README
¶
NAME
fsift - File Sifter: Powerful tool to scan, compute digests and compare files. Files can be scanned in directory trees or loaded from previously saved results.
SYNOPSIS
fsift [ options ] [ left-roots... ] [ : right-roots... ]
DESCRIPTION
File Sifter is a utility that has features inspired by the *NIX utilities find and diff, as well as file digest programs such as sha256sum. It can be used for a variety of tasks involving the management of large numbers of files, such as determining which files may be out-of-date in off-site backups.
OVERVIEW
File Sifter able to scan file systems like find, but instead of immediately printing entries, it loads the information into an internal index.
Once loaded, File Sifter performs analysis on the index. It then optionally sorts the entries and prints them out in tabular format, followed by summary statistics. This output may be saved to an FSIFT file, and entries from that file may be loaded during a later run for comparisons or similar analysis.
The non-option arguments given to File Sifter are called roots. Roots are generally either the top of a directory tree to scan, or a previously saved FSIFT file. Each root is loaded into one of two sides of the index, which are called left and right. During analysis, File Sifter can compare the contents of the left and right sides of the index based on user-specified criteria, and it can generate output and reports about the comparisons. Multiple roots may be specified for each side.
File entries may be subjected to user-defined filters at two points: before loading into the index, or before the final output. These filters allow tailoring the analysis to specific requirements.
Each class of information about a file processed by the program is called a column. Columns are defined for the usual file system attributes such as path, size and permissions. There are also columns for digests of file data. Finally, some columns can by added by the File Sifter program as the result of analysis, such as whether the file matches at least one other file belonging to the opposite side of the index.
Columns are only computed and stored if they are made necessary by the user-specified command line options. For example, the contents of files are only read if the user specifies operations on a digest column.
EXAMPLES
- Scan and print information about the current directory tree:
fsift
Example output (can also be saved into an FSIFT file):
| File Sifter output file - V1 |
| Command line:
| Current working directory: /home/user/projects/file-sifter
| Compare keys: path,size,mtime,modestr
| Evaluated columns: path,size,mtime,modestr
| Run start time: 2017-02-06T15:36:04Z
|
| Columns: modestr,size,mtime,path
|
-rw-rw-r-- 15028 2017-02-06T03:41:10Z fsift_manpage.html
-rw-rw-r-- 10476 2017-02-06T03:41:03Z manpage.md
-rw-rw-r-- 12383 2017-02-06T03:41:10Z fsift.1
-rw-r--r-- 24576 2017-02-06T03:41:07Z .manpage.md.swp
drwxrwxr-x 62463 2017-02-06T03:41:03Z ./
|
| Run end time: 2017-02-06T15:36:04Z
| Elapsed time: 254.186µs
|
| STATISTICS: Count Size
| Scanned: 5 62463
| Indexed: 5 62463
| Output: 5 62463- Save the information about a directory tree to an FSIFT file; add md5sum digests:
fsift /path/to/mydir --md5 --out mydir.FSIFT
or
fsift /path/to/mydir -5omydir.FSIFT
- Compare the contents of a directory tree to a previously saved FSIFT file:
fsift t.FSIFT : t/ --diff --md5 --sort path
Example output:
| File Sifter output file - V1 |
| Command line: t.FSIFT : t -5 -sp
| Current working directory: /home/user/temp
| Compare keys: md5,path,size,mtime,modestr
| Sort keys: path
| Evaluated columns: path,size,mtime,side,matched,membership,modestr,md5
| Run start time: 2017-02-10T19:39:18Z
|
| Columns: membership,modestr,size,mtime,md5,path
|
>! drwxrwxr-x 3 2017-02-10T19:38:00Z \- ./
<! drwxrwxr-x 5 2017-02-10T19:36:24Z \- ./
>! -rw-rw-r-- 0 2017-02-10T19:38:00Z d41d8cd98f00b204e9800998ecf8427e added.txt
<= drwxrwxr-x 0 2017-02-10T19:35:39Z \- dir/
>= drwxrwxr-x 0 2017-02-10T19:35:39Z \- dir/
<= -rw-rw-r-- 0 2017-02-10T19:35:39Z d41d8cd98f00b204e9800998ecf8427e dir/no-change.txt
>= -rw-rw-r-- 0 2017-02-10T19:35:39Z d41d8cd98f00b204e9800998ecf8427e dir/no-change.txt
<= -rw-rw-r-- 0 2017-02-10T19:34:53Z d41d8cd98f00b204e9800998ecf8427e no-change.txt
>= -rw-rw-r-- 0 2017-02-10T19:34:53Z d41d8cd98f00b204e9800998ecf8427e no-change.txt
<! -rw-rw-r-- 0 2017-02-10T19:36:24Z d41d8cd98f00b204e9800998ecf8427e remove-me.txt
<! -rw-rw-r-- 5 2017-02-10T19:36:06Z 9387a56cd6f2b6bd4a42f07329b93fca update-me.txt
>! -rw-rw-r-- 3 2017-02-10T19:37:52Z a2141a5ddbd5295d9ec96788790cf1bf update-me.txt
|
| Run end time: 2017-02-10T19:39:18Z
| Elapsed time: 1.03466ms
|
| STATISTICS: L:Count L:Size R:Count R:Size
| Scanned: 6 5 6 3
| Indexed: 6 5 6 3
| Unmatched: 3 5 3 3
| Matching: 3 0 3 0
| Output: 6 5 6 3- Compare the contents of a directory tree to a previously saved FSIFT file, showing just differences:
fsift t.FSIFT : t/ --diff
- Find which direct child subdirectories are using the most storage:
fsift top/dir --sort size --postfilter filetype=d --postfilter 'depth<2'
or
fsift top/dir -ss -ff=d -fd\<2
- Find which git repositories are most in need of garbage collection:
fsift my/projects --prefilter 'path *=**/.git/objects/' --sort nlinks --columns +nlinks
or
fsift my/projects '-ep*=**/.git/objects/' -sL -c+L
- List what kinds of files are in a directory tree by unique extension:
fsift top/dir --columns extension --key extension --postfilter redunidx=1
or
fsift top/dir -cx -kx -fI=1
- Assume lists of previously archived files have been saved in a set of FSIFT files. Before decommissioning a disk, scan it to check against the archives to find any files that may need to be added to archives:
fsift archive-index/*.FSIFT : path/to/disk --membership R --key base,md5
or
fsift archive-index/*.FSIFT : path/to/disk -mR -kb5
- Find all files that have redundant data content:
fsift top/dir --postfilter 'redundancy >1' --key md5 --columns +redundancy --sort size --regular-only
or
fsift top/dir -fr\>1 -k5 -c+r -Rss
OPTIONS
Options may be freely intermixed with roots. Most options have both a long version and a short version. Short boolean options may be concatenated (such as -qS), and short options which take an argument may have the argument concatenated to the option. Long options can have the argument joined with an "=" character. The following are all equivalent: -st, -s t, -smtime, --sort mtime, --sort=mtime, --sort t.
- :
- A single colon on the command line is a special marker that divides the left-side roots from the right-side roots. If no colon is present, all roots are assigned to the left side. Otherwise, any roots on the command line after the colon are assigned to the right side.
Field selection, comparing and sorting:
- -c, --columns=COLUMNS
- Select output columns. The default is "modestr,size,mtime,path" (alternatively, "ostp"). If there are roots on both sides, then "membership" is also added to the default column set.
- -s, --sort=COLUMNS
- Sort output using these fields, in order of precedence. By default, the output is not sorted. For this option, any column name may be precede by a "/" character to sort in inverse order.
- -k, --key=COLUMNS
- Specify which fields used to compare files on each side for equivalence. The default is "modestr,size,mtime,path".
- -5, --md5
- Shortcut to add md5 column to compare key and output.
- -2, --sha256
- Shortcut to add sha256 column to compare key and output.
- -A, --sha512
- Shortcut to add sha512 column to compare key and output.
- -1, --sha1
- Shortcut to add sha1 column to compare key and output.
Pre-analysis filtering:
- -e, --prefilter=FILTER-EXP
- Filter to screen files before they are loaded into the index. Multiple filters may be specified. (When scanning file systems, even if a directory is rejected by this filter, its contents are still scanned. See --prunefilter for controlling recursive scanning.)
- -P, --prunefilter=FILTER-EXP
- Filter that is only applied to directories, and then only while scanning actual file systems (not while loading FSIFT files). Any directory that fails to match this filter will be ignored, and the scan will not descend into that directory.
- -b, --base-match=GLOB-PAT
- Filter files by base name glob pattern. Shortcut for --prefilter 'base*=*GLOB-PAT*'.
- -x, --exclude=GLOB-PAT
- Exclude file system files and/or directory trees if their path matches the given glob pattern. If a directory name matches an exclude pattern, do not descend into it. This option only applies to scans of file systems; it does not filter entries loaded from FSIFT files.
- -R, --regular-only
- Only load regular files into the index while scanning. This option only applies to scans of file systems; it does not filter entries loaded from FSIFT files.
- -L, --follow-links
- Follow symbolic links while scanning the file system. By default, when a symbolic link is found, an entry is created about the symbolic link itself. When this option is in effect, an entry is created with information about the link target. If a symbolic link points to a directory, the program will descend into that directory when this option is in effect.
- -X, --xdev
- Stay within one file system. If a subdirectory is mount point, don't descend into it. (This option currently does not work on Windows systems.)
Post-analysis filtering:
- -f, --postfilter=FILTER-EXP
- After analysis, any entries rejected by this filter are not output. Multiple filters may be specified.
- -m, --membership=CHARS
- Filter output by membership code. The code must be a string containing only a subset of the characters l,r,L or R. The L and l codes only allow entries from left-side roots, and the other two codes allow entries from the right-side roots. The lower case codes only allow files that were matched by files on the other side, and the upper case codes only allow files that were unmatched by files on the other side. For example, the option --membership=Lr only prints files from the left that were unmatched, as well as files from the right that were matched.
- -d, --diff
- Show unmatched entries only. This is a shortcut for --membership=LR. (Which in turn is a shortcut for -f OR -f 'm=<!' -f 'm=>!'.)
- --nodetect
- Don't try to detect whether regular files specified as roots are FSIFT files. By default, if a file looks like it is an FSIFT file, entries are parsed and loaded from the contents of the file instead of loading information about the file itself. If this option is in effect, the detection step is not done, no FSIFT parsing is attempted, and for any regular file given as a root, an entry is created about the file itself.
Output formatting
- -o, --out=PATH
- Output to the given file path instead of the default stdout.
- -Y, --verify
- Checks that all left entries under left-sided roots are matched by at least one entry under a right-sided root. If the left side root is a FSIFT file, this is somewhat similar to running a program like sha1sum -c. If a mismatched left-side file is found, the program exits with a nonzero status code.
- -S, --summary
- Only the header and footer summary info. No file entries are output.
- -p, --plain
- Only output the file entries. No header or footer summary info is printed.
- -0, --plain0
- Like 'plain', but also separate all output fields with ASCII NUL characters. Newlines betweeen file entries are also replaced with NUL characters. The usual FSIFT format escaping is not done.
- -G, --group-nums
- For numeric output, separate groups of three decimal digits with commas.
- -N, --ignore-nulls
- File entry fields get compared during filtering, membership analysis and sorting. By default, if any comparison uses a field that is missing (in other words, null), then an error message will be generated and the program will have a nonzero exit status. If this option is in effect, then no error is generated. In any case, all null comparisons are considered to have a false result.
- -J, --json-out
- Output file entries in JSON format. The output will be an array of JSON objects, each with a key-value pair for each column defined in the output. Numeric fields are output as JSON integers, string fields are output as JSON strings, and missing fields are JSON null values. No header or footer information is output.
- -Z, --out-zone
- Format output times for given location. The default is UTC. Locations can be specified as fixed offsets like "+06:00", or as locations recognized by the Go language time package, such as "Local" or "America/Chicago". Times are always output in RFC3339 format, such as "2017-02-03T04:52:13Z".
- -v, --verbose
- Increase verbosity.
- -q, --quiet
- Decrease verbosity.
- -V, --version
- Show program version and exit.
- -h, --help
- Print a help message and exit.
COLUMN CODES
Each column has a full name and a single-character short name alias.
For options that take a list of columns as an argument, the columns can be specified with a comma-separated list of names, long or short. If no commas are in the argument and it does not match a long name, then the program tries assuming that the argument is a concatenation of short name characters. The argument must not contain whitespace.
For example, size,mtime,path can be shortened to stp.
- p path
- The path of this file relative to the given root.
- b base
- The base name of this file.
- x ext
- The extension of this filename, if any.
- D dir
- The directory part of the 'path' field.
- d depth
- How many subdirectories this file is below its root.
- s size
- Regular files: size in bytes. Dirs: cumulative size. Other: 0. Note that the cumulative sizes of directories do not figure into statistics roundups.
- t mtime
- Modification time as a string in RFC3339 format.
- T mstamp
- Modification time as seconds since Jan 1, 1970.
- V device
- The ID of the device this file resides on.
- S side
- The side of this file's root: 0=left 1=right.
- M matched
- True if this file matches any file from the other side, according to the fields in the --key option.
- m membership
Visual representation of 'side' and 'matched' columns:
side matched membership 0 0 "<!" 0 1 "<=" 1 0 ">!" 1 1 ">="- r redundancy
- Count of all files on this side matching this file.
- I redunidx
- Ordinal of this file amongst equivalents on this side. If a postfilter is set to only select entries where the redunidx value is 1, then only one entry per group of equivalent files will be output on each side, so such a filter can be used to enumerate unique values.
- o modestr
- Mode and permission bits as a human readable string, in *NIX-like format. See the documentation of the Go language os.FileMode object for exact details on the format.
- f filetype
- A single-character code indicating the type of this file. This is usually the same as the first character of modestr, except that for regular files '-' is changed to 'f', and for block devices 'D' is changed to 'b'.
- U uid
- The user ID of this file's owner.
- G gid
- The group ID of this file's group.
- u user
- The name of this file's owner.
- g group
- The name of this file's group.
- L nlinks
- The number of hard links to this file.
- 3 crc32
- The CRC32 checksum of this file. Note: for all checksum and digest fields, directories and other nonregular files get an empty string for a value instead of null. This avoids excessive null comparison errors when these fields are compared. Important: The crc32 checksum should not be used for security purposes.
- 1 sha1
- The SHA1 digest of this file. Important: The sha1 digest probably should not be used for security purposes.
- 2 sha256
- The SHA256 digest of this file.
- A sha512
- The SHA512 digest of this file.
- 5 md5
- The MD5 digest of this file. Important: The md5 digest should not be used for security purposes.
FILTER SPECIFICATIONS
Filters allow the rejection of file entries based on user-defined criteria.
Any column can be used in a filter specification. The general filter syntax is as follows:
<column-name><operator><value>
The column name may be any long or short column identifier. The value is an arbitrary string for comparison to an entry's value. There should be no whitespace between the operator and the value.
The supported operators are:
- =
- Equals
- < <= > >=
- Ordered comparisons (numeric or lexicographic, depending on the column type)
- *=
- Glob pattern match (see below).
- ~=
- Regular expression match. For detailed syntax, see the Go language regexp package documentation.
- .isnull
- Matches if the value is missing in this file entry
- != !*= !~= !.isnull
- Negated versions of the above operators
Combining Filters
Multiple filters of the same type may be specified. In addition to the above filters, there are two special combining filters:
or Matches if either child filter matches.
and Matches if both child filters match.
Combining filters act as binary operators between other filters. They are applied in prefix order (also known as normal Polish notation). In other words, a combining filter appears directly in front of its two child filters.
For example, the following set of filters selects all the files larger than one megabyte that belong to Jack or Jill:
-f and -f 'size>1000000' -f or -f user=jack -f user=jill
If there are any filters of a given type remaining after combining filters have been assigned children, then the remaining filters are assigned to implicit and filters. So for the common case of multiple filters defined with no combining filters, they are anded together.
Glob Patterns
Glob patterns are similar to glob expansion in shell interpreters: "?" matches any single character except "/". "*" matches any number of non-"/" characters. In this implementation, the contents of brackets [...] are fed directly to the underlying regular expression evaluator; the result is similar to many glob implementations, but there are some differences. See the Go language regexp package documentation for details.
As a special case, a ** matches any number of characters, including "/". This can be used to match entire segments of file paths. For example, the filter: path*=**/.config/** will select all files under .config directories below the top level.
When using the ** operator, note that directory paths are always stored with a trailing "/" character. Also note that files directly under the root will not have a "/" preceding them. If this creates problems with glob matching entire paths, a regular expression pattern may be a more flexible alternative.
OTHER FEATURES
FSIFT Files
The output of File Sifter may be saved to a file (an extension of .FSIFT is recommended). If this file is later specified as a root during another run of File Sifter, then by default entries will parsed and loaded from that file.
Note that if the header information or the file entries are suppressed using command line options (such as --summary or --plain), then the output will not be useful for loading later.
Syntax
FSIFT files are text files with two kinds of lines: directive lines and entry lines. Directive lines start with a | character.
The file starts with a directive line identifying it as an FSIFT file. After that is a group of informational header directives. The only other directive that is relevant to parsing FSIFT files is the Columns directive. This specifies which fields are present in each entry line. The Columns directive must appear before the first entry line. All other directive lines are ignored by the parser.
Entry lines follow the header, with one entry line per output file entry. These lines start with a space, and are followed by one or more fields. The number of fields matches the number of names in the Columns directive, and each item is the field information from the corresponding column name.
After the entries, a footer is output. This contains a set of directive lines with summary information such as run time, file and byte counts.
In entry lines, fields are separated by space characters. Certain characters within a field are escaped with backslashes: spaces, newlines, carriage returns and backslashes. In addition, there are two special escape sequences: \- indicates a zero-length string, and \~ indicates a missing value (called a null). These escapes are removed when FSIFT files are parsed.
As a special exception, space characters are not escaped in the last column. This is possible because the parser knows that no other fields will follow this one before the next newline. In the common case where path is the last field on each line, this makes the output look cleaner when there are spaces in file names. (However, even in the last column, any spaces at the beginning or end of a field are still escaped.)
When the --plain0 option is specified, there is no escaping performed, and all data is separated by ASCII NUL characters. Then the --json option is specified, the output is escaped according to JSON rules. File Sifter does not support later loading from either of these formats.
Summary Statistics
At the end of the run, a footer is printed by default which summarizes the analysis of the files. If both left and right roots were specified, it breaks out the statistics by left and right files. It shows file count and total size for the files processed.
The entry lines for directories show the cumulative size of all the files indexed under the directory. These cumulative sizes are not included in the summary statistics because they would cause double-counting.
The Scanned line shows all of the files considered (which does not include those files rejected by --exclude or --regular-only). The Indexed line shows all of the files that pass the prefilter stage and get loaded into the index.
The Unmatched line shows all files that did not have a match on the other side, and the Matching line shows the files that did have a match. The previous two lines are only output if there were roots on both sides. The Output line shows all of the files that passed the postfilter stage and were printed to the output (or if --summary is specified, would have been output).
| Run end time: 2017-02-10T02:58:56Z
| Elapsed time: 732.146µs
|
| STATISTICS: L:Count L:Size R:Count R:Size
| Scanned: 26 241927 21 177969
| Indexed: 26 241927 21 177969
| Unmatched: 7 92064 2 45056
| Matching: 19 149863 19 132913
| Output: 26 241927 21 177969Interactive Status Output
While scanning the file system, File Sifter can print temporary interactive messages that show the current status of the scan. This includes the initial scan phase, as well as any required digest scan phases. This output can be suppressed with the --quiet option.
Character Encodings
All characters are processed assuming UTF-8 encoding. File names with characters that are not decodable as valid Unicode may produce unexpected results. Such characters are likely to pass through to the output unchanged, but comparisons and analysis might have problems. Note that in some cases, file systems can be mounted with options that automatically translate characters which cannot be converted to Unicode to "safe" substitute sequences.
Platform Specific Differences
On all platforms, path separators are internally represented and output as "/", regardless of what the OS uses.
On all platforms, FSIFT files always use *NIX-style line endings.
On Windows, the following columns do not currently get populated with meaningful values: uid, user, gid, group, nlinks and device.
On windows, the modestr column contains a simplified approximation of permissions.
On Windows, the program is not currently able to detect the console width and assumes a fixed value of 80. This may affect the appearance of interactive status messages.
HISTORY
File Sifter is the result of a long evolution of personal utilities that I wrote over the years to help keep track of files from various computer systems.
The first utilities were simple Perl scripts that did a simple scan/sort/diff on directories. Eventually, I wrote an program in C++ that used SQLite for an internal engine that had features somewhat similar to this implementation. However, it was hard to use the SQL-oriented features of that version, and although I found it very useful and used it for many years, I was never very happy with it.
I recently decided to pare down the program to its most useful features, clean up the user interface, drop the embedded database, and port it to Go. The result of that effort is this rendition of File Sifter.
Documentation
¶
Overview ¶
Package sifter contains the core functionality of the file sifter application.
Index ¶
Constants ¶
const ( ColPath = iota // path relative to scan root ColBase // basename of file ColExt // extention of file ColDir // directory part of path ColDepth // depth below scan root ColSize // size in bytes ColMtime // file mod time, RFC3339 format in UTC ColMstamp // mtime, unix timestamp ColDevice // ID of device file resides in ColSide // "side" of the scan, false=left true=right ColMatched // true if this file matches any on the other side ColMembership // matching flags in string format ColRedundancy // number of matches of this file on this side ColRedunIdx // the index of this entry within equivalent entries on this side ColModestr // file type and permissions in unix format ColFileType // 1-char file type code ColUid // user ID of file owner ColGid // group ID of file ColUser // user name of file owner ColGroup // group name of file ColNlinks // number of hard links to file ColCrc32 // crc32 digest ColSha1 // sha1 digest ColSha256 // sha256 digest ColSha512 // sha512 digest ColMd5 // md5 digest ColLAST // dummy end marker; must be last // Flag for inverse sort ColInvertFlag = 0x1000000 )
IDs for column types
Variables ¶
This section is empty.
Functions ¶
func GetColumnHelp ¶
func GetColumnHelp() []string
Return a list of strings holding help text describing all columns
Types ¶
type ColSelector ¶
type ColSelector struct {
// contains filtered or unexported fields
}
Holds a list of columns and the default list
type Column ¶
type Column int
func ParseColumnsList ¶
Parse an argument that has a comma-separated list of long and/or short column names, return a corresponding slice of column IDs. If allowInverse is set, inverse sort columns (prefixed with '/' in the argument) are identified in the returned inverse list, If the argument has no commas and it doesn't match a long name, also try assuming each char is an individual short name. If an error occurs, return nil and the error.
type Context ¶
type Context struct {
// The following fields are set by the command line program before starting
OutCols ColSelector // columns to show in output
SortCols ColSelector // columns to sort by, in order of precedence
KeyCols ColSelector // columns to use for compare key
PreFilterArgs []*Filter // filter objects as parsed from command line --prefilter args
PostFilterArgs []*Filter // filter objects as parsed from command line --postfilter args
PruneFilterArgs []*Filter // filter objects as parsed from command line --prunefilter args
Roots map[bool][]string // lists of root paths, [false] = left side, [true] = right side
CurSide bool // current side to add roots to (true after ":" command line arg)
Verbosity int // verbosity level, default=0
SummaryOnly bool // true to suppress file entry output
GroupNumerics bool // true to group numbers with commas, like 1,234
Plain bool // true to suppress header and footer output
Plain0 bool // like Plain, but also use '0x00' to separate fields
FollowLinks bool // true to follow/use targets of symbolic links
RegularOnly bool // true to only index regular files
AddMd5 bool // true to add digest columns to output and compare key...
AddSha1 bool // "
AddSha256 bool // "
AddSha512 bool // "
JsonOut bool // true to output data in JSON format
MembershipFilt string // add a postfilter based on membership codes [lrLR]
IgnoreNullCmps bool // suppress warnings about null comparisons
Excludes []*regexp.Regexp // pattern to exclude files, dir trees by path match
OutputPath string // output file, if any (default=stdout)
NoDetect bool // true to suppress autodetect of FSIFT files for roots
XDev bool // true to prevent descending into directories on different file systems
Verify bool // true to check that all files on left are matched on right
OutputTimezone *time.Location // if set, translate output dates to given timezone
// contains filtered or unexported fields
}
Context is the main object that holds the state of a running instance of the file sifter program.
func NewContext ¶
func NewContext() *Context
NewContext creates a newly initialized context object with default settings. Note that unless the global flag unitTest is set, the shutDown method must eventually be called to avoid leaking a goroutine.
func (*Context) Run ¶
Run the scan defined for this context. Assumes that the exported option fields have been set as necessary before this method is called. Returns zero on success, nonzero if any errors occurred during the run.
func (*Context) SetOutputStreams ¶
Set the output and error streams. Used for unit testing.
func (*Context) UpdateColumnsCmdlineArg ¶
func (self *Context) UpdateColumnsCmdlineArg(colSel *ColSelector, posn int, arg string)
Update a column-based option by adding columns at the indicated position. An argument prefixed with '+' causes the columns to be appended to any current columns. The posn index may be negative to insert near the end. The string in arg is parsed into a list of columns.
type Filter ¶
type Filter struct {
// contains filtered or unexported fields
}
Filter holds the definition of a prefilter or postfilter.
func ParseFilter ¶
ParseFilter parses a filter specification from a command line argument and returns a new filter object implementing it, or an error if it couldn't be parsed.
Source Files
Directories