Merge pull request #145 from lightpanda-io/libdom-rewrite-wrapper

netsurf: rewrite wrapper.c in pure Zig

build.zig

@@ -151,13 +151,4 @@ fn linkNetSurf(step: *std.build.LibExeObjStep) void {
         step.addIncludePath(.{ .path = ns ++ lib ++ "/src" });
     }
     step.addIncludePath(.{ .path = ns ++ "/include" });
-
-    // wrapper
-    const flags = [_][]const u8{};
-    const files: [1][]const u8 = .{ns ++ "wrapper/wrapper.c"};
-    step.addCSourceFiles(.{
-        .files = &files,
-        .flags = &flags,
-    });
-    step.addIncludePath(.{ .path = ns ++ "wrapper" });
 }

src/main.zig

@@ -61,7 +61,10 @@ pub fn main() !void {
     defer arena.deinit();
 
     // document
-    doc = try parser.documentHTMLParseFromFileAlloc(arena.allocator(), "test.html");
+    const file = try std.fs.cwd().openFile("test.html", .{});
+    defer file.close();
+
+    doc = try parser.documentHTMLParseFromFile(file);
     defer parser.documentHTMLClose(doc) catch |err| {
         std.debug.print("documentHTMLClose error: {s}\n", .{@errorName(err)});
     };

src/main_shell.zig

@@ -42,7 +42,10 @@ pub fn main() !void {
     defer arena.deinit();
 
     // document
-    doc = try parser.documentHTMLParseFromFileAlloc(arena.allocator(), "test.html");
+    const file = try std.fs.cwd().openFile("test.html", .{});
+    defer file.close();
+
+    doc = try parser.documentHTMLParseFromFile(file);
     defer parser.documentHTMLClose(doc) catch |err| {
         std.debug.print("documentHTMLClose error: {s}\n", .{@errorName(err)});
     };

src/netsurf.zig

@@ -1,7 +1,8 @@
 const std = @import("std");
 
 const c = @cImport({
-    @cInclude("wrapper.h");
+    @cInclude("dom/dom.h");
+    @cInclude("dom/bindings/hubbub/parser.h");
 });
 
 // Vtable
@@ -1280,42 +1281,100 @@ fn documentHTMLVtable(doc_html: *DocumentHTML) c.dom_html_document_vtable {
     return getVtable(c.dom_html_document_vtable, DocumentHTML, doc_html);
 }
 
-// documentHTMLParseFromFileAlloc parses the file.
+const ParserError = error{
-// The allocator is required to create a null terminated string from filename.
+    Reprocess,
-// The buffer is freed by the function.
+    EncodingChange,
-// The caller is responsible for closing the document.
+    Paused,
-pub fn documentHTMLParseFromFileAlloc(allocator: std.mem.Allocator, filename: []const u8) !*DocumentHTML {
+    NoMemory,
-    const cstr = try allocator.dupeZ(u8, filename);
+    BadParameter,
-    defer allocator.free(cstr);
+    BadEncoding,
+    Invalid,
+    FileNotFound,
+    NeedData,
+    Unknown,
+};
 
-    return documentHTMLParseFromFile(cstr);
+const HubbubErr = c.hubbub_error;
+
+fn parserErr(err: HubbubErr) ParserError!void {
+    return switch (err) {
+        c.HUBBUB_OK => {},
+        c.HUBBUB_REPROCESS => ParserError.Reprocess,
+        c.HUBBUB_ENCODINGCHANGE => ParserError.EncodingChange,
+        c.HUBBUB_PAUSED => ParserError.Paused,
+        c.HUBBUB_NOMEM => ParserError.NoMemory,
+        c.HUBBUB_BADPARM => ParserError.BadParameter,
+        c.HUBBUB_BADENCODING => ParserError.BadEncoding,
+        c.HUBBUB_INVALID => ParserError.Invalid,
+        c.HUBBUB_FILENOTFOUND => ParserError.FileNotFound,
+        c.HUBBUB_NEEDDATA => ParserError.NeedData,
+        c.HUBBUB_UNKNOWN => ParserError.Unknown,
+        else => unreachable,
+    };
 }
 
-// documentHTMLParseFromFile parses the given filename c string (ie. with 0 sentinel).
+// documentHTMLParseFromFile parses the given HTML file.
 // The caller is responsible for closing the document.
-pub fn documentHTMLParseFromFile(filename: [:0]const u8) !*DocumentHTML {
+pub fn documentHTMLParseFromFile(file: std.fs.File) !*DocumentHTML {
-    // create a null terminated c string.
+    var parser: ?*c.dom_hubbub_parser = undefined;
-    const doc = c.wr_create_doc_dom_from_file(filename.ptr);
+    var doc: ?*c.dom_document = undefined;
-    if (doc == null) return error.ParserError;
+    var err: c.hubbub_error = undefined;
+
+    var params = c.dom_hubbub_parser_params{
+        .enc = null,
+        .fix_enc = true,
+        .msg = null,
+        .script = null,
+        .enable_script = false,
+        .ctx = null,
+        .daf = null,
+    };
+
+    err = c.dom_hubbub_parser_create(&params, &parser, &doc);
+    try parserErr(err);
+    defer c.dom_hubbub_parser_destroy(parser);
+
+    var buffer: [1024]u8 = undefined;
+    var ln = buffer.len;
+    while (ln == buffer.len) {
+        ln = try file.readAll(&buffer);
+        err = c.dom_hubbub_parser_parse_chunk(parser, &buffer, ln);
+        try parserErr(err);
+    }
+
+    err = c.dom_hubbub_parser_completed(parser);
+    try parserErr(err);
+
     return @as(*DocumentHTML, @ptrCast(doc.?));
 }
 
-// documentHTMLParseFromStrAlloc the given string.
+// documentHTMLParseFromStr parses the given HTML 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: []const u8) !*DocumentHTML {
+pub fn documentHTMLParseFromStr(str: []const u8) ParserError!*DocumentHTML {
-    // create a null terminated c string.
+    var parser: ?*c.dom_hubbub_parser = undefined;
-    const cstr = try allocator.dupeZ(u8, str);
+    var doc: ?*c.dom_document = undefined;
-    defer allocator.free(cstr);
+    var err: c.hubbub_error = undefined;
-    return documentHTMLParseFromStr(cstr);
+
-}
+    var params = c.dom_hubbub_parser_params{
+        .enc = null,
+        .fix_enc = true,
+        .msg = null,
+        .script = null,
+        .enable_script = false,
+        .ctx = null,
+        .daf = null,
+    };
+
+    err = c.dom_hubbub_parser_create(&params, &parser, &doc);
+    try parserErr(err);
+    defer c.dom_hubbub_parser_destroy(parser);
+
+    err = c.dom_hubbub_parser_parse_chunk(parser, str, str.len);
+    try parserErr(err);
+
+    err = c.dom_hubbub_parser_completed(parser);
+    try parserErr(err);
 
-// 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

@@ -37,7 +37,10 @@ fn testExecFn(
     try js_env.attachObject(try js_env.getGlobal(), "window", null);
 
     // document
-    doc = try parser.documentHTMLParseFromFileAlloc(std.testing.allocator, "test.html");
+    const file = try std.fs.cwd().openFile("test.html", .{});
+    defer file.close();
+
+    doc = try parser.documentHTMLParseFromFile(file);
     defer parser.documentHTMLClose(doc) catch |err| {
         std.debug.print("documentHTMLClose error: {s}\n", .{@errorName(err)});
     };

src/wpt/run.zig

@@ -17,7 +17,10 @@ pub fn run(arena: *std.heap.ArenaAllocator, comptime apis: []jsruntime.API, comp
     const alloc = arena.allocator();
 
     // document
-    const html_doc = try parser.documentHTMLParseFromFileAlloc(alloc, f);
+    const file = try std.fs.cwd().openFile(f, .{});
+    defer file.close();
+
+    const html_doc = try parser.documentHTMLParseFromFile(file);
     const doc = parser.documentHTMLToDocument(html_doc);
 
     const dirname = fspath.dirname(f[dir.len..]) orelse unreachable;

vendor/netsurf/wrapper/wrapper.c

@@ -1,127 +0,0 @@
-#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(const char *filename)
-{
-	size_t buffer_size = 1024;
-	dom_hubbub_parser *parser = NULL;
-	FILE *handle;
-	int chunk_length;
-	dom_hubbub_error error;
-	dom_hubbub_parser_params params;
-	dom_document *doc;
-	unsigned char buffer[buffer_size];
-
-	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;
-	}
-
-	/* Open input file */
-	handle = fopen(filename, "rb");
-	if (handle == NULL) {
-		dom_hubbub_parser_destroy(parser);
-		printf("Can't open test input file: %s\n", filename);
-		return NULL;
-	}
-
-	/* Parse input file in chunks */
-	chunk_length = buffer_size;
-	while (chunk_length == buffer_size) {
-		chunk_length = fread(buffer, 1, buffer_size, handle);
-		error = dom_hubbub_parser_parse_chunk(parser, buffer,
-				chunk_length);
-		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);
-
-	/* Close input file */
-	if (fclose(handle) != 0) {
-	  printf("Can't close test input file: %s\n", filename);
-		return NULL;
-	}
-
-	return doc;
-}

vendor/netsurf/wrapper/wrapper.h

@@ -1,9 +0,0 @@
-#ifndef wrapper_dom_h_
-#define wrapper_dom_h_
-
-#include <dom/dom.h>
-
-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_ */