Get Started & Guides#

pw_string: Efficient, easy, and safe string manipulation

Get Started#

Add @pigweed//pw_string to the deps list in your Bazel target:

cc_library("...") {
  # ...
  deps = [
    # ...
    "@pigweed//pw_string",
    # ...
  ]
}

If only one part of the module is needed, depend only on it; for example @pigweed//pw_string:format.

This assumes @pigweed is the name you pulled Pigweed into your Bazel WORKSPACE as.

Add $dir_pw_string to the deps list in your pw_executable() build target:

pw_executable("...") {
  # ...
  deps = [
    # ...
    "$dir_pw_string",
    # ...
  ]
}

See //source/BUILD.gn in the Pigweed Sample Project for an example.

Add pw_string to your pw_add_library or similar CMake target:

pw_add_library(my_library STATIC
  HEADERS
    ...
  PRIVATE_DEPS
    # ...
    pw_string
    # ...
)

For a narrower dependency, depend on subtargets like pw_string.builder, etc.

There are two ways to use pw_string from a Zephyr project:

  1. Depend on pw_string in your CMake target (see CMake tab). This is Pigweed Team’s suggested approach since it enables precise CMake dependency analysis.

  2. Add CONFIG_PIGWEED_STRING=y to the Zephyr project’s configuration, which causes pw_string to become a global dependency and have the includes exposed to all targets. Pigweed team does not recommend this approach, though it is the typical Zephyr solution.

Choose between pw::InlineString and pw::StringBuilder#

pw::InlineString is intended to replace typical null terminated character arrays in embedded data structures. Use pw::InlineString if you need:

  • Compatibility with std::string

  • Storage internal to the object

  • A string object to persist in other data structures

  • Lower code size overhead

pw::StringBuilder is intended to ease constructing strings in external data; typically created on the stack and disposed of in the same function. Use pw::StringBuilder if you need:

  • Compatibility with std::ostringstream, including custom object support

  • Storage external to the object

  • Non-fatal handling of failed append/format operations

  • Tracking of the status of a series of operations

  • A temporary stack object to aid string construction

  • Medium code size overhead

An example of when to prefer pw::InlineString is wrapping a length-delimited string (e.g. std::string_view) for APIs that require null termination:

#include <string>
#include "pw_log/log.h"
#include "pw_string/string_builder.h"

void ProcessName(std::string_view name) {
  // %s format strings require null terminated strings, so create one on the
  // stack with size up to kMaxNameLen, copy the string view `name` contents
  // into it, add a null terminator, and log it.
  PW_LOG_DEBUG("The name is %s",
               pw::InlineString<kMaxNameLen>(name).c_str());
}

An example of when to prefer pw::StringBuilder is when constructing a string for external use.

#include "pw_string/string_builder.h"

pw::Status FlushSensorValueToUart(int32_t sensor_value) {
  pw::StringBuffer<42> sb;
  sb << "Sensor value: ";
  sb << sensor_value;  // Formats as int.
  FlushCStringToUart(sb.c_str());

  if (!sb.status().ok) {
    format_error_metric.Increment();  // Track overflows.
  }
  return sb.status();
}

Build a string with pw::StringBuilder#

The following shows basic use of a pw::StringBuilder.

#include "pw_log/log.h"
#include "pw_string/string_builder.h"

pw::Status LogProducedData(std::string_view func_name,
                           span<const std::byte> data) {
  // pw::StringBuffer allocates a pw::StringBuilder with a built-in buffer.
  pw::StringBuffer<42> sb;

  // Append a std::string_view to the buffer.
  sb << func_name;

  // Append a format string to the buffer.
  sb.Format(" produced %d bytes of data: ", static_cast<int>(data.data()));

  // Append bytes as hex to the buffer.
  sb << data;

  // Log the final string.
  PW_LOG_DEBUG("%s", sb.c_str());

  // Errors encountered while mutating the string builder are tracked.
  return sb.status();
}

Build a string with pw::InlineString#

pw::InlineString objects must be constructed by specifying a fixed capacity for the string.

#include "pw_string/string.h"

// Initialize from a C string.
pw::InlineString<32> inline_string = "Literally";
inline_string.append('?', 3);   // contains "Literally???"

// Supports copying into known-capacity strings.
pw::InlineString<64> other = inline_string;

// Supports various helpful std::string functions
if (inline_string.starts_with("Lit") || inline_string == "not\0literally"sv) {
  other += inline_string;
}

// Like std::string, InlineString is always null terminated when accessed
// through c_str(). InlineString can be used to null-terminate
// length-delimited strings for APIs that expect null-terminated strings.
std::string_view file(".gif");
if (std::fopen(pw::InlineString<kMaxNameLen>(file).c_str(), "r") == nullptr) {
  return;
}

// pw::InlineString integrates well with std::string_view. It supports
// implicit conversions to and from std::string_view.
inline_string = std::string_view("not\0literally", 12);

FunctionThatTakesAStringView(inline_string);

FunctionThatTakesAnInlineString(std::string_view("1234", 4));

Build a string inside an pw::InlineString with a pw::StringBuilder#

pw::StringBuilder can build a string in a pw::InlineString:

#include "pw_string/string.h"

void DoFoo() {
  InlineString<32> inline_str;
  StringBuilder sb(inline_str);
  sb << 123 << "456";
  // inline_str contains "456"
}

Pass an pw::InlineString object as a parameter#

pw::InlineString objects can be passed to non-templated functions via type erasure. This saves code size in most cases, since it avoids template expansions triggered by string size differences.

Unknown size strings#

To operate on pw::InlineString objects without knowing their type, use the pw::InlineString<> type, shown in the examples below:

// Note that the first argument is a generically-sized InlineString.
void RemoveSuffix(pw::InlineString<>& string, std::string_view suffix) {
  if (string.ends_with(suffix)) {
     string.resize(string.size() - suffix.size());
  }
}

void DoStuff() {
  pw::InlineString<32> str1 = "Good morning!";
  RemoveSuffix(str1, " morning!");

  pw::InlineString<40> str2 = "Good";
  RemoveSuffix(str2, " morning!");

  PW_ASSERT(str1 == str2);
}

However, generically sized pw::InlineString objects don’t work in constexpr contexts.

Known size strings#

pw::InlineString operations on known-size strings may be used in constexpr expressions.

static constexpr pw::InlineString<64> kMyString = [] {
  pw::InlineString<64> string;

  for (int i = 0; i < 10; ++i) {
    string += "Hello";
  }

  return string;
}();

Initialization of pw::InlineString objects#

pw::InlineBasicString supports class template argument deduction (CTAD) in C++17 and newer. Since pw::InlineString is an alias, CTAD is not supported until C++20.

// Deduces a capacity of 5 characters to match the 5-character string literal
// (not counting the null terminator).
pw::InlineBasicString inline_string = "12345";

// In C++20, CTAD may be used with the pw::InlineString alias.
pw::InlineString my_other_string("123456789");

Custom types with pw::StringBuilder#

As with std::ostream, pw::StringBuilder supports printing custom types by overriding the << operator. This is is done by defining operator<< in the same namespace as the custom type. For example:

namespace my_project {

struct MyType {
  int foo;
  const char* bar;
};

pw::StringBuilder& operator<<(pw::StringBuilder& sb, const MyType& value) {
  return sb << "MyType(" << value.foo << ", " << value.bar << ')';
}

}  // namespace my_project

Internally, StringBuilder uses the ToString function to print. The ToString template function can be specialized to support custom types with StringBuilder, though it is recommended to overload operator<< instead. This example shows how to specialize pw::ToString:

#include "pw_string/to_string.h"

namespace pw {

template <>
StatusWithSize ToString<MyStatus>(MyStatus value, span<char> buffer) {
  return Copy(MyStatusString(value), buffer);
}

}  // namespace pw