874,18 @@ push_marked_content(fz_context *ctx, pdf_run_processor *proc, const char *tagstr
 
 		/* Structure */
 		if (mc_dict)
-			send_begin_structure(ctx, proc, mc_dict);
-		else
 		{
-			/* Maybe drop this entirely? */
+			fz_try(ctx)
+				send_begin_structure(ctx, proc, mc_dict);
+			fz_catch(ctx)
+			{
+				fz_warn(ctx, "structure tree broken, assume tree is missing: %s", fz_caught_message(ctx));
+				fallback = 1;
+			}
+		}
+
+		if (!mc_dict || fallback)
+		{
 			standard = structure_type(ctx, proc, tag);
 			if (standard != FZ_STRUCTURE_INVALID)
 			{