Skip to content

pandoc-ext/logging

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Pandoc lua logging

This library provides pandoc-aware functions for dumping and logging lua objects. It can be used standalone but is primarily intended for using within pandoc lua filters.

Getting started

Put logging.lua somewhere where your pandoc lua filters can access it. For example, if your lua filters are in $HOME/.local/share/pandoc/filters you could put logging.lua in the same place and set the LUA_PATH environment variable:

export LUA_PATH="$HOME/.local/share/pandoc/filters/?.lua;;"

Now try this simple.lua filter:

local logging = require 'logging'

function Pandoc(pandoc)
    logging.temp('pandoc', pandoc)
end

...with this simple.md input:

text

...to get this output on stderr:

(#) pandoc Pandoc {
  blocks: Blocks[1] {
    [1] Para {
      content: Inlines[1] {
        [1] Str "text"
      }
    }
  }
  meta: Meta {}
}

The logging.temp() function is intended for temporary debug output (hence its name) and always generates output. You can use logging.error(), logging.warning(), logging.info() etc. to generate output that's conditional on the log level.

The initial log level is derived from the pandoc --quiet, --verbose and --trace command-line options. By default (i.e., if none of these options are specified) the log level is 0 and only errors and warnings will be output.

With this para.lua filter:

local logging = require 'logging'

function Para(para)
    logging.info('para', para)
end

...by default there's no output:

% pandoc simple.md -L para.lua >/dev/null

...but --verbose sets the log level to 1 (info):

% pandoc simple.md -L para.lua >/dev/null --verbose
[INFO] Running filter para.lua
(I) para Para {content: Inlines[1] {[1] Str "text"}}
[INFO] Completed filter para.lua in 7 ms

All lua objects can be passed to logging.info() etc., and they will be output in a form that should be useful for lua filter development and debugging. The output is intended to be a faithful representation of the pandoc lua types and should make it easy to "dig down". For example, you can see that:

  • para is a Para instance
  • para.content is an Inlines instance with a single element
  • para.content[1] is a Str instance
  • para.content[1].text is a string

...and you could reference all of these directly in the filter. For example, with this para2.lua filter:

local logging = require 'logging'

function Para(para)
    logging.info('para', para)
    logging.info('para.content', para.content)
    logging.info('para.content[1]', para.content[1])
    logging.info('para.content[1].text', para.content[1].text)
end

...you get this:

% pandoc simple.md -L para2.lua >/dev/null --verbose
[INFO] Running filter para2.lua
(I) para Para {content: Inlines[1] {[1] Str "text"}}
(I) para.content Inlines[1] {[1] Str "text"}
(I) para.content[1] Str "text"
(I) para.content[1].text text
[INFO] Completed filter para2.lua in 7 ms

The last text is not quoted, because lua strings are always output literally.

Module contents

logging.type(value)

Returns whatever pandoc.utils.type() returns, modified as follows:

  • Spaces are replaced with periods, e.g., pandoc Row becomes pandoc.Row
  • Inline and Block are replaced with the corresponding tag value, e.g. Emph or Table

logging.spairs(list [, comp])

Like pairs() but with sorted keys. Keys are converted to strings and sorted using table.sort(keys, comp) so by default they're visited in alphabetical order.

logging.dump(value [, maxlen])

Returns a pandoc-aware string representation of value, which can be an arbitrary lua object.

The returned string is a single line if not longer than maxlen (default 70), and is otherwise multiple lines (with two character indentation). The string is not terminated with a newline.

Map keys are sorted alphabetically in order to ensure that output is repeatable.

See the Getting started section for examples.

logging.output(...)

Pass each argument to logging.dump() and output the results to stderr, separated by single spaces and terminated (if necessary) with a newline.

Note: Only table and userdata arguments are passed to logging.dump(). Other arguments are passed to the built-in tostring() function. This is partly an optimization and partly to prevent string arguments from being quoted.

logging.loglevel

Integer log level, which controls which of logging.error(), logging.warning(), logging.info() will generate output when called.

  • -2 : (or less) suppress all logging (apart from logging.temp())
  • -1 : output only error messages
  • 0 : output error and warning messages
  • 1 : output error, warning and info messages
  • 2 : output error, warning, info and debug messages
  • 3 : (or more) output error, warning, info, debug and trace messages

The initial log level is 0, unless the following pandoc command-line options are specified:

  • --trace : 3 if --verbose is also specified; otherwise 2
  • --verbose : 1
  • --quiet : -1

logging.setloglevel(loglevel)

Sets the log level and returns the previous level.

Calling this function is preferable to setting logging.loglevel directly.

logging.error(...)

If the log level is >= -1, calls logging.output() with (E) and the supplied arguments.

logging.warning(...)

If the log level is >= 0, calls logging.output() with (W) and the supplied arguments.

logging.info(...)

If the log level is >= 1, calls logging.output() with (I) and the supplied arguments.

logging.debug(...)

If the log level is >= 2, calls logging.output() with (D) and the supplied arguments.

logging.trace(...)

If the log level is >= 3, calls logging.output() with (T) and the supplied arguments.

logging.temp(...)

Unconditionally calls logging.output() with (#) and the supplied arguments.