mcap/modules/jaison/module.jai

// This file contains just the JSON serialization functions. See generic.jai and typed.jai for parse fuctions.

// Generates a JSON string from either a JSON_Value or any custom type.
// "indent_char" does what it says on the tin.
// "ignore" is only used for custom types to determine which properties of your custom type should be serialized.
//		The default ignore function ignores all struct members that have the note @JsonIgnore.
// "rename" is used for renaming certain members.
//      It gets called with the Type_Info_Struct_Member and must return the new name of the field.
// 		The default procedure rename members by their @JsonName note. Eg: @JsonName(renamed_member).
json_write_string :: (value: $T, indent_char := "\t", ignore := ignore_by_note, rename := rename_by_note, float_handling := Special_Float_Handling.SUPPORT_NULL) -> string {
	builder: String_Builder;
	defer free_buffers(*builder);
	json_append_value(*builder, value, indent_char, ignore, rename, float_handling);
	return builder_to_string(*builder);
}

json_write_file :: (filename: string, value: $T, indent_char := "\t", ignore := ignore_by_note, rename := rename_by_note, float_handling := Special_Float_Handling.SUPPORT_NULL) -> bool {
	builder: String_Builder;
	defer free_buffers(*builder);
	json_append_value(*builder, value, indent_char, ignore, rename, float_handling);
	return write_entire_file(filename, *builder);
}

json_append_value :: (builder: *String_Builder, val: $T, indent_char := "\t", ignore := ignore_by_note, rename := rename_by_note, float_handling := Special_Float_Handling.SUPPORT_NULL) {
	#if T == JSON_Value {
		json_write_json_value(builder, val, indent_char, float_handling=float_handling);
	} else {
		info := type_info(T);
		json_write_native(builder, *val, info, indent_char, ignore, rename, float_handling=float_handling);
	}
}

// This function is useful if you have a JSON template string and just want to
// safely insert a value without having to replicate the full json structure in Jai.
// The return value does NOT include quotes around the string.
//
// Example:
// JSON_TEMPLATE :: #string END
// {
// 	"complicated": {
// 		"json": {
// 			"structure": {
// 				"for_a_stupid_api": {
// 					"that_needs": [
// 						{"a_deeply_nested_value": "%1"}
// 					]
// 				}
// 			}
// 		}
// 	}
// }
// END
// escaped_value := json_escape_string(my_unsafe_value);
// defer free(escaped_value);
// json_str := print(JSON_TEMPLATE, escaped_value);
json_escape_string :: (str: string) -> string {
	builder: String_Builder;
	defer	free_buffers(*builder);
	json_append_escaped(*builder, str);
	escaped := builder_to_string(*builder);
	return escaped;
}

// How shall NaN and Infinity be serialized / deserialized?
Special_Float_Handling :: enum_flags {
    SUPPORT_NULL    :: 0x01;
    SUPPORT_STRINGS :: 0x02;
    SUPPORT_ALL     :: SUPPORT_NULL | SUPPORT_STRINGS;
}

Ignore_Proc :: #type (member: *Type_Info_Struct_Member) -> bool;
Rename_Proc :: #type (member: *Type_Info_Struct_Member) -> string;

ignore_by_note :: (member: *Type_Info_Struct_Member) -> bool {
	for note: member.notes {
		if note == "JsonIgnore"	return true;
	}
	return false;
}

rename_by_note :: (member: *Type_Info_Struct_Member) -> string {
	for note: member.notes {
		if !begins_with(note, "JsonName(") continue;
        if note.count <= 10 || note[note.count-1] != #char ")" {
            log_error("Invalid JsonName note format. Expected a name in parenthesis, but the note was \"%\".", note);
            continue;
        }

		return slice(note, 9, note.count-10);
	}

	return member.name;
}

#scope_module

WHITESPACE_CHARS :: " \t\n\r";

#load "generic.jai";
#load "typed.jai";

json_append_escaped :: (builder: *String_Builder, str: string) {
	remaining := str;
	next_pos := index_of_illegal_string_char(remaining);
	append(builder, "\"");
	while (next_pos >= 0) {
		append(builder, slice(remaining, 0, next_pos));
		if remaining[next_pos] == {
			case #char "\\";
				append(builder,  "\\\\");
			case #char "\"";
				append(builder,  "\\\"");
			case #char "\n";
				append(builder, "\\n");
			case #char "\r";
				append(builder, "\\r");
			case #char "\t";
				append(builder, "\\t");
			case;
				// ToDo: handle illegal multi-byte characters
				// print("Escaping: %\n\n", slice(remaining, next_pos, remaining.count - next_pos));
				print_to_builder(builder, "\\u%", formatInt(remaining[next_pos], base=16, minimum_digits=4));
		}
		remaining = advance(remaining, next_pos + 1);
		next_pos = index_of_illegal_string_char(remaining);
	}
	append(builder, remaining);
	append(builder, "\"");
}

index_of_illegal_string_char :: (str: string) -> s64 {
	for 0..str.count - 1 {
		if str[it] == #char "\\" || str[it] == #char "\"" || str[it] <= 0x1F {
			return it;
		}
	}
	return -1;
}

expect_and_slice :: (str: string, expected: string) -> remainder: string, success: bool {
	if str.count < expected.count || !equal(slice(str, 0, expected.count), expected) {
		log_error("Unexpected token. Expected \"%\" but got: %", expected, str);
		return str, false;
	}
	remainder := advance(str, expected.count);
	return remainder, true;
}

parse_string :: (str: string) -> result: string, remainder: string, success: bool {
	assert(str[0] == #char "\"", "Invalid string start %", str);
	inside := advance(str);
    if !inside return "", "", false;
	needsUnescape := false;
	while true {
		if !inside || (inside.count < 2 && inside[0] != #char "\"")	return "", str, false;
		if inside[0] == #char "\""	break;
		if inside[0] == #char "\\" {
			needsUnescape = true;
			if inside.count < 2	return "", str, false;
			advance(*inside);
		}
		advance(*inside);
	}

	length := inside.data - str.data - 1;
	result := slice(str, 1, length);
	if needsUnescape {
		success: bool;
		result, success = unescape(result);
		if !success		return "", str, false;
	} else {
		result = copy_string(result);
	}

	remainder := slice(str, length + 2, str.count - length - 2);
	return result, remainder, true;
}

unescape :: (str: string) -> result: string, success: bool {
	result := alloc_string(str.count);
	rc := 0;
    success := true;
	for i: 0..str.count-1 {
		if str[i] != #char "\\" {
			// Check for invalid characters for JSON
			if str[i] < 0x20	{
                success = false;
                break;
            }

			result[rc] = str[i];
			rc += 1;
		} else {
			if i == str.count - 1	{
                success = false;
                break;
            }

			i += 1;
			if str[i] == {
				case #char "\""; #through;
				case #char "/"; #through;
				case #char "\\";
					result[rc] = str[i];
					rc += 1;
				case #char "b";
					result[rc] = 0x08;
					rc += 1;
				case #char "f";
					result[rc] = 0x0c;
					rc += 1;
				case #char "n";
					result[rc] = #char "\n";
					rc += 1;
				case #char "r";
					result[rc] = #char "\r";
					rc += 1;
				case #char "t";
					result[rc] = #char "\t";
					rc += 1;
				case #char "u";
					if i + 4 >= str.count	{
                        success = false;
                        break;
                    }
					unicode_char, unicode_success := parse_unicode(slice(str, i + 1, 4));
					if !unicode_success	{
                        success = false;
                        break;
                    }
					// A high surrogate followed by \u and a low one is a single character beyond
					// U+FFFF, such as an emoji: four bytes of UTF-8, not two 3-byte halves (which are
					// not valid UTF-8). A surrogate without its other half is encoded on its own, as before.
					if unicode_char >= 0xD800 && unicode_char <= 0xDBFF && i + 10 < str.count && str[i + 5] == #char "\\" && str[i + 6] == #char "u" {
						low, low_success := parse_unicode(slice(str, i + 7, 4));
						if low_success && low >= 0xDC00 && low <= 0xDFFF {
							code_point := 0x10000 + ((cast(u32) unicode_char - 0xD800) << 10) + (cast(u32) low - 0xDC00);
							result[rc]     = xx (0xF0 | (code_point >> 18));
							result[rc + 1] = xx (0x80 | ((code_point >> 12) & 0x3F));
							result[rc + 2] = xx (0x80 | ((code_point >> 6) & 0x3F));
							result[rc + 3] = xx (0x80 | (code_point & 0x3F));
							rc += 4;
							i += 10;
							continue;
						}
					}
					utf8_len := encode_utf8(unicode_char, *(result.data[rc]));
					rc += utf8_len;
					i += 4;
				case;
                    success = false;
                    break;
			}
		}
	}
    if !success {
        free(result);
        return "", false;
    }
	result.count = rc;
	return result, true;
}

append_special_float :: (builder: *String_Builder, float_handling: Special_Float_Handling, is_nan: bool, is_negative: bool) {
    if !(float_handling & .SUPPORT_STRINGS) {
        append(builder, "null");
    } else if is_nan {
        append(builder, "\"NaN\"");
    } else if is_negative {
        append(builder, "\"-Infinity\"");
    } else {
        append(builder, "\"Infinity\"");
    }
}

get_nan_or_infinity :: (value: u32) -> is_nan: bool, is_infinity: bool, is_negative: bool {
    is_special := (value & 0x7F80_0000) == 0x7F80_0000;
    mant := (value & 0x007F_FFFF);
    return (is_special && mant), is_special && !mant, !!(value & 0x8000_0000);
}

get_nan_or_infinity :: (value: u64) -> is_nan: bool, is_infinity: bool, is_negative: bool {
    is_special := (value & 0x7FF0_0000_0000_0000) == 0x7FF0_0000_0000_0000;
    mant := (value & 0xF_FFFF_FFFF_FFF);
    return (is_special && mant), is_special && !mant, !!(value & 0x8000_0000_0000_0000);
}

NAN               :: 0h7FFF_FFFF_FFFF_FFFF;
INFINITY          :: 0h7FF0_0000_0000_0000;
NEGATIVE_INFINITY :: 0h8FF0_0000_0000_0000;

#import "Basic";
#import "File";
#import "String";

#import "Hash_Table";
#import,dir "./unicode_utils";
#import "IntroSort";