roland/xo-umbrella2
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

xo-tokenizer: example tokenrepl restored to wokring order

Browse source 
Now with CBufferedInput in Tokenizer
This commit is contained in:
Roland Conybeare 2026-01-11 19:10:42 -05:00
commit c7486ba674
6 changed files with 45 additions and 29 deletions
Download patch file Download diff file Expand all files Collapse all files

6
xo-arena/include/xo/arena/DCircularBuffer.hpp
View file

@ -39,8 +39,8 @@ namespace xo {
using size_type = std::size_t;
using byte = std::byte;
/** a contiguous addres range **/
using span_type = span<byte>;
using const_span_type = span<const byte>;
using span_type = span<char>;
using const_span_type = span<const char>;
///@}
@ -162,7 +162,7 @@ namespace xo {
*
* Require: @p hi < @ref reserved_range_.hi
**/
bool _expand_to(byte * hi);
bool _expand_to(char * hi);
/** shrink occupied rnage to the smallest contiguous range that contains both:
* all of .input_range_, and all pinned ranges in .pinned_spans_

8
xo-arena/include/xo/arena/span.hpp
View file

@ -88,6 +88,14 @@ namespace xo {
return span(lo, hi);
}
/** @brief create span from raw memory **/
static span from_memory(span<std::byte> span_memory) {
CharT * lo = (CharT *)span_memory.lo();
CharT * hi = (CharT *)span_memory.hi();
return span(lo, hi);
}
/** @brief concatenate two contiguous spans */
static span concat(const span & span1, const span & span2) {
if (span1.is_null())

19
xo-arena/src/arena/DCircularBuffer.cpp
View file

@ -48,17 +48,19 @@ namespace xo {
log && log(xtag("page_z", page_z),
xtag("align_z", align_z));
auto span = mmap_util::map_aligned_range(config.max_capacity_,
align_z,
enable_hugepage_flag,
config.debug_flag_);
auto mapped_span
= span<char>::from_memory(mmap_util::map_aligned_range
(config.max_capacity_,
align_z,
enable_hugepage_flag,
config.debug_flag_));
if (!span.lo()) {
if (!mapped_span.lo()) {
throw std::runtime_error(tostr("DCircularBuffer: reserve address range failed",
xtag("size", config.max_capacity_)));
}
return DCircularBuffer(config, page_z, align_z, span);
return DCircularBuffer(config, page_z, align_z, mapped_span);
}
DCircularBuffer::DCircularBuffer(const CircularBufferConfig & config,
@ -302,7 +304,8 @@ namespace xo {
}
bool
DCircularBuffer::_expand_to(byte * hi) {
DCircularBuffer::_expand_to(char * hi)
{
scope log(XO_DEBUG(config_.debug_flag_));
if (hi < mapped_range_.hi()) {
@ -312,7 +315,7 @@ namespace xo {
size_t add_z = hi - mapped_range_.hi();
size_t add_commit_z = padding::with_padding(add_z, buffer_align_z_);
byte * commit_start = mapped_range_.hi();
char * commit_start = mapped_range_.hi();
if (::mprotect(commit_start,
add_commit_z,

10
xo-arena/utest/DCircularBuffer.test.cpp
View file

@ -34,18 +34,16 @@ namespace xo {
REQUIRE(buf.occupied_range().size() == 0);
REQUIRE(buf.input_range().size() == 0);
std::string_view s0 = "abcdefghijk";
auto s0 = DCircularBuffer::const_span_type::from_cstr("abcdefghijk");
/* return value is unaccepted suffix of input */
REQUIRE(buf.append(DCircularBuffer::span_type((byte *)s0.begin(),
(byte *)s0.end())).empty());
REQUIRE(buf.append(s0).empty());
REQUIRE(buf.verify_ok(verify_policy::log_only()));
REQUIRE(buf.mapped_range().size() == getpagesize());
REQUIRE(buf.occupied_range().size() == s0.size());
REQUIRE(buf.input_range().size() == s0.size());
std::string_view s1 = "lmnopq";
REQUIRE(buf.append(DCircularBuffer::span_type((byte *)s1.begin(),
(byte *)s1.end())).empty());
auto s1 = DCircularBuffer::const_span_type::from_cstr("lmnopq");
REQUIRE(buf.append(s1).empty());
REQUIRE(buf.mapped_range().size() == getpagesize());
REQUIRE(buf.occupied_range().size() == s0.size() + s1.size());

15
xo-tokenizer2/example/tokenrepl/tokenrepl.cpp
View file

@ -51,6 +51,8 @@ main() {
using xo::scm::operator<<;
using xo::mm::CircularBufferConfig;
using xo::mm::span;
using xo::scope;
using xo::xtag;
using replxx::Replxx;
using namespace std;
@ -65,10 +67,13 @@ main() {
rx.set_max_history_size(1000);
rx.history_load("repl_history.txt");
constexpr bool c_debug_flag = true;
scope log(XO_DEBUG(c_debug_flag));
Tokenizer tkz(CircularBufferConfig{.name_ = "tokenrepl-input",
.max_capacity_ = 4*1024,
.max_captured_span_ = 128},
true /*debug_flag*/);
c_debug_flag);
const char * input_cstr = nullptr;;
@ -84,9 +89,17 @@ main() {
if (input_cstr && *input_cstr) {
auto [error, input] = tkz.buffer_input_line(input_cstr, false /*!eof*/);
if (log) {
log(xtag("msg", "buffered input line"));
log(xtag("input", input));
}
while (!input.empty())
{
auto [tk, consumed, error] = tkz.scan(input);
log && log(xtag("consumed", consumed), xtag("tk", tk));
if (tk.is_valid()) {
cout << tk << endl;
} else if (error.is_error()) {

14
xo-tokenizer2/src/tokenizer2/Tokenizer.cpp
View file

@ -622,15 +622,9 @@ namespace xo {
auto buf_input_0 = input_buffer_.input_range().hi();
auto remainder = input_buffer_.append
(DCircularBuffer::const_span_type
((const byte *)input_cstr,
(const byte *)input_cstr + strlen(input_cstr)));
const char * newline_cstr = "\n";
(DCircularBuffer::const_span_type::from_cstr(input_cstr));
auto remainder2 = input_buffer_.append
(DCircularBuffer::const_span_type
((const byte *)newline_cstr,
(const byte *)newline_cstr + strlen(newline_cstr)));
(DCircularBuffer::const_span_type::from_cstr("\n"));
if (!remainder.empty() || !remainder2.empty()) {
throw std::runtime_error(tostr("Tokenizer::buffer_line: line too long!",
@ -639,8 +633,8 @@ namespace xo {
auto buf_input_1 = input_buffer_.input_range().hi();
span_type input = span_type((const char *)buf_input_0,
(const char *)buf_input_1);
span_type input = span_type(buf_input_0,
buf_input_1);
return this->input_state_.capture_current_line(input, eof_flag);
}