/*

strip.d
From the dig library by Burton Radons, http://www.opend.org/dig/index.html

Author:       Burton Radons (and J C Calvarese, )
License:      Public Domain
Purpose:      Strip all whitespace from D source files.

Slightly modified by:  J C Calvarese, http://jcc_7.tripod.com/d/

Suggested compilation options:
dmd -unittest -version=dstripcopy -O -inline strip.d


Usage Example:
strip ..\..\win32\*.d dest (Note: the "dest" directory must already exist.)

*/

import std.ctype, std.file;
import std.string; /* for memmove */
import std.c.windows.windows; /* for WIN32_FIND_DATA */


char spaceCharacter = '\n';

/* Strip comments from a block of code. */
char [] dstripComments (char [] data)
{
    char* s = data, e = s + data.length;
    char* b = data, o = s;

    bit start(char[] string)
    {
        if (s > e - string.length)
            return false;
        if (s [0..string.length] != string)
            return false;
        s += string.length;
        return true;
    }

    bit match(char[] string)
    {
        if (s > e - string.length)
            return false;
        if (s [0..string.length] != string)
            return false;
        s += string.length;
        return true;
    }

    bit skipString()
    {
        if (match ("\""))
        {
            *o ++ = '\"';
            while (s < e)
            {
                *o ++ = *s ++;
                if (s [-1] == '\\')
                    *o ++ = *s ++;
                else if (s [-1] == '\"')
                    break;
            }
        }
        else if (match ("\'"))
        {
            *o ++ = '\'';
            while (s < e)
            {
                *o ++ = *s ++;
                if (s [-1] == '\'')
                    break;
            }
        }
        else
            return false;
        return true;
    }

    while (s < e)
    {
        if (skipString ())
            continue;
        if (start ("/*"))
        {
            while (s < e - 1)
            {
                if (*s == '*' && s [1] == '/')
                {
                    s += 2;
                    break;
                }
                else
                    s ++;
            }
        }
        else if (start ("//"))
        {
            while (s < e)
            {
                if (*s == '\n')
                {
                    s ++;
                    break;
                }
                else
                    s ++;
            }
        }
        else if (start ("/+"))
        {
            int depth = 1;

            while (s < e)
            {
                if (match ("/+"))
                    depth ++;
                else if (match ("+/"))
                {
                    depth --;
                    if (depth == 0)
                        break;
                }
                else
                    s ++;
            }
        }
        else
            *o ++ = *s ++;
    }

    return data [0..cast (int) (o - b)];
}

/* Strip lines with no content from the code. */
char[] dstripEmptyLines(char[] data)
{
    char* s = data, e = s + data.length;
    char* b = data, o = s;

    void remove ()
    {
        memmove (o, s, cast (int) (e - s));
        data = data [0 .. data.length - cast (int) (s - o)];
        e -= cast (int) (s - o);
        s = o;
    }

    bit iswhite = true;

    while (s < e)
    {
        s ++;
        if (s [-1] == '\n')
        {
            if (iswhite)
                remove ();
            iswhite = true;
            o = s;
        }
        else if (!isspace (s [-1]))
        {
            if (iswhite)
            {
                s --;
                remove ();
                s ++;
            }
            iswhite = false;
        }
    }

    return data;
}

/* Strip function bodies. */
char[] dstripBodies (char [] data)
{
    char* s = data, e = s + data.length;
    char* b = data, o = s;

    void copy ()
    {
        *o ++ = *s ++;
    }

    void skipSpaces ()
    {
        while (s < e && isspace (*s))
            s ++;
    }

    void cskipSpaces ()
    {
        while (s < e && isspace (*s))
            copy ();
    }

    bit isIdStart (char ch) { return isalpha (ch) || ch == '_'; }
    bit isId (char ch) { return isalnum (ch) || ch == '_'; }

    bit start (char [] string)
    {
        if (s > e - string.length)
            return false;
        if (s [0 .. string.length] != string)
            return false;
        if (isIdStart (s [0]) && s != e - string.length && isId (s [string.length]))
            return false;
        //o = s;
        s += string.length;
        return true;
    }

    bit match (char [] string)
    {
        if (s + string.length > e)
            return false;
        if (s [0 .. string.length] != string)
            return false;
        if (isIdStart (s [0]) && s != e - string.length && isId (s [string.length]))
            return false;
        s += string.length;
        return true;
    }

    bit cmatch (char [] string)
    {
        if (!match (string))
            return false;
        o [0 .. string.length] = string;
        o += string.length;
        return true;
    }

    void remove()
    {
        /*memmove (o, s, cast (int) (e - s));
        data = data [0 .. data.length - cast (int) (s - o)];
        e -= cast (int) (s - o);
        s = o;*/
    }

    bit skipString()
    {
        if (match ("\""))
        {
            while (s < e)
            {
                s ++;
                if (s [-1] == '\\')
                    s ++;
                else if (s [-1] == '\"')
                    break;
            }
        }
        else if (match ("\'"))
        {
            while (s < e)
            {
                s ++;
                if (s [-1] == '\'')
                    break;
            }
        }
        else
            return false;
        return true;
    }

    bit cskipString()
    {
        char* p = s;

        if (!skipString ())
            return false;
        o [0 .. (int) (s - p)] = p [0 .. (int) (s - p)];
        o += (int) (s - p);
        return true;
    }

    void removeBody(int depth)
    {
        if (o [-1] == '\n')
        {
            o [-1] = ';';
            *o ++ = '\r';
        }
        else
            *o ++ = ';';

        while (s < e)
        {
            if (skipString())
                continue;
            if (match ("{"))
                depth ++;
            else if (match ("}"))
            {
                depth --;
                if (depth == 0)
                    break;
            }
            else
                s ++;
        }

        remove ();
    }

    void skipInitializer(char closea, char closeb)
    {
        while (s < e)
        {
            copy ();
            if (s [-1] == '[')
                skipInitializer (']', 0);
            else if (s [-1] == '{')
                skipInitializer ('}', 0);
            else if (s [-1] == '(')
                skipInitializer (')', 0);
            else if (s [-1] == closea || s [-1] == closeb)
                break;
        }
    }

    void scan(bit inner)
    {
    outer:
        while (s < e)
        {
            char *p = s;

            if (skipString ())
                continue;
            else if (start ("in") || start ("body"))
            {
                skipSpaces ();
                if (!match ("{")) continue;
                removeBody (1);
            }
            else if (start ("out"))
            {
                skipSpaces ();
                if (!match ("(")) continue;
                while (s < e && *s != ')')
                    s ++;
                if (!match (")")) continue;
                skipSpaces ();
                if (!match ("{")) continue;
                removeBody (1);
            }
            else if (cmatch ("="))
                skipInitializer (';', ',');
            else if (cmatch ("template"))
            {
                int depth = 0;

                while (s < e)
                {
                    if (cmatch (";") && depth == 0)
                        break;
                    else if (cskipString ())
                        continue;
                    else if (cmatch ("{"))
                        depth ++;
                    else if (cmatch ("}"))
                    {
                        depth --;
                        if (depth == 0)
                            break;
                    }
                    else
                        copy ();
                }
            }
            else if (start ("static"))
            {
                skipSpaces ();
                if (match ("this") || (match ("~") && match ("this")))
                    removeBody(0);
                else
                {
                    for (int c; c < (int) (s - p); c ++)
                        *o ++ = p [c];
                    if (cmatch ("{"))
                        scan (true);
                }
            }
            else if (cmatch ("public") || cmatch ("private") || cmatch ("protected")
                || cmatch ("final") || cmatch ("override")
                || cmatch ("deprecated"))
            {
                cskipSpaces ();
                if (cmatch ("{"))
                    scan (true);
            }
            else if (start ("unittest"))
            {
                skipSpaces ();
                match ("{");
                removeBody (1);
            }
            else if (cmatch ("version"))
            {
                skipSpaces ();
                if (!cmatch ("(")) continue;
                while (s < e && *s != ')') copy ();
                if (!cmatch (")")) continue;
                skipSpaces ();
                if (!cmatch ("{")) continue;
                scan (true);
                cskipSpaces ();
                if (cmatch ("else"))
                {
                    skipSpaces ();
                    if (cmatch ("{"))
                        scan (true);
                }
            }
            else if (cmatch ("extern") || cmatch ("export"))
            {
                skipSpaces ();
                if (!cmatch ("(")) continue;
                while (s < e && *s != ')') copy ();
                if (!cmatch (")")) continue;
                skipSpaces ();
                if (!cmatch ("{")) continue;
            }
            else if (cmatch ("class") || cmatch ("struct") || cmatch ("union") || cmatch ("enum") || cmatch ("interface"))
            {
                while (s < e)
                {
                    if (cmatch ("{"))
                    {
                        scan (true);
                        break;
                    }
                        
                    if (cmatch (";"))
                        break;
                    copy ();
                }
            }
            else if (start ("{"))
                removeBody (1);
            else if (inner && cmatch ("}"))
                return;
            else if (isIdStart (*s))
            {
                copy ();
                while (s < e && isId (*s))
                    copy ();
            }
            else if (cmatch ("("))
            {
                int depth = 1;

                while (s < e)
                {
                    if (*s == '(')
                        depth ++;
                    else if (*s == ')')
                        depth --;
                    copy ();
                    if (depth == 0)
                        break;
                }
            }
            else
               copy ();
        }
    }

    scan (false);
    return data [0 .. (int) (o - b)];
}

/* Strip all possible whitespace from the text. */
char[] dstripWhitespace(char[] data)
{
    char* s = data, e = s + data.length;
    char* o = s;

    bit skipString ()
    {
        if (*s == '\"')
        {
            *o ++ = *s ++;
            while (s < e)
            {
                *o ++ = *s ++;
                if (s [-1] == '\\')
                    *o ++ = *s ++;
                else if (s [-1] == '\"')
                    break;
            }
        }
        else if (*s == '\'')
        {
            *o ++ = *s ++;
            while (s < e)
            {
                *o ++ = *s ++;
                if (s [-1] == '\'')
                    break;
            }
        }
        else
            return false;
        return true;
    }

    while (s < e)
    {
        if (skipString ())
            continue;
        if (isspace (*s))
        {
            if (s === data)
            {
                s ++;
                while (s < e && isspace (*s))
                    s ++;
            }
            else if (isalnum (s [-1]) || s [-1] == '_')
            {
                *o ++ = spaceCharacter;
                while (s < e && isspace (*s))
                    s ++;
                if (!isalnum (*s) && *s != '_')
                    o --;
            }
            else
            {
                while (s < e && isspace (*s))
                    s ++;
            }
        }
        else
            *o ++ = *s ++;
    }

    return data[0..cast (int) (o - (char *) data)];
}

/* Strip comments and function bodies from a block of code. */
char[] dstrip (char [] data)
{
    data = dstripComments (data);
    data = dstripBodies (data);
    data = dstripWhitespace (data);
    return data;
}

private void dstripTest(char[] input, char[] output)
{
    char [1] space;
    char [] initial = input.dup;

    space [0] = spaceCharacter;
    output = replace (output, " ", space);
    if (dstrip (input) != output)
        throw new Error ("dstrip (\"" ~ initial ~ "\") should return \"" ~ replace (output, space, " ") ~ "\", but returned \"" ~ replace (dstrip (input), space, " ") ~ "\".");
}

unittest
{
    dstripTest ("version (xyz) { struct { foo } } else { ... }", "version(xyz){struct{foo}}else{...}");
    dstripTest ("void foo () body { }", "void foo();");
    dstripTest ("void foo () { }", "void foo();");
    dstripTest ("template Bar(T) { void foo (int z, int y) { carp () ; } }", "template Bar(T){void foo(int z,int y){carp();}}");

    dstripTest ("extern (C) { foo }", "extern(C){foo}");
    dstripTest ("public static { }", "public static{}");
    dstripTest ("version = foobar;", "version=foobar;");
    dstripTest ("version (Fopo) { ... }", "version(Fopo){...}");
    dstripTest ("version (Fopo) { ... }else{}", "version(Fopo){...}else{}");
    dstripTest ("static this() { }", ";");
    dstripTest ("interface Foo { }", "interface Foo{}");
    dstripTest ("struct x { enum { } void foo () { } }", "struct x{enum{}void foo();}");
    dstripTest ("enum MB { OK = 1 << 0, } char [] messageBox (char [] title, char [] message, MB flags) { }", "enum MB{OK=1<<0,}char[]messageBox(char[]title,char[]message,MB flags);");
}

bit readWildcard (inout char [] [] files, char [] arg)
{
    WIN32_FIND_DATA data;
    HANDLE handle;

    arg = replace (arg, "/", "\\");

    if (std.string.find (arg, '?') < 0 && std.string.find (arg, '*') < 0)
    {
        files ~= arg;
        return true;
    }

    handle = FindFirstFileA (arg, &data);
    if (handle == (HANDLE) 0)
    {
        printf ("couldn't find files matching '%.*s'\n", arg);
        return false;
    }

    while (1)
    {
        char [] path;
        
        for (int d = arg.length - 1; d >= 0; d --)
            if (arg [d] == '\\')
            {
                path = arg [0 .. d + 1];
                break;
            }

        char [] filename = path ~ data.cFileName [0 .. strlen (data.cFileName)];

        files ~= filename.dup;
        if (!FindNextFileA (handle, &data))
            break;
    }

    FindClose (handle);
    return true;
}

version (dstripcopy)
{
    import std.path, std.string;

    int main (char [] [] args)
    {
        bit showHelp = false;
        char[][] files;
        char[] dest;

        if (args.length < 3)
            showHelp = true;

        dest = args [args.length - 1];
        for (int c = 1; c < args.length - 1; c ++)
            if (!readWildcard(files, args[c]))
                showHelp = true;

        if (showHelp)
        {
			int p;
			char[] program;
			p = rfind(args[0], '\\');
			if(p > -1)
				program = args[0][p+1..args[0].length];
			else
				program = args[0];
	
            printf ("\n%.*s FILES... DESTINATION\n"
                "\n"
                "Copy the D files to the destination, stripping out all whitespace from them.\n", program); //args [0]);
            return 1;
        }

        printf ("%.*s\n", dest);
        for (int c; c < files.length; c ++)
        {
            char [] file = files [c];
            char [] fileResult = std.path.join (dest, getBaseName (file));

            write (fileResult, (byte []) dstrip ((char []) read (file)));
            printf ("    %.*s\n", file);
        }

        return 0;
    }
}