xplorer˛ Quick Start Guide
File exploring reinvented: Feel like home miles away from home

[PRO] Advanced programming and scripting

One step up the ladder from DOS scripts, come user interface macros and file property algebra. These advanced topics are closer to computer programming than casual file management, but they don't require a degree in computer science to take advantage of. If you are used to MS Excel cell expressions, you should get the hang of it quickly. Note that xplorer˛ doesn't offer a fully scriptable programming model with file and user interface objects, but a simpler macro language for user interface elements, that should be within the reach of xplorer˛ medium-to-advanced users. As for the programmers amongst you, I refer you to the windows scripting host for the full monty <g>

User interface automation with macros macros for UI automation
TIP: There is a much easier way to automate xplorer˛ with macros, using the macro recorder.

A simple in-app "programming language" allows you to automate many program tasks. It is not a full blown scripting language (as e.g. the one in MS Excel) but it is surprisingly versatile and easy to use. You write scripts, that is sequences of simple text commands that manipulate the program window as if a robot was in control. You can do these sorts of tasks:

All in all you can do whatever any user could do using the keyboard and the mouse to operate the program, only do it in a tightly controlled fashion. You can string commands together that operate on a series of dialogs to achieve complex results. Here is an example script:

macro editor
Figure 20a. Macro editor: compose new macro

COMMAND "copy to"
WAITWND
FINDWND filter
FINDCLASS combobox
SETTEXT *.pdf
OK

As you can tell from the command names, this macro is equivalent to clicking on the Copy To toolbar button, waiting for the dialog to appear, then finding the Filter box and setting its value to "*.pdf" ; and finally "clicking" the OK button to start copying the files with the selected options (only PDF files will be copied).

Press <F12> key to start the macro editor (or use Customize | Macros | Edit macro menu), then copy/paste the above script, then click Run button to see it working. When a macro is running you cannot work with the main program window. A macro control window appears on top of other windows that shows the progress of the execution. To abort a macro mid-way through its execution click on the Cancel button.

NOTE: While a macro is running you should not switch to another window or application, otherwise the macro will probably fail. Nor should you interfere with random clicks or keypresses. Just hang in there till the macro is finished!

Macro command syntax

Each line of a macro contains just one command and its arguments, if any. For example:

COMMAND "copy to" # start the copy to dialog

COMMAND is the reserved command name that tells the macro what to do, "copy to" is the argument (which program command to execute), and the rest is the comment. Comments are ignored by the script but help you to make sense what each command is doing. Inline comments begin with the hash # symbol.

Arguments are necessary for many commands. They are usually constant text strings as the name of the toolbar button you are trying to click. Use "quotes" whenever you have spaces in arguments (many words). Arguments can also be numbers e.g. 32793 is the equivalent identifier for "Copy to" toolbar button (use Help | Command finder menu to find numeric equivalents to commands). Most of the time it will be easier to use the text label, but sometimes there will be no alternative than to use the number, e.g. for hidden commands. Arguments can also be variables, which will be discussed below.

NOTE: Commands should be searched as they appear in xplorer˛ menus or toolbar. If you are using a translation, search commands in your native tongue.

Many commands return a value, e.g. FINDWND returns the handle to the window found. You don't see this result immediately but it is stored in the last result register ($0). Subsequent macro commands may rely on the previous result to function properly, e.g. CLICK needs a window handle found earlier. The last result is piped from one command to its successor automatically.

9 elementary variables are supported to be used as arguments, called $1 to $9. Most of the time you will use $0 if you want to visualize the result of the last command (MESSAGE $0).

Finding windows and menu commands

The most important operations you will be using all the time is finding control windows (buttons, edit controls etc) to operate, and issuing message commands (using menu or toolbars) to the program — using FINDWND and COMMAND respectively.

If you are familiar with windows programming, you will know that each menu command and toolbar button appearing in a program's user interface correspond to a message number. Likewise, controls that appear in dialogs have an identifier number. For instance 32793 is the message number corresponding to "Copy to" toolbar button; and the OK button in most dialogs translates to number 1. Unless you are a big time geek <g> that can use Resource hacker or Spy++ you cannot be expected to know these numbers!

Laymen needn't worry. Instead of fumbling with cryptic numbers, you can use the text that appears on a menu, toolbar or dialog button to tell the macro which command you are after. Take note which text appears on the menu command you want to issue, and use it as the argument to COMMAND. For toolbars use the tooltip if the button you are after has no text showing. It is best to use the full command text as it appears in the GUI, which will ensure that the intended command will be issued. Partial names can be used too, but keep in mind that the first matching name will be used, which may be some other command or window!

TIP: When troubleshooting macros, insert a MESSAGE $0 instruction under a FINDWND or COMMAND. This will show you which window or command was found ($0 is the last result variable). Once you are done debugging you can delete or comment out these MESSAGE $0 instructions. Alternatively you can run the macro in debug mode where you single step the execution and see detailed information about each command, see which window is operated upon etc.

Finding buttons on dialogs is easy, just use the name of the button with FINDWND. Other controls have some descriptive text label before them, which may be used to locate the control. Sometimes a control has no useful text or label; in these cases locate the control nearest to the unlabeled one, then move forward or backward with NEXTWND or PREVWND instructions till you hit the desired dialog control. The directions "forward" and "backward" correspond to the tab order in the dialog — or in plain english, what happens when you press <TAB> key to move around in a dialog window.

TIP: When searching for commands and windows, keep in mind that the search is within the active window. If no dialogs show, that means the main program window; if a dialog is showing, the search is within the dialog controls. Use the wait instructions (e.g. WAITWND) to make sure the window you want to manipulate has had time to show up.

Dealing with syntax and execution errors

When you first start working with macros, inevitably you will make mistakes. For example you can misspell an instruction name or forget a necessary argument. In that case a syntax error will result and the macro editor will highlight the token that caused the trouble. Here are some pointers to keep in mind:

Once the basic syntax is correct, you can try running the macro. Here you may encounter subtle problems that will cause execution errors. The macro will stop, highlight the command that faulted and show an error message. Here are the runtime error codes explained:

Error code 2 (wrong argument) is raised also when an instruction cannot find the correct last result from the previous command. For example you cannot use CLICK without a preceding FINDWND. Most commands that operate on dialog controls require a window handle in the last result $0.

When you get an unexplained error 3 (not found), it could be that the dialog didn't have time to appear. If you use COMMAND and expect to operate the dialog controls, use WAITWND to make sure that the dialog is visible before you start manipulating its controls. FINDWND instruction is searching the dialog that is currently active (excluding the macro dialog itself).

Finally you may realize that a running macro is doing crazy things that were not intended. In that case click on Cancel button to abort the macro execution.

Conditional execution

By default all execution errors are terminal and stop the macro dead. However you may want to take responsibility and manage the execution despite previous errors. For example, if a control cannot be found you may want to try to search for another. If you want to ignore all errors, use the instruction: ERRORSTOP 0.

Once you turn off the automatic error handling, the last result $0 variable will reflect the execution status of the last instruction. You can test for success using the JUMPOK instruction, which will jump to a label (i.e. move the point of execution) only if everything goes well — otherwise the instruction immediately below it will execute. Here's an example that looks for a window called test and stops execution manually if not found:

ERRORSTOP 0
FINDWND test
JUMPOK found
MESSAGE "window not found"
EXIT
LABEL found
CLICK

Using conditional jumps and GOTO instructions you can modify the order that script lines execute. Instead of linearly executing all lines in sequence, you can jump to earlier or later lines depending on your logic. Jump destination points (LABEL) can be any text you like to make sense of your script. If you have multiple jump points, make sure each label gets a unique name.

Simple loops are also possible using a control variable as a countdown counter, as such:

POP 5 # repeat 5 times
STORE $1
LABEL repeat
MESSAGE $1
DECR $1
JUMPOK repeat
# end of loop

The trick is that DECR instruction updates (decreases by 1) the counter variable $1 and the last result, so when it reaches zero, you break the loop (JUMPOK fails). Also note the roundabout way to initialize the counter variable using POP then STORE.

Macro command reference

The macro composition dialog has a listbox with the most common commands so you don't need to remember the names. Below you will find the full list of supported macro instructions and their expected arguments.

A typical instruction accepts a single argument, but there are some that apparently work on thin air without arguments. This is because they use the result of the previous command as their argument — this is called pipe argument. For example CLICK operates on the result of an earlier FINDWND command. If a command fails the result is zero (void).

Arguments can be either constants (strings or numbers) or variables ($0 to $9). Variables represent strings, numbers or window handles, depending on what is stored in them. The last result is kept in variable $0. Unless otherwise stated, the following table uses name for either a string or number, furnished either as a constant or variable.

Table 2. Macro instruction reference
Instruction Argument Pipe arg. Result Description
FINDWND name window Find a window by name on it or next to it (if labelled). Searching for part of the name is ok if you are sure that it will uniquely match the window. Also finds by window numeric identifier. The search is in the topmost active dialog or the main window. FINDWND 0 returns the currently active dialog.
FINDCLASS name window window Find a window by its exact class name (e.g. edit, button etc). If the prior result is a window, it finds the next window of the same class, if any, possibly cycling from the top.
FINDFRAME title window Find a top level window by its titlebar text and switch to it (make it active). The window must be selectable, e.g. not blocked by a modal dialog.
GETFOCUS window Returns the active window (where you last clicked).
SETFOCUS window Activates the window
NEXTWND window window Returns the next window in the tab order
PREVWND window window Returns the previous window in the tab order
CHILDWND window window Returns the first child window (of whatever was found earlier)
PARENTWND window window Returns the parent window
CLICK window Clicks on the window found earlier. May fail if the button is disabled
CLICKDEF name window Executes the default accessibility command for child name (if using a number, start from 1). This is for non-standard buttons that don't have window handles (exposed through MSAA). For many standard controls it's equivalent to double-click on a list item.
SETTEXT name window Sets the window text to name. Only meant for edit controls and combo boxes.
GETTEXT $var window Gets the text of the last result window and stores it in variable $0 - $9
GETCOUNT $var window Gets the number of items in listview control and stores it in variable $0 - $9
GETITEMTEXT index window text Extracts individual item text from windows like listview controls. Items should be specified by a number from 1 to GETCOUNT.
SELECT name window Selects an item in a combo, list or listview control either by name or by number (1 → max items present). SELECT 0 will unselect everything
GETSEL $var window Gets the selected item number in selectable controls and stores it in variable $0 - $9
SETCHECK 1/0 window Checks or clears a checkbox or radio button, if enabled. Special value 2 means flip previous state
OK Terminates the active dialog. It is a combination of FINDWND "ok" followed by CLICK. It may fail if the dialog won't close due to validation errors.
CANCEL Simulates pressing ESC key. If a dialog is active this will be equivalent to clicking on Cancel button

COMMAND name help text Issues a program command (WM_COMMAND) by number or by name. It is equivalent to using a menu command or toolbar button. Searching by name is more intuitive; it finds the command from a menu or toolbar that corresponds to the (partial) name. For toolbar buttons without names use the tooltip (popup) name. If the active window is a dialog, it finds the named button and simulates a click. If the command is disabled (e.g. ghosted toolbar button), it fails. If successful, returns the explanation text of a the command issued (use MESSAGE $0 to see it).
CMENU name window number Simulate a context menu (right click) on the window last found, and find the menu command that corresponds to name. Returns the command identifier of the menu item found. Can be used for the shell context menu or any other background menu
WAITWND Wait till the active window changes, e.g. a new dialog window opens
WAITIDLE Wait till the active window finishes processing the last command.
WAIT period Wait a fixed amount of time in milliseconds (1000 = 1 sec). The argument may also be a window handle in a variable, where it behaves like WAITIDLE for that particular window.

ERRORSTOP 1/0 Turns the "resume on error" policy on/off (default is to stop on any error).
LABEL name This instruction doesn't do anything but marks the destination of a jump/goto instruction
GOTO label Unconditional jump to label name
JUMPOK label any Conditional jump to label name. Checks the result of last command and if it is failure/empty/0 then does nothing. So it only jumps if the last command was successful
JUMPERR label any Conditional jump to label name. Checks the result of last command and only jumps if it was 0/failed/empty — the opposite of JUMPOK
EQUAL name any 0/1 Compares its argument with the last result. If the 2 values are compatible and equal, the result is 1, else 0
DECR $var $var-1 Decreases the variable ($1 - $9) by one and updates last result too. Used for loop control.
INCR $var $var+1 Adds 1 to a variable ($1 - $9) and updates last result too.
MATH expr number Calculate the expr argument and store the integer result. Allows performing simple integer arithmetic e.g. ($1+3)/2
EXIT Stops execution; any instructions below this are ignored
JSR label Unconditional jump to label name as a subroutine. There is no stack and no nesting is supported. Use regular $0-$9 variables for information exchange
RET code =code Resume execution at line following the last JSR instruction. Used to return from a "subroutine" with a code inserted in $0

MESSAGE name Pauses execution showing a message. Debugging aid mainly used to visualise the result of the last command as MESSAGE $0
ADDSTR name old old+ name Adds name to the old string result. If you need a newline use "\n". Intended for composing longer messages
CHOPSTR format full part Splits the last result string into tokens and returns one of them. The format argument includes the split character and the substring number to return. For example \2 would separate a path on \ points and return the second part; \t-1 would split on tab characters and return the last part (negative numbers count from the end)
REM any Comment line that doesn't do anything but explain something. For inline comments use #
INPUT prompt text Pops up an input dialog where you enter a string, which is then stored in the result. Prompt is used as the question to ask. Tip: use INPUT 0 to import the text from the clipboard.
STORE $var any Stores the last result in a variable $1 — $9 for future use
POP any =any Copies the constant or stored variable into the $0 register to be used in a subsequent command
SENDKEY keys window Send keyboard input to a window. E.g. CTRL+LEFT will send a combination of control and left arrow keys. Most special keys are recognized by name (see VK_xxx names), and for the rest you can use the equivalent virtual key code number, e.g F1=112 (from hex 0x70)
NOPOPUPS 1/0 Turn silent mode on/off, i.e. don't show any of the dialogs that would normally appear during macro execution. This mode is on by default if you repeat a macro.

Harnessing macros will take some practice. Below you will find some sample macros you can play with. Try running them in debug mode (line by line) to see what they do with xplorer˛. Observe the inline comments for extra explanations.

Click here to show sample macros

Automatic macro recorder

If you don't need the full power (and learning hassle) of the scripting language, you can record simple macros pressing <F12> key to start the macro recorder (or use Customize | Macros | Record macro menu command). Use the mouse and the keyboard to do what you like (issue commands, tweak dialog controls and buttons, etc), then click on FINISH button to stop recording. The equivalent script representing your actions will be generated automatically and presented to you, just as if you had typed in the instuctions.

The recorder will only generate plain macros, and the resulting code looks spaghetti, but it should be adequate for the average macro requirements — and you don't need to do any "programming". If you start the recorder and click Finish immediately, without any actions, you will be presented with the last macro used.

[ULT] Programmable column: file property algebra desktop detective

You may be familiar with DeskRule's desktop detective panel that enables searching with combinations of file properties and attributes. xplorer˛ (ultimate edition only) new programmable column is an extension to this concept: it allows you to combine file properties, boolean and arithmetic operators and a few functions into expressions, whose result is the value of the Programmable [S] stock column. This result can be then used as any other column, for display, sorting, filtering and searching for matching files. It is really versatile and the possible uses are too many to enumerate — just use your imagination!

programmable column definition
Figure 20b. Define programmable column expression

To define or change the programmable column, right click on the detailed view column header row and pick Programmable from the menu. This will allow you to change the expression and will show the programmable column too. The programmable column definition dialog has boxes that guide you to create expressions, and it also supports autocompletion of file property names for the more experienced users. If something goes wrong, it will pop a syntax error with the likely error location.

All file properties (columns) can be used in the calculations. The rules can also search in files (GREP function) and extract specific file bytes (FBYTES function). The result of the expression can be either text or number or even a date, depending on the properties and operators used. If you use conjunctions of boolean operators, the result will be 1 or 0, for true or false respectively.

NOTE: programmable expressions are not programs. The expression can be as long as you like but it is evaluated as a single line of "code". Think of cell expressions in MS Excel
=SUM(A1:A5)

You can save expressions you use frequently by supplying a name in the Predefined box in the above dialog and clicking Save button. This is similar to storing hyperfilters for search commands. Then expressions become available for future use from the history drop-down dialog control.

File attribute algebra

This dialog allows you to do amazing feats which would normally require you to write a program or advanced shell script. You can combine any file property with arithmetic (add/subtract/multiply etc), relational (equal, greater than etc) and boolean operators (and/or etc) to create arbitrary expressions that can be evaluated for each file. For example take the expression:

(${Modified} > ${Created} + 2) & ${Size} + 5KB < ${Size on disk}

It will match files whose modification date is later than the creation date + 2 days, and (&) their reported size plus 5 kilobytes is less than its total size on disk. When you become familiar with the "official" property names and operators, you can type such expressions manually (press <Ctrl+DownArrow> key to autocomplete properties), or you can use the dialog drop-down lists to help you compose a complex rule with properties, operators and functions. A few clarifications are in order:

Table 3. Supported helper functions
Function Description
num=LEN(str) Return the number of characters in string, e.g. len('abc') returns 3
substr=MID(str, nStart, nLen) Extract a substring out of a bigger text string, starting at position nStart and extracting nLen characters, for example mid('abc12', 2, 3) would return the string 'bc1'. Can be also used to access individual characters if nLen is 1.
str=TRIM(str) Eliminate whitespace around a string, if any. It cuts off any leading or trailing spaces, e.g. trim(' ab ') would return 'ab'
num=FIND(strHaystack, strNeedle) Find a needle in a haystack and return the position of the first occurrence — or 0 if nothing is found. find(abc12,bc) would return 2. The search isn't case sensitive.
num=DPART(date,str) Access parts of the date, e.g. dpart('march 2016', year) would return 2016. The second argument str can return other parts of the date e.g. month or day (or even just d).
xxx=MIN(xxx,xxx) Compare the two arguments and return the smaller one. This function accepts all kind of argument types (numbers, dates, etc) and returns the same kind. So min(-1,2) would return -1.
xxx=MAX(xxx,xxx) Similar to MIN function, this returns the larger of the 2 arguments
xxx=IF(term, if_true, if_false) Evaluates term and if it is true returns the second argument, else the third. The 3 arguments can be whatever makes sense for your conditional logic. term is true if it isn't empty or if it is non-zero.
num=ABS(num) Returns the absolute value of a number, e.g. abs(-1.3) would be 1.3
str=STR(xxx) Convert any type of argument into a string, useful for passing dates and numeric variables to other string functions. It can also convert hex strings obtained by FBYTES into plain text
num=GREP(str) Search file contents for occurrence of text (can be a wildcard). If successful a positive value is returned, else 0 if keyword is not present in file.
str=FBYTES(nStart,nLen) Extract nLen bytes from file, starting at file offset nStart (beginning=0). The returned string is the hexadecimal representation of the file bytes. For example PNG files can be identified because FBYTES(1,3)="50.4E.47" (use a hex file viewer to understand why)
num=INT(xxx) Convert any type of argument into a number; it can also convert hex strings from FBYTES if up to 4 bytes long.
substr=TOKEN(str, separ, nTok) Break up a string at separ points, and return the nTok entry (1=first token, -1=last). Special separator values are " " to break at spaces, \n which breaks at newlines and \t that splits on tabs. Similar in concept to CHOPSTR macro instruction. Example: token("hello world"," ",2) returns 'world'.
stdout=SYSTEM(strCmd) Execute DOS command strCmd and collect its console output. For example system('dir "' + ${path} + ${name} + '"') will execute the DIR command on each item (this example isn't particularly useful :). Note the use of 'single ears' to compose the command including the "double quotes" around the path name -- in case it contains spaces -- and the + operator used to compile the command string. With a little imagination it can be used as a poor man's (slow) column handler! Commands executed should be in %PATH% or full paths must be specified. External commands that don't write to a console (e.g. GUI programs) are not supported.

If there is some problem with the complex expression like a misspelled file property or an operation between incompatible arguments, xplorer˛ will report a syntax error, highlighting the trouble spot. Correct the typo and you are ready to search.

Sometimes a property name on its own can be a useful rule, e.g. ${Keywords} without any other operators will match files that have some tags defined. Or combine it with a leading NOT (!) operator to match files that do not have tags! Bare property names are also convenient for yes/no properties like ${IsFolder}, otherwise you must be careful to compare them only against zero (=false or no).

The programmable column is useful for many advanced search tasks, e.g. find pictures whose width/height ratio is 4/3. It will also allow general tomfoolery like (${Modified} - ${Date taken})/2 = ${Size}, which arguably is not much use <g>. A final note is that extracting this column is rather slow; the more properties you use in an expression, the slower it gets.




DOS commands Contents Customization

Download xplorer2



© 2002-2019 ZABKAT, All Rights Reserved
http://zabkat.com