Struct ToolOutputPath

pub struct ToolOutputPath {
    pub baseline_kind: BaselineKind,
    pub dir: PathBuf,
    pub kind: ToolOutputPathKind,
    pub modifiers: Vec<String>,
    pub name: String,
    pub tool: ValgrindTool,
}

Available on crate feature runner only.

The tool specific output path(s)

Fields

baseline_kind: BaselineKind

The BaselineKind

dir: PathBuf

The final directory of all the output files

kind: ToolOutputPathKind

The ToolOutputPathKind

modifiers: Vec<String>

The modifiers which are prepended to the extension

name: String

The name of this output path

tool: ValgrindTool

The tool

Implementations

impl ToolOutputPath

pub fn new( kind: ToolOutputPathKind, tool: ValgrindTool, baseline_kind: &BaselineKind, base_dir: &Path, module: &ModulePath, name: &str, ) -> Self

Create a new ToolOutputPath.

The base_dir is supposed to be the same as crate::runner::meta::Metadata::target_dir. The name is supposed to be the name of the benchmark function. If a benchmark id is present join both with a dot as separator to get the final name.

pub fn with_init( kind: ToolOutputPathKind, tool: ValgrindTool, baseline_kind: &BaselineKind, base_dir: &Path, module: &str, name: &str, ) -> Result<Self>

Initialize and create the output directory and organize files

This method moves the old output to $TOOL_ID.*.out.old

pub fn init(&self) -> Result<()>

Initialize the directory in which the files are stored

pub fn clear(&self) -> Result<()>

Remove the files of this output path

pub fn shift(&self) -> Result<()>

Remove the old or base files and rename the present files to “old” files

pub fn exists(&self) -> bool

Return true if a real file of this output path exists

pub fn is_multiple(&self) -> bool

Return true if there are multiple real files of this output path

pub fn to_base_path(&self) -> Self

Convert this output path to a base output path

pub fn to_tool_output(&self, tool: ValgrindTool) -> Self

Convert this tool output path to the output of another tool output path

A tool with no *.out file is log-file based. If the other tool is a out-file based tool the ToolOutputPathKind will be converted and vice-versa. The “old” (base) type (a tool output converted with ToolOutputPath::to_base_path) will be converted to a new ToolOutputPath.

pub fn to_log_output(&self) -> Self

Convert this tool output to the according log output

All tools have a log output even the ones which are out-file based.

pub fn to_xtree_output(&self) -> Option<Self>

If possible, convert this tool output to the according xtree output

Not all tools support xtree output files

pub fn to_xleak_output(&self) -> Option<Self>

If possible, convert this tool output to the according xleak output

Not all tools support xleak output files

pub fn log_path_of(&self, path: &Path) -> Option<PathBuf>

Return the path to the log file for the given path

path is supposed to be a path to a valid file in the directory of this ToolOutputPath.

pub fn dump_log<W>(&self, log_level: Level, writer: &mut W) -> Result<()>

where W: Write,

If the log::Level matches dump the content of all output files into the writer

pub fn extension(&self) -> String

This method can only be used to create the path passed to the tools

The modifiers are extrapolated by the tools and won’t match any real path name.

pub fn with_modifiers<I, T>(&self, modifiers: T) -> Self

where I: Into<String>, T: IntoIterator<Item = I>,

Create new ToolOutputPath with modifiers

pub fn to_path(&self) -> PathBuf

Return the unexpanded path usable as input for --callgrind-out-file, …

The path returned by this method does not necessarily have to exist and can include modifiers like %p. Use Self::real_paths to get the real and existing (possibly multiple) paths to the output files of the respective tool.

pub fn walk_dir(&self) -> Result<impl Iterator<Item = DirEntry>>

Walk the benchmark directory (non-recursive)

pub fn strip_prefix<'a>(&self, file_name: &'a str) -> Option<&'a str>

Strip the <tool>.<name> prefix from a file_name

pub fn prefix(&self) -> String

Return the file name prefix as in <tool>.<name>

pub fn real_paths(&self) -> Result<Vec<PathBuf>>

Return the real paths of a tool’s output files

A tool can have many output files so Self::to_path is not enough

pub fn real_paths_with_modifier(&self) -> Result<Vec<(PathBuf, Option<String>)>>

Return the real paths with their respective modifiers if present

pub fn sanitize_callgrind(&self) -> Result<()>

Sanitize callgrind output file names

This method will remove empty files which are occasionally produced by callgrind and only cause problems in the parser. The files are renamed from the callgrind file naming scheme to ours which is easier to handle.

The information about pids, parts and threads is obtained by parsing the header from the callgrind output files instead of relying on the sometimes flaky file names produced by callgrind. The header is around 10-20 lines, so this method should be still sufficiently fast. Additionally, callgrind might change the naming scheme of its files, so using the headers makes us more independent of a specific valgrind/callgrind version.

pub fn sanitize_bbv(&self) -> Result<()>

Sanitize bbv file names

The original output files of bb have a .<number> suffix if there are multiple threads. We need the threads as t<number> in the modifier part of the final file names.

For example: (orig -> sanitized)

If there are multiple threads, the bb output file name doesn’t include the first thread:

exp-bbv.bench_thread_in_subprocess.548365.bb.out -> exp-bbv.bench_thread_in_subprocess.548365.t1.bb.out

exp-bbv.bench_thread_in_subprocess.548365.bb.out.2 -> exp-bbv.bench_thread_in_subprocess.548365.t2.bb.out

pub fn sanitize_generic(&self) -> Result<()>

Sanitize file names of all tools if not sanitized by a more specific method

The pids are removed from the file name if there was only a single process (pid). Additionally, we check for empty files and remove them.

pub fn sanitize(&self) -> Result<()>

Sanitize file names for a specific tool

Empty files are cleaned up. For more details on a specific tool see the respective sanitize_ method.

Trait Implementations

impl Clone for ToolOutputPath

fn clone(&self) -> ToolOutputPath

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ToolOutputPath

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Display for ToolOutputPath

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl PartialEq for ToolOutputPath

fn eq(&self, other: &ToolOutputPath) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Eq for ToolOutputPath

impl StructuralPartialEq for ToolOutputPath