Extracting labels from records has been discussed extensively in previous posts. The following is a partial list of proposed solutions: (1) processing the record through the clipboard, (2) processing the record through the Automator Runner application, (3) extracting the labels after coercing the record to an NSDictionary via Cocoa Applescript, (4) osax solutions such as the get user property names command of Late Night Software’s List & Record Tools, (5) parsing the text representation of the record, and (6) the elegant method described by Nigel Garvey of directly parsing the content of the record at the byte level (http://macscripter.net/viewtopic.php?id=35884). Some methods (#'s 1, 2, and 4) have problems with labels that are reserved Applescript words. To my knowledge, all of the methods except #5 are unable to extract labels from application records. Previously described solutions of type #5 have exhibited some difficulties handling certain “challenging” records: those with complex nesting of lists and records and those whose labels and/or values, after transformation into text form, contain the record property delimiter string ", " and/or the label:value delimiter character “:”.
The current handler is an attempt at a vanilla Applescript solution of type #5. It proceeds through the following steps: (1) generates a text representation of the record, (2) tokenizes all text elements enclosed in double-quotes, pipes, and curly braces (i.e., those elements that may contain ", " or “:” character sequences), (3) extracts the record labels and values from the tokenized text, and (4) detokenizes the labels and values. Detokenizing is done recursively to handle any nested tokens. The record’s text representation is gotten with the well known technique of parsing a forced error message. The current code is a bit more complex than customary in order to handle extremely rare outlier cases where simpler methods of parsing the forced error message have failed. Text item delimiters are used for string searching and replacing. Thanks to text item delimiter’s capabilities, the handler executes reasonably quickly.
A known limitation of the handler is an inability to properly parse a record with a piped record label that contains an odd number of unescaped double-quote characters. This should be a rare occurrence in real-life usage. Another limitation affects only the extraction of record values and not record labels. It is encountered when the forced error message technique renders an application record value into a text form that does not permit the reconstruction of the original value with a command of the form: run script “tell application "[application name]" to return [text representation of record value]”. This problem has been encountered very infrequently and only with application values of isolated non-Applescript classes.
A principal goal with the current handler is the extraction of application record labels and values. Yet only a tiny subset of applications have been tested. Any feedback from failure of the handler in the wide world of applications out there, and any comments or suggestions about the handler in general, are most welcome.
USAGE NOTES:
The handler takes as input an Applescript or application record. It returns an Applescript record with the following properties:
[b]recordLabels[/b] - list of the record's labels as text strings
[b]recordLabelsPiped[/b] - list the record's labels as text strings with enclosing pipes for those labels requiring pipes to be valid (those not requiring pipes are rendered without pipes and are identical to the corresponding [b]recordLabels[/b] items)
[b]recordValuesAsText[/b] - list of the text representations of the record's values corresponding to the labels in [b]recordLabels[/b] and [b]recordLabelsPiped[/b]
[b]recordValues[/b] - list of the record's original values corresponding to the labels in [b]recordLabels[/b] and [b]recordLabelsPiped[/b]
The handler uses the Unicode code point 60000 character in the formation of tokens. That character was chosen for its obscurity and low likelihood of appearing in an input record. In the rare circumstance where it does appear, the delimiting character may be changed to any non-conflicting character or character string of the user’s choice by changing the value of the util script’s tokenChar property.
EXAMPLES:
Vanilla Applescript record example:
(Includes labels consisting of Applescript reserved words and pipes without content, and labels and values containing “challenging” text elements: nested lists and records, double-quotes, curly braces, and ", " and “:” character sequences)
set theRecord to {name:1, item:2.2, ww:"foo", xx:true, yy:(current date), zz:{2, {aa:{3, {4, 5}, {bb:{6, {7, 8}, {cc:9, dd:10}}, ee:11}}, ff:12}}, |""|:true, ||:false, |1 " 2 " 3|:true, |, |:", ", |:|:":", |{a:1, b:2, c:3}|:"{a:1, b:2, c:3}"}
tell recordLabelsAndValues(theRecord)
its recordLabels --> {"name", "item", "ww", "xx", "yy", "zz", "\"\"", "", "1 \" 2 \" 3", ", ", ":", "{a:1, b:2, c:3}"}
its recordLabelsPiped --> {"name", "item", "ww", "xx", "yy", "zz", "|\"\"|", "||", "|1 \" 2 \" 3|", "|, |", "|:|", "|{a:1, b:2, c:3}|"}
its recordValuesAsText --> {"1", "2.2", "\"foo\"", "true", "date \"[...date...]\"", "{2, {aa:{3, {4, 5}, {bb:{6, {7, 8}, {cc:9, dd:10}}, ee:11}}, ff:12}}", "true", "false", "true", "\", \"", "\":\"", "\"{a:1, b:2, c:3}\""}
its recordValues --> {1, 2.2, "foo", true, date "[...date...]", {2, {aa:{3, {4, 5}, {bb:{6, {7, 8}, {cc:9, dd:10}}, ee:11}}, ff:12}}, true, false, true, ", ", ":", "{a:1, b:2, c:3}"}
end tell
Application record example:
tell application "Finder" to set theRecord to properties of desktop
tell recordLabelsAndValues(theRecord) -->
its recordLabels --> {"class", "name", "index", "displayed name", "name extension", "extension hidden", "container", "disk", "position", "desktop position", "bounds", "kind", "label index", "locked", "description", "comment", "size", "physical size", "creation date", "modification date", "icon", "URL", "owner", "group", "owner privileges", "group privileges", "everyones privileges", "container window"}
its recordLabelsPiped --> {"class", "name", "index", "displayed name", "name extension", "extension hidden", "container", "disk", "position", "desktop position", "bounds", "kind", "label index", "locked", "description", "comment", "size", "physical size", "creation date", "modification date", "icon", "URL", "owner", "group", "owner privileges", "group privileges", "everyones privileges", "container window"}
its recordValuesAsText --> {"desktop-object", "\"Desktop\"", "missing value", "\"Desktop\"", "\"\"", "false", "folder \"[...username...]\" of folder \"Users\" of startup disk of application \"Finder\"", "startup disk of application \"Finder\"", "{-1, -1}", "missing value", "{-33, -33, 31, 31}", "\"Desktop\"", "0", "false", "missing value", "\"\"", "4.3351161719E+10", "4.3355721728E+10", "date \"[...date...]\"", "date \"[...date...]\"", "missing value", "\"file:///Users/[...username...]/Desktop/\"", "\"[...username...]\"", "\"(unknown)\"", "read write", "none", "none", "window of desktop of application \"Finder\""}
its recordValues --> {desktop-object, "Desktop", missing value, "Desktop", "", false, folder "[...username...]" of folder "Users" of startup disk of application "Finder", startup disk of application "Finder", {-1, -1}, missing value, {-33, -33, 31, 31}, "Desktop", 0, false, missing value, "", 4.3351161719E+10, 4.3355721728E+10, date "[...date...]", date "[...date...]", missing value, "file:///Users/[...username...]/Desktop/", "[...username...]", "(unknown)", read write, none, none, window of desktop of application "Finder"}
end tell
HANDLER:
on recordLabelsAndValues(theRecord)
-- Returns the unpiped and piped forms of a record's labels and the text and value forms of a record's values
-- Utility properties and handlers
script util
property tokenChar : character id 60000 -- obscure character chosen for the low likelihood of its appearance in an input record's text representation; this may be substituted by any character or character string that does not appear in the text representation of the input record
property tokenizedStrings : {}
on detokenizeString(tokenizedString)
-- Convert any tokens of the form "[token char][token index number][token char]" to their original values; handle nested tokens with recursive handler calls
set tid to AppleScript's text item delimiters
try
set AppleScript's text item delimiters to my tokenChar
tell (get tokenizedString's text items)
if length < 3 then
set originalString to tokenizedString
else
set originalString to (item 1) & my detokenizeString((my tokenizedStrings's item ((item 2) as integer)) & (items 3 thru -1))
end if
end tell
end try
set AppleScript's text item delimiters to tid
return originalString
end detokenizeString
on representValueAsText(theValue)
-- Parse a forced error message for the text representation of the input value
try
|| of {theValue}
on error m
try
if m does not contain "{" then error
repeat while m does not start with "{"
set m to m's text 2 thru -1
end repeat
if m does not contain "}" then error
repeat while m does not end with "}"
set m to m's text 1 thru -2
end repeat
if m = "{}" then error
set valueAsText to m's text 2 thru -2
on error
try
-- Try an alternative method of generating a text representation from a forced error message if the first method fails
{||:{theValue}} as null
on error m
try
if m does not contain "{" then error
repeat while m does not start with "{"
set m to m's text 2 thru -1
end repeat
set m to m's text 2 thru -1
repeat while m does not start with "{"
set m to m's text 2 thru -1
end repeat
if m does not contain "}" then error
repeat while m does not end with "}"
set m to m's text 1 thru -2
end repeat
set m to m's text 1 thru -2
repeat while m does not end with "}"
set m to m's text 1 thru -2
end repeat
if m = "{}" then error
set valueAsText to m's text 2 thru -2
on error
error "Can't get a text representation of the value."
end try
end try
end try
end try
return valueAsText
end representValueAsText
end script
-- Perform the handler's actions inside a try block to capture any errors; if an error is encountered, restore AppleScript's text item delimiters to their baseline value
set tid to AppleScript's text item delimiters
try
-- Handle the special case of an empty record
if theRecord = {} then return {recordLabels:{}, recordLabelsPiped:{}, recordValuesAsText:{}, recordValues:{}}
-- Get the text representation of the input record
set textValue to util's representValueAsText(theRecord)
-- Partially validate the text representation, and remove the leading and trailing curly braces
tell textValue
-- This test does not exclude an input argument in the form of an Applescript list; however, a list will result in a parsing error in the code below
if (it does not start with "{") or (it does not end with "}") then error "The input value is not a record."
set textValue to text 2 thru -2
end tell
-- Initialize return values and the token counter
set {recordLabels, recordLabelsPiped, recordValuesAsText, recordValues, iToken} to {{}, {}, {}, {}, 0}
-- Tokenize text elements that could potentially contain record property (", ") or label/value (":") delimiter characters in order to avoid errors while parsing record properties
-- Tokens are of the form "[token char][token index number][token char]", where the index numbers increase sequentially 1, 2, 3, ... with each new token
-- Tokenize escaped double-quote characters (to facilitate tokenizing double-quoted items)
set AppleScript's text item delimiters to "\\\""
tell (get textValue's text items)
if length > 1 then
set iToken to iToken + 1
set AppleScript's text item delimiters to (util's tokenChar) & iToken & (util's tokenChar)
set {textValue, end of util's tokenizedStrings} to {it as text, "\\\""}
end if
end tell
-- Tokenize double-quoted items
set AppleScript's text item delimiters to "\""
tell (get textValue's text items)
if length > 2 then
set textValue to ""
repeat with i from 1 to (length - 2) by 2
set iToken to iToken + 1
set {textValue, end of util's tokenizedStrings} to {textValue & (item i) & (util's tokenChar) & iToken & (util's tokenChar), "\"" & item (i + 1) & "\""}
end repeat
set textValue to textValue & item -1
end if
end tell
-- Tokenize piped items
set AppleScript's text item delimiters to "|"
tell (get textValue's text items)
if length > 2 then
set textValue to ""
repeat with i from 1 to (length - 2) by 2
set iToken to iToken + 1
set {textValue, end of util's tokenizedStrings} to {textValue & (item i) & (util's tokenChar) & iToken & (util's tokenChar), "|" & item (i + 1) & "|"}
end repeat
set textValue to textValue & item -1
end if
end tell
-- Tokenize curly-braced items
set AppleScript's text item delimiters to "{"
tell (get textValue's text items)
if length > 1 then
set {AppleScript's text item delimiters, textValue, iNestedLevel, currBracedString} to {"}", item 1, 1, ""}
repeat with s in rest
tell (get s's text items) to if length > 1 then set {t, iNestedLevel} to {it, iNestedLevel + 1 - length}
if iNestedLevel > 0 then
set {iNestedLevel, currBracedString} to {iNestedLevel + 1, currBracedString & s & "{"}
else
set iToken to iToken + 1
set {textValue, iNestedLevel, currBracedString, end of util's tokenizedStrings} to {textValue & (util's tokenChar) & iToken & (util's tokenChar) & t's item -1, 1, "", "{" & currBracedString & (t's items 1 thru -2) & "}"}
end if
end repeat
end if
end tell
-- Get the unpiped and piped record labels and text representations of the record values
set AppleScript's text item delimiters to ", " -- delimits record properties
tell (get textValue's text items)
set AppleScript's text item delimiters to ":" -- delimits the label and value for a given record property
repeat with s in it
-- Extract and detokenize the current record label and value
tell (get s's text items) to set {end of recordLabelsPiped, end of recordValuesAsText} to {util's detokenizeString(item 1), util's detokenizeString(item 2)}
-- Create an unpiped version of the current record label
tell recordLabelsPiped's item -1
if it = "||" then
set end of recordLabels to ""
else if (it starts with "|") and (it ends with "|") then
set end of recordLabels to text 2 thru -2
else
set end of recordLabels to it
end if
end tell
end repeat
end tell
-- Get the record values
set recordValues to theRecord as list
on error m number n
set AppleScript's text item delimiters to tid
if n = -128 then error number -128
if n ≠-2700 then set m to "(" & n & ") " & m -- -2700 = purposely thrown error
error "Handler recordLabelsAndValues error:" & return & return & m
end try
set AppleScript's text item delimiters to tid
-- Return the results
return {recordLabels:recordLabels, recordLabelsPiped:recordLabelsPiped, recordValuesAsText:recordValuesAsText, recordValues:recordValues}
end recordLabelsAndValues
Edit note: The original submission mistakenly described the record in the application record example above as a pseudo-record. It is in fact a conventional record but simply has overridden Applescript’s class property. The term “pseudo” has been removed.
It only handles top-level labels in nested records, but that’s probably best anyway.