| ... | ... | @@ -245,8 +245,6 @@ pub fn fetchAndAddDependencies( |
| 245 | 245 | error.FileNotFound => { |
| 246 | 246 | // Handle the same as no dependencies. |
| 247 | 247 | if (this_hash) |hash| { |
| 248 | | const pkg_dir_sub_path = "p" ++ fs.path.sep_str ++ hash[0..hex_multihash_len]; |
| 249 | | const build_root = try global_cache_directory.join(arena, &.{pkg_dir_sub_path}); |
| 250 | 248 | try dependencies_source.writer().print( |
| 251 | 249 | \\ pub const {} = struct {{ |
| 252 | 250 | \\ pub const build_root = "{}"; |
| ... | ... | @@ -256,7 +254,7 @@ pub fn fetchAndAddDependencies( |
| 256 | 254 | \\ |
| 257 | 255 | , .{ |
| 258 | 256 | std.zig.fmtId(hash), |
| 259 | | std.zig.fmtEscapes(build_root), |
| 257 | std.zig.fmtEscapes(pkg.root_src_directory.path.?), |
| 260 | 258 | std.zig.fmtEscapes(hash), |
| 261 | 259 | }); |
| 262 | 260 | } else { |
| ... | ... | @@ -312,66 +310,85 @@ pub fn fetchAndAddDependencies( |
| 312 | 310 | try dependencies_source.writer().writeAll("pub const packages = struct {\n"); |
| 313 | 311 | } |
| 314 | 312 | |
| 315 | | const deps_list = manifest.dependencies.values(); |
| 316 | | for (manifest.dependencies.keys(), 0..) |name, i| { |
| 317 | | const dep = deps_list[i]; |
| 318 | | |
| 319 | | const sub = try fetchAndUnpack( |
| 320 | | thread_pool, |
| 321 | | http_client, |
| 322 | | global_cache_directory, |
| 323 | | dep, |
| 324 | | report, |
| 325 | | all_modules, |
| 326 | | root_prog_node, |
| 327 | | name, |
| 328 | | ); |
| 329 | | |
| 330 | | if (sub.mod) |mod| { |
| 331 | | if (!sub.found_existing) { |
| 332 | | try mod.fetchAndAddDependencies( |
| 333 | | deps_pkg, |
| 334 | | arena, |
| 313 | for (manifest.dependencies.keys(), manifest.dependencies.values()) |name, *dep| { |
| 314 | var fetch_location = try FetchLocation.init(gpa, dep.*, directory, report); |
| 315 | defer fetch_location.deinit(gpa); |
| 316 | |
| 317 | // Directories do not provide a hash in build.zig.zon. |
| 318 | // Hash the path to the module rather than its contents. |
| 319 | const sub_mod, const found_existing = if (fetch_location == .directory) |
| 320 | try getDirectoryModule(gpa, fetch_location, directory, all_modules, dep, report) |
| 321 | else |
| 322 | try getCachedPackage( |
| 323 | gpa, |
| 324 | global_cache_directory, |
| 325 | dep.*, |
| 326 | all_modules, |
| 327 | root_prog_node, |
| 328 | ) orelse .{ |
| 329 | try fetchAndUnpack( |
| 330 | fetch_location, |
| 335 | 331 | thread_pool, |
| 336 | 332 | http_client, |
| 337 | | mod.root_src_directory, |
| 333 | directory, |
| 338 | 334 | global_cache_directory, |
| 339 | | local_cache_directory, |
| 340 | | dependencies_source, |
| 341 | | error_bundle, |
| 335 | dep.*, |
| 336 | report, |
| 342 | 337 | all_modules, |
| 343 | 338 | root_prog_node, |
| 344 | | dep.hash.?, |
| 345 | | ); |
| 346 | | } |
| 339 | name, |
| 340 | ), |
| 341 | false, |
| 342 | }; |
| 347 | 343 | |
| 348 | | try pkg.add(gpa, name, mod); |
| 349 | | if (deps_pkg.table.get(dep.hash.?)) |other_sub| { |
| 350 | | // This should be the same package (and hence module) since it's the same hash |
| 351 | | // TODO: dedup multiple versions of the same package |
| 352 | | assert(other_sub == mod); |
| 353 | | } else { |
| 354 | | try deps_pkg.add(gpa, dep.hash.?, mod); |
| 355 | | } |
| 356 | | } else if (!sub.found_existing) { |
| 357 | | const pkg_dir_sub_path = "p" ++ fs.path.sep_str ++ (dep.hash.?)[0..hex_multihash_len]; |
| 358 | | const build_root = try global_cache_directory.join(arena, &.{pkg_dir_sub_path}); |
| 359 | | try dependencies_source.writer().print( |
| 360 | | \\ pub const {} = struct {{ |
| 361 | | \\ pub const build_root = "{}"; |
| 362 | | \\ pub const deps: []const struct {{ []const u8, []const u8 }} = &.{{}}; |
| 363 | | \\ }}; |
| 364 | | \\ |
| 365 | | , .{ |
| 366 | | std.zig.fmtId(dep.hash.?), |
| 367 | | std.zig.fmtEscapes(build_root), |
| 368 | | }); |
| 344 | assert(dep.hash != null); |
| 345 | |
| 346 | switch (sub_mod) { |
| 347 | .zig_pkg => |sub_pkg| { |
| 348 | if (!found_existing) { |
| 349 | try sub_pkg.fetchAndAddDependencies( |
| 350 | deps_pkg, |
| 351 | arena, |
| 352 | thread_pool, |
| 353 | http_client, |
| 354 | sub_pkg.root_src_directory, |
| 355 | global_cache_directory, |
| 356 | local_cache_directory, |
| 357 | dependencies_source, |
| 358 | error_bundle, |
| 359 | all_modules, |
| 360 | root_prog_node, |
| 361 | dep.hash.?, |
| 362 | ); |
| 363 | } |
| 364 | |
| 365 | try pkg.add(gpa, name, sub_pkg); |
| 366 | if (deps_pkg.table.get(dep.hash.?)) |other_sub| { |
| 367 | // This should be the same package (and hence module) since it's the same hash |
| 368 | // TODO: dedup multiple versions of the same package |
| 369 | assert(other_sub == sub_pkg); |
| 370 | } else { |
| 371 | try deps_pkg.add(gpa, dep.hash.?, sub_pkg); |
| 372 | } |
| 373 | }, |
| 374 | .non_zig_pkg => |sub_pkg| { |
| 375 | if (!found_existing) { |
| 376 | try dependencies_source.writer().print( |
| 377 | \\ pub const {} = struct {{ |
| 378 | \\ pub const build_root = "{}"; |
| 379 | \\ pub const deps: []const struct {{ []const u8, []const u8 }} = &.{{}}; |
| 380 | \\ }}; |
| 381 | \\ |
| 382 | , .{ |
| 383 | std.zig.fmtId(dep.hash.?), |
| 384 | std.zig.fmtEscapes(sub_pkg.root_src_directory.path.?), |
| 385 | }); |
| 386 | } |
| 387 | }, |
| 369 | 388 | } |
| 370 | 389 | } |
| 371 | 390 | |
| 372 | 391 | if (this_hash) |hash| { |
| 373 | | const pkg_dir_sub_path = "p" ++ fs.path.sep_str ++ hash[0..hex_multihash_len]; |
| 374 | | const build_root = try global_cache_directory.join(arena, &.{pkg_dir_sub_path}); |
| 375 | 392 | try dependencies_source.writer().print( |
| 376 | 393 | \\ pub const {} = struct {{ |
| 377 | 394 | \\ pub const build_root = "{}"; |
| ... | ... | @@ -380,7 +397,7 @@ pub fn fetchAndAddDependencies( |
| 380 | 397 | \\ |
| 381 | 398 | , .{ |
| 382 | 399 | std.zig.fmtId(hash), |
| 383 | | std.zig.fmtEscapes(build_root), |
| 400 | std.zig.fmtEscapes(pkg.root_src_directory.path.?), |
| 384 | 401 | std.zig.fmtEscapes(hash), |
| 385 | 402 | }); |
| 386 | 403 | for (manifest.dependencies.keys(), manifest.dependencies.values()) |name, dep| { |
| ... | ... | @@ -490,15 +507,296 @@ const Report = struct { |
| 490 | 507 | } |
| 491 | 508 | }; |
| 492 | 509 | |
| 510 | const FetchLocation = union(enum) { |
| 511 | /// The relative path to a file or directory. |
| 512 | /// This may be a file that requires unpacking (such as a .tar.gz), |
| 513 | /// or the path to the root directory of a package. |
| 514 | file: []const u8, |
| 515 | directory: []const u8, |
| 516 | http_request: std.Uri, |
| 517 | |
| 518 | pub fn init(gpa: Allocator, dep: Manifest.Dependency, root_dir: Compilation.Directory, report: Report) !FetchLocation { |
| 519 | switch (dep.location) { |
| 520 | .url => |url| { |
| 521 | const uri = std.Uri.parse(url) catch |err| switch (err) { |
| 522 | error.UnexpectedCharacter => return report.fail(dep.location_tok, "failed to parse dependency location as URI", .{}), |
| 523 | else => return err, |
| 524 | }; |
| 525 | if (ascii.eqlIgnoreCase(uri.scheme, "file")) { |
| 526 | return report.fail(dep.location_tok, "'file' scheme is not allowed for URLs. Use '.path' instead", .{}); |
| 527 | } |
| 528 | return .{ .http_request = uri }; |
| 529 | }, |
| 530 | .path => |path| { |
| 531 | if (fs.path.isAbsolute(path)) { |
| 532 | return report.fail(dep.location_tok, "Absolute paths are not allowed. Use a relative path instead", .{}); |
| 533 | } |
| 534 | |
| 535 | const is_dir = isDirectory(root_dir, path) catch |err| switch (err) { |
| 536 | error.FileNotFound => return report.fail(dep.location_tok, "File not found: {s}", .{path}), |
| 537 | else => return err, |
| 538 | }; |
| 539 | |
| 540 | return if (is_dir) |
| 541 | .{ .directory = try gpa.dupe(u8, path) } |
| 542 | else |
| 543 | .{ .file = try gpa.dupe(u8, path) }; |
| 544 | }, |
| 545 | } |
| 546 | } |
| 547 | |
| 548 | pub fn deinit(f: *FetchLocation, gpa: Allocator) void { |
| 549 | switch (f.*) { |
| 550 | inline .file, .directory => |path| gpa.free(path), |
| 551 | .http_request => {}, |
| 552 | } |
| 553 | f.* = undefined; |
| 554 | } |
| 555 | |
| 556 | pub fn fetch( |
| 557 | f: FetchLocation, |
| 558 | gpa: Allocator, |
| 559 | root_dir: Compilation.Directory, |
| 560 | http_client: *std.http.Client, |
| 561 | dep: Manifest.Dependency, |
| 562 | report: Report, |
| 563 | ) !ReadableResource { |
| 564 | switch (f) { |
| 565 | .file => |file| { |
| 566 | const owned_path = try gpa.dupe(u8, file); |
| 567 | errdefer gpa.free(owned_path); |
| 568 | return .{ |
| 569 | .path = owned_path, |
| 570 | .resource = .{ .file = try root_dir.handle.openFile(file, .{}) }, |
| 571 | }; |
| 572 | }, |
| 573 | .http_request => |uri| { |
| 574 | var h = std.http.Headers{ .allocator = gpa }; |
| 575 | defer h.deinit(); |
| 576 | |
| 577 | var req = try http_client.request(.GET, uri, h, .{}); |
| 578 | errdefer req.deinit(); |
| 579 | |
| 580 | try req.start(.{}); |
| 581 | try req.wait(); |
| 582 | |
| 583 | if (req.response.status != .ok) { |
| 584 | return report.fail(dep.location_tok, "Expected response status '200 OK' got '{} {s}'", .{ |
| 585 | @intFromEnum(req.response.status), |
| 586 | req.response.status.phrase() orelse "", |
| 587 | }); |
| 588 | } |
| 589 | |
| 590 | return .{ |
| 591 | .path = try gpa.dupe(u8, uri.path), |
| 592 | .resource = .{ .http_request = req }, |
| 593 | }; |
| 594 | }, |
| 595 | .directory => unreachable, // Directories do not require fetching |
| 596 | } |
| 597 | } |
| 598 | }; |
| 599 | |
| 600 | const ReadableResource = struct { |
| 601 | path: []const u8, |
| 602 | resource: union(enum) { |
| 603 | file: fs.File, |
| 604 | http_request: std.http.Client.Request, |
| 605 | }, |
| 606 | |
| 607 | /// Unpack the package into the global cache directory. |
| 608 | /// If `ps` does not require unpacking (for example, if it is a directory), then no caching is performed. |
| 609 | /// In either case, the hash is computed and returned along with the path to the package. |
| 610 | pub fn unpack( |
| 611 | rr: *ReadableResource, |
| 612 | allocator: Allocator, |
| 613 | thread_pool: *ThreadPool, |
| 614 | global_cache_directory: Compilation.Directory, |
| 615 | dep: Manifest.Dependency, |
| 616 | report: Report, |
| 617 | pkg_prog_node: *std.Progress.Node, |
| 618 | ) !PackageLocation { |
| 619 | switch (rr.resource) { |
| 620 | inline .file, .http_request => |*r| { |
| 621 | const s = fs.path.sep_str; |
| 622 | const rand_int = std.crypto.random.int(u64); |
| 623 | const tmp_dir_sub_path = "tmp" ++ s ++ Manifest.hex64(rand_int); |
| 624 | |
| 625 | const actual_hash = h: { |
| 626 | var tmp_directory: Compilation.Directory = d: { |
| 627 | const path = try global_cache_directory.join(allocator, &.{tmp_dir_sub_path}); |
| 628 | errdefer allocator.free(path); |
| 629 | |
| 630 | const iterable_dir = try global_cache_directory.handle.makeOpenPathIterable(tmp_dir_sub_path, .{}); |
| 631 | errdefer iterable_dir.close(); |
| 632 | |
| 633 | break :d .{ |
| 634 | .path = path, |
| 635 | .handle = iterable_dir.dir, |
| 636 | }; |
| 637 | }; |
| 638 | defer tmp_directory.closeAndFree(allocator); |
| 639 | |
| 640 | const opt_content_length = try rr.getSize(); |
| 641 | |
| 642 | var prog_reader: ProgressReader(@TypeOf(r.reader())) = .{ |
| 643 | .child_reader = r.reader(), |
| 644 | .prog_node = pkg_prog_node, |
| 645 | .unit = if (opt_content_length) |content_length| unit: { |
| 646 | const kib = content_length / 1024; |
| 647 | const mib = kib / 1024; |
| 648 | if (mib > 0) { |
| 649 | pkg_prog_node.setEstimatedTotalItems(@intCast(mib)); |
| 650 | pkg_prog_node.setUnit("MiB"); |
| 651 | break :unit .mib; |
| 652 | } else { |
| 653 | pkg_prog_node.setEstimatedTotalItems(@intCast(@max(1, kib))); |
| 654 | pkg_prog_node.setUnit("KiB"); |
| 655 | break :unit .kib; |
| 656 | } |
| 657 | } else .any, |
| 658 | }; |
| 659 | pkg_prog_node.context.refresh(); |
| 660 | |
| 661 | switch (try rr.getFileType(dep, report)) { |
| 662 | .@"tar.gz" => try unpackTarball(allocator, prog_reader, tmp_directory.handle, std.compress.gzip), |
| 663 | // I have not checked what buffer sizes the xz decompression implementation uses |
| 664 | // by default, so the same logic applies for buffering the reader as for gzip. |
| 665 | .@"tar.xz" => try unpackTarball(allocator, prog_reader, tmp_directory.handle, std.compress.xz), |
| 666 | } |
| 667 | |
| 668 | // Unpack completed - stop showing amount as progress |
| 669 | pkg_prog_node.setEstimatedTotalItems(0); |
| 670 | pkg_prog_node.setCompletedItems(0); |
| 671 | pkg_prog_node.context.refresh(); |
| 672 | |
| 673 | // TODO: delete files not included in the package prior to computing the package hash. |
| 674 | // for example, if the ini file has directives to include/not include certain files, |
| 675 | // apply those rules directly to the filesystem right here. This ensures that files |
| 676 | // not protected by the hash are not present on the file system. |
| 677 | |
| 678 | break :h try computePackageHash(thread_pool, .{ .dir = tmp_directory.handle }); |
| 679 | }; |
| 680 | |
| 681 | const pkg_dir_sub_path = "p" ++ s ++ Manifest.hexDigest(actual_hash); |
| 682 | const unpacked_path = try global_cache_directory.join(allocator, &.{pkg_dir_sub_path}); |
| 683 | defer allocator.free(unpacked_path); |
| 684 | |
| 685 | const relative_unpacked_path = try fs.path.relative(allocator, global_cache_directory.path.?, unpacked_path); |
| 686 | errdefer allocator.free(relative_unpacked_path); |
| 687 | try renameTmpIntoCache(global_cache_directory.handle, tmp_dir_sub_path, relative_unpacked_path); |
| 688 | |
| 689 | return .{ |
| 690 | .hash = actual_hash, |
| 691 | .relative_unpacked_path = relative_unpacked_path, |
| 692 | }; |
| 693 | }, |
| 694 | } |
| 695 | } |
| 696 | |
| 697 | const FileType = enum { |
| 698 | @"tar.gz", |
| 699 | @"tar.xz", |
| 700 | }; |
| 701 | |
| 702 | pub fn getSize(rr: ReadableResource) !?u64 { |
| 703 | switch (rr.resource) { |
| 704 | // TODO: Handle case of chunked content-length |
| 705 | .http_request => |req| return req.response.content_length, |
| 706 | .file => |f| return (try f.metadata()).size(), |
| 707 | } |
| 708 | } |
| 709 | |
| 710 | pub fn getFileType(rr: ReadableResource, dep: Manifest.Dependency, report: Report) !FileType { |
| 711 | switch (rr.resource) { |
| 712 | .file => { |
| 713 | return fileTypeFromPath(rr.path) orelse |
| 714 | return report.fail(dep.location_tok, "Unknown file type", .{}); |
| 715 | }, |
| 716 | .http_request => |req| { |
| 717 | const content_type = req.response.headers.getFirstValue("Content-Type") orelse |
| 718 | return report.fail(dep.location_tok, "Missing 'Content-Type' header", .{}); |
| 719 | |
| 720 | // If the response has a different content type than the URI indicates, override |
| 721 | // the previously assumed file type. |
| 722 | return if (ascii.eqlIgnoreCase(content_type, "application/gzip") or |
| 723 | ascii.eqlIgnoreCase(content_type, "application/x-gzip") or |
| 724 | ascii.eqlIgnoreCase(content_type, "application/tar+gzip")) |
| 725 | .@"tar.gz" |
| 726 | else if (ascii.eqlIgnoreCase(content_type, "application/x-xz")) |
| 727 | .@"tar.xz" |
| 728 | else if (ascii.eqlIgnoreCase(content_type, "application/octet-stream")) ty: { |
| 729 | // support gitlab tarball urls such as https://gitlab.com/<namespace>/<project>/-/archive/<sha>/<project>-<sha>.tar.gz |
| 730 | // whose content-disposition header is: 'attachment; filename="<project>-<sha>.tar.gz"' |
| 731 | const content_disposition = req.response.headers.getFirstValue("Content-Disposition") orelse |
| 732 | return report.fail(dep.location_tok, "Missing 'Content-Disposition' header for Content-Type=application/octet-stream", .{}); |
| 733 | break :ty getAttachmentType(content_disposition) orelse |
| 734 | return report.fail(dep.location_tok, "Unsupported 'Content-Disposition' header value: '{s}' for Content-Type=application/octet-stream", .{content_disposition}); |
| 735 | } else return report.fail(dep.location_tok, "Unrecognized value for 'Content-Type' header: {s}", .{content_type}); |
| 736 | }, |
| 737 | } |
| 738 | } |
| 739 | |
| 740 | fn fileTypeFromPath(file_path: []const u8) ?FileType { |
| 741 | return if (ascii.endsWithIgnoreCase(file_path, ".tar.gz")) |
| 742 | .@"tar.gz" |
| 743 | else if (ascii.endsWithIgnoreCase(file_path, ".tar.xz")) |
| 744 | .@"tar.xz" |
| 745 | else |
| 746 | null; |
| 747 | } |
| 748 | |
| 749 | fn getAttachmentType(content_disposition: []const u8) ?FileType { |
| 750 | const disposition_type_end = ascii.indexOfIgnoreCase(content_disposition, "attachment;") orelse return null; |
| 751 | |
| 752 | var value_start = ascii.indexOfIgnoreCasePos(content_disposition, disposition_type_end + 1, "filename") orelse return null; |
| 753 | value_start += "filename".len; |
| 754 | if (content_disposition[value_start] == '*') { |
| 755 | value_start += 1; |
| 756 | } |
| 757 | if (content_disposition[value_start] != '=') return null; |
| 758 | value_start += 1; |
| 759 | |
| 760 | var value_end = mem.indexOfPos(u8, content_disposition, value_start, ";") orelse content_disposition.len; |
| 761 | if (content_disposition[value_end - 1] == '\"') { |
| 762 | value_end -= 1; |
| 763 | } |
| 764 | return fileTypeFromPath(content_disposition[value_start..value_end]); |
| 765 | } |
| 766 | |
| 767 | pub fn deinit(rr: *ReadableResource, gpa: Allocator) void { |
| 768 | gpa.free(rr.path); |
| 769 | switch (rr.resource) { |
| 770 | .file => |file| file.close(), |
| 771 | .http_request => |*req| req.deinit(), |
| 772 | } |
| 773 | rr.* = undefined; |
| 774 | } |
| 775 | }; |
| 776 | |
| 777 | pub const PackageLocation = struct { |
| 778 | /// For packages that require unpacking, this is the hash of the package contents. |
| 779 | /// For directories, this is the hash of the absolute file path. |
| 780 | hash: [Manifest.Hash.digest_length]u8, |
| 781 | relative_unpacked_path: []const u8, |
| 782 | |
| 783 | pub fn deinit(pl: *PackageLocation, allocator: Allocator) void { |
| 784 | allocator.free(pl.relative_unpacked_path); |
| 785 | pl.* = undefined; |
| 786 | } |
| 787 | }; |
| 788 | |
| 493 | 789 | const hex_multihash_len = 2 * Manifest.multihash_len; |
| 494 | 790 | const MultiHashHexDigest = [hex_multihash_len]u8; |
| 791 | |
| 792 | const DependencyModule = union(enum) { |
| 793 | zig_pkg: *Package, |
| 794 | non_zig_pkg: *Package, |
| 795 | }; |
| 495 | 796 | /// This is to avoid creating multiple modules for the same build.zig file. |
| 496 | 797 | /// If the value is `null`, the package is a known dependency, but has not yet |
| 497 | 798 | /// been fetched. |
| 498 | | pub const AllModules = std.AutoHashMapUnmanaged(MultiHashHexDigest, ?union(enum) { |
| 499 | | zig_pkg: *Package, |
| 500 | | non_zig_pkg: void, |
| 501 | | }); |
| 799 | pub const AllModules = std.AutoHashMapUnmanaged(MultiHashHexDigest, ?DependencyModule); |
| 502 | 800 | |
| 503 | 801 | fn ProgressReader(comptime ReaderType: type) type { |
| 504 | 802 | return struct { |
| ... | ... | @@ -542,29 +840,27 @@ fn ProgressReader(comptime ReaderType: type) type { |
| 542 | 840 | }; |
| 543 | 841 | } |
| 544 | 842 | |
| 545 | | fn fetchAndUnpack( |
| 546 | | thread_pool: *ThreadPool, |
| 547 | | http_client: *std.http.Client, |
| 843 | /// Get a cached package if it exists. |
| 844 | /// Returns `null` if the package has not been cached |
| 845 | /// If the package exists in the cache, returns a pointer to the package and a |
| 846 | /// boolean indicating whether this package has already been seen in the build |
| 847 | /// (i.e. whether or not its transitive dependencies have been fetched). |
| 848 | fn getCachedPackage( |
| 849 | gpa: Allocator, |
| 548 | 850 | global_cache_directory: Compilation.Directory, |
| 549 | 851 | dep: Manifest.Dependency, |
| 550 | | report: Report, |
| 551 | 852 | all_modules: *AllModules, |
| 552 | 853 | root_prog_node: *std.Progress.Node, |
| 553 | | /// This does not have to be any form of canonical or fully-qualified name: it |
| 554 | | /// is only intended to be human-readable for progress reporting. |
| 555 | | name_for_prog: []const u8, |
| 556 | | ) !struct { mod: ?*Package, found_existing: bool } { |
| 557 | | const gpa = http_client.allocator; |
| 854 | ) !?struct { DependencyModule, bool } { |
| 558 | 855 | const s = fs.path.sep_str; |
| 559 | | |
| 560 | 856 | // Check if the expected_hash is already present in the global package |
| 561 | 857 | // cache, and thereby avoid both fetching and unpacking. |
| 562 | | if (dep.hash) |h| cached: { |
| 858 | if (dep.hash) |h| { |
| 563 | 859 | const hex_digest = h[0..hex_multihash_len]; |
| 564 | 860 | const pkg_dir_sub_path = "p" ++ s ++ hex_digest; |
| 565 | 861 | |
| 566 | 862 | var pkg_dir = global_cache_directory.handle.openDir(pkg_dir_sub_path, .{}) catch |err| switch (err) { |
| 567 | | error.FileNotFound => break :cached, |
| 863 | error.FileNotFound => return null, |
| 568 | 864 | else => |e| return e, |
| 569 | 865 | }; |
| 570 | 866 | errdefer pkg_dir.close(); |
| ... | ... | @@ -574,162 +870,99 @@ fn fetchAndUnpack( |
| 574 | 870 | const gop = try all_modules.getOrPut(gpa, hex_digest.*); |
| 575 | 871 | if (gop.found_existing) { |
| 576 | 872 | if (gop.value_ptr.*) |mod| { |
| 577 | | return switch (mod) { |
| 578 | | .zig_pkg => |pkg| .{ |
| 579 | | .mod = pkg, |
| 580 | | .found_existing = true, |
| 581 | | }, |
| 582 | | .non_zig_pkg => .{ |
| 583 | | .mod = null, |
| 584 | | .found_existing = true, |
| 585 | | }, |
| 586 | | }; |
| 873 | return .{ mod, true }; |
| 587 | 874 | } |
| 588 | 875 | } |
| 589 | 876 | |
| 590 | | pkg_dir.access(build_zig_basename, .{}) catch { |
| 591 | | gop.value_ptr.* = .non_zig_pkg; |
| 592 | | return .{ |
| 593 | | .mod = null, |
| 594 | | .found_existing = false, |
| 595 | | }; |
| 596 | | }; |
| 597 | | |
| 598 | | const build_root = try global_cache_directory.join(gpa, &.{pkg_dir_sub_path}); |
| 599 | | errdefer gpa.free(build_root); |
| 600 | | |
| 601 | 877 | root_prog_node.completeOne(); |
| 602 | 878 | |
| 603 | | const ptr = try gpa.create(Package); |
| 604 | | errdefer gpa.destroy(ptr); |
| 879 | const is_zig_mod = if (pkg_dir.access(build_zig_basename, .{})) |_| true else |_| false; |
| 880 | const basename = if (is_zig_mod) build_zig_basename else ""; |
| 881 | const pkg = try createWithDir(gpa, global_cache_directory, pkg_dir_sub_path, basename); |
| 605 | 882 | |
| 606 | | const owned_src_path = try gpa.dupe(u8, build_zig_basename); |
| 607 | | errdefer gpa.free(owned_src_path); |
| 883 | const module: DependencyModule = if (is_zig_mod) |
| 884 | .{ .zig_pkg = pkg } |
| 885 | else |
| 886 | .{ .non_zig_pkg = pkg }; |
| 608 | 887 | |
| 609 | | ptr.* = .{ |
| 610 | | .root_src_directory = .{ |
| 611 | | .path = build_root, |
| 612 | | .handle = pkg_dir, |
| 613 | | }, |
| 614 | | .root_src_directory_owned = true, |
| 615 | | .root_src_path = owned_src_path, |
| 616 | | }; |
| 617 | | |
| 618 | | gop.value_ptr.* = .{ .zig_pkg = ptr }; |
| 619 | | return .{ |
| 620 | | .mod = ptr, |
| 621 | | .found_existing = false, |
| 622 | | }; |
| 888 | try all_modules.put(gpa, hex_digest.*, module); |
| 889 | return .{ module, false }; |
| 623 | 890 | } |
| 624 | 891 | |
| 625 | | var pkg_prog_node = root_prog_node.start(name_for_prog, 0); |
| 626 | | defer pkg_prog_node.end(); |
| 627 | | pkg_prog_node.activate(); |
| 628 | | pkg_prog_node.context.refresh(); |
| 629 | | |
| 630 | | const uri = try std.Uri.parse(dep.url); |
| 631 | | |
| 632 | | const rand_int = std.crypto.random.int(u64); |
| 633 | | const tmp_dir_sub_path = "tmp" ++ s ++ Manifest.hex64(rand_int); |
| 892 | return null; |
| 893 | } |
| 634 | 894 | |
| 635 | | const actual_hash = a: { |
| 636 | | var tmp_directory: Compilation.Directory = d: { |
| 637 | | const path = try global_cache_directory.join(gpa, &.{tmp_dir_sub_path}); |
| 638 | | errdefer gpa.free(path); |
| 895 | fn getDirectoryModule( |
| 896 | gpa: Allocator, |
| 897 | fetch_location: FetchLocation, |
| 898 | directory: Compilation.Directory, |
| 899 | all_modules: *AllModules, |
| 900 | dep: *Manifest.Dependency, |
| 901 | report: Report, |
| 902 | ) !struct { DependencyModule, bool } { |
| 903 | assert(fetch_location == .directory); |
| 639 | 904 | |
| 640 | | const iterable_dir = try global_cache_directory.handle.makeOpenPathIterable(tmp_dir_sub_path, .{}); |
| 641 | | errdefer iterable_dir.close(); |
| 905 | if (dep.hash != null) { |
| 906 | return report.fail(dep.hash_tok, "hash not allowed for directory package", .{}); |
| 907 | } |
| 642 | 908 | |
| 643 | | break :d .{ |
| 644 | | .path = path, |
| 645 | | .handle = iterable_dir.dir, |
| 646 | | }; |
| 647 | | }; |
| 648 | | defer tmp_directory.closeAndFree(gpa); |
| 909 | const hash = try computePathHash(gpa, directory, fetch_location.directory); |
| 910 | const hex_digest = Manifest.hexDigest(hash); |
| 911 | dep.hash = try gpa.dupe(u8, &hex_digest); |
| 649 | 912 | |
| 650 | | var h = std.http.Headers{ .allocator = gpa }; |
| 651 | | defer h.deinit(); |
| 913 | // There is no fixed location to check for directory modules. |
| 914 | // Instead, check whether it is already listed in all_modules. |
| 915 | if (all_modules.get(hex_digest)) |mod| return .{ mod.?, true }; |
| 652 | 916 | |
| 653 | | var req = try http_client.request(.GET, uri, h, .{}); |
| 654 | | defer req.deinit(); |
| 917 | var pkg_dir = directory.handle.openDir(fetch_location.directory, .{}) catch |err| switch (err) { |
| 918 | error.FileNotFound => return report.fail(dep.location_tok, "File not found: {s}", .{fetch_location.directory}), |
| 919 | else => |e| return e, |
| 920 | }; |
| 921 | defer pkg_dir.close(); |
| 655 | 922 | |
| 656 | | try req.start(.{}); |
| 657 | | try req.wait(); |
| 923 | const is_zig_mod = if (pkg_dir.access(build_zig_basename, .{})) |_| true else |_| false; |
| 924 | const basename = if (is_zig_mod) build_zig_basename else ""; |
| 658 | 925 | |
| 659 | | if (req.response.status != .ok) { |
| 660 | | return report.fail(dep.url_tok, "Expected response status '200 OK' got '{} {s}'", .{ |
| 661 | | @intFromEnum(req.response.status), |
| 662 | | req.response.status.phrase() orelse "", |
| 663 | | }); |
| 664 | | } |
| 926 | const pkg = try createWithDir(gpa, directory, fetch_location.directory, basename); |
| 927 | const module: DependencyModule = if (is_zig_mod) |
| 928 | .{ .zig_pkg = pkg } |
| 929 | else |
| 930 | .{ .non_zig_pkg = pkg }; |
| 665 | 931 | |
| 666 | | const content_type = req.response.headers.getFirstValue("Content-Type") orelse |
| 667 | | return report.fail(dep.url_tok, "Missing 'Content-Type' header", .{}); |
| 668 | | |
| 669 | | var prog_reader: ProgressReader(std.http.Client.Request.Reader) = .{ |
| 670 | | .child_reader = req.reader(), |
| 671 | | .prog_node = &pkg_prog_node, |
| 672 | | .unit = if (req.response.content_length) |content_length| unit: { |
| 673 | | const kib = content_length / 1024; |
| 674 | | const mib = kib / 1024; |
| 675 | | if (mib > 0) { |
| 676 | | pkg_prog_node.setEstimatedTotalItems(@intCast(mib)); |
| 677 | | pkg_prog_node.setUnit("MiB"); |
| 678 | | break :unit .mib; |
| 679 | | } else { |
| 680 | | pkg_prog_node.setEstimatedTotalItems(@intCast(@max(1, kib))); |
| 681 | | pkg_prog_node.setUnit("KiB"); |
| 682 | | break :unit .kib; |
| 683 | | } |
| 684 | | } else .any, |
| 685 | | }; |
| 686 | | pkg_prog_node.context.refresh(); |
| 687 | | |
| 688 | | if (ascii.eqlIgnoreCase(content_type, "application/gzip") or |
| 689 | | ascii.eqlIgnoreCase(content_type, "application/x-gzip") or |
| 690 | | ascii.eqlIgnoreCase(content_type, "application/tar+gzip")) |
| 691 | | { |
| 692 | | // I observed the gzip stream to read 1 byte at a time, so I am using a |
| 693 | | // buffered reader on the front of it. |
| 694 | | try unpackTarball(gpa, prog_reader.reader(), tmp_directory.handle, std.compress.gzip); |
| 695 | | } else if (ascii.eqlIgnoreCase(content_type, "application/x-xz")) { |
| 696 | | // I have not checked what buffer sizes the xz decompression implementation uses |
| 697 | | // by default, so the same logic applies for buffering the reader as for gzip. |
| 698 | | try unpackTarball(gpa, prog_reader.reader(), tmp_directory.handle, std.compress.xz); |
| 699 | | } else if (ascii.eqlIgnoreCase(content_type, "application/octet-stream")) { |
| 700 | | // support gitlab tarball urls such as https://gitlab.com/<namespace>/<project>/-/archive/<sha>/<project>-<sha>.tar.gz |
| 701 | | // whose content-disposition header is: 'attachment; filename="<project>-<sha>.tar.gz"' |
| 702 | | const content_disposition = req.response.headers.getFirstValue("Content-Disposition") orelse |
| 703 | | return report.fail(dep.url_tok, "Missing 'Content-Disposition' header for Content-Type=application/octet-stream", .{}); |
| 704 | | if (isTarAttachment(content_disposition)) { |
| 705 | | try unpackTarball(gpa, prog_reader.reader(), tmp_directory.handle, std.compress.gzip); |
| 706 | | } else return report.fail(dep.url_tok, "Unsupported 'Content-Disposition' header value: '{s}' for Content-Type=application/octet-stream", .{content_disposition}); |
| 707 | | } else { |
| 708 | | return report.fail(dep.url_tok, "Unsupported 'Content-Type' header value: '{s}'", .{content_type}); |
| 709 | | } |
| 932 | try all_modules.put(gpa, hex_digest, module); |
| 933 | return .{ module, false }; |
| 934 | } |
| 710 | 935 | |
| 711 | | // Download completed - stop showing downloaded amount as progress |
| 712 | | pkg_prog_node.setEstimatedTotalItems(0); |
| 713 | | pkg_prog_node.setCompletedItems(0); |
| 714 | | pkg_prog_node.context.refresh(); |
| 936 | fn fetchAndUnpack( |
| 937 | fetch_location: FetchLocation, |
| 938 | thread_pool: *ThreadPool, |
| 939 | http_client: *std.http.Client, |
| 940 | directory: Compilation.Directory, |
| 941 | global_cache_directory: Compilation.Directory, |
| 942 | dep: Manifest.Dependency, |
| 943 | report: Report, |
| 944 | all_modules: *AllModules, |
| 945 | root_prog_node: *std.Progress.Node, |
| 946 | /// This does not have to be any form of canonical or fully-qualified name: it |
| 947 | /// is only intended to be human-readable for progress reporting. |
| 948 | name_for_prog: []const u8, |
| 949 | ) !DependencyModule { |
| 950 | assert(fetch_location == .file or fetch_location == .http_request); |
| 715 | 951 | |
| 716 | | // TODO: delete files not included in the package prior to computing the package hash. |
| 717 | | // for example, if the ini file has directives to include/not include certain files, |
| 718 | | // apply those rules directly to the filesystem right here. This ensures that files |
| 719 | | // not protected by the hash are not present on the file system. |
| 952 | const gpa = http_client.allocator; |
| 720 | 953 | |
| 721 | | // TODO: raise an error for files that have illegal paths on some operating systems. |
| 722 | | // For example, on Linux a path with a backslash should raise an error here. |
| 723 | | // Of course, if the ignore rules above omit the file from the package, then everything |
| 724 | | // is fine and no error should be raised. |
| 954 | var pkg_prog_node = root_prog_node.start(name_for_prog, 0); |
| 955 | defer pkg_prog_node.end(); |
| 956 | pkg_prog_node.activate(); |
| 957 | pkg_prog_node.context.refresh(); |
| 725 | 958 | |
| 726 | | break :a try computePackageHash(thread_pool, .{ .dir = tmp_directory.handle }); |
| 727 | | }; |
| 959 | var readable_resource = try fetch_location.fetch(gpa, directory, http_client, dep, report); |
| 960 | defer readable_resource.deinit(gpa); |
| 728 | 961 | |
| 729 | | const pkg_dir_sub_path = "p" ++ s ++ Manifest.hexDigest(actual_hash); |
| 730 | | try renameTmpIntoCache(global_cache_directory.handle, tmp_dir_sub_path, pkg_dir_sub_path); |
| 962 | var package_location = try readable_resource.unpack(gpa, thread_pool, global_cache_directory, dep, report, &pkg_prog_node); |
| 963 | defer package_location.deinit(gpa); |
| 731 | 964 | |
| 732 | | const actual_hex = Manifest.hexDigest(actual_hash); |
| 965 | const actual_hex = Manifest.hexDigest(package_location.hash); |
| 733 | 966 | if (dep.hash) |h| { |
| 734 | 967 | if (!mem.eql(u8, h, &actual_hex)) { |
| 735 | 968 | return report.fail(dep.hash_tok, "hash mismatch: expected: {s}, found: {s}", .{ |
| ... | ... | @@ -743,9 +976,9 @@ fn fetchAndUnpack( |
| 743 | 976 | const eb = report.error_bundle; |
| 744 | 977 | const notes_len = 1; |
| 745 | 978 | try Report.addErrorMessage(report.ast.*, file_path, eb, notes_len, .{ |
| 746 | | .tok = dep.url_tok, |
| 979 | .tok = dep.location_tok, |
| 747 | 980 | .off = 0, |
| 748 | | .msg = "url field is missing corresponding hash field", |
| 981 | .msg = "dependency is missing hash field", |
| 749 | 982 | }); |
| 750 | 983 | const notes_start = try eb.reserveNotes(notes_len); |
| 751 | 984 | eb.extra.items[notes_start] = @intFromEnum(try eb.addErrorMessage(.{ |
| ... | ... | @@ -754,35 +987,28 @@ fn fetchAndUnpack( |
| 754 | 987 | return error.PackageFetchFailed; |
| 755 | 988 | } |
| 756 | 989 | |
| 757 | | const build_zig_path = try std.fs.path.join(gpa, &.{ pkg_dir_sub_path, build_zig_basename }); |
| 990 | const build_zig_path = try fs.path.join(gpa, &.{ package_location.relative_unpacked_path, build_zig_basename }); |
| 758 | 991 | defer gpa.free(build_zig_path); |
| 759 | 992 | |
| 760 | | global_cache_directory.handle.access(build_zig_path, .{}) catch |err| switch (err) { |
| 761 | | error.FileNotFound => { |
| 762 | | try all_modules.put(gpa, actual_hex, .non_zig_pkg); |
| 763 | | return .{ |
| 764 | | .mod = null, |
| 765 | | .found_existing = false, |
| 766 | | }; |
| 767 | | }, |
| 768 | | else => return err, |
| 769 | | }; |
| 993 | const is_zig_mod = if (global_cache_directory.handle.access(build_zig_path, .{})) |_| true else |_| false; |
| 994 | const basename = if (is_zig_mod) build_zig_basename else ""; |
| 995 | const pkg = try createWithDir(gpa, global_cache_directory, package_location.relative_unpacked_path, basename); |
| 996 | const module: DependencyModule = if (is_zig_mod) |
| 997 | .{ .zig_pkg = pkg } |
| 998 | else |
| 999 | .{ .non_zig_pkg = pkg }; |
| 770 | 1000 | |
| 771 | | const mod = try createWithDir(gpa, global_cache_directory, pkg_dir_sub_path, build_zig_basename); |
| 772 | | try all_modules.put(gpa, actual_hex, .{ .zig_pkg = mod }); |
| 773 | | return .{ |
| 774 | | .mod = mod, |
| 775 | | .found_existing = false, |
| 776 | | }; |
| 1001 | try all_modules.put(gpa, actual_hex, module); |
| 1002 | return module; |
| 777 | 1003 | } |
| 778 | 1004 | |
| 779 | 1005 | fn unpackTarball( |
| 780 | 1006 | gpa: Allocator, |
| 781 | | req_reader: anytype, |
| 1007 | reader: anytype, |
| 782 | 1008 | out_dir: fs.Dir, |
| 783 | 1009 | comptime compression: type, |
| 784 | 1010 | ) !void { |
| 785 | | var br = std.io.bufferedReaderSize(std.crypto.tls.max_ciphertext_record_len, req_reader); |
| 1011 | var br = std.io.bufferedReaderSize(std.crypto.tls.max_ciphertext_record_len, reader); |
| 786 | 1012 | |
| 787 | 1013 | var decompress = try compression.decompress(gpa, br.reader()); |
| 788 | 1014 | defer decompress.deinit(); |
| ... | ... | @@ -873,6 +1099,24 @@ fn computePackageHash( |
| 873 | 1099 | return hasher.finalResult(); |
| 874 | 1100 | } |
| 875 | 1101 | |
| 1102 | /// Compute the hash of a file path. |
| 1103 | fn computePathHash(gpa: Allocator, dir: Compilation.Directory, path: []const u8) ![Manifest.Hash.digest_length]u8 { |
| 1104 | const resolved_path = try std.fs.path.resolve(gpa, &.{ dir.path.?, path }); |
| 1105 | defer gpa.free(resolved_path); |
| 1106 | var hasher = Manifest.Hash.init(.{}); |
| 1107 | hasher.update(resolved_path); |
| 1108 | return hasher.finalResult(); |
| 1109 | } |
| 1110 | |
| 1111 | fn isDirectory(root_dir: Compilation.Directory, path: []const u8) !bool { |
| 1112 | var dir = root_dir.handle.openDir(path, .{}) catch |err| switch (err) { |
| 1113 | error.NotDir => return false, |
| 1114 | else => return err, |
| 1115 | }; |
| 1116 | defer dir.close(); |
| 1117 | return true; |
| 1118 | } |
| 1119 | |
| 876 | 1120 | /// Make a file system path identical independently of operating system path inconsistencies. |
| 877 | 1121 | /// This converts backslashes into forward slashes. |
| 878 | 1122 | fn normalizePath(arena: Allocator, fs_path: []const u8) ![]const u8 { |
| ... | ... | @@ -953,36 +1197,18 @@ fn renameTmpIntoCache( |
| 953 | 1197 | } |
| 954 | 1198 | } |
| 955 | 1199 | |
| 956 | | fn isTarAttachment(content_disposition: []const u8) bool { |
| 957 | | const disposition_type_end = ascii.indexOfIgnoreCase(content_disposition, "attachment;") orelse return false; |
| 958 | | |
| 959 | | var value_start = ascii.indexOfIgnoreCasePos(content_disposition, disposition_type_end + 1, "filename") orelse return false; |
| 960 | | value_start += "filename".len; |
| 961 | | if (content_disposition[value_start] == '*') { |
| 962 | | value_start += 1; |
| 963 | | } |
| 964 | | if (content_disposition[value_start] != '=') return false; |
| 965 | | value_start += 1; |
| 966 | | |
| 967 | | var value_end = mem.indexOfPos(u8, content_disposition, value_start, ";") orelse content_disposition.len; |
| 968 | | if (content_disposition[value_end - 1] == '\"') { |
| 969 | | value_end -= 1; |
| 970 | | } |
| 971 | | return ascii.endsWithIgnoreCase(content_disposition[value_start..value_end], ".tar.gz"); |
| 972 | | } |
| 973 | | |
| 974 | | test "isTarAttachment" { |
| 975 | | try std.testing.expect(isTarAttachment("attaChment; FILENAME=\"stuff.tar.gz\"; size=42")); |
| 976 | | try std.testing.expect(isTarAttachment("attachment; filename*=\"stuff.tar.gz\"")); |
| 977 | | try std.testing.expect(isTarAttachment("ATTACHMENT; filename=\"stuff.tar.gz\"")); |
| 978 | | try std.testing.expect(isTarAttachment("attachment; FileName=\"stuff.tar.gz\"")); |
| 979 | | try std.testing.expect(isTarAttachment("attachment; FileName*=UTF-8\'\'xyz%2Fstuff.tar.gz")); |
| 980 | | |
| 981 | | try std.testing.expect(!isTarAttachment("attachment FileName=\"stuff.tar.gz\"")); |
| 982 | | try std.testing.expect(!isTarAttachment("attachment; FileName=\"stuff.tar\"")); |
| 983 | | try std.testing.expect(!isTarAttachment("attachment; FileName\"stuff.gz\"")); |
| 984 | | try std.testing.expect(!isTarAttachment("attachment; size=42")); |
| 985 | | try std.testing.expect(!isTarAttachment("inline; size=42")); |
| 986 | | try std.testing.expect(!isTarAttachment("FileName=\"stuff.tar.gz\"; attachment;")); |
| 987 | | try std.testing.expect(!isTarAttachment("FileName=\"stuff.tar.gz\";")); |
| 1200 | test "getAttachmentType" { |
| 1201 | try std.testing.expectEqual(@as(?ReadableResource.FileType, .@"tar.gz"), ReadableResource.getAttachmentType("attaChment; FILENAME=\"stuff.tar.gz\"; size=42")); |
| 1202 | try std.testing.expectEqual(@as(?ReadableResource.FileType, .@"tar.gz"), ReadableResource.getAttachmentType("attachment; filename*=\"stuff.tar.gz\"")); |
| 1203 | try std.testing.expectEqual(@as(?ReadableResource.FileType, .@"tar.xz"), ReadableResource.getAttachmentType("ATTACHMENT; filename=\"stuff.tar.xz\"")); |
| 1204 | try std.testing.expectEqual(@as(?ReadableResource.FileType, .@"tar.xz"), ReadableResource.getAttachmentType("attachment; FileName=\"stuff.tar.xz\"")); |
| 1205 | try std.testing.expectEqual(@as(?ReadableResource.FileType, .@"tar.gz"), ReadableResource.getAttachmentType("attachment; FileName*=UTF-8\'\'xyz%2Fstuff.tar.gz")); |
| 1206 | |
| 1207 | try std.testing.expect(ReadableResource.getAttachmentType("attachment FileName=\"stuff.tar.gz\"") == null); |
| 1208 | try std.testing.expect(ReadableResource.getAttachmentType("attachment; FileName=\"stuff.tar\"") == null); |
| 1209 | try std.testing.expect(ReadableResource.getAttachmentType("attachment; FileName\"stuff.gz\"") == null); |
| 1210 | try std.testing.expect(ReadableResource.getAttachmentType("attachment; size=42") == null); |
| 1211 | try std.testing.expect(ReadableResource.getAttachmentType("inline; size=42") == null); |
| 1212 | try std.testing.expect(ReadableResource.getAttachmentType("FileName=\"stuff.tar.gz\"; attachment;") == null); |
| 1213 | try std.testing.expect(ReadableResource.getAttachmentType("FileName=\"stuff.tar.gz\";") == null); |
| 988 | 1214 | } |
Andrew Kelley <andrew@ziglang.org> 2023-09-29 00:34:07-07:00
GitHub <noreply@github.com> 2023-09-29 00:34:07-07:00
Signed by PGP key 4AEE18F83AFDEB23
lib