Aegisub Automation documentation Version 3 Copyright 2005 Niels Martin Hansen. --- This document describes version 3 of the automation system used in Aegisub. The automation system uses the Lua language for scripting engine. See <http://www.lua.org/> for more information. --- What is Automation? Aegisub Automation is a scripting system designed to automate many processing tasks of ASS subtitles, instead of using tedious, error-prone manual processing. The primary purpose is creating karaoke effects for anime fansubs. The Automation script is given the complete subtitle data from a subtitle file, in a format suited for creating special effects. The script will return a complete substiture for the original subtitle data, allowing full freedom of processing. A number of helper functions are provided, to aid in finding errors in scripts, as well as retrieve further data about the subtitles, needed to created advanced effects. --- Scripts, files, functions: A script is a file containing Lua code. One file can define just one script, but several scripts can share code by the help of including other files. All scripts are run in a separate interpreter, and as such don't have any way of interacting with other loaded scripts. All strings in a script should be in UTF-8 encoding, without byte-order mark. All strings input to a script are encoded as UTF-8. Script files may start with an UTF-8 BOM (byte-order mark) or not, but this is currently not well tested. A script must define certain global variables: version Number. Version of the scripting interface used. The version described in this file is 3. To comply with version 3, version must be: 3 <= version < 4 kind String. Not used, but mandatory. Set it to "basic_ass" for now. name String. Displayed name of the script. description String. Optional. Long description of the script. process_lines Function. The main script function. configuration Table. Optional. Configuration options for the script. The functions are described in detail in the following. The script may define further global variables, but they do not have any special meaning. Be aware, however, that later versions of the scripting system might define further global variables with special meanings, so be careful choosing names for private use globals. It's recommended to prefix private global variables with "p_"; the scripting system will never assign special meanings to global variables with that prefix. The scripting system defines a global variable with name "aegisub", which contains important values. You should not hide the "aegisub" variable. --- The processing function: The processing function is the heart of the script. It takes as input some meta-information about the subtitles, the styles used in the subtitles, as well as the actual subtitle data to process. The output is a set of subtitle data in the same format as the input. The output subtitle data will be used as a complete replacement of the input data. Future versions might allow modifying style data and meta data as well. The processing function is defined as follows: function process_lines(meta, styles, lines, config) The arguments are: @meta Table. Meta information about the script. (Script Info section.) @styles Table. Style definitions. (V4+ Styles section.) @lines Table. Subtitle events. (Events section.) @config Table. Values set for the configuration options provided. If no configuration options were provided, this will be an empty table. Returns: One value. This value must be a table, using the same format as @lines. Note that the indexes in the return value may be either zero-based or one-based, to allow for greater compatibility. You are encouraged to use one-based indexes. Description of @meta: This is a table with the following keys: res_x Horizontal resolution of the script. res_y Vertical resolution of the script. Description of @styles: This is a table with the following keys: -1 Number. The amount of styles defined, called "n". 0 -> n-1 Table. The actual style definitions. <string "name"> Table. The style definition with the specified name. The key -1 is used for count rather than "n", since one might have a style definition with the name "n". A style definition is a table with the following keys: name String. Name of the style. fontname String. Name of the font used. fontsize Number. Size of the font used. color1 String. Primary color. All color fields use raw hexadecimal format, that is, no special characters before or after the hex string. color2 String. Secondary color. color3 String. Outline color. color4 String. Shadow color. bold Boolean. Bold text or not. italic Boolean. Italic text or not. underline Boolean. Underlined text or not. strikeout Boolean. Striked-out text or not. scale_x Number. Horizontal scale. scale_y Number. Vertical scale. spacing Number. Spacing between characters. angle Number. Rotation angle in degrees. borderstyle Number. 1=Outline + drop shadow, 3=Opaque box (not really used???) outline Number. Thickness of outline. shadow Number. Distance of shadow from text. align Number. Numpad style alignment. margin_l Number. Left margin in pixels. margin_r Number. Right margin in pixels. margin_v Number. Vertical margin in pixels. encoding Number. Font encoding used. Description of @lines: This is a table with the following keys: n Number. The amount of lines. 0 -> n-1 Table. The actual lines. A line is a table with the following key: kind String. Can be "blank", "scomment", "comment" or "dialogue". The keys otherwise defined depends on the kind of the line. If the kind if "blank", no further fields are defined. If the kind is "scomment", the line is a "semicolon comment", and the following key is defined: text String. Text following the semicolon until end of line. EOL not included. If the kind is "comment" or "dialogue", the line is either a Comment: or a Dialogue: line. In both cases, the following keys are defined: layer Number. start_time Number. Start time of line in centiseconds. (Might change to userdata later.) end_time Number. End time of line in centiseconds. (Might change to userdata later.) style String. Style used for this line. name String. Character name speaking this line. margin_l Number. Left margin override, in pixels. (0=no override) margin_r Number. Right margin override, in pixels. (0=no override) margin_v Number. Right margin override, in pixels. (0=no override) effect String. Effect to apply to the line. (No error checking done.) text String. Text to display. text_stripped String. Same as text, but stripped for all tags, and newline/hardspace tags are converted to real newlines/spaces. Non-hard spaces at the start/ end of lines are stripped. karaoke Table. Line split into karaoke syllables. See below for more information. Note about output: Neither text_stripped nor karaoke are used when the results are parsed, they are only passed to simplify processing. You should set text to the final text of the line, you want in the output. It is encouraged to entirely leave text_stripped and karaoke out of the tables in the result. Karaoke tables: A karaoke table has a number of values indexed by numbers. Each value represents a karaoke syllable. Key "n" holds the number of syllables. The syllables can be accessed from index 0 and up. The syllables are indexed chronologically. A karaoke table always has at least one syllable. The first syllable (index 0) contains all data before the first timed syllable. Each syllable is a table containing the following keys: duration Number. Duration of the syllable in centiseconds. Always 0 for first syllable. kind String. "Kind" of the karaoke, the name of the tag. For a \k type syllable, kind is "k", for a \kf syllable kind is "kf". Freeform tags can be used, as long as they start with the letter "k" or "K". Always the empty string ("") for the first syllable. text String. Text of the syllable. This includes formatting tags. For the first syllable, this contains everything before the first karaoke timing tag. text_stripped String. Same as text, but with all formatting tags stripped. Description of @config: This is a table. The keys are the names for the options defined in the global "configuration" table. The values are the values provided by the user. --- Script configuration: An automation script can provide a configuration set, allowing the user to set certain options before the script is called. This is performed through the "configuration" value. Scripts can define configuration options of the following types: label A static, non-editable text displayed to the user. (Useful for adding additional explanations for some options.) text Freeform text entry. int Integer numbers. A range of valid values can be specified. float Any kind of number. A range of valid values can be specified. bool A boolean on/off value. colour An RGB colour value. style The name of a style defined in the subtitles. The "configuration" table: The "configuration" table contains a number of values indexed by numbers. Each value defines a configuration option. The configuration options must be in keys numbered from 1 to n, where n is the number of options. No "n" key is required. The configuration options will be presented to the user in the order defined. Each configuration option is a table containing the following keys: name String. The internal name used to refer to the configuration option. Must not contain the colon or pipe characters. (ASCII 58 and 124.) kind String. One of "label", "text", "int", "float", "bool", "colour" and "style". Defines what kind of option this is. label String. Name of the option, presented to the user. Should be very short. hint String. Longer description of the option, presented to the user as a tooltip. Ignored for "label" kind options. min Number. Optional. Lowest value allowed. Only used for "int" and "float" kinds. max Number. Optional. Highest value allowed. Only used for "int" and "float" kinds. default. Type depends on "kind". The value given to this configuration option before the user has entered another value. Ignored for "label" kind options. Data types for the different kinds: label None. A label doesn't have a value, and won't be present in the @config table in the process_lines function. text String. You might want to do some kind of extra validation on text input, as it might be anything. int Number. Guaranteed to always be integer. float Number. Can be integer or not. bool Boolean. colour String. An ASS hex colourcode in "&HBBGGRR&" format. style String. The name of the style. The style can't be guaranteed to exist, as another export filter in Aegisub might have removed it before your script gets to run. --- Script environment and registration: A script is assigned to a subtitle file by adding it to the "Automation Scripts" extra header in the [Script Info] section. This header contains a list of script filenames, separated by pipe characters. Example: Automation Scripts: test1.lua|test2.lua All scripts run in their own separate interpreter. This means there is no risk of name collisions, though also that scripts can't easily share code. If you need to share code between several scripts, you should create a subdirectory to the script directory, and place include files there. The settings for the configuration options for a script are stored in the ASS file in the following way: Each script gets one line for configuration, named "Automation Settings" plus a space plus the filename of the script. The filename used is stripped of all path specifiers. (Use unique filenames for your scripts!) The value of the line is a pipe-separated list of "name:value" pairs. The name is the internal name given by the "name" key. It is not mangled in any way. The way the value is stored depends on the kind of the option. label Not stored. text The string is stored in an URL-encoding like manner. Some unsafe characters are replaced with escape-sequences of the form #xx, where xx is a two-digit hexadecimal number for the ASCII code of the escaped character. Only ASCII- characters can be escaped this way, Unicode characters aren't supported. int Stored in ASCII base 10 without any group separators. float Stored in exponential notation, using ASCII base 10. (As the %e sprintf() argument.) bool True is stored as "1", false as "0". colour Stored as in ASS hex format without any mangling. style Stored in the same manner as "text" kind options. --- Helper functions: There is a gloabl variable names "aegisub". This is a table containing various helper functions. The following helper functions are defined: function aegisub.set_status(text) Sets the current status-message. (Used for progress-reporting.) @text String. The status message. Returns: nothing. function aegisub.output_debug(text) Output text to a debug console. @text String. The text to output. Returns: nothing. function aegisub.colorstring_to_rgb(colorstring) Convert an ASS color-string to a set of RGB values. @colorstring String. The color-string to convert. Returns: Four values, all numbers, being the color components in this order: Red, Green, Blue, Alpha-channel function aegisub.report_progress(percent) Report the progress of the processing. @percent Number. How much of the data have been processed so far. (Percent) Returns: nothing. function aegisub.text_extents(style, text) Calculate the on-screen pixel size of the given text using the given style. @style Table. A single style definition like those passed to process_lines. @text String. The text to calculate the extents for. This should not contain formatting codes, as they will be treated as part of the text. Returns 4 values: 1: Number. Width of the text, in pixels. 2: Number. Height of the text, in pixels. 3: Number. Descent of the text, in pixels. 4: Number. External leading for the text, in pixels. Short description of the values returned: Width: The X advance of the text, how much the "cursor" moves forward when this text is rendered. Height: The total height of the text, including internal leading. Descent: How far below the baseline a character can extend. The ascent of the text can be calculated as (height - descent). External leading: How much vertical spacing will be added between the lines of text rendered with this font. The total height of a line is (height + external_leading). function aegisub.frame_from_ms(ms) Return the video frame-number for the given time. @ms Number. Time in miliseconds to get the frame number for. Returns: A number, the frame numer. If there is no framerate data, returns nil. function aegisub.ms_from_frame(frame) Returns the start-time for the given video frame-number. @frame Number. Frame-number to get start-time from. Returns: A number, the start-time of the frame. If there is no framerate data, returns nil. function include(filename) Include the named script. The script search-path defined in Aegisub will be used, searching for the script. If the filename is relative, the regular search path will not be used, but instead the filename will be taken as relative to the directory the current script is located in. Note that if you use include() inside an included script, relative paths will still be taken relative to the original script, and not relative to the current included script. This is a design limitation. The included script is loaded as an anonymous function, which is executed in the current environment. This has two implications: You can include files based on conditional statements, and even in loops, and included files can return values using the "return" statement. @filename String. Name of the file to include. Returns: Depends on the script included. Note that if the file couldn't be found, the script will be terminated (or fail to load.) --- Versions of the scripting interface: Here's a quick history of the scripting interface: Version 1 Using Lua as engine. The scripts used in the Karaoke Effector application, avaible at: <http://www.jiifurusu.dk/files/programming/effector/> Version 2 Using Python as engine. The first draft for an Aegisub automation engine. Never implemented. Version 3 Using Lua as engine. The current version.