pw_tokenizer/
lib.rs

1// Copyright 2023 The Pigweed Authors
2//
3// Licensed under the Apache License, Version 2.0 (the "License"); you may not
4// use this file except in compliance with the License. You may obtain a copy of
5// the License at
6//
7//     https://www.apache.org/licenses/LICENSE-2.0
8//
9// Unless required by applicable law or agreed to in writing, software
10// distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
11// WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
12// License for the specific language governing permissions and limitations under
13// the License.
14
15//! `pw_tokenizer` - Efficient string handling and printf style encoding.
16//!
17//! Logging is critical, but developers are often forced to choose between
18//! additional logging or saving crucial flash space. The `pw_tokenizer` crate
19//! helps address this by replacing printf-style strings with binary tokens
20//! during compilation. This enables extensive logging with substantially less
21//! memory usage.
22//!
23//! For a more in depth explanation of the systems design and motivations,
24//! see [Pigweed's pw_tokenizer module documentation](https://pigweed.dev/pw_tokenizer/).
25//!
26//! # Examples
27//!
28//! Pigweed's tokenization database uses `printf` style strings internally so
29//! those are supported directly.
30//!
31//! ```
32//! use pw_tokenizer::tokenize_printf_to_buffer;
33//!
34//! let mut buffer = [0u8; 1024];
35//! let len = tokenize_printf_to_buffer!(&mut buffer, "The answer is %d", 42)?;
36//!
37//! // 4 bytes used to encode the token and one to encode the value 42.  This
38//! // is a **3.5x** reduction in size compared to the raw string!
39//! assert_eq!(len, 5);
40//! # Ok::<(), pw_status::Error>(())
41//! ```
42//!
43//! We also support Rust's `core::fmt` style syntax.  These format strings are
44//! converted to `printf` style at compile time to maintain compatibly with the
45//! rest of the Pigweed tokenizer ecosystem.  The below example produces the
46//! same token and output as the above one.
47//!
48//! ```
49//! use pw_tokenizer::tokenize_core_fmt_to_buffer;
50//!
51//! let mut buffer = [0u8; 1024];
52//! let len = tokenize_core_fmt_to_buffer!(&mut buffer, "The answer is {}", 42 as i32)?;
53//! assert_eq!(len, 5);
54//! # Ok::<(), pw_status::Error>(())
55//! ```
56#![cfg_attr(feature = "nightly", feature(type_alias_impl_trait))]
57#![cfg_attr(not(feature = "std"), no_std)]
58#![deny(missing_docs)]
59
60use pw_status::Result;
61
62#[doc(hidden)]
63pub mod internal;
64
65#[doc(hidden)]
66// Creating a __private namespace allows us a way to get to the modules
67// we need from macros by doing:
68//     use $crate::__private as __pw_tokenizer_crate;
69//
70// This is how proc macro generated code can reliably reference back to
71// `pw_tokenizer` while still allowing a user to import it under a different
72// name.
73pub mod __private {
74    pub use pw_bytes::concat_static_strs;
75    pub use pw_format_core::{PrintfFormatter, PrintfHexFormatter, PrintfUpperHexFormatter};
76    pub use pw_status::Result;
77    pub use pw_stream::{Cursor, Seek, WriteInteger, WriteVarint};
78    pub use pw_tokenizer_core::hash_string;
79    pub use pw_tokenizer_macro::{
80        _token, _tokenize_core_fmt_to_buffer, _tokenize_core_fmt_to_writer,
81        _tokenize_printf_to_buffer, _tokenize_printf_to_writer,
82    };
83
84    pub use crate::*;
85}
86
87/// Return the [`u32`] token for the specified string and add it to the token
88/// database.
89///
90/// This is where the magic happens in `pw_tokenizer`!   ... and by magic
91/// we mean hiding information in a special linker section that ends up in the
92/// final elf binary but does not get flashed to the device.
93///
94/// Two things are accomplished here:
95/// 1) The string is hashed into its stable `u32` token.  This is the value that
96///    is returned from the macro.
97/// 2) A [token database entry](https://pigweed.dev/pw_tokenizer/design.html#binary-database-format)
98///    is generated, assigned to a unique static symbol, placed in a linker
99///    section named `pw_tokenizer.entries.<TOKEN_HASH>`.  A
100///    [linker script](https://pigweed.googlesource.com/pigweed/pigweed/+/refs/heads/main/pw_tokenizer/pw_tokenizer_linker_sections.ld)
101///    is responsible for picking these symbols up and aggregating them into a
102///    single `.pw_tokenizer.entries` section in the final binary.
103///
104/// # Example
105/// ```
106/// use pw_tokenizer::token;
107///
108/// let token = token!("hello, \"world\"");
109/// assert_eq!(token, 3537412730);
110/// ```
111///
112/// Currently there is no support for encoding tokens to specific domains
113/// or with "fixed lengths" per [`pw_tokenizer_core::hash_bytes_fixed`].
114#[macro_export]
115macro_rules! token {
116    ($string:literal) => {{
117        use $crate::__private as __pw_tokenizer_crate;
118        $crate::__private::_token!($string)
119    }};
120}
121
122/// Tokenize a `core::fmt` style format string and arguments to an [`AsMut<u8>`]
123/// buffer.  The format string is converted in to a `printf` and added token to
124/// the token database.
125///
126/// See [`token`] for an explanation on how strings are tokenized and entries
127/// are added to the token database.  The token's domain is set to `""`.
128///
129/// Returns a [`pw_status::Result<usize>`] the number of bytes written to the buffer.
130///
131/// `tokenize_to_buffer!` supports concatenation of format strings as described
132/// in [`pw_format::macros::FormatAndArgs`].
133///
134/// # Errors
135/// - [`pw_status::Error::OutOfRange`] - Buffer is not large enough to fit
136///   tokenized data.
137/// - [`pw_status::Error::InvalidArgument`] - Invalid buffer was provided.
138///
139/// # Example
140///
141/// ```
142/// use pw_tokenizer::tokenize_core_fmt_to_buffer;
143///
144/// // Tokenize a format string and argument into a buffer.
145/// let mut buffer = [0u8; 1024];
146/// let len = tokenize_core_fmt_to_buffer!(&mut buffer, "The answer is {}", 42 as i32)?;
147///
148/// // 4 bytes used to encode the token and one to encode the value 42.
149/// assert_eq!(len, 5);
150///
151/// // The format string can be composed of multiple strings literals using
152/// // the custom`PW_FMT_CONCAT` operator.
153/// let len = tokenize_core_fmt_to_buffer!(&mut buffer, "Hello " PW_FMT_CONCAT "Pigweed")?;
154///
155/// // Only a single 4 byte token is emitted after concatenation of the string
156/// // literals above.
157/// assert_eq!(len, 4);
158/// # Ok::<(), pw_status::Error>(())
159/// ```
160#[macro_export]
161macro_rules! tokenize_core_fmt_to_buffer {
162    ($buffer:expr, $($format_string:literal)PW_FMT_CONCAT+ $(, $args:expr)* $(,)?) => {{
163      use $crate::__private as __pw_tokenizer_crate;
164      __pw_tokenizer_crate::_tokenize_core_fmt_to_buffer!($buffer, $($format_string)PW_FMT_CONCAT+, $($args),*)
165    }};
166}
167
168/// Tokenize a printf format string and arguments to an [`AsMut<u8>`] buffer
169/// and add the format string's token to the token database.
170///
171/// See [`token`] for an explanation on how strings are tokenized and entries
172/// are added to the token database.  The token's domain is set to `""`.
173///
174/// Returns a [`pw_status::Result<usize>`] the number of bytes written to the buffer.
175///
176/// `tokenize_to_buffer!` supports concatenation of format strings as described
177/// in [`pw_format::macros::FormatAndArgs`].
178///
179/// # Errors
180/// - [`pw_status::Error::OutOfRange`] - Buffer is not large enough to fit
181///   tokenized data.
182/// - [`pw_status::Error::InvalidArgument`] - Invalid buffer was provided.
183///
184/// # Example
185///
186/// ```
187/// use pw_tokenizer::tokenize_printf_to_buffer;
188///
189/// // Tokenize a format string and argument into a buffer.
190/// let mut buffer = [0u8; 1024];
191/// let len = tokenize_printf_to_buffer!(&mut buffer, "The answer is %d", 42)?;
192///
193/// // 4 bytes used to encode the token and one to encode the value 42.
194/// assert_eq!(len, 5);
195///
196/// // The format string can be composed of multiple strings literals using
197/// // the custom`PW_FMT_CONCAT` operator.
198/// let len = tokenize_printf_to_buffer!(&mut buffer, "Hello " PW_FMT_CONCAT "Pigweed")?;
199///
200/// // Only a single 4 byte token is emitted after concatenation of the string
201/// // literals above.
202/// assert_eq!(len, 4);
203/// # Ok::<(), pw_status::Error>(())
204/// ```
205#[macro_export]
206macro_rules! tokenize_printf_to_buffer {
207    ($buffer:expr, $($format_string:literal)PW_FMT_CONCAT+ $(, $args:expr)* $(,)?) => {{
208      use $crate::__private as __pw_tokenizer_crate;
209      __pw_tokenizer_crate::_tokenize_printf_to_buffer!($buffer, $($format_string)PW_FMT_CONCAT+, $($args),*)
210    }};
211}
212
213/// Deprecated alias for [`tokenize_printf_to_buffer!`].
214#[macro_export]
215macro_rules! tokenize_to_buffer {
216    ($buffer:expr, $($format_string:literal)PW_FMT_CONCAT+ $(, $args:expr)* $(,)?) => {{
217      $crate::tokenize_printf_to_buffer!($buffer, $($format_string)PW_FMT_CONCAT+, $($args),*)
218    }};
219}
220
221/// Tokenize a `core::fmt` format string and arguments to a [`MessageWriter`].
222/// The format string is converted in to a `printf` and added token to the token
223/// database.
224///
225/// `tokenize_core_fmt_to_writer!` and the accompanying [`MessageWriter`] trait
226/// provide an optimized API for use cases like logging where the output of the
227/// tokenization will be written to a shared/ambient resource like stdio, a
228/// UART, or a shared buffer.
229///
230/// See [`token`] for an explanation on how strings are tokenized and entries
231/// are added to the token database.  The token's domain is set to `""`.
232///
233/// Returns a [`pw_status::Result<()>`].
234///
235/// `tokenize_core_fmt_to_writer!` supports concatenation of format strings as
236///  described in [`pw_format::macros::FormatAndArgs`].
237///
238/// # Errors
239/// - [`pw_status::Error::OutOfRange`] - [`MessageWriter`] does not have enough
240///   space to fit tokenized data.
241/// - others - `tokenize_core_fmt_to_writer!` will pass on any errors returned
242///   by the [`MessageWriter`].
243///
244/// # Code Size
245///
246/// This data was collected by examining the disassembly of a test program
247/// built for a Cortex M0.
248///
249/// | Tokenized Message   | Per Call-site Cost (bytes) |
250/// | --------------------| -------------------------- |
251/// | no arguments        | 10                         |
252/// | one `i32` argument  | 18                         |
253///
254/// # Example
255///
256/// ```
257/// use pw_status::Result;
258/// use pw_stream::{Cursor, Write};
259/// use pw_tokenizer::{MessageWriter, tokenize_core_fmt_to_writer};
260///
261/// const BUFFER_LEN: usize = 32;
262///
263/// // Declare a simple MessageWriter that uses a [`pw_status::Cursor`] to
264/// // maintain an internal buffer.
265/// struct TestMessageWriter {
266///   cursor: Cursor<[u8; BUFFER_LEN]>,
267/// }
268///
269/// impl TestMessageWriter {
270///   fn new() -> Self {
271///       Self {
272///           cursor: Cursor::new([0u8; BUFFER_LEN]),
273///       }
274///   }
275/// }
276///
277/// impl MessageWriter for TestMessageWriter {
278///   fn write(&mut self, data: &[u8]) -> Result<()> {
279///       self.cursor.write_all(data)
280///   }
281///
282///   fn remaining(&self) -> usize {
283///       self.cursor.remaining()
284///   }
285///
286///   fn finalize(self) -> Result<()> {
287///       let len = self.cursor.position();
288///       // 4 bytes used to encode the token and one to encode the value 42.
289///       assert_eq!(len, 5);
290///       Ok(())
291///   }
292/// }
293///
294/// // Tokenize a format string and argument into the writer.  Note how we
295/// // pass in the message writer's type, not an instance of it.
296/// let len = tokenize_core_fmt_to_writer!(TestMessageWriter::new(), "The answer is {}", 42 as i32)?;
297/// # Ok::<(), pw_status::Error>(())
298/// ```
299#[macro_export]
300macro_rules! tokenize_core_fmt_to_writer {
301    ($writer:expr, $($format_string:literal)PW_FMT_CONCAT+ $(, $args:expr)* $(,)?) => {{
302      use $crate::__private as __pw_tokenizer_crate;
303      __pw_tokenizer_crate::_tokenize_core_fmt_to_writer!($writer, $($format_string)PW_FMT_CONCAT+, $($args),*)
304    }};
305}
306
307/// Tokenize a `printf` format string and arguments to a [`MessageWriter`] and
308/// add the format string's token to the token database.
309///
310/// `tokenize_printf_fmt_to_writer!` and the accompanying [`MessageWriter`] trait
311/// provide an optimized API for use cases like logging where the output of the
312/// tokenization will be written to a shared/ambient resource like stdio, a
313/// UART, or a shared buffer.
314///
315/// See [`token`] for an explanation on how strings are tokenized and entries
316/// are added to the token database.  The token's domain is set to `""`.
317///
318/// Returns a [`pw_status::Result<()>`].
319///
320/// `tokenize_core_fmt_to_writer!` supports concatenation of format strings as
321///  described in [`pw_format::macros::FormatAndArgs`].
322///
323/// # Errors
324/// - [`pw_status::Error::OutOfRange`] - [`MessageWriter`] does not have enough
325///   space to fit tokenized data.
326/// - others - `tokenize_printf_to_writer!` will pass on any errors returned
327///   by the [`MessageWriter`].
328///
329/// # Code Size
330///
331/// This data was collected by examining the disassembly of a test program
332/// built for a Cortex M0.
333///
334/// | Tokenized Message   | Per Call-site Cost (bytes) |
335/// | --------------------| -------------------------- |
336/// | no arguments        | 10                         |
337/// | one `i32` argument  | 18                         |
338///
339/// # Example
340///
341/// ```
342/// use pw_status::Result;
343/// use pw_stream::{Cursor, Write};
344/// use pw_tokenizer::{MessageWriter, tokenize_printf_to_writer};
345///
346/// const BUFFER_LEN: usize = 32;
347///
348/// // Declare a simple MessageWriter that uses a [`pw_status::Cursor`] to
349/// // maintain an internal buffer.
350/// struct TestMessageWriter {
351///   cursor: Cursor<[u8; BUFFER_LEN]>,
352/// }
353///
354/// impl TestMessageWriter {
355///   fn new() -> Self {
356///       Self {
357///           cursor: Cursor::new([0u8; BUFFER_LEN]),
358///       }
359///   }
360/// }
361///
362///
363/// impl MessageWriter for TestMessageWriter {
364///   fn write(&mut self, data: &[u8]) -> Result<()> {
365///       self.cursor.write_all(data)
366///   }
367///
368///   fn remaining(&self) -> usize {
369///       self.cursor.remaining()
370///   }
371///
372///   fn finalize(self) -> Result<()> {
373///       let len = self.cursor.position();
374///       // 4 bytes used to encode the token and one to encode the value 42.
375///       assert_eq!(len, 5);
376///       Ok(())
377///   }
378/// }
379///
380/// // Tokenize a format string and argument into the writer.  Note how we
381/// // pass in the message writer's type, not an instance of it.
382/// let len = tokenize_printf_to_writer!(TestMessageWriter::new(), "The answer is %d", 42)?;
383/// # Ok::<(), pw_status::Error>(())
384/// ```
385#[macro_export]
386macro_rules! tokenize_printf_to_writer {
387    ($writer:expr, $($format_string:literal)PW_FMT_CONCAT+ $(, $args:expr)* $(,)?) => {{
388      use $crate::__private as __pw_tokenizer_crate;
389      __pw_tokenizer_crate::_tokenize_printf_to_writer!($writer, $($format_string)PW_FMT_CONCAT+, $($args),*)
390    }};
391}
392
393/// Deprecated alias for [`tokenize_printf_to_writer!`].
394#[macro_export]
395macro_rules! tokenize_to_writer {
396  ($writer:expr, $($format_string:literal)PW_FMT_CONCAT+ $(, $args:expr)* $(,)?) => {{
397    $crate::tokenize_printf_to_writer!($writer, $($format_string)PW_FMT_CONCAT+, $($args),*)
398  }};
399}
400
401/// A trait used by [`tokenize_to_writer!`] to output tokenized messages.
402///
403/// For more details on how this type is used, see the [`tokenize_to_writer!`]
404/// documentation.
405pub trait MessageWriter {
406    /// Append `data` to the message.
407    fn write(&mut self, data: &[u8]) -> Result<()>;
408
409    /// Return the remaining space in this message instance.
410    ///
411    /// If there are no space constraints, return `usize::MAX`.
412    fn remaining(&self) -> usize;
413
414    /// Finalize message.
415    ///
416    /// `finalize()` is called when the tokenized message is complete.
417    fn finalize(self) -> Result<()>;
418}
419
420#[cfg(test)]
421// Untyped prints code rely on as casts to annotate type information.
422#[allow(clippy::unnecessary_cast)]
423#[allow(clippy::literal_string_with_formatting_args)]
424mod tests {
425    use super::*;
426    extern crate self as pw_tokenizer;
427    use std::cell::RefCell;
428
429    use pw_stream::{Cursor, Write};
430
431    // This is not meant to be an exhaustive test of tokenization which is
432    // covered by `pw_tokenizer_core`'s unit tests.  Rather, this is testing
433    // that the `tokenize!` macro connects to that correctly.
434    #[test]
435    fn test_token() {}
436
437    macro_rules! tokenize_to_buffer_test {
438      ($expected_data:expr, $buffer_len:expr, $printf_fmt:literal, $core_fmt:literal $(, $args:expr)* $(,)?) => {{
439        if $printf_fmt != "" {
440          let mut buffer = [0u8; $buffer_len];
441          let len = tokenize_printf_to_buffer!(&mut buffer, $printf_fmt, $($args),*).unwrap();
442          assert_eq!(
443              &buffer[..len],
444              $expected_data,
445              "printf style input does not produce expected output",
446          );
447        }
448        if $core_fmt != "" {
449           let mut buffer = [0u8; $buffer_len];
450           let len = tokenize_core_fmt_to_buffer!(&mut buffer, $core_fmt, $($args),*).unwrap();
451           assert_eq!(
452               &buffer[..len],
453               $expected_data,
454              "core::fmt style input does not produce expected output",
455           );
456        }
457      }}
458    }
459
460    macro_rules! tokenize_to_writer_test {
461      ($expected_data:expr, $buffer_len:expr, $printf_fmt:literal, $core_fmt:literal $(, $args:expr)* $(,)?) => {{
462        // The `MessageWriter` API is used in places like logging where it
463        // accesses an shared/ambient resource (like stdio or an UART).  To test
464        // it in a hermetic way we declare test specific `MessageWriter` that
465        // writes it's output to a scoped static variable that can be checked
466        // after the test is run.
467
468        // Since these tests are not multi-threaded, we can use a thread_local!
469        // instead of a mutex.
470        thread_local!(static TEST_OUTPUT: RefCell<Option<Vec<u8>>> = RefCell::new(None));
471
472        struct TestMessageWriter {
473            cursor: Cursor<[u8; $buffer_len]>,
474        }
475
476        impl TestMessageWriter {
477          fn new() -> Self {
478              Self {
479                  cursor: Cursor::new([0u8; $buffer_len]),
480              }
481          }
482        }
483
484        impl MessageWriter for TestMessageWriter {
485          fn write(&mut self, data: &[u8]) -> Result<()> {
486              self.cursor.write_all(data)
487          }
488
489          fn remaining(&self) -> usize {
490              self.cursor.remaining()
491          }
492
493          fn finalize(self) -> Result<()> {
494              let write_len = self.cursor.position();
495              let data = self.cursor.into_inner();
496              TEST_OUTPUT.with(|output| *output.borrow_mut() = Some(data[..write_len].to_vec()));
497
498              Ok(())
499          }
500        }
501
502        if $printf_fmt != "" {
503          TEST_OUTPUT.with(|output| *output.borrow_mut() = None);
504          tokenize_printf_to_writer!(TestMessageWriter::new(), $printf_fmt, $($args),*).unwrap();
505          TEST_OUTPUT.with(|output| {
506              assert_eq!(
507                  *output.borrow(),
508                  Some($expected_data.to_vec()),
509              )
510          });
511        }
512
513        if $core_fmt != "" {
514          TEST_OUTPUT.with(|output| *output.borrow_mut() = None);
515          tokenize_core_fmt_to_writer!(TestMessageWriter::new(), $core_fmt, $($args),*).unwrap();
516          TEST_OUTPUT.with(|output| {
517              assert_eq!(
518                  *output.borrow(),
519                  Some($expected_data.to_vec()),
520              )
521          });
522        }
523      }}
524    }
525
526    macro_rules! tokenize_test {
527        ($expected_data:expr, $buffer_len:expr, $printf_fmt:literal, $core_fmt:literal $(, $args:expr)* $(,)?) => {{
528            tokenize_to_buffer_test!($expected_data, $buffer_len, $printf_fmt, $core_fmt, $($args),*);
529            tokenize_to_writer_test!($expected_data, $buffer_len, $printf_fmt, $core_fmt, $($args),*);
530        }};
531    }
532
533    #[test]
534    fn bare_string_encodes_correctly() {
535        tokenize_test!(
536            &[0xe0, 0x92, 0xe0, 0xa], // expected buffer
537            64,                       // buffer size
538            "Hello Pigweed",          // printf style
539            "Hello Pigweed",          // core::fmt style
540        );
541    }
542
543    #[test]
544    fn test_decimal_format() {
545        // "as casts" are used for the integer arguments below.  They are only
546        // need for the core::fmt style arguments but are added so that we can
547        // check that the printf and core::fmt style equivalents encode the same.
548        tokenize_test!(
549            &[0x52, 0x1c, 0xb0, 0x4c, 0x2], // expected buffer
550            64,                             // buffer size
551            "The answer is %d!",            // printf style
552            "The answer is {}!",            // core::fmt style
553            1 as i32
554        );
555
556        tokenize_test!(
557            &[0x36, 0xd0, 0xfb, 0x69, 0x1], // expected buffer
558            64,                             // buffer size
559            "No! The answer is %d!",        // printf style
560            "No! The answer is {}!",        // core::fmt style
561            -1 as i32
562        );
563
564        tokenize_test!(
565            &[0xa4, 0xad, 0x50, 0x54, 0x0],               // expected buffer
566            64,                                           // buffer size
567            "I think you'll find that the answer is %d!", // printf style
568            "I think you'll find that the answer is {}!", // core::fmt style
569            0 as i32
570        );
571    }
572
573    #[test]
574    fn test_misc_integer_format() {
575        // %d, %i, %o, %u, %x, %X all encode integers the same.
576        tokenize_test!(
577            &[0x52, 0x1c, 0xb0, 0x4c, 0x2], // expected buffer
578            64,                             // buffer size
579            "The answer is %d!",            // printf style
580            "",                             // no equivalent core::fmt style
581            1
582        );
583
584        // Because %i is an alias for %d, it gets converted to a %d by the
585        // `pw_format` macro infrastructure.
586        tokenize_test!(
587            &[0x52, 0x1c, 0xb0, 0x4c, 0x2], // expected buffer
588            64,                             // buffer size
589            "The answer is %i!",            // printf style
590            "",                             // no equivalent core::fmt style
591            1
592        );
593
594        tokenize_test!(
595            &[0x5d, 0x70, 0x12, 0xb4, 0x2], // expected buffer
596            64,                             // buffer size
597            "The answer is %o!",            // printf style
598            "",                             // no equivalent core::fmt style
599            1u32
600        );
601
602        tokenize_test!(
603            &[0x63, 0x58, 0x5f, 0x8f, 0x2], // expected buffer
604            64,                             // buffer size
605            "The answer is %u!",            // printf style
606            "",                             // no equivalent core::fmt style
607            1u32
608        );
609
610        tokenize_test!(
611            &[0x66, 0xcc, 0x05, 0x7d, 0x2], // expected buffer
612            64,                             // buffer size
613            "The answer is %x!",            // printf style
614            "",                             // no equivalent core::fmt style
615            1u32
616        );
617
618        tokenize_test!(
619            &[0x46, 0x4c, 0x16, 0x96, 0x2], // expected buffer
620            64,                             // buffer size
621            "The answer is %X!",            // printf style
622            "",                             // no equivalent core::fmt style
623            1u32
624        );
625    }
626
627    #[test]
628    fn test_string_format() {
629        tokenize_test!(
630            b"\x25\xf6\x2e\x66\x07Pigweed", // expected buffer
631            64,                             // buffer size
632            "Hello: %s!",                   // printf style
633            "",                             // no equivalent core::fmt style
634            "Pigweed"
635        );
636    }
637
638    #[test]
639    fn test_string_format_overflow() {
640        tokenize_test!(
641            b"\x25\xf6\x2e\x66\x83Pig", // expected buffer
642            8,                          // buffer size
643            "Hello: %s!",               // printf style
644            "",                         // no equivalent core::fmt style
645            "Pigweed"
646        );
647    }
648
649    #[test]
650    fn test_char_format() {
651        tokenize_test!(
652            &[0x2e, 0x52, 0xac, 0xe4, 0xa0, 0x1], // expected buffer
653            64,                                   // buffer size
654            "Hello: %cigweed",                    // printf style
655            "",                                   // no equivalent core::fmt style
656            "P".as_bytes()[0]
657        );
658    }
659
660    #[test]
661    fn test_untyped_format() {
662        tokenize_test!(
663            &[0x63, 0x58, 0x5f, 0x8f, 0x2], // expected buffer
664            64,                             // buffer size
665            "The answer is %u!",            // printf style
666            "The answer is {}!",            // core::fmt style
667            1 as u32
668        );
669
670        tokenize_test!(
671            &[0x36, 0xd0, 0xfb, 0x69, 0x1], // expected buffer
672            64,                             // buffer size
673            "No! The answer is %v!",        // printf style
674            "No! The answer is {}!",        // core::fmt style
675            -1 as i32
676        );
677
678        tokenize_test!(
679            b"\x25\xf6\x2e\x66\x07Pigweed", // expected buffer
680            64,                             // buffer size
681            "Hello: %v!",                   // printf style
682            "Hello: {}!",                   // core::fmt style
683            "Pigweed" as &str
684        );
685    }
686
687    #[test]
688    fn test_field_width_and_zero_pad_format() {
689        tokenize_test!(
690            &[0x3a, 0xc2, 0x1a, 0x05, 0xfc, 0xab, 0x06], // expected buffer
691            64,                                          // buffer size
692            "Lets go to the %x",                         // printf style
693            "Lets go to the {:x}",                       // core::fmt style
694            0xcafe as u32
695        );
696
697        tokenize_test!(
698            &[0xf3, 0x16, 0x03, 0x99, 0xfc, 0xab, 0x06], // expected buffer
699            64,                                          // buffer size
700            "Lets go to the %8x",                        // printf style
701            "Lets go to the {:8x}",                      // core::fmt style
702            0xcafe as u32
703        );
704
705        tokenize_test!(
706            &[0x44, 0xce, 0xa3, 0x7e, 0xfc, 0xab, 0x06], // expected buffer
707            64,                                          // buffer size
708            "Lets go to the %08x",                       // printf style
709            "Lets go to the {:08x}",                     // core::fmt style
710            0xcafe as u32
711        );
712    }
713
714    #[test]
715    fn tokenizer_supports_concatenated_printf_format_strings() {
716        // Since the no argument and some arguments cases are handled differently
717        // by `tokenize_to_buffer!` we need to test both.
718        let mut buffer = [0u8; 64];
719        let len =
720            tokenize_printf_to_buffer!(&mut buffer, "Hello" PW_FMT_CONCAT " Pigweed").unwrap();
721        assert_eq!(&buffer[..len], &[0xe0, 0x92, 0xe0, 0xa]);
722
723        let len = tokenize_printf_to_buffer!(&mut buffer, "Hello: " PW_FMT_CONCAT "%cigweed",
724          "P".as_bytes()[0])
725        .unwrap();
726        assert_eq!(&buffer[..len], &[0x2e, 0x52, 0xac, 0xe4, 0xa0, 0x1]);
727    }
728
729    #[test]
730    fn tokenizer_supports_concatenated_core_fmt_format_strings() {
731        // Since the no argument and some arguments cases are handled differently
732        // by `tokenize_to_buffer!` we need to test both.
733        let mut buffer = [0u8; 64];
734        let len =
735            tokenize_core_fmt_to_buffer!(&mut buffer, "Hello" PW_FMT_CONCAT " Pigweed").unwrap();
736        assert_eq!(&buffer[..len], &[0xe0, 0x92, 0xe0, 0xa]);
737
738        let len = tokenize_core_fmt_to_buffer!(&mut buffer, "The answer is " PW_FMT_CONCAT "{}!",
739          1 as i32)
740        .unwrap();
741        assert_eq!(&buffer[..len], &[0x52, 0x1c, 0xb0, 0x4c, 0x2]);
742    }
743}