tawk - awk but in tcl

awk is a great tool for working with columnar data a line at a time, but the langauge itself is a bit limited. Usually when it's not quite powerful enough, I turn to perl. But... choice is good. Enter tawk, which uses tcl as the scripting language. It's designed to be very familiar to anyone coming from an awk background.

Running tawk

Much like awk:

tawk [OPTIONS] ['script'] [var=value | filename] ...

Reads from standard input if no filenames are given on command line.

Installing

Dependencies are tcl 8.6, and tcllib. Copy the tawk script to /usr/local/bin or wherever - it's a single, self-contained script.

Options

• -F regexp Sets the field seperator (FS).
• -f filename Read the script from the given file instead of it being the first non-option command line argument.
• -safe Run the script in a safe tcl interpreter. Meant for untrusted code.
• -timeout N Exit with an error if a script takes more than N seconds to complete.
• -csv Turn on CSV line parsing. Prefer this over setting FS to a comma.
• -quotechar C Use the given character instead of double quote for quoted CSV fields.
• -quoteall Always quote every CSV field when printing.

The language

Commands

tawk adds the following commands on top of basic tcl:

These are available as top-level commands.

• BEGIN script Executed at the beginning of processing, before any data.
• END script Executed at the end, after processing all data.
• BEGINFILE script Executed at the beginning of each file.
• ENDFILE script Executed at the end of reading each file.
• line script Executed for every line read.
• line test script If test returns true when evaluated by expr, execute the script.
• rline [-field N] re script If the regular expression re matches the line, (Or the specified field), execute the script.

These are available everywhere

• print [arg ...] Print out all its arguments joined by $OFS, or $F(0) if called with no arguments.
• csv_join arglist [delim] [quotechar] [quotemode] Return the list joined into a CSV-formatted string.
• csv_split string [delim] [quotechar] Split a CSV-formatted string into a list.

Changes to existing commands

• continue stops processing the current line and goes to the next. Like next in awk.
• break stops processing the current file and goes on to the next.

Variables

Most of these are lifted straight from awk names.

• F An array holding the columns of the line. $F(0) is the whole line. Setting a new element above NF fills in the missing interval with empty strings. Setting F(0) rebuilds the rest of the array based on splitting the new value.
• NF The number of fields in the current line. Modifying this adjusts F.
• NR The current line number.
• FNR The line number of the current file.
• FILENAME The name of the current file, - for standard input.
• INFILE The file handle of the current file. Only set in BEGINFILE, line and rline blocks.
• FS If set, a single character, or regular expression that is used to indicate field delimiters. If a a single space, or not set, any amount of whitespace is used, and leading and trailing whitespace is first stripped. If an empty string, splits every character into its own field. Can only be a single character in CSV mode.
• OFS Used to separate fields in F(0) when other elements of F are written to or NF is changed. Also used to seperate arguments of print. Can only be a single character in CSV mode.
• CSV 1 if in CSV mode, 0 if in normal mode. (Read-only)
• CSVQUOTECHAR when in CSV mode, the character used to quote fields. Set by the -quotechar option. Defaults to a double quote.
• CSVQUOTE Set to always to always quote CSV fields (Turned on by the -quoteall argument), or auto to only quote when needed. Attempting to set other values raises an error. Defaults to auto.

CSV Mode

If invoked with the -csv option, the output field separator (OFS) is set to comma instead of a space, and print joins its arguments with CSV escaping.

When reading fields, the default field separator (FS) if not explicitly set is a comma, and only single-character separators are supported. Lines are split by a CSV-aware parser - so commas in quoted fields don't count, unlike if just setting FS to a comma in normal mode. The CSVQUOTECHAR variable controls the character used to quote fields (Defaults to double quote, set by the -quotechar option.)

Also, the print command CSV-escapes its arguments, and gets reads a full CSV record, which may be multiple lines.