Merge pull request #52 from Browsercore/netsurf-parse

refacto netsurf parse document

src/main.zig

@@ -52,8 +52,12 @@ pub fn main() !void {
     const vm = jsruntime.VM.init();
     defer vm.deinit();
 
+    // alloc
+    var arena = std.heap.ArenaAllocator.init(std.heap.page_allocator);
+    defer arena.deinit();
+
     // document
-    doc = parser.documentHTMLParse("test.html");
+    doc = try parser.documentHTMLParseFromFileAlloc(arena.allocator(), "test.html");
     defer parser.documentHTMLClose(doc);
 
     // remove socket file of internal server
@@ -66,10 +70,6 @@ pub fn main() !void {
         }
     };
 
-    // alloc
-    var arena = std.heap.ArenaAllocator.init(std.heap.page_allocator);
-    defer arena.deinit();
-
     // server
     var addr = try std.net.Address.initUnix(socket_path);
     server = std.net.StreamServer.init(.{});

src/main_shell.zig

@@ -31,20 +31,20 @@ pub fn main() !void {
     // generate APIs
     const apis = jsruntime.compile(DOM.Interfaces);
 
+    // allocator
+    var gpa = std.heap.GeneralPurposeAllocator(.{}){};
+    defer _ = gpa.deinit();
+    var arena = std.heap.ArenaAllocator.init(gpa.allocator());
+    defer arena.deinit();
+
     // document
-    doc = parser.documentHTMLParse("test.html");
+    doc = try parser.documentHTMLParseFromFileAlloc(arena.allocator(), "test.html");
     defer parser.documentHTMLClose(doc);
 
     // create JS vm
     const vm = jsruntime.VM.init();
     defer vm.deinit();
 
-    // alloc
-    var gpa = std.heap.GeneralPurposeAllocator(.{}){};
-    defer _ = gpa.deinit();
-    var arena = std.heap.ArenaAllocator.init(gpa.allocator());
-    defer arena.deinit();
-
     // launch shell
     try jsruntime.shell(&arena, apis, execJS, .{ .app_name = "browsercore" });
 }

src/netsurf.zig

@@ -553,11 +553,46 @@ fn documentHTMLVtable(doc_html: *DocumentHTML) c.dom_html_document_vtable {
     return getVtable(c.dom_html_document_vtable, DocumentHTML, doc_html);
 }
 
-pub fn documentHTMLParse(filename: []const u8) *DocumentHTML {
+// documentHTMLParseFromFileAlloc parses the file.
-    var f: []u8 = @constCast(filename);
+// The allocator is required to create a null terminated string from filename.
-    const doc = c.wr_create_doc_dom_from_file(f.ptr);
+// The buffer is freed by the function.
+// The caller is responsible for closing the document.
+pub fn documentHTMLParseFromFileAlloc(allocator: std.mem.Allocator, filename: []const u8) !*DocumentHTML {
+    const cstr = try allocator.dupeZ(u8, filename);
+    defer allocator.free(cstr);
+
+    return documentHTMLParseFromFile(cstr);
+}
+
+// documentHTMLParseFromFile parses the given filename c string (ie. with 0 sentinel).
+// The caller is responsible for closing the document.
+pub fn documentHTMLParseFromFile(filename: [:0]const u8) !*DocumentHTML {
+    // create a null terminated c string.
+    const doc = c.wr_create_doc_dom_from_file(filename.ptr);
     if (doc == null) {
-        @panic("error parser");
+        return error.ParserError;
+    }
+    return @as(*DocumentHTML, @ptrCast(doc.?));
+}
+
+// documentHTMLParseFromStrAlloc the given string.
+// The allocator is required to create a null terminated string.
+// The c string allocated is freed by the function.
+// The caller is responsible for closing the document.
+pub fn documentHTMLParseFromStrAlloc(allocator: std.mem.Allocator, str: [:0]const u8) !*DocumentHTML {
+    // create a null terminated c string.
+    const cstr = try allocator.dupeZ(u8, str);
+    defer allocator.free(cstr);
+
+    return documentHTMLParseFromStr(cstr);
+}
+
+// documentHTMLParseFromStr parses the given c string (ie. with 0 sentinel).
+// The caller is responsible for closing the document.
+pub fn documentHTMLParseFromStr(cstr: [:0]const u8) !*DocumentHTML {
+    const doc = c.wr_create_doc_dom_from_string(cstr.ptr);
+    if (doc == null) {
+        return error.ParserError;
     }
     return @as(*DocumentHTML, @ptrCast(doc.?));
 }

src/run_tests.zig

@@ -38,10 +38,9 @@ test {
     const apis = jsruntime.compile(DOM.Interfaces);
 
     // document
-    doc = parser.documentHTMLParse("test.html");
+    doc = try parser.documentHTMLParseFromFileAlloc(std.testing.allocator, "test.html");
     defer parser.documentHTMLClose(doc);
 
-
     // create JS vm
     const vm = jsruntime.VM.init();
     defer vm.deinit();

vendor/netsurf/wrapper/wrapper.c

@@ -1,15 +1,65 @@
 #include <stdio.h>
+#include <string.h>
 
 #include <dom/dom.h>
 #include <dom/bindings/hubbub/parser.h>
 
+/**
+ * Generate a LibDOM document DOM from an HTML string
+ *
+ * \param string The HTML string
+ * \return  pointer to DOM document, or NULL on error
+ */
+dom_document *wr_create_doc_dom_from_string(const char *html)
+{
+	dom_hubbub_parser *parser = NULL;
+	dom_hubbub_error error;
+	dom_hubbub_parser_params params;
+	dom_document *doc;
+
+	params.enc = NULL;
+	params.fix_enc = true;
+	params.enable_script = false;
+	params.msg = NULL;
+	params.script = NULL;
+	params.ctx = NULL;
+	params.daf = NULL;
+
+	/* Create Hubbub parser */
+	error = dom_hubbub_parser_create(&params, &parser, &doc);
+	if (error != DOM_HUBBUB_OK) {
+		printf("Can't create Hubbub Parser\n");
+		return NULL;
+	}
+
+	error = dom_hubbub_parser_parse_chunk(parser, html, strlen(html));
+	if (error != DOM_HUBBUB_OK) {
+		dom_hubbub_parser_destroy(parser);
+		printf("Parsing errors occur\n");
+		return NULL;
+	}
+
+	/* Done parsing file */
+	error = dom_hubbub_parser_completed(parser);
+	if (error != DOM_HUBBUB_OK) {
+		dom_hubbub_parser_destroy(parser);
+		printf("Parsing error when construct DOM\n");
+		return NULL;
+	}
+
+	/* Finished with parser */
+	dom_hubbub_parser_destroy(parser);
+
+	return doc;
+}
+
 /**
  * Generate a LibDOM document DOM from an HTML file
  *
  * \param file  The file path
  * \return  pointer to DOM document, or NULL on error
  */
-dom_document *wr_create_doc_dom_from_file(char *filename)
+dom_document *wr_create_doc_dom_from_file(const char *filename)
 {
 	size_t buffer_size = 1024;
 	dom_hubbub_parser *parser = NULL;

vendor/netsurf/wrapper/wrapper.h

@@ -3,6 +3,7 @@
 
 #include <dom/dom.h>
 
-dom_document *wr_create_doc_dom_from_file(char *filename);
+dom_document *wr_create_doc_dom_from_string(const char *html);
+dom_document *wr_create_doc_dom_from_file(const char *filename);
 
 #endif	/* wrapper_dom_h_ */