| ctags(1) | Generate tag files for source code |
| ctags(1P, 1p) | create a tags file (DEVELOPMENT, FORTRAN) |
| etags, ctags(1) | generate tag file for Emacs, vi |
| ctags(1) | Universal Ctags | ctags(1) |
ctags - Generate tag files for source code
ctags [<options>] [<source_file(s)>] etags [<options>] [<source_file(s)>]
The ctags and etags (see -e option) programs (hereinafter collectively referred to as ctags, except where distinguished) generate an index (or "tag") file for a variety of language objects found in source file(s). This tag file allows these items to be quickly and easily located by a text editor or other utilities (client tools). A tag signifies a language object for which an index entry is available (or, alternatively, the index entry created for that object).
Alternatively, ctags can generate a cross reference file which lists, in human readable form, information about the various language objects found in a set of source files.
Tag index files are supported by numerous editors, which allow the user to locate the object associated with a name appearing in a source file and jump to the file and line which defines the name. See the manual of your favorite editor about utilizing ctags command and the tag index files in the editor.
ctags is capable of generating different kinds of tags for each of many different languages. For a complete list of supported languages, the names by which they are recognized, and the kinds of tags which are generated for each, see the --list-languages and --list-kinds-full options.
This man page describes Universal Ctags, an implementation of ctags derived from Exuberant Ctags. The major incompatible changes between Universal Ctags and Exuberant Ctags are enumerated in ctags-incompatibilities(7).
One of the advantages of Exuberant Ctags is that it allows a user to define a new parser from the command line. Extending this capability is one of the major features of Universal Ctags. ctags-optlib(7) describes how the capability is extended.
Newly introduced experimental features are not explained here. If you are interested in such features and ctags internals, visit https://docs.ctags.io/.
Despite the wealth of available options, defaults are set so that ctags is most commonly executed without any options (e.g. "ctags *", or "ctags -R"), which will create a tag file in the current directory for all recognized source files. The options described below are provided merely to allow custom tailoring to meet special needs.
Note that spaces separating the single-letter options from their parameters are optional.
Note also that the boolean parameters to the long form options (those beginning with -- and that take a [=(yes|no)] parameter) may be omitted, in which case =yes is implied. (e.g. --sort is equivalent to --sort=yes). Note further that =1, =on, and =true are considered synonyms for =yes, and that =0, =off, and =false are considered synonyms for =no.
Some options are either ignored or useful only when used while running in etags mode (see -e option). Such options will be noted.
<options> must precede the <source_file(s)> following the standard POSIX convention.
Options taking language names will accept those names in either upper or lower case. See the --list-languages option for a complete list of the built-in language names.
Some options take one-letter flags as parameters (e.g. --kinds-<LANG> option). Specifying just letters help a user create a complicated command line quickly. However, a command line including sequences of one-letter flags becomes difficult to understand.
Universal Ctags accepts long-name flags in addition to such one-letter flags. The long-name and one-letter flags can be mixed in an option parameter by surrounding each long-name by braces. Thus, for an example, the following three notations for --kinds-C option have the same meaning:
--kinds-C=+pLl
--kinds-C=+{prototype}{label}{local}
--kinds-C=+{prototype}L{local}
Note that braces may be meta characters in your shell. Put single quotes in such case.
--list-... options shows one-letter flags and associated long-name flags.
Universal Ctags introduces many --list-... options that provide the internal data of Universal Ctags (See "Listing Options"). Both users and client tools may use the data. --with-list-header and --machinable options adjust the output of the most of --list-... options.
The default setting (--with-list-header=yes and --machinable=no) is for using interactively from a terminal. The header that explains the meaning of columns is simply added to the output, and each column is aligned in all lines. The header line starts with a hash ('#') character.
For scripting in a client tool, --with-list-header=no and --machinable=yes may be useful. The header is not added to the output, and each column is separated by tab characters.
Note the order of columns will change in the future release. However, labels in the header will not change. So by scanning the header, a client tool can find the index for the target column.
ctags has more options than listed here. Options starting with an underscore character, such as --_echo=<msg>, are not listed here. They are experimental or for debugging purpose.
Notation: <foo> is for a variable string foo, [ ... ] for optional, | for selection, and ( ... ) for grouping. For example --foo[=(yes|no)]'' means ``--foo, -foo=yes, or -foo=no.
If appropriate support is available from the runtime library of your C compiler, then pattern may contain the usual shell wildcards (not regular expressions) common on Unix (be sure to quote the option parameter to protect the wildcards from being expanded by the shell before being passed to ctags; also be aware that wildcards can match the slash character, '/'). You can determine if shell wildcards are available on your platform by examining the output of the --list-features option, which will include wildcards in the compiled feature list; otherwise, pattern is matched against file names using a simple textual comparison.
If <pattern> begins with the character '@', then the rest of the string is interpreted as a file name from which to read exclusion patterns, one per line. If pattern is empty, the list of excluded patterns is cleared.
Note that at program startup, the default exclude list contains names of common hidden and system files, patterns for binary files, and directories for which it is generally not desirable to descend while processing the --recurse option. To see the list of built-in exclude patterns, use --list-excludes.
See also the description for --exclude-exception= option.
For an example, you want ctags to ignore all files under foo directory except foo/main.c, use the following command line: --exclude=foo/* --exclude-exception=foo/main.c.
Note that if the file name read is a directory and --recurse is enabled, this string will be printed only once at the end of all tags found for by descending the directory. This string will always be separated from the last tag line for the file by its terminating newline.
This option is quite esoteric and is empty by default.
If the list of supplied files is empty and no file list is specified with the -L option, then the current directory (i.e. '.') is assumed. Symbolic links are followed by default (See --links option). If you don't like these behaviors, either explicitly specify the files or pipe the output of find(1) into "ctags -L -" instead. See, also, the --exclude and --maxdepth to limit recursion.
Note: This option is not supported on all platforms at present. It is available if the output of the --help option includes this option.
If file is specified as '-', then file names are read from standard input. File names read using this option are processed following file names appearing on the command line. Options are also accepted in this input. If this option is specified more than once, only the last will apply.
Note: file is read in line-oriented mode, where a new line is the only delimiter and non-trailing white space is considered significant, in order that file names containing spaces may be supplied (however, trailing white space is stripped from lines); this can affect how options are parsed if included in the input.
ctags will stubbornly refuse to take orders if tagfile exists and its first line contains something other than a valid tags line. This will save your neck if you mistakenly type "ctags -f *.c", which would otherwise overwrite your first C file with the tags generated by the rest! It will also refuse to accept a multi-character file name which begins with a '-' (dash) character, since this most likely means that you left out the tag file name and this option tried to grab the next option as the file name. If you really want to name your output tag file -ugly, specify it as "-f ./-ugly".
This option must appear before the first file name. If this option is specified more than once, only the last will apply.
Example applications for this feature are generating a listing of all functions located in a source file (e.g. "ctags -x --kinds-c=f file"), or generating a list of all externally visible global variables located in a source file (e.g. "ctags -x --kinds-c=v --extras=-F file").
In addition <encoding> is specified at the top the tags file as the value for the TAG_FILE_ENCODING pseudo-tag. The default value of <encoding> is UTF-8.
In addition, the special value auto indicates that the language should be automatically selected (which effectively disables this option).
If the first language of <list> is not preceded by either a '+' or '-', the current list (the current settings of enabled/disabled languages managed in ctags internally) will be cleared before adding or removing the languages in <list>. Until a '-' is encountered, each language in the <list> will be added to the current list.
As either the '+' or '-' is encountered in the <list>, the languages following it are added or removed from the current list, respectively. Thus, it becomes simple to replace the current list with a new one, or to add or remove languages from the current list.
The actual list of files for which tags will be generated depends upon the language extension mapping in effect (see the --langmap option). Note that the most of languages, including user-defined languages, are enabled unless explicitly disabled using this option. Language names included in list may be any built-in language or one previously defined with --langdef.
The default is all, which is also accepted as a valid argument. See the --list-languages option for a list of the all (built-in and user-defined) language names.
Note --languages= option works cumulative way; the option can be specified with different arguments multiple times in a command line.
The parameter <pattern> is not a list. Use this option multiple times in a command line to add or remove multiple alias patterns.
To restore the default language aliases, specify default.
Using all for <LANG> has meaning in following two cases:
If appropriate support is available from the runtime library of your C compiler, then the file name pattern may contain the usual shell wildcards common on Unix (be sure to quote the option parameter to protect the wildcards from being expanded by the shell before being passed to ctags). You can determine if shell wildcards are available on your platform by examining the output of the --list-features option, which will include wildcards in the compiled feature list; otherwise, the file name patterns are matched against file names using a simple textual comparison.
When mapping a file extension with --langmap option, it will first be unmapped from any other languages. (--map-<LANG> option provides more fine-grained control.)
If the first character in a <map> is a plus sign ('+'), then the extensions and file name patterns in that map will be appended to the current map for that language; otherwise, the map will replace the current map. For example, to specify that only files with extensions of .c and .x are to be treated as C language files, use --langmap=c:.c.x; to also add files with extensions of .j as Java language files, specify --langmap=c:.c.x,java:+.j. To map makefiles (e.g. files named either Makefile, makefile, or having the extension .mak) to a language called make, specify --langmap=make:([Mm]akefile).mak. To map files having no extension, specify a period not followed by a non-period character (e.g. '.', ..x, .x.).
To clear the mapping for a particular language (thus inhibiting automatic generation of tags for that language), specify an empty extension list (e.g. --langmap=fortran:). To restore the default language mappings for a particular language, supply the keyword default for the mapping. To specify restore the default language mappings for all languages, specify --langmap=default.
Note that file name patterns are tested before file extensions when inferring the language of a file. This order of Universal Ctags is different from Exuberant Ctags. See ctags-incompatibilities(7) for the background of this incompatible change.
In ctags, more than one language can map to a file name <pattern> or file <extension> (N:1 map). Alternatively, --langmap option handle only 1:1 map, only one language mapping to one file name <pattern> or file <extension>. A typical N:1 map is seen in C++ and ObjectiveC language; both languages have a map to .h as a file extension.
A file extension is specified by preceding the extension with a period (e.g. .c). A file name pattern is specified by enclosing the pattern in parentheses (e.g. ([Mm]akefile)). A prefixed plus ('+') sign is for adding, and minus ('-') is for removing. No prefix means replacing the map of <LANG>.
Unlike --langmap, <extension> (or <pattern>) is not a list. --map-<LANG> takes one extension (or pattern). However, the option can be specified with different arguments multiple times in a command line.
See "TAG ENTRIES" about fields, kinds, roles, and extras.
The valid values for type (either the entire word or the first letter is accepted) are:
However, this option has one significant drawback: changes to the source files can cause the line numbers recorded in the tag file to no longer correspond to the lines in the source file, causing jumps to some tags to miss the target definition by one or more lines. Basically, this option is best used when the source code to which it is applied is not subject to change. Selecting this option type causes the following options to be ignored: -B, -F.
number type is ignored in Xref and JSON output formats. Use --_xformat="...%n" for Xref output format, or --fields=+n-P for JSON output format.
This was the default format generated by the original ctags and is, therefore, retained as the default for this option.
The parameter <flags> is a set of one-letter flags (and/or long-name flags), each representing one kind of extra tag entry to include in the tag file. If flags is preceded by either the '+' or '-' character, the effect of each flag is added to, or removed from, those currently enabled; otherwise the flags replace any current settings. All entries are included if '*' is given.
This --extras= option is for controlling extras common in all languages (or language-independent extras). Universal Ctags also supports language-specific extras. (See "Language-specific fields and extras" about the concept). Use --extras-<LANG>= option for controlling them.
Specifies all as <LANG> to apply the parameter <flags> to all languages; all extras are enabled with specifying '*' as the parameter flags. If specifying nothing as the parameter flags (--extras-all=), all extras are disabled. These two combinations are useful for testing.
Check the output of the --list-extras=<LANG> option for the extras of specific language <LANG>.
The parameter <flags> is a set of one-letter or long-name flags, each representing one type of extension field to include. Each flag or group of flags may be preceded by either '+' to add it to the default set, or '-' to exclude it. In the absence of any preceding '+' or '-' sign, only those fields explicitly listed in flags will be included in the output (i.e. overriding the default set). All fields are included if '*' is given.
This option is ignored if the option --format=1 (legacy tag file format) has been specified.
Use --fields-<LANG>= option for controlling language-specific fields.
Specify all as <LANG> to apply the parameter <flags> to all languages; all fields are enabled with specifying '*' as the parameter flags. If specifying nothing as the parameter <flags> (i.e. --fields-all=), all fields are disabled. These two combinations are useful for testing.
See the description of --fields=[+|-][<flags>|*] about <flags>.
Use --fields= option for controlling language-independent fields.
The parameter <kinds> is a group of one-letter or long-name flags designating kinds of tags (particular to the language) to either include or exclude from the output. The specific sets of flags recognized for each language, their meanings and defaults may be list using the --list-kinds-full option.
Each letter or group of letters may be preceded by either '+' to add it to, or '-' to remove it from, the default set. In the absence of any preceding '+' or '-' sign, only those kinds explicitly listed in kinds will be included in the output (i.e. overriding the default for the specified language).
Specify '*' as the parameter to include all kinds implemented in <LANG> in the output. Furthermore if all is given as <LANG>, specification of the parameter kinds affects all languages defined in ctags. Giving all makes sense only when '*' or 'F' is given as the parameter kinds.
As an example for the C language, in order to add prototypes and external variable declarations to the default set of tag kinds, but exclude macros, use --kinds-c=+px-d; to include only tags for functions, use --kinds-c=f.
Some kinds of C and C++ languages are synchronized; enabling (or disabling) a kind in one language enables the kind having the same one-letter and long-name in the other language. See also the description of MASTER column of --list-kinds-full.
An input source file with long lines and multiple tag matches per line can generate an excessively large tags file with an unconstrained pattern length. For example, running ctags on a minified JavaScript source file often exhibits this behavior.
The truncation avoids cutting in the middle of a UTF-8 code point spanning multiple bytes to prevent writing invalid byte sequences from valid input files. This handling allows for an extra 3 bytes above the configured limit in the worse case of a 4 byte code point starting right before the limit. Please also note that this handling is fairly naive and fast, and although it is resistant against any input, it requires a valid input to work properly; it is not guaranteed to work as the user expects when dealing with partially invalid UTF-8 input. This also partially affect non-UTF-8 input, if the byte sequence at the truncation length looks like a multibyte UTF-8 sequence. This should however be rare, and in the worse case will lead to including up to an extra 3 bytes above the limit.
Some fields are newly introduced in Universal Ctags and more will be introduced in the future. Other tags generators may also introduce their specific fields.
In such a situation, there is a concern about conflicting field names; mixing tags files generated by multiple tags generators including Universal Ctags is difficult. This option provides a workaround for such station.
$ ctags --fields='{line}{end}' -o - hello.c
main hello.c /^main(int argc, char **argv)$/;" f line:3 end:6
$ ctags --put-field-prefix --fields='{line}{end}' -o - hello.c
main hello.c /^main(int argc, char **argv)$/;" f line:3 UCTAGSend:6
In the above example, the prefix is put to end field which is newly introduced in Universal Ctags.
Like --kinds-<LANG> option, '+' is for adding the role to the list, and '-' is for removing from the list. '*' is for including all roles of the kind to the list. The option with no argument makes the list empty.
Both a one-letter flag or a long name flag surrounded by braces are acceptable for specifying a kind (e.g. --roles-C.h=+{system}{local} or --roles-C.{header}=+{system}{local}). '*' can be used for <KIND> only for adding/removing all roles of all kinds in a language to/from the list (e.g. --roles-C.*=* or --roles-C.*=).
all can be used for <LANG> only for adding/removing all roles of all kinds in all languages to/from the list (e.g. --roles-all.*=* or --roles-all.*=).
This option is available on MS Windows only.
ctags searches <pathname> in the optlib path list first. If ctags cannot find a file or directory in the list, ctags reads a file or directory at the specified <pathname>.
If a file is specified, it should contain one option per line. If a directory is specified, files suffixed with .ctags under it are read in alphabetical order.
As a special case, if --options=NONE is specified as the first option on the command line, preloading is disabled; the option will disable the automatic reading of any configuration options from a file (see "FILES").
See ctags-optlib(7) for details of each option.
Note that this indicates a preference only and does not guarantee skipping code within an "#if 0" branch, since the fall-back algorithm used to generate tags when preprocessor conditionals are too complex follows all branches of a conditional.
When enabled, this option will cause ctags to generate tag entries marked with the file names and line numbers of their locations original source file(s), instead of their actual locations in the preprocessor output. The actual file names placed into the tag file will have the same leading path components as the preprocessor output file, since it is assumed that the original source files are located relative to the preprocessor output file (unless, of course, the #line directive specifies an absolute path).
Note: This option is generally only useful when used together with the --excmd=number (-n) option. Also, you may have to use either the --langmap or --language-force option if the extension of the preprocessor output file is not known to ctags.
This option only affects how the scoping of particular kinds of tags are interpreted (i.e. whether or not they are considered as globally visible or visible only within the file in which they are defined); it does not map the extension to any particular language. Any tag which is located in a non-include file and cannot be seen (e.g. linked to) from another file is considered to have file-limited (e.g. static) scope. No kind of tag appearing in an include file will be considered to have file-limited scope.
If the first character in the list is '+', then the extensions in the list will be appended to the current list; otherwise, the list will replace the current list. See, also, the fileScope/F flag of --extras option.
The default list is .h.H.hh.hpp.hxx.h++.inc.def. To restore the default list, specify "-h default".
Note that if an extension supplied to this option is not already mapped to a particular language (see "Determining file language", above), you will also need to use either the --map-<LANG>, --langmap or --language-force option.
If an identifier is suffixed with a '+' character (i.e. "-I FOO+"), ctags will also ignore any parenthesis-enclosed argument list which may immediately follow the identifier in the source files. See the example of "-I MODULE_VERSION+" below.
If two identifiers are separated with the '=' character (i.e. -I FOO=BAR), the first identifiers is replaced by the second identifiers for parsing purposes. The list of identifiers may be supplied directly on the command line or read in from a separate file. See the example of "-I CLASS=class" below.
If the first character of <identifier-list> is '@', '.' or a pathname separator ('/' or '\'), or the first two characters specify a drive letter (e.g. C:), the parameter <identifier-list> will be interpreted as a filename from which to read a list of identifiers, one per input line.
Otherwise, <identifier-list> is a list of identifiers (or identifier pairs) to be specially handled, each delimited by either a comma or by white space (in which case the list should be quoted to keep the entire list as one command line argument).
Multiple -I options may be supplied. To clear the list of ignore identifiers, supply a single dash ('-') for <identifier-list>.
This feature is useful when preprocessor macros are used in such a way that they cause syntactic confusion due to their presence. Indeed, this is the best way of working around a number of problems caused by the presence of syntax-busting macros in source files (see "CAVEATS"). Some examples will illustrate this point.
int foo ARGDECL4(void *, ptr, long int, nbytes)
In the above example, the macro ARGDECL4 would be mistakenly interpreted to be the name of the function instead of the correct name of foo. Specifying "-I ARGDECL4" results in the correct behavior.
/* creates an RCS version string in module */
MODULE_VERSION("$Revision$")
In the above example the macro invocation looks too much like a function definition because it is not followed by a semicolon (indeed, it could even be followed by a global variable definition that would look much like a K&R style function parameter declaration). In fact, this seeming function definition could possibly even cause the rest of the file to be skipped over while trying to complete the definition. Specifying "-I MODULE_VERSION+" would avoid such a problem.
CLASS Example {
// your content here
};
The example above uses CLASS as a preprocessor macro which expands to something different for each platform. For instance CLASS may be defined as class __declspec(dllexport) on Win32 platforms and simply class on UNIX. Normally, the absence of the C++ keyword class would cause the source file to be incorrectly parsed. Correct behavior can be restored by specifying "-I CLASS=class".
Available parameters can be listed with --list-params.
An extra can be enabled or disabled with --extras= for common extras in all languages, or --extras-<LANG>= for the specified language. These option takes one-letter flag or long-name flag as a parameter for specifying an extra.
The meaning of columns in output are as follows:
The meaning of columns are as follows:
Some fields are printed always in tags output. They have yes as the value for this column.
Unlike the tag output mode, JSON output mode allows disabling any fields.
This option prints only LETTER, DESCRIPTION, and ENABLED fields of --list-kinds-full output. However, the presentation of ENABLED column is different from that of --list-kinds-full option; [off] follows after description if the kind is disabled, and nothing follows if enabled. The most of all kinds are enabled by default.
The critical weakness of this option is that this option does not print the name of kind. Universal Ctags introduces --list-kinds-full because it considers that names are important.
This option does not work with --machinable nor --with-list-header.
Each kind of tag recorded in the tag file is represented by a one-letter flag, or a long-name flag. They are also used to filter the tags placed into the output through use of the --kinds-<LANG> option.
The meaning of columns are as follows:
However, Exuberant Ctags does not separate the kinds of C and C++. Enabling/disabling kindX in C language enables/disables a kind in C++ language having the same long-name flag with kindX. To emulate this behavior in Universal Ctags, a concept named master parser is introduced. Enabling/disabling some kinds are synchronized under the control of a master language.
$ ctags --kinds-C=+'{local}' --list-kinds-full \
| grep -E '^(#|C\+\+ .* local)'
#LANGUAGE LETTER NAME ENABLED REFONLY NROLES MASTER DESCRIPTION
C++ l local yes no 0 C local variables
$ ctags --kinds-C=-'{local}' --list-kinds-full \
| grep -E '^(#|C\+\+ .* local)'
#LANGUAGE LETTER NAME ENABLED REFONLY NROLES MASTER DESCRIPTION
C++ l local no no 0 C local variables
You see ENABLED field of local kind of C++ language is changed Though local kind of C language is enabled/disabled. If you swap the languages, you see the same result.
Each language listed is disabled if followed by [disabled]. To use the parser for such a language, specify the language as an argument of --languages=+ option.
--machinable and --with-list-header options are ignored if they are specified with this option.
To list the file extensions or file name patterns individually, use --list-map-extensions or --list-map-patterns option. See the --langmap option, and "Determining file language", above.
This option does not work with --machinable nor --with-list-header.
If the parameter <kindspecs> is given after the parameter <language> or all with concatenating with '.', list only roles defined in the kinds. Both one-letter flags and long name flags surrounded by braces are acceptable as the parameter <kindspecs>.
The meaning of columns are as follows:
The extra value prints parser specific statistics for parsers gathering such information.
These options are kept for backward-compatibility with Exuberant Ctags.
As ctags considers each source file name in turn, it tries to determine the language of the file by applying tests described in "Determining file language".
If a language was identified, the file is opened and then the appropriate language parser is called to operate on the currently open file. The parser parses through the file and adds an entry to the tag file for each language object it is written to handle. See "TAG FILE FORMAT", below, for details on these entries.
This implementation of ctags imposes no formatting requirements on C code as do legacy implementations. Older implementations of ctags tended to rely upon certain formatting assumptions in order to help it resolve coding dilemmas caused by preprocessor conditionals.
In general, ctags tries to be smart about conditional preprocessor directives. If a preprocessor conditional is encountered within a statement which defines a tag, ctags follows only the first branch of that conditional (except in the special case of #if 0, in which case it follows only the last branch). The reason for this is that failing to pursue only one branch can result in ambiguous syntax, as in the following example:
#ifdef TWO_ALTERNATIVES
struct {
#else
union {
#endif
short a;
long b;
}
Both branches cannot be followed, or braces become unbalanced and ctags would be unable to make sense of the syntax.
If the application of this heuristic fails to properly parse a file, generally due to complicated and inconsistent pairing within the conditionals, ctags will retry the file using a different heuristic which does not selectively follow conditional preprocessor branches, but instead falls back to relying upon a closing brace ('}') in column 1 as indicating the end of a block once any brace imbalance results from following a #if conditional branch.
ctags will also try to specially handle arguments lists enclosed in double sets of parentheses in order to accept the following conditional construct:
extern void foo __ARGS((int one, char two));
Any name immediately preceding the '((' will be automatically ignored and the previous name will be used.
C++ operator definitions are specially handled. In order for consistency with all types of operators (overloaded and conversion), the operator name in the tag file will always be preceded by the string "operator " (i.e. even if the actual operator definition was written as "operator<<").
After creating or appending to the tag file, it is sorted by the tag name, removing identical tag lines.
Unless the --language-force option is specified, the language of each source file is automatically selected based upon a mapping of file names to languages. The mappings in effect for each language may be displayed using the --list-maps option and may be changed using the --langmap or --map-<LANG> options.
If the name of a file is not mapped to a language, ctags tries to heuristically guess the language for the file by inspecting its content.
All files that have no file name mapping and no guessed parser are ignored. This permits running ctags on all files in either a single directory (e.g. "ctags *"), or on all files in an entire source directory tree (e.g. "ctags -R"), since only those files whose names are mapped to languages will be scanned.
An extension may be mapped to multiple parsers. For example, .h are mapped to C++, C and ObjectiveC. These mappings can cause issues. ctags tries to select the proper parser for the source file by applying heuristics to its content, however it is not perfect. In case of issues one can use --language-force=<language>, --langmap=<map>[,<map>[...]], or the --map-<LANG>=[+|-]<extension>|<pattern> options. (Some of the heuristics are applied whether --guess-language-eagerly is given or not.)
If ctags cannot select a parser from the mapping of file names, various heuristic tests are conducted to determine the language:
If ctags finds no such parser, ctags looks for the name in alias lists. For example, consider if the first line is #!/bin/sh. Though ctags has a "shell" parser, it doesn't have a "sh" parser. However, sh is listed as an alias for shell, therefore ctags selects the "shell" parser for the file.
An exception is env. If env is specified (for example "#!/usr/bin/env python"), ctags reads more lines to find real interpreter specification.
To display the list of aliases, use --list-aliases option. To add an item to the list or to remove an item from the list, use the --alias-<LANG>=+<pattern> or --alias-<LANG>=-<pattern> option respectively.
ctags treats MODE as a name of interpreter and applies the same rule of "interpreter" testing if the first line has one of the following patterns:
-*- mode: MODE -*-
or
-*- MODE -*-
ctags treats MODE as a name of an interpreter and applies the same rule of "interpreter" heuristic testing, if the lines at the tail of the file have the following pattern:
Local Variables: ... mode: MODE ... End:
3000 characters are sought from the end of file to find the pattern.
filetype=TYPE
or
ft=TYPE
Looking into the file contents is a more expensive operation than file name matching. So ctags runs the testings in limited conditions. "interpreter" testing is enabled only when a file is an executable or the --guess-language-eagerly (-G in short) option is given. The other heuristic tests are enabled only when -G option is given.
The --print-language option can be used just to print the results of parser selections for given files instead of generating a tags file.
Examples:
$ ctags --print-language config.h.in input.m input.unknown config.h.in: C++ input.m: MatLab input.unknown: NONE
NONE means that ctags does not select any parser for the file.
This section describes the tag file format briefly. See tags(5) and ctags-client-tools(7) for more details.
When not running in etags mode, each entry in the tag file consists of a separate line, each looking like this, called regular tags, in the most general case:
<tag_name><TAB><file_name><TAB><ex_cmd>;"<TAB><extension_fields>
The fields and separators of these lines are specified as follows:
Tag file format 2 (see --format) extends the EX command to include the extension fields embedded in an EX comment immediately appended to the EX command, which leaves it backward-compatible with original vi(1) implementations.
A few special tags, called pseudo tags, are written into the tag file for internal purposes.
!_TAG_FILE_FORMAT 2 /extended format; --format=1 will not append ;" to lines/ !_TAG_FILE_SORTED 1 /0=unsorted, 1=sorted, 2=foldcase/ ...
--pseudo-tags=[+|-](<pseudo-tag>|*) option enables or disables emitting pseudo-tags.
See the output of "ctags --list-pseudo-tags" for the list of the kinds. See also tags(5) and ctags-client-tools(7) for more details of the pseudo tags.
These tags are composed in such a way that they always sort to the top of the file. Therefore, the first two characters of these tags are used a magic number to detect a tag file for purposes of determining whether a valid tag file is being overwritten rather than a source file.
Note that the name of each source file will be recorded in the tag file exactly as it appears on the command line. Therefore, if the path you specified on the command line was relative to the current directory, then it will be recorded in that same manner in the tag file. See, however, the --tag-relative=(yes|no|always|never) option for how this behavior can be modified.
A tag is an index for a language object. The concept of a tag and related items in Exuberant Ctags are refined and extended in Universal Ctags.
A tag is categorized into definition tags or reference tags. In general, Exuberant Ctags only tags definitions of language objects: places where newly named language objects are introduced. Universal Ctags, on the other hand, can also tag references of language objects: places where named language objects are used. However, support for generating reference tags is new and limited to specific areas of specific languages in the current version.
A tag can record various information, called extension fields.
Extension fields are tab-separated key-value pairs appended to the end of the EX command as a comment, as described above. These key value pairs appear in the general form key:value.
In addition, information on the scope of the tag definition may be available, with the key portion equal to some language-dependent construct name and its value the name declared for that construct in the program. This scope entry indicates the scope in which the tag was found. For example, a tag generated for a C structure member would have a scope looking like struct:myStruct.
--fields=[+|-][<flags>|*] and --fields-(<LANG>|all)=[+|-][<flags>|*] options specifies which available extension fields are to be included in the tag entries.
See the output of "ctags --list-fields" for the list of extension fields. The essential fields are name, input, pattern, and line. The meaning of major fields is as follows (long-name flag/one-letter flag):
kind is a field which represents the kind of language object specified by a tag. Kinds used and defined are very different between parsers. For example, C language defines macro, function, variable, typedef, etc.
--kinds-(<LANG>|all)=[+|-](<kinds>|*) option specifies a list of language-specific kinds of tags (or kinds) to include in the output file for a particular language.
See the output of "ctags --list-kinds-full" for the complete list of the kinds.
Its value is either one of the corresponding one-letter flags or a long-name flag. It is permitted (and is, in fact, the default) for the key portion of this field to be omitted. The optional behaviors are controlled with the --fields option as follows.
$ ctags -o - kinds.c
foo kinds.c /^int foo() {$/;" f typeref:typename:int
$ ctags --fields=+k -o - kinds.c
foo kinds.c /^int foo() {$/;" f typeref:typename:int
$ ctags --fields=+K -o - kinds.c
foo kinds.c /^int foo() {$/;" function typeref:typename:int
$ ctags --fields=+z -o - kinds.c
foo kinds.c /^int foo() {$/;" kind:f typeref:typename:int
$ ctags --fields=+zK -o - kinds.c
foo kinds.c /^int foo() {$/;" kind:function typeref:typename:int
Role is a newly introduced concept in Universal Ctags. Role is a concept associated with reference tags, and is not implemented widely yet.
As described previously in "Kinds", the kind field represents the type of language object specified with a tag, such as a function vs. a variable. Specific kinds are defined for reference tags, such as the C++ kind header for header file, or Java kind package for package statements. For such reference kinds, a roles field can be added to distinguish the role of the reference kind. In other words, the kind field identifies the what of the language object, whereas the roles field identifies the how of a referenced language object. Roles are only used with specific kinds.
For a definition tag, this field takes def as a value.
For example, Baz is tagged as a reference tag with kind package and with role imported with the following code.
package Bar;
import Baz;
class Foo {
// ...
}
$ ctags --fields=+KEr -uo - roles.java
Bar roles.java /^package Bar;$/;" package roles:def
Foo roles.java /^class Foo {$/;" class roles:def
$ ctags --fields=+EKr --extras=+r -uo - roles.java
Bar roles.java /^package Bar;$/;" package roles:def
Baz roles.java /^import Baz;$/;" package roles:imported extras:reference
Foo roles.java /^class Foo {$/;" class roles:def
--roles-(<LANG>|all).(<kind>|all)=[+|-][<roles>|*] option specifies a list of kind-specific roles of tags to include in the output file for a particular language.
Inquire the output of "ctags --list-roles" for the list of roles.
Generally, ctags tags only language objects appearing in source files, as is. In other words, a value for a name: field should be found on the source file associated with the name:. An extra type tag (extra) is for tagging a language object with a processed name, or for tagging something not associated with a language object. A typical extra tag is qualified, which tags a language object with a class-qualified or scope-qualified name.
--extras-(<LANG>|all)=[+|-][<flags>|*] option specifies whether to include extra tag entries for certain kinds of information.
Inquire the output of ctags --list-extras for the list of extras. The meaning of major extras is as follows (long-name flag/one-letter flag):
The extra tag is useful as a placeholder to fill scope fields for language objects defined in a language object with no name.
struct {
double x, y;
} p = { .x = 0.0, .y = 0.0 };
'x' and 'y' are the members of a structure. When filling the scope fields for them, ctags has trouble because the struct where 'x' and 'y' belong to has no name. For overcoming the trouble, ctags generates an anonymous extra tag for the struct and fills the scope fields with the name of the extra tag.
$ ctags --fields=-f -uo - input.c
__anon9f26d2460108 input.c /^struct {$/;" s
x input.c /^ double x, y;$/;" m struct:__anon9f26d2460108
y input.c /^ double x, y;$/;" m struct:__anon9f26d2460108
p input.c /^} p = { .x = 0.0, .y = 0.0 };$/;" v typeref:struct:__anon9f26d2460108
The above tag output has __anon9f26d2460108 as an anonymous extra tag. The typeref field of 'p' also receives the benefit of it.
This extra tag is enabled by default. Add --extras=-F option not to output tags scoped only for a single-file. This is the replacement for --file-scope option of Exuberant Ctags.
static int f() {
return 0;
}
int g() {
return 0;
}
$ ctags -uo - filescope.c
f filescope.c /^static int f() {$/;" f typeref:typename:int file:
g filescope.c /^int g() {$/;" f typeref:typename:int
$ ctags --extras=-F -uo - filescope.c
g filescope.c /^int g() {$/;" f typeref:typename:int
If the end: field is enabled, the end line number of the file can be attached to the tag. (However, ctags omits the end: field if no newline is in the file like an empty file.)
By default, ctags doesn't create the inputFile/f extra tag for the source file when ctags doesn't find a parser for it. Enabling Unknown parser with --languages=+Unknown forces ctags to create the extra tags for any source files.
The etags mode enables the Unknown parser implicitly.
This may allow easier location of a specific tags when multiple occurrences of a tag name occur in the tag file. Note, however, that this could potentially more than double the size of the tag file.
The actual form of the qualified tag depends upon the language from which the tag was derived (using a form that is most natural for how qualified calls are specified in the language). For C++ and Perl, it is in the form class::member; for Eiffel and Java, it is in the form class.member.
Note: Using backslash characters as separators forming qualified name in PHP. However, in tags output of Universal Ctags, a backslash character in a name is escaped with a backslash character. See tags(5) about the escaping.
The following example demonstrates the qualified extra tag.
class point {
double x;
};
For the above source file, ctags tags point and x by default. If the qualified extra is enabled from the command line (--extras=+q), then point.x is also tagged even though the string "point.x" is not in the source code.
$ ctags --fields=+K -uo - qualified.java
point qualified.java /^class point {$/;" class
x qualified.java /^ double x;$/;" field class:point
$ ctags --fields=+K --extras=+q -uo - qualified.java
point qualified.java /^class point {$/;" class
x qualified.java /^ double x;$/;" field class:point
point.x qualified.java /^ double x;$/;" field class:point
The following example demonstrates the reference extra tag.
#include <stdio.h> #include "utils.h" #define X #undef X
The roles:system or roles:local fields will be added depending on whether the include file name begins with '<' or not.
"#define X" emits a definition tag. On the other hand "#undef X" emits a reference tag.
$ ctags --fields=+EKr -uo - inc.c X inc.c /^#define X$/;" macro file: roles:def extras:fileScope $ ctags --fields=+EKr --extras=+r -uo - inc.c stdio.h inc.c /^#include <stdio.h>/;" header roles:system extras:reference utils.h inc.c /^#include "utils.h"/;" header roles:local extras:reference X inc.c /^#define X$/;" macro file: roles:def extras:fileScope X inc.c /^#undef X$/;" macro file: roles:undef extras:fileScope,reference
Exuberant Ctags has the concept of fields and extras. They are common between parsers of different languages. Universal Ctags extends this concept by providing language-specific fields and extras.
vi(1) will, by default, expect a tag file by the name tags in the current directory. Once the tag file is built, the following commands exercise the tag indexing feature:
emacs(1) will, by default, expect a tag file by the name TAGS in the current directory. Once the tag file is built, the following commands exercise the tag indexing feature:
For more commands, see the Tags topic in the Emacs info document.
NEdit version 5.1 and later can handle the new extended tag file format (see --format).
NEdit 5.1 can read multiple tag files from different directories. Setting the X resource nedit.tagFile to the name of a tag file instructs NEdit to automatically load that tag file at startup time.
Because ctags is neither a preprocessor nor a compiler, use of preprocessor macros can fool ctags into either missing tags or improperly generating inappropriate tags. Although ctags has been designed to handle certain common cases, this is the single biggest cause of reported problems. In particular, the use of preprocessor constructs which alter the textual syntax of C can fool ctags. You can work around many such problems by using the -I option.
Note that since ctags generates patterns for locating tags (see the --excmd option), it is entirely possible that the wrong line may be found by your editor if there exists another source line which is identical to the line containing the tag. The following example demonstrates this condition:
int variable;
/* ... */
void foo(variable)
int variable;
{
/* ... */
}
Depending upon which editor you use and where in the code you happen to be, it is possible that the search pattern may locate the local parameter declaration before it finds the actual global variable definition, since the lines (and therefore their search patterns) are identical.
This can be avoided by use of the --excmd=n option.
ctags has more options than ls(1).
ctags assumes the input file is written in the correct grammar. Otherwise output of ctags is undefined. In other words it has garbage in, garbage out (GIGO) feature.
When parsing a C++ member function definition (e.g. className::function), ctags cannot determine whether the scope specifier is a class name or a namespace specifier and always lists it as a class name in the scope portion of the extension fields. Also, if a C++ function is defined outside of the class declaration (the usual case), the access specification (i.e. public, protected, or private) and implementation information (e.g. virtual, pure virtual) contained in the function declaration are not known when the tag is generated for the function definition. It will, however be available for prototypes (e.g. --kinds-c++=+p).
No qualified tags are generated for language objects inherited into a class.
ctags creates temporary files only if either (1) an emacs-style tag file is being generated, (2) the tag file is being sent to standard output, or (3) the program was compiled to use an internal sort algorithm to sort the tag files instead of the sort(1) utility of the operating system. If the sort(1) utility of the operating system is being used, it will generally observe this variable also.
Note that if ctags is setuid, the value of TMPDIR will be ignored.
$XDG_CONFIG_HOME/ctags/*.ctags, or $HOME/.config/ctags/*.ctags if $XDG_CONFIG_HOME is not defined (on other than MS Windows)
$HOME/.ctags.d/*.ctags
$HOMEDRIVE$HOMEPATH/ctags.d/*.ctags (on MS Windows only)
.ctags.d/*.ctags
ctags.d/*.ctags
It is possible to compile ctags to read an additional configuration file before any of those shown above, which will be indicated if the output produced by the --version option lists the custom-conf feature.
Options appearing on the command line will override options specified in these files. Only options will be read from these files.
Note that the option files are read in line-oriented mode in which spaces are significant (since shell quoting is not possible) but spaces at the beginning of a line are ignored. Each line of the file is read as one command line parameter (as if it were quoted with single quotes). Therefore, use new lines to indicate separate command-line arguments.
A line starting with '#' is treated as a comment.
*.ctags files in a directory are loaded in alphabetical order.
See ctags-optlib(7) for defining (or extending) a parser in a configuration file.
See tags(5) for the format of tag files.
See ctags-incompatibilities(7) about known incompatible changes with Exuberant Ctags.
See ctags-client-tools(7) if you are interested in writing a tool for processing tags files.
See ctags-lang-python(7) about python input specific notes.
See readtags(1) about a client tool for binary searching a name in a sorted tags file.
The official Universal Ctags web site at: https://ctags.io/
Also ex(1), vi(1), elvis(1), or, better yet, vim(1), the official editor of ctags. For more information on vim(1), see the Vim web site at: https://www.vim.org/
Universal Ctags project https://ctags.io/
Darren Hiebert <dhiebert@users.sourceforge.net> http://DarrenHiebert.com/
"Think ye at all times of rendering some service to every member of the human race."
"All effort and exertion put forth by man from the fullness of his heart is worship, if it is prompted by the highest motives and the will to do service to humanity."
-- From the Baha'i Writings
This version of ctags (Universal Ctags) derived from the repository, known as fishman-ctags, started by Reza Jelveh.
The fishman-ctags was derived from Exuberant Ctags.
Some parsers are taken from tagmanager of the Geany (https://www.geany.org/) project.
Exuberant Ctags was originally derived from and inspired by the ctags program by Steve Kirkendall <kirkenda@cs.pdx.edu> that comes with the Elvis vi clone (though virtually none of the original code remains).
Credit is also due Bram Moolenaar <Bram@vim.org>, the author of vim, who has devoted so much of his time and energy both to developing the editor as a service to others, and to helping the orphans of Uganda.
The section entitled "HOW TO USE WITH GNU EMACS" was shamelessly stolen from the info page for GNU etags.
| 5.9.0 |