From c4f2de4933ae014cf81d5c4e350ca2eb25c67916 Mon Sep 17 00:00:00 2001
From: Victor Zverovich <viz@meta.com>
Date: Sun, 24 Dec 2023 07:32:27 -0800
Subject: [PATCH] Improve scan

---
 test/scan-test.cc | 12 ++++++++----
 test/scan.h       | 30 ++++++++++++++++++------------
 2 files changed, 26 insertions(+), 16 deletions(-)

diff --git a/test/scan-test.cc b/test/scan-test.cc
index 2f3cd6a2..96ae3746 100644
--- a/test/scan-test.cc
+++ b/test/scan-test.cc
@@ -127,11 +127,15 @@ TEST(scan_test, example) {
 TEST(scan_test, file) {
   fmt::file read_end, write_end;
   fmt::file::pipe(read_end, write_end);
-  fmt::string_view input = "42";
+
+  fmt::string_view input = "10 20";
   write_end.write(input.data(), input.size());
   write_end.close();
-  int value = 0;
-  fmt::scan(read_end.fdopen("r").get(), "{}", value);
-  EXPECT_EQ(value, 42);
+
+  int n1 = 0, n2 = 0;
+  fmt::buffered_file f = read_end.fdopen("r");
+  fmt::scan(f.get(), "{} {}", n1, n2);
+  EXPECT_EQ(n1, 10);
+  EXPECT_EQ(n2, 20);
 }
 #endif  // FMT_USE_FCNTL
diff --git a/test/scan.h b/test/scan.h
index 2ea7bfc2..f7d3e1da 100644
--- a/test/scan.h
+++ b/test/scan.h
@@ -14,6 +14,10 @@
 FMT_BEGIN_NAMESPACE
 namespace detail {
 
+inline bool is_whitespace(char c) {
+  return c == ' ' || c == '\n';
+}
+
 struct maybe_contiguous_range {
   const char* begin;
   const char* end;
@@ -39,21 +43,19 @@ class scan_buffer {
     end_ = end;
   }
 
-  auto peek() -> int {
-    if (ptr_ == end_) {
-      // TODO: refill buffer
-      return EOF;
-    }
-    return *ptr_;
-  }
+  const char* ptr() const { return ptr_; }
 
-  // Fills the buffer with more input if available.
-  virtual void consume() = 0;
+  auto peek() -> int {
+    return ptr_ != end_ ? *ptr_ : EOF;
+  }
 
  public:
   scan_buffer(const scan_buffer&) = delete;
   void operator=(const scan_buffer&) = delete;
 
+  // Fills the buffer with more input if available.
+  virtual void consume() = 0;
+
   class iterator {
    private:
     const char** ptr_;
@@ -76,6 +78,7 @@ class scan_buffer {
 
     iterator(scan_buffer* buf)
         : ptr_(&buf->ptr_), buf_(buf), value_(static_cast<char>(buf->peek())) {
+      // TODO: fix check
       if (value_ == EOF) ptr_ = sentinel();
     }
 
@@ -235,7 +238,8 @@ class file_scan_buffer : public scan_buffer {
   void consume() override {
     // Consume the current buffer content.
     // TODO: do it more efficiently
-    for (size_t i = 0, n = file_.buffer().size(); i != n; ++i) file_.get();
+    size_t n = to_unsigned(ptr() - file_.buffer().begin());
+    for (size_t i = 0; i != n; ++i) file_.get();
     fill();
   }
 
@@ -284,8 +288,7 @@ struct scan_context {
   auto end() const -> iterator { return buf_.end(); }
 
   void advance_to(iterator) {
-    // The scan_buffer iterator automatically updates the buffer position when
-    // incremented.
+    buf_.consume();
   }
 };
 
@@ -418,6 +421,7 @@ struct scan_handler : error_handler {
   auto pos() const -> scan_buffer::iterator { return scan_ctx_.begin(); }
 
   void on_text(const char* begin, const char* end) {
+    if (begin == end) return;
     auto it = scan_ctx_.begin(), scan_end = scan_ctx_.end();
     for (; begin != end; ++begin, ++it) {
       if (it == scan_end || *begin != *it) on_error("invalid input");
@@ -438,6 +442,8 @@ struct scan_handler : error_handler {
 
   void on_replacement_field(int, const char*) {
     auto it = scan_ctx_.begin(), end = scan_ctx_.end();
+    while (it != end && is_whitespace(*it)) ++it;
+    scan_ctx_.advance_to(it);
     switch (arg_.type) {
     case scan_type::int_type:
       *arg_.int_value = read_int();