citnames - Deduce command semantic
citnames [options] --input <file> --output <file>
The name citnames comes from to reverse the word “semantic”.
Because when you type a command, you know your intent. The command execution is just a thing to achieve your goal. This program takes the command which was executed, and try to find out what the intent was to run that command. It deduces the semantic of the command.
This is useful to generate a compilation database. Citnames get a list of commands, and it creates a JSON compilation database. (This is currently the only output of the tool.)
- Print version number.
- Print help message.
- Enable verbose logging.
- --input file
- Specify input file. (Default file name provided.) The input is a command execution list, with some extra information. The syntax is detailed in a separate section.
- --output file
- Specify output file. (Default file name provided.) The output is currently a JSON compilation database.
- Use previously generated output file and append the new entries to it. This way you can run continuously during work, and it keeps the compilation database up to date. File deletion and addition are both considered. But build process change (compiler flags change) might cause duplicate entries.
- Allow the program to verify file location checks on the current machine it runs. (Default value provided. Run help to query it.) This is important if the execution list is not from the current host.
- --config file
- Specify a configuration file. The configuration file captures how the output should be formatted and which entries it shall contain.
Citnames exit status is non zero in case of IO problems, otherwise it’s zero. The exit status is independent from how many command it recognized or was it recognized at all.
Any insight about the command recognition logic can be observed with --verbose flag on the standard error. Otherwise the command is silent.
It’s a JSON file, with the command execution history. (Plus some metadata, that is useful for debugging the application which was produced it.) This file can be produced by the intercept command, which records the process executions of a build.
Read more about the syntax of the file in the intercept man page.
Currently the only output format is the JSON compilation database. Read more about the syntax of that in the bear man page.
The config file influences the command recognition (by the section “compilation”) and the output format (by the section “output”).
The config file is optional. The program will use default values, which can be dumped with the --verbose flags.
Some parts of the file has overlap with the command line arguments. If both present the command line argument overrides the config file values.
- where compiler can be specified, which are not yet recognized by default. The executable is an absolute path to the compiler. The additional_flags is an optional attribute, which contains flags which will append to the final output. (It’s a good candidate to use this for adding OpenMPI compiler wrapper flags from the mpicc --showme:compile output.)
- this is an optional list of executables (with absolute path) which needs to be removed from the output.
- The paths_to_include and paths_to_exclude are for filter out entries from these directories. (Directory names has to be absolute paths.) The include_only_existing_source allows or disables file check for the output.
- The command_as_array controls which command field is emitted in the output. True produces arguments, false produces command field. The drop_output_field will disable the output field from the output.
Copyright (C) 2012-2021 by László Nagy <https://github.com/rizsotto/Bear>
|September 14, 2020||Bear User Manuals|