SLT: SymlinkTool R6 class

SLTR Documentation

SymlinkTool R6 class

Description

Class for lightweight file-system level data versioning, logs and reports without need for a database.

Methods

Public methods


Method new()

Initialize the SymlinkTool object - an R6 class

The constructor function.

Usage
SLT$new(
  user_root_list = NULL,
  user_central_log_root = NULL,
  schema_repair = TRUE,
  verbose = TRUE,
  verbose_startup = FALSE,
  csv_reader = "fread_quiet",
  timezone = Sys.timezone()
)
Arguments
user_root_list

[list] Named list of root directories for pipeline outputs. This is where 'version_name' folders live - these are iterative runs of an analysis pipeline.

user_central_log_root

[path] Root directory for the central log. If you have multiple roots in the 'user_root_list', you probably want the central log to live one level above those roots.

schema_repair

[logical] Default 'TRUE'. If 'TRUE', the tool will attempt to repair any schema mismatches it finds in the logs when reading and writing e.g. add new columns if the tool schema has columns that existing logs do not. If 'FALSE', the tool will stop and throw an error if it finds a schema mismatch.

verbose

[lgl: default TRUE] control message verbosity - if TRUE, standard message, if FALSE, warn only if something is irregular.

verbose_startup

[lgl] see start up warnings, if relevant?

csv_reader

[chr] The CSV reader to use (also assigns matching CSV writer). CAUTION: DO NOT USE 'data.table::fread' if you have any quotation marks (") in log comments (these lead to exploding series of quotations). https://github.com/Rdatatable/data.table/issues/4779. Otherwise use 'read.csv[2]'. Options:

  • fread_quiet - 'data.table::fread' and suppress warnings (default)

  • fread - 'data.table::fread'

  • read.csv - 'utils::read.csv' - safer

  • read.csv2 - 'utils::read.csv2' - safer, comma as decimal point, semicolon as field separator

timezone

[chr] Default 'America/Los_Angeles'. The timezone to use for datestamps in logs. Must be a valid 'OlsonNames()' string.

Returns

[symlink_tool] A symlink tool object. You can instantiate a.k.a. create multiple objects, each of which has different roots and central logs.

Examples
try(SLT$new()) # call with no arguments to see instructions
# Tool will not instantiate on Windows unless running with Admin permissions
# - requirement for symlink creation on Windows

Method return_dictionaries()

Return the contents of all private dictionaries.

Usage
SLT$return_dictionaries(item_names = NULL)
Arguments
item_names

[chr] Default 'NULL'. If 'NULL', show all static internal fields. Otherwise, vector of static field names you want to see.

Returns

[list] of all static internal fields


Method return_dynamic_fields()

Print the contents of all dynamic fields.

Usage
SLT$return_dynamic_fields(item_names = NULL)
Arguments
item_names

[chr] Default 'NULL'. If 'NULL', show all dynamic internal fields. Otherwise, vector of dynamic field names you want to see.

Returns

[std_out] Print dynamic field values to std_out.


Method mark_best()

Mark an output folder with a "best" symlink.

Enforces: - maximum of one best model - does not go back through history to make a best model from a prior version (not capable, this is what log_tool is for)

Writes: - appends to a log file in the output folder with a date and time stamp - appends a line to the central log file with a date and time stamp

Usage
SLT$mark_best(version_name, user_entry)
Arguments
version_name

[chr] The directory name of the output folder that lives directly under one of the 'root's you define when you instantiate the tool.

user_entry

[list] Named list of user-defined fields to append to the log. After making a tool called e.g. slt, call 'slt$return_dictionaries("log_fields_user")' to find which fields a user may add. If you want to make your own version of this class, you may update 'log_schema' in the 'private$DICT' section to allow for them.

Returns

[ste_err] Messages about actions taken.


Method mark_keep()

Mark an output folder with a "keep_<version_name>" symlink

Writes: - appends to a log file in the output folder with a date and time stamp - appends a line to the central log file with a date and time stamp

Usage
SLT$mark_keep(version_name, user_entry)
Arguments
version_name

[chr] The directory name of the output folder that lives directly under one of the 'root's you define when you instantiate the tool.

user_entry

[list] Named list of user-defined fields to append to the log. After making a tool called e.g. slt, call 'slt$return_dictionaries("log_fields_user")' to find which fields a user may add. If you want to make your own version of this class, you may update 'log_schema' in the 'private$DICT' section to allow for them.

Returns

[std_err] Messages about actions taken.


Method mark_remove()

Mark an output folder with a "remove_<version_name>" symlink

Indication that the results can be deleted - In the future, this will be used to remove old versions of the output, and provide a list of ST-GPR models to delete

Writes: - appends to a log file in the output folder with a date and time stamp - appends a line to the central log file with a date and time stamp

Usage
SLT$mark_remove(version_name, user_entry)
Arguments
version_name

[chr] The directory name of the output folder that lives directly under one of the 'root's you define when you instantiate the tool.

user_entry

[list] Named list of user-defined fields to append to the log. After making a tool called e.g. slt, call 'slt$return_dictionaries("log_fields_user")' to find which fields a user may add. If you want to make your own version of this class, you may update 'log_schema' in the 'private$DICT' section to allow for them.

Returns

[std_err] Messages about actions taken.


Method unmark()

Remove all symlinks for a single 'version_name' in all 'roots'

Writes: - appends to a log file in the output folder with a date and time stamp - does _not_ append to the central log file

Usage
SLT$unmark(version_name, user_entry)
Arguments
version_name

[chr] The directory name of the output folder that lives directly under one of the 'root's you define when you instantiate the tool.

user_entry

[list] Named list of user-defined fields to append to the log. After making a tool called e.g. slt, call 'slt$return_dictionaries("log_fields_user")' to find which fields a user may add. If you want to make your own version of this class, you may update 'log_schema' in the 'private$DICT' section to allow for them.

Returns

[std_err] Messages about the symlinks removed.


Method roundup_best()

Find all 'best_' symlinks in all 'roots'

Return both the symlink and the resolved symlink (folder the symlink points to)

Usage
SLT$roundup_best()
Returns

[list] list of data.tables - one for each 'root'


Method roundup_keep()

Find all 'keep_' symlinks in all 'roots'

Return both the symlink and the resolved symlink (folder the symlink points to)

Usage
SLT$roundup_keep()
Returns

[list] list of data.tables - one for each 'root'


Method roundup_remove()

Find all 'remove_' symlinks in all 'roots'

Return both the symlink and the resolved symlink (folder the symlink points to)

Usage
SLT$roundup_remove()
Returns

[list] list of data.tables - one for each 'root'


Method roundup_unmarked()

Find all folders without symlinks in all 'roots'

Useful if you're rapidly iterating, have only marked a couple folders, and want to remove the rest.

Usage
SLT$roundup_unmarked()
Returns

[list] list of data.tables - one for each 'root'


Method roundup_by_date()

Find all 'version_name' folders by creation date

Only finds folders that _have a log_, and reads creation date on first row. User may select dates by (using the 'date_selector' argument): - greater than - 'gt' - greater than or equal to - 'gte' - less than - 'nt' - less than or equal to 'nte' - equal to 'e'

Usage
SLT$roundup_by_date(user_date, date_selector)
Arguments
user_date

[c("character", "Date", POSIXct", "POSIXt")] A date with class requirements - must be formatted "2020-01-01 or 2020_01_01 or 2020/01/01"

date_selector

[chr] See docstring explanation.

Returns

[list] list of data.tables - one for each 'root'


Method get_common_new_version_name()

Get a new YYYY_MM_DD.VV version compatible with _ALL THE TOOL'S ROOTS_

If root1 has 2025_01_01.01 and root2 has 2025_01_01.03, then a new folder would need to be 2025_01_01.04

Usage
SLT$get_common_new_version_name(date = "today", root_list = private$DICT$ROOTS)
Arguments
date

[chr] Default "today". The date to use for the new version name. Must be formatted "2020_01_01"

root_list

[list] named list of root directories for pipeline

Returns

[chr] format YYYY_MM_DD.VV


Method make_new_version_folder()

Create a new ‘version_name' folder in _ALL THE TOOL’S ROOTS_

Create a new log in each folder. No symlinks are created. No 'user_entry' is used.

Usage
SLT$make_new_version_folder(version_name = self$get_common_new_version_name())
Arguments
version_name

[chr] The directory name of the output folder that lives directly under one of the 'root's you define when you instantiate the tool. For convenience, user may leave NULL (default) and 'get_common_new_version_name()' is used on that root.

Returns

[std_err] Messages about the folder creation.


Method make_new_log()

Safely write an empty log file for first pipeline runs

When you start a new pipeline run, make an empty log - helpful if you let this tool manage all your versions - you can roundup version_names by creation date using the log's first entry - the file system doesn't track directory creation dates (at time of writing)

Usage
SLT$make_new_log(version_name)
Arguments
version_name

[chr] The directory name of the output folder that lives directly under one of the 'root's you define when you instantiate the tool.

Returns

[std_err] Messages about the log creation.


Method delete_version_folders()

Delete a 'version_name' folder marked with a 'remove_' symlink from _ALL ITS ROOTS_

Removes the symlink(s) and the underlying folder(s), and updates central log if folders were removed.

Writes: - appends a line to the central log file with a date and time stamp

Usage
SLT$delete_version_folders(version_name, user_entry, require_user_input = TRUE)
Arguments
version_name

[chr] The directory name of the output folder that lives directly under one of the 'root's you define when you instantiate the tool.

user_entry

[list] Named list of user-defined fields to append to the log. After making a tool called e.g. slt, call 'slt$return_dictionaries("log_fields_user")' to find which fields a user may add. If you want to make your own version of this class, you may update 'log_schema' in the 'private$DICT' section to allow for them.

require_user_input

[lgl] if 'TRUE', will prompt user to confirm deletion.

Returns

[std_err] Messages about deletion events.


Method make_reports()

Make all reports

Writes all reports to a summary .csv for every 'root' defined in the tool.

Usage
SLT$make_reports()
Returns

[std_err] Messages about where reports were written.


Method clone()

The objects of this class are cloneable with this method.

Usage
SLT$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Examples


## ------------------------------------------------
## Method `SLT$new`
## ------------------------------------------------


try(SLT$new()) # call with no arguments to see instructions
# Tool will not instantiate on Windows unless running with Admin permissions
# - requirement for symlink creation on Windows

vmTools documentation built on Aug. 8, 2025, 7:28 p.m.