Merge pull request #19094 from truemedian/std-http-fields

std.http: fix http field parsing

4 files changed, 230 insertions, 19 deletions

diff --git a/lib/std/http/Client.zig b/lib/std/http/Client.zig
index 7eade117d0..10088e7627 100644
--- a/lib/std/http/Client.zig
+++ b/lib/std/http/Client.zig
@@ -489,9 +489,9 @@ pub const Response = struct {
                 else => {},
             }
 
-            var line_it = mem.splitSequence(u8, line, ": ");
+            var line_it = mem.splitScalar(u8, line, ':');
             const header_name = line_it.next().?;
-            const header_value = line_it.rest();
+            const header_value = mem.trim(u8, line_it.rest(), " \t");
             if (header_name.len == 0) return error.HttpHeadersInvalid;
 
             if (std.ascii.eqlIgnoreCase(header_name, "connection")) {
@@ -551,6 +551,43 @@ pub const Response = struct {
         return error.HttpHeadersInvalid; // missing empty line
     }
 
+    test parse {
+        const response_bytes = "HTTP/1.1 200 OK\r\n" ++
+            "LOcation:url\r\n" ++
+            "content-tYpe: text/plain\r\n" ++
+            "content-disposition:attachment; filename=example.txt \r\n" ++
+            "content-Length:10\r\n" ++
+            "TRansfer-encoding:\tdeflate, chunked \r\n" ++
+            "connectioN:\t keep-alive \r\n\r\n";
+
+        var header_buffer: [1024]u8 = undefined;
+        var res = Response{
+            .status = undefined,
+            .reason = undefined,
+            .version = undefined,
+            .keep_alive = false,
+            .parser = proto.HeadersParser.init(&header_buffer),
+        };
+
+        @memcpy(header_buffer[0..response_bytes.len], response_bytes);
+        res.parser.header_bytes_len = response_bytes.len;
+
+        try res.parse(response_bytes);
+
+        try testing.expectEqual(.@"HTTP/1.1", res.version);
+        try testing.expectEqualStrings("OK", res.reason);
+        try testing.expectEqual(.ok, res.status);
+
+        try testing.expectEqualStrings("url", res.location.?);
+        try testing.expectEqualStrings("text/plain", res.content_type.?);
+        try testing.expectEqualStrings("attachment; filename=example.txt", res.content_disposition.?);
+
+        try testing.expectEqual(true, res.keep_alive);
+        try testing.expectEqual(10, res.content_length.?);
+        try testing.expectEqual(.chunked, res.transfer_encoding);
+        try testing.expectEqual(.deflate, res.transfer_compression);
+    }
+
     inline fn int64(array: *const [8]u8) u64 {
         return @bitCast(array.*);
     }
@@ -575,6 +612,67 @@ pub const Response = struct {
     pub fn iterateHeaders(r: Response) http.HeaderIterator {
         return http.HeaderIterator.init(r.parser.get());
     }
+
+    test iterateHeaders {
+        const response_bytes = "HTTP/1.1 200 OK\r\n" ++
+            "LOcation:url\r\n" ++
+            "content-tYpe: text/plain\r\n" ++
+            "content-disposition:attachment; filename=example.txt \r\n" ++
+            "content-Length:10\r\n" ++
+            "TRansfer-encoding:\tdeflate, chunked \r\n" ++
+            "connectioN:\t keep-alive \r\n\r\n";
+
+        var header_buffer: [1024]u8 = undefined;
+        var res = Response{
+            .status = undefined,
+            .reason = undefined,
+            .version = undefined,
+            .keep_alive = false,
+            .parser = proto.HeadersParser.init(&header_buffer),
+        };
+
+        @memcpy(header_buffer[0..response_bytes.len], response_bytes);
+        res.parser.header_bytes_len = response_bytes.len;
+
+        var it = res.iterateHeaders();
+        {
+            const header = it.next().?;
+            try testing.expectEqualStrings("LOcation", header.name);
+            try testing.expectEqualStrings("url", header.value);
+            try testing.expect(!it.is_trailer);
+        }
+        {
+            const header = it.next().?;
+            try testing.expectEqualStrings("content-tYpe", header.name);
+            try testing.expectEqualStrings("text/plain", header.value);
+            try testing.expect(!it.is_trailer);
+        }
+        {
+            const header = it.next().?;
+            try testing.expectEqualStrings("content-disposition", header.name);
+            try testing.expectEqualStrings("attachment; filename=example.txt", header.value);
+            try testing.expect(!it.is_trailer);
+        }
+        {
+            const header = it.next().?;
+            try testing.expectEqualStrings("content-Length", header.name);
+            try testing.expectEqualStrings("10", header.value);
+            try testing.expect(!it.is_trailer);
+        }
+        {
+            const header = it.next().?;
+            try testing.expectEqualStrings("TRansfer-encoding", header.name);
+            try testing.expectEqualStrings("deflate, chunked", header.value);
+            try testing.expect(!it.is_trailer);
+        }
+        {
+            const header = it.next().?;
+            try testing.expectEqualStrings("connectioN", header.name);
+            try testing.expectEqualStrings("keep-alive", header.value);
+            try testing.expect(!it.is_trailer);
+        }
+        try testing.expectEqual(null, it.next());
+    }
 };
 
 /// A HTTP request that has been sent.
diff --git a/lib/std/http/HeaderIterator.zig b/lib/std/http/HeaderIterator.zig
index 515058859d..518950a9ab 100644
--- a/lib/std/http/HeaderIterator.zig
+++ b/lib/std/http/HeaderIterator.zig
@@ -12,30 +12,43 @@ pub fn init(bytes: []const u8) HeaderIterator {
 
 pub fn next(it: *HeaderIterator) ?std.http.Header {
     const end = std.mem.indexOfPosLinear(u8, it.bytes, it.index, "\r\n").?;
-    var kv_it = std.mem.splitSequence(u8, it.bytes[it.index..end], ": ");
-    const name = kv_it.next().?;
-    const value = kv_it.rest();
-    if (name.len == 0 and value.len == 0) {
+    if (it.index == end) { // found the trailer boundary (\r\n\r\n)
         if (it.is_trailer) return null;
+
         const next_end = std.mem.indexOfPosLinear(u8, it.bytes, end + 2, "\r\n") orelse
             return null;
+
+        var kv_it = std.mem.splitScalar(u8, it.bytes[end + 2 .. next_end], ':');
+        const name = kv_it.first();
+        const value = kv_it.rest();
+
         it.is_trailer = true;
         it.index = next_end + 2;
-        kv_it = std.mem.splitSequence(u8, it.bytes[end + 2 .. next_end], ": ");
+        if (name.len == 0)
+            return null;
+
+        return .{
+            .name = name,
+            .value = std.mem.trim(u8, value, " \t"),
+        };
+    } else { // normal header
+        var kv_it = std.mem.splitScalar(u8, it.bytes[it.index..end], ':');
+        const name = kv_it.first();
+        const value = kv_it.rest();
+
+        it.index = end + 2;
+        if (name.len == 0)
+            return null;
+
         return .{
-            .name = kv_it.next().?,
-            .value = kv_it.rest(),
+            .name = name,
+            .value = std.mem.trim(u8, value, " \t"),
         };
     }
-    it.index = end + 2;
-    return .{
-        .name = name,
-        .value = value,
-    };
 }
 
 test next {
-    var it = HeaderIterator.init("200 OK\r\na: b\r\nc: \r\nd: e\r\n\r\nf: g\r\n\r\n");
+    var it = HeaderIterator.init("200 OK\r\na: b\r\nc:  \r\nd:e\r\n\r\nf: g\r\n\r\n");
     try std.testing.expect(!it.is_trailer);
     {
         const header = it.next().?;
@@ -62,7 +75,23 @@ test next {
         try std.testing.expectEqualStrings("g", header.value);
     }
     try std.testing.expectEqual(null, it.next());
+
+    it = HeaderIterator.init("200 OK\r\n: ss\r\n\r\n");
+    try std.testing.expect(!it.is_trailer);
+    try std.testing.expectEqual(null, it.next());
+
+    it = HeaderIterator.init("200 OK\r\na:b\r\n\r\n: ss\r\n\r\n");
+    try std.testing.expect(!it.is_trailer);
+    {
+        const header = it.next().?;
+        try std.testing.expect(!it.is_trailer);
+        try std.testing.expectEqualStrings("a", header.name);
+        try std.testing.expectEqualStrings("b", header.value);
+    }
+    try std.testing.expectEqual(null, it.next());
+    try std.testing.expect(it.is_trailer);
 }
 
 const HeaderIterator = @This();
 const std = @import("../std.zig");
+const assert = std.debug.assert;
diff --git a/lib/std/http/Server.zig b/lib/std/http/Server.zig
index bfc97157f6..5290241b6e 100644
--- a/lib/std/http/Server.zig
+++ b/lib/std/http/Server.zig
@@ -211,9 +211,9 @@ pub const Request = struct {
                     else => {},
                 }
 
-                var line_it = mem.splitSequence(u8, line, ": ");
+                var line_it = mem.splitScalar(u8, line, ':');
                 const header_name = line_it.next().?;
-                const header_value = line_it.rest();
+                const header_value = mem.trim(u8, line_it.rest(), " \t");
                 if (header_name.len == 0) return error.HttpHeadersInvalid;
 
                 if (std.ascii.eqlIgnoreCase(header_name, "connection")) {
@@ -271,6 +271,29 @@ pub const Request = struct {
             return error.MissingFinalNewline;
         }
 
+        test parse {
+            const request_bytes = "GET /hi HTTP/1.0\r\n" ++
+                "content-tYpe: text/plain\r\n" ++
+                "content-Length:10\r\n" ++
+                "expeCt:   100-continue \r\n" ++
+                "TRansfer-encoding:\tdeflate, chunked \r\n" ++
+                "connectioN:\t keep-alive \r\n\r\n";
+
+            const req = try parse(request_bytes);
+
+            try testing.expectEqual(.GET, req.method);
+            try testing.expectEqual(.@"HTTP/1.0", req.version);
+            try testing.expectEqualStrings("/hi", req.target);
+
+            try testing.expectEqualStrings("text/plain", req.content_type.?);
+            try testing.expectEqualStrings("100-continue", req.expect.?);
+
+            try testing.expectEqual(true, req.keep_alive);
+            try testing.expectEqual(10, req.content_length.?);
+            try testing.expectEqual(.chunked, req.transfer_encoding);
+            try testing.expectEqual(.deflate, req.transfer_compression);
+        }
+
         inline fn int64(array: *const [8]u8) u64 {
             return @bitCast(array.*);
         }
@@ -280,6 +303,66 @@ pub const Request = struct {
         return http.HeaderIterator.init(r.server.read_buffer[0..r.head_end]);
     }
 
+    test iterateHeaders {
+        const request_bytes = "GET /hi HTTP/1.0\r\n" ++
+            "content-tYpe: text/plain\r\n" ++
+            "content-Length:10\r\n" ++
+            "expeCt:   100-continue \r\n" ++
+            "TRansfer-encoding:\tdeflate, chunked \r\n" ++
+            "connectioN:\t keep-alive \r\n\r\n";
+
+        var read_buffer: [500]u8 = undefined;
+        @memcpy(read_buffer[0..request_bytes.len], request_bytes);
+
+        var server: Server = .{
+            .connection = undefined,
+            .state = .ready,
+            .read_buffer = &read_buffer,
+            .read_buffer_len = request_bytes.len,
+            .next_request_start = 0,
+        };
+
+        var request: Request = .{
+            .server = &server,
+            .head_end = request_bytes.len,
+            .head = undefined,
+            .reader_state = undefined,
+        };
+
+        var it = request.iterateHeaders();
+        {
+            const header = it.next().?;
+            try testing.expectEqualStrings("content-tYpe", header.name);
+            try testing.expectEqualStrings("text/plain", header.value);
+            try testing.expect(!it.is_trailer);
+        }
+        {
+            const header = it.next().?;
+            try testing.expectEqualStrings("content-Length", header.name);
+            try testing.expectEqualStrings("10", header.value);
+            try testing.expect(!it.is_trailer);
+        }
+        {
+            const header = it.next().?;
+            try testing.expectEqualStrings("expeCt", header.name);
+            try testing.expectEqualStrings("100-continue", header.value);
+            try testing.expect(!it.is_trailer);
+        }
+        {
+            const header = it.next().?;
+            try testing.expectEqualStrings("TRansfer-encoding", header.name);
+            try testing.expectEqualStrings("deflate, chunked", header.value);
+            try testing.expect(!it.is_trailer);
+        }
+        {
+            const header = it.next().?;
+            try testing.expectEqualStrings("connectioN", header.name);
+            try testing.expectEqualStrings("keep-alive", header.value);
+            try testing.expect(!it.is_trailer);
+        }
+        try testing.expectEqual(null, it.next());
+    }
+
     pub const RespondOptions = struct {
         version: http.Version = .@"HTTP/1.1",
         status: http.Status = .ok,
@@ -1060,5 +1143,6 @@ const mem = std.mem;
 const net = std.net;
 const Uri = std.Uri;
 const assert = std.debug.assert;
+const testing = std.testing;
 
 const Server = @This();
diff --git a/lib/std/http/test.zig b/lib/std/http/test.zig
index 15e7542d19..e2aa810d58 100644
--- a/lib/std/http/test.zig
+++ b/lib/std/http/test.zig
@@ -328,8 +328,8 @@ test "receiving arbitrary http headers from the client" {
     defer test_server.destroy();
 
     const request_bytes = "GET /bar HTTP/1.1\r\n" ++
-        "CoNneCtIoN: close\r\n" ++
-        "aoeu: asdf\r\n" ++
+        "CoNneCtIoN:close\r\n" ++
+        "aoeu:  asdf \r\n" ++
         "\r\n";
     const gpa = std.testing.allocator;
     const stream = try std.net.tcpConnectToHost(gpa, "127.0.0.1", test_server.port());