dlls/shlwapi/url.c

   1 /*
   2  * Url functions
   3  *
   4  * Copyright 2000 Huw D M Davies for CodeWeavers.
   5  */
   6
   7 #include <string.h>
   8 #include <stdlib.h>
   9 #include "windef.h"
  10 #include "winnls.h"
  11 #include "winbase.h"
  12 #include "winerror.h"
  13 #include "wine/unicode.h"
  14 #include "wininet.h"
  15 #include "winreg.h"
  16 #define NO_SHLWAPI_STREAM
  17 #include "shlwapi.h"
  18 #include "debugtools.h"
  19 #include "ordinal.h"
  20
  21 DEFAULT_DEBUG_CHANNEL(shell);
  22
  23 typedef struct {
  24     LPCWSTR pScheme;      /* [out] start of scheme                     */
  25     DWORD   szScheme;     /* [out] size of scheme (until colon)        */
  26     LPCWSTR pUserName;    /* [out] start of Username                   */
  27     DWORD   szUserName;   /* [out] size of Username (until ":" or "@") */
  28     LPCWSTR pPassword;    /* [out] start of Password                   */
  29     DWORD   szPassword;   /* [out] size of Password (until "@")        */
  30     LPCWSTR pHostName;    /* [out] start of Hostname                   */
  31     DWORD   szHostName;   /* [out] size of Hostname (until ":" or "/") */
  32     LPCWSTR pPort;        /* [out] start of Port                       */
  33     DWORD   szPort;       /* [out] size of Port (until "/" or eos)     */
  34     LPCWSTR pQuery;       /* [out] start of Query                      */
  35     DWORD   szQuery;      /* [out] size of Query (until eos)           */
  36 } WINE_PARSE_URL;
  37
  38 typedef enum {
  39     SCHEME,
  40     HOST,
  41     PORT,
  42     USERPASS,
  43 } WINE_URL_SCAN_TYPE;
  44
  45 static const WCHAR fileW[] = {'f','i','l','e','\0'};
  46
  47 static const unsigned char HashDataLookup[256] = {
  48  0x01, 0x0E, 0x6E, 0x19, 0x61, 0xAE, 0x84, 0x77, 0x8A, 0xAA, 0x7D, 0x76, 0x1B,
  49  0xE9, 0x8C, 0x33, 0x57, 0xC5, 0xB1, 0x6B, 0xEA, 0xA9, 0x38, 0x44, 0x1E, 0x07,
  50  0xAD, 0x49, 0xBC, 0x28, 0x24, 0x41, 0x31, 0xD5, 0x68, 0xBE, 0x39, 0xD3, 0x94,
  51  0xDF, 0x30, 0x73, 0x0F, 0x02, 0x43, 0xBA, 0xD2, 0x1C, 0x0C, 0xB5, 0x67, 0x46,
  52  0x16, 0x3A, 0x4B, 0x4E, 0xB7, 0xA7, 0xEE, 0x9D, 0x7C, 0x93, 0xAC, 0x90, 0xB0,
  53  0xA1, 0x8D, 0x56, 0x3C, 0x42, 0x80, 0x53, 0x9C, 0xF1, 0x4F, 0x2E, 0xA8, 0xC6,
  54  0x29, 0xFE, 0xB2, 0x55, 0xFD, 0xED, 0xFA, 0x9A, 0x85, 0x58, 0x23, 0xCE, 0x5F,
  55  0x74, 0xFC, 0xC0, 0x36, 0xDD, 0x66, 0xDA, 0xFF, 0xF0, 0x52, 0x6A, 0x9E, 0xC9,
  56  0x3D, 0x03, 0x59, 0x09, 0x2A, 0x9B, 0x9F, 0x5D, 0xA6, 0x50, 0x32, 0x22, 0xAF,
  57  0xC3, 0x64, 0x63, 0x1A, 0x96, 0x10, 0x91, 0x04, 0x21, 0x08, 0xBD, 0x79, 0x40,
  58  0x4D, 0x48, 0xD0, 0xF5, 0x82, 0x7A, 0x8F, 0x37, 0x69, 0x86, 0x1D, 0xA4, 0xB9,
  59  0xC2, 0xC1, 0xEF, 0x65, 0xF2, 0x05, 0xAB, 0x7E, 0x0B, 0x4A, 0x3B, 0x89, 0xE4,
  60  0x6C, 0xBF, 0xE8, 0x8B, 0x06, 0x18, 0x51, 0x14, 0x7F, 0x11, 0x5B, 0x5C, 0xFB,
  61  0x97, 0xE1, 0xCF, 0x15, 0x62, 0x71, 0x70, 0x54, 0xE2, 0x12, 0xD6, 0xC7, 0xBB,
  62  0x0D, 0x20, 0x5E, 0xDC, 0xE0, 0xD4, 0xF7, 0xCC, 0xC4, 0x2B, 0xF9, 0xEC, 0x2D,
  63  0xF4, 0x6F, 0xB6, 0x99, 0x88, 0x81, 0x5A, 0xD9, 0xCA, 0x13, 0xA5, 0xE7, 0x47,
  64  0xE6, 0x8E, 0x60, 0xE3, 0x3E, 0xB3, 0xF6, 0x72, 0xA2, 0x35, 0xA0, 0xD7, 0xCD,
  65  0xB4, 0x2F, 0x6D, 0x2C, 0x26, 0x1F, 0x95, 0x87, 0x00, 0xD8, 0x34, 0x3F, 0x17,
  66  0x25, 0x45, 0x27, 0x75, 0x92, 0xB8, 0xA3, 0xC8, 0xDE, 0xEB, 0xF8, 0xF3, 0xDB,
  67  0x0A, 0x98, 0x83, 0x7B, 0xE5, 0xCB, 0x4C, 0x78, 0xD1 };
  68
  69 static BOOL URL_NeedEscapeA(CHAR ch, DWORD dwFlags)
  70 {
  71
  72     if (isalnum(ch))
  73         return FALSE;
  74
  75     if(dwFlags & URL_ESCAPE_SPACES_ONLY) {
  76         if(ch == ' ')
  77             return TRUE;
  78         else
  79             return FALSE;
  80     }
  81
  82     if ((dwFlags & URL_ESCAPE_PERCENT) && (ch == '%'))
  83         return TRUE;
  84
  85     if (ch <= 31 || ch >= 127)
  86         return TRUE;
  87
  88     else {
  89         switch (ch) {
  90         case ' ':
  91         case '<':
  92         case '>':
  93         case '\"':
  94         case '{':
  95         case '}':
  96         case '|':
  97         case '\\':
  98         case '^':
  99         case ']':
 100         case '[':
 101         case '`':
 102         case '&':
 103             return TRUE;
 104
 105         case '/':
 106         case '?':
 107             if (dwFlags & URL_ESCAPE_SEGMENT_ONLY) return TRUE;
 108         default:
 109             return FALSE;
 110         }
 111     }
 112 }
 113
 114 static BOOL URL_NeedEscapeW(WCHAR ch, DWORD dwFlags)
 115 {
 116
 117     if (isalnumW(ch))
 118         return FALSE;
 119
 120     if(dwFlags & URL_ESCAPE_SPACES_ONLY) {
 121         if(ch == L' ')
 122             return TRUE;
 123         else
 124             return FALSE;
 125     }
 126
 127     if ((dwFlags & URL_ESCAPE_PERCENT) && (ch == L'%'))
 128         return TRUE;
 129
 130     if (ch <= 31 || ch >= 127)
 131         return TRUE;
 132
 133     else {
 134         switch (ch) {
 135         case L' ':
 136         case L'<':
 137         case L'>':
 138         case L'\"':
 139         case L'{':
 140         case L'}':
 141         case L'|':
 142         case L'\\':
 143         case L'^':
 144         case L']':
 145         case L'[':
 146         case L'`':
 147         case L'&':
 148             return TRUE;
 149
 150         case L'/':
 151         case L'?':
 152             if (dwFlags & URL_ESCAPE_SEGMENT_ONLY) return TRUE;
 153         default:
 154             return FALSE;
 155         }
 156     }
 157 }
 158
 159 static BOOL URL_JustLocation(LPCWSTR str)
 160 {
 161     while(*str && (*str == L'/')) str++;
 162     if (*str) {
 163         while (*str && ((*str == L'-') ||
 164                         (*str == L'.') ||
 165                         isalnumW(*str))) str++;
 166         if (*str == L'/') return FALSE;
 167     }
 168     return TRUE;
 169 }
 170
 171
 172 /*************************************************************************
 173  *        UrlCanonicalizeA     [SHLWAPI.@]
 174  *
 175  * Uses the W version to do job.
 176  */
 177 HRESULT WINAPI UrlCanonicalizeA(LPCSTR pszUrl, LPSTR pszCanonicalized,
 178         LPDWORD pcchCanonicalized, DWORD dwFlags)
 179 {
 180     LPWSTR base, canonical;
 181     DWORD ret, len, len2;
 182
 183     TRACE("(%s %p %p 0x%08lx) using W version\n",
 184           debugstr_a(pszUrl), pszCanonicalized,
 185           pcchCanonicalized, dwFlags);
 186
 187     base = (LPWSTR) HeapAlloc(GetProcessHeap(), 0,
 188                               (2*INTERNET_MAX_URL_LENGTH) * sizeof(WCHAR));
 189     canonical = base + INTERNET_MAX_URL_LENGTH;
 190
 191     MultiByteToWideChar(0, 0, pszUrl, -1, base, INTERNET_MAX_URL_LENGTH);
 192     len = INTERNET_MAX_URL_LENGTH;
 193
 194     ret = UrlCanonicalizeW(base, canonical, &len, dwFlags);
 195     if (ret != S_OK) {
 196         HeapFree(GetProcessHeap(), 0, base);
 197         return ret;
 198     }
 199
 200     len2 = WideCharToMultiByte(0, 0, canonical, len, 0, 0, 0, 0);
 201     if (len2 > *pcchCanonicalized) {
 202         *pcchCanonicalized = len;
 203         HeapFree(GetProcessHeap(), 0, base);
 204         return E_POINTER;
 205     }
 206     WideCharToMultiByte(0, 0, canonical, len+1, pszCanonicalized,
 207                         *pcchCanonicalized, 0, 0);
 208     *pcchCanonicalized = len2;
 209     HeapFree(GetProcessHeap(), 0, base);
 210     return S_OK;
 211 }
 212
 213 /*************************************************************************
 214  *        UrlCanonicalizeW     [SHLWAPI.@]
 215  *
 216  *
 217  * MSDN is wrong (at 10/30/01 - go figure). This should support the
 218  * following flags:                                      GLA
 219  *    URL_DONT_ESCAPE_EXTRA_INFO    0x02000000
 220  *    URL_ESCAPE_SPACES_ONLY        0x04000000
 221  *    URL_ESCAPE_PERCENT            0x00001000
 222  *    URL_ESCAPE_UNSAFE             0x10000000
 223  *    URL_UNESCAPE                  0x10000000
 224  *    URL_DONT_SIMPLIFY             0x08000000
 225  *    URL_ESCAPE_SEGMENT_ONLY       0x00002000
 226  */
 227 HRESULT WINAPI UrlCanonicalizeW(LPCWSTR pszUrl, LPWSTR pszCanonicalized,
 228                                 LPDWORD pcchCanonicalized, DWORD dwFlags)
 229 {
 230     HRESULT hr = S_OK;
 231     DWORD EscapeFlags;
 232     LPWSTR lpszUrlCpy, wk1, wk2, mp, root;
 233     INT nLen, nByteLen, state;
 234
 235     TRACE("(%s %p %p 0x%08lx)\n", debugstr_w(pszUrl), pszCanonicalized,
 236           pcchCanonicalized, dwFlags);
 237
 238     nByteLen = (lstrlenW(pszUrl) + 1) * sizeof(WCHAR); /* length in bytes */
 239     lpszUrlCpy = HeapAlloc(GetProcessHeap(), 0, nByteLen);
 240
 241     if (dwFlags & URL_DONT_SIMPLIFY)
 242         memcpy(lpszUrlCpy, pszUrl, nByteLen);
 243     else {
 244
 245         /*
 246          * state =
 247          *         0   initial  1,3
 248          *         1   have 2[+] alnum  2,3
 249          *         2   have scheme (found :)  4,6,3
 250          *         3   failed (no location)
 251          *         4   have //  5,3
 252          *         5   have 1[+] alnum  6,3
 253          *         6   have location (found /) save root location
 254          */
 255
 256         wk1 = (LPWSTR)pszUrl;
 257         wk2 = lpszUrlCpy;
 258         state = 0;
 259         while (*wk1) {
 260             switch (state) {
 261             case 0:
 262                 if (!isalnumW(*wk1)) {state = 3; break;}
 263                 *wk2++ = *wk1++;
 264                 if (!isalnumW(*wk1)) {state = 3; break;}
 265                 *wk2++ = *wk1++;
 266                 state = 1;
 267                 break;
 268             case 1:
 269                 *wk2++ = *wk1;
 270                 if (*wk1++ == L':') state = 2;
 271                 break;
 272             case 2:
 273                 if (*wk1 != L'/') {state = 3; break;}
 274                 *wk2++ = *wk1++;
 275                 if (*wk1 != L'/') {state = 6; break;}
 276                 *wk2++ = *wk1++;
 277                 state = 4;
 278                 break;
 279             case 3:
 280                 strcpyW(wk2, wk1);
 281                 wk1 += strlenW(wk1);
 282                 wk2 += strlenW(wk2);
 283                 break;
 284             case 4:
 285                 if (!isalnumW(*wk1) && (*wk1 != L'-')) {state = 3; break;}
 286                 while(isalnumW(*wk1) || (*wk1 == L'-')) *wk2++ = *wk1++;
 287                 state = 5;
 288                 break;
 289             case 5:
 290                 if (*wk1 != L'/') {state = 3; break;}
 291                 *wk2++ = *wk1++;
 292                 state = 6;
 293                 break;
 294             case 6:
 295                 /* Now at root location, cannot back up any more. */
 296                 /* "root" will point at the '/' */
 297                 root = wk2-1;
 298                 while (*wk1) {
 299                     TRACE("wk1=%c\n", (CHAR)*wk1);
 300                     mp = strchrW(wk1, L'/');
 301                     if (!mp) {
 302                         strcpyW(wk2, wk1);
 303                         wk1 += strlenW(wk1);
 304                         wk2 += strlenW(wk2);
 305                         continue;
 306                     }
 307                     nLen = mp - wk1 + 1;
 308                     strncpyW(wk2, wk1, nLen);
 309                     wk2 += nLen;
 310                     wk1 += nLen;
 311                     if (*wk1 == L'.') {
 312                         TRACE("found '/.'\n");
 313                         if (*(wk1+1) == L'/') {
 314                             /* case of /./ -> skip the ./ */
 315                             wk1 += 2;
 316                         }
 317                         else if (*(wk1+1) == L'.') {
 318                             /* found /..  look for next / */
 319                             TRACE("found '/..'\n");
 320                             if (*(wk1+2) == L'/') {
 321                                 /* case /../ -> need to backup wk2 */
 322                                 TRACE("found '/../'\n");
 323                                 *(wk2-1) = L'\0';  /* set end of string */
 324                                 mp = strrchrW(root, L'/');
 325                                 if (mp && (mp >= root)) {
 326                                     /* found valid backup point */
 327                                     wk2 = mp + 1;
 328                                     wk1 += 3;
 329                                 }
 330                                 else {
 331                                     /* did not find point, restore '/' */
 332                                     *(wk2-1) = L'/';
 333                                 }
 334                             }
 335                         }
 336                     }
 337                 }
 338                 *wk2 = L'\0';
 339                 break;
 340             default:
 341                 FIXME("how did we get here - state=%d\n", state);
 342                 return E_INVALIDARG;
 343             }
 344         }
 345         *wk2 = L'\0';
 346         TRACE("Simplified, orig <%s>, simple <%s>\n",
 347               debugstr_w(pszUrl), debugstr_w(lpszUrlCpy));
 348     }
 349
 350     if(dwFlags & URL_UNESCAPE)
 351         UrlUnescapeW(lpszUrlCpy, NULL, NULL, URL_UNESCAPE_INPLACE);
 352
 353     if((EscapeFlags = dwFlags & (URL_ESCAPE_UNSAFE |
 354                                  URL_ESCAPE_SPACES_ONLY |
 355                                  URL_ESCAPE_PERCENT |
 356                                  URL_DONT_ESCAPE_EXTRA_INFO |
 357                                  URL_ESCAPE_SEGMENT_ONLY ))) {
 358         EscapeFlags &= ~URL_ESCAPE_UNSAFE;
 359         hr = UrlEscapeW(lpszUrlCpy, pszCanonicalized, pcchCanonicalized,
 360                         EscapeFlags);
 361     } else { /* No escaping needed, just copy the string */
 362         nLen = lstrlenW(lpszUrlCpy);
 363         if(nLen < *pcchCanonicalized)
 364             memcpy(pszCanonicalized, lpszUrlCpy, (nLen + 1)*sizeof(WCHAR));
 365         else {
 366             hr = E_POINTER;
 367             nLen++;
 368         }
 369         *pcchCanonicalized = nLen;
 370     }
 371
 372     HeapFree(GetProcessHeap(), 0, lpszUrlCpy);
 373
 374     if (hr == S_OK)
 375         TRACE("result %s\n", debugstr_w(pszCanonicalized));
 376
 377     return hr;
 378 }
 379
 380 /*************************************************************************
 381  *        UrlCombineA     [SHLWAPI.@]
 382  *
 383  * Uses the W version to do job.
 384  */
 385 HRESULT WINAPI UrlCombineA(LPCSTR pszBase, LPCSTR pszRelative,
 386                            LPSTR pszCombined, LPDWORD pcchCombined,
 387                            DWORD dwFlags)
 388 {
 389     LPWSTR base, relative, combined;
 390     DWORD ret, len, len2;
 391
 392     TRACE("(base %s, Relative %s, Combine size %ld, flags %08lx) using W version\n",
 393           debugstr_a(pszBase),debugstr_a(pszRelative),
 394           *pcchCombined,dwFlags);
 395
 396     base = (LPWSTR) HeapAlloc(GetProcessHeap(), 0,
 397                               (3*INTERNET_MAX_URL_LENGTH) * sizeof(WCHAR));
 398     relative = base + INTERNET_MAX_URL_LENGTH;
 399     combined = relative + INTERNET_MAX_URL_LENGTH;
 400
 401     MultiByteToWideChar(0, 0, pszBase, -1, base, INTERNET_MAX_URL_LENGTH);
 402     MultiByteToWideChar(0, 0, pszRelative, -1, relative, INTERNET_MAX_URL_LENGTH);
 403     len = INTERNET_MAX_URL_LENGTH;
 404
 405     ret = UrlCombineW(base, relative, combined, &len, dwFlags);
 406     if (ret != S_OK) {
 407         HeapFree(GetProcessHeap(), 0, base);
 408         return ret;
 409     }
 410
 411     len2 = WideCharToMultiByte(0, 0, combined, len, 0, 0, 0, 0);
 412     if (len2 > *pcchCombined) {
 413         *pcchCombined = len2;
 414         HeapFree(GetProcessHeap(), 0, base);
 415         return E_POINTER;
 416     }
 417     WideCharToMultiByte(0, 0, combined, len+1, pszCombined, *pcchCombined,
 418                         0, 0);
 419     *pcchCombined = len2;
 420     HeapFree(GetProcessHeap(), 0, base);
 421     return S_OK;
 422 }
 423
 424 /*************************************************************************
 425  *        UrlCombineW     [SHLWAPI.@]
 426  */
 427 HRESULT WINAPI UrlCombineW(LPCWSTR pszBase, LPCWSTR pszRelative,
 428                            LPWSTR pszCombined, LPDWORD pcchCombined,
 429                            DWORD dwFlags)
 430 {
 431     UNKNOWN_SHLWAPI_2 base, relative;
 432     DWORD myflags, sizeloc = 0;
 433     DWORD len, res1, res2, process_case = 0;
 434     LPWSTR work, preliminary, mbase, mrelative;
 435     WCHAR myfilestr[] = {'f','i','l','e',':','/','/','/','\0'};
 436     WCHAR single_slash[] = {'/','\0'};
 437     HRESULT ret;
 438
 439     TRACE("(base %s, Relative %s, Combine size %ld, flags %08lx)\n",
 440           debugstr_w(pszBase),debugstr_w(pszRelative),
 441           *pcchCombined,dwFlags);
 442
 443     base.size = 24;
 444     relative.size = 24;
 445
 446     /* Get space for duplicates of the input and the output */
 447     preliminary = HeapAlloc(GetProcessHeap(), 0, (3*INTERNET_MAX_URL_LENGTH) *
 448                             sizeof(WCHAR));
 449     mbase = preliminary + INTERNET_MAX_URL_LENGTH;
 450     mrelative = mbase + INTERNET_MAX_URL_LENGTH;
 451     *preliminary = L'\0';
 452
 453     /* Canonicalize the base input prior to looking for the scheme */
 454     myflags = dwFlags & (URL_DONT_SIMPLIFY | URL_UNESCAPE);
 455     len = INTERNET_MAX_URL_LENGTH;
 456     ret = UrlCanonicalizeW(pszBase, mbase, &len, myflags);
 457
 458     /* Canonicalize the relative input prior to looking for the scheme */
 459     len = INTERNET_MAX_URL_LENGTH;
 460     ret = UrlCanonicalizeW(pszRelative, mrelative, &len, myflags);
 461
 462     /* See if the base has a scheme */
 463     res1 = SHLWAPI_2(mbase, &base);
 464     if (res1) {
 465         /* if pszBase has no scheme, then return pszRelative */
 466         TRACE("no scheme detected in Base\n");
 467         process_case = 1;
 468     }
 469     else do {
 470
 471         /* get size of location field (if it exists) */
 472         work = (LPWSTR)base.ap2;
 473         sizeloc = 0;
 474         if (*work++ == L'/') {
 475             if (*work++ == L'/') {
 476                 /* At this point have start of location and
 477                  * it ends at next '/' or end of string.
 478                  */
 479                 while(*work && (*work != L'/')) work++;
 480                 sizeloc = work - base.ap2;
 481             }
 482         }
 483
 484         /* Change .sizep2 to not have the last leaf in it,
 485          * Note: we need to start after the location (if it exists)
 486          */
 487         work = strrchrW((base.ap2+sizeloc), L'/');
 488         if (work) {
 489             len = work - base.ap2 + 1;
 490             base.sizep2 = len;
 491         }
 492         /*
 493          * At this point:
 494          *    .ap2      points to location (starting with '//')
 495          *    .sizep2   length of location (above) and rest less the last
 496          *              leaf (if any)
 497          *    sizeloc   length of location (above) up to but not including
 498          *              the last '/'
 499          */
 500
 501         res2 = SHLWAPI_2(mrelative, &relative);
 502         if (res2) {
 503             /* no scheme in pszRelative */
 504             TRACE("no scheme detected in Relative\n");
 505             relative.ap2 = mrelative;  /* case 3,4,5 depends on this */
 506             relative.sizep2 = strlenW(mrelative);
 507             if (*pszRelative  == L':') {
 508                 /* case that is either left alone or uses pszBase */
 509                 if (dwFlags & URL_PLUGGABLE_PROTOCOL) {
 510                     process_case = 5;
 511                     break;
 512                 }
 513                 process_case = 1;
 514                 break;
 515             }
 516             if (isalnum(*mrelative) && (*(mrelative + 1) == L':')) {
 517                 /* case that becomes "file:///" */
 518                 strcpyW(preliminary, myfilestr);
 519                 process_case = 1;
 520                 break;
 521             }
 522             if ((*mrelative == L'/') && (*(mrelative+1) == L'/')) {
 523                 /* pszRelative has location and rest */
 524                 process_case = 3;
 525                 break;
 526             }
 527             if (*mrelative == L'/') {
 528                 /* case where pszRelative is root to location */
 529                 process_case = 4;
 530                 break;
 531             }
 532             process_case = (*base.ap2 == L'/') ? 5 : 3;
 533             break;
 534         }
 535
 536         /* handle cases where pszRelative has scheme */
 537         if ((base.sizep1 == relative.sizep1) &&
 538             (strncmpW(base.ap1, relative.ap1, base.sizep1) == 0)) {
 539
 540             /* since the schemes are the same */
 541             if ((*relative.ap2 == L'/') && (*(relative.ap2+1) == L'/')) {
 542                 /* case where pszRelative replaces location and following */
 543                 process_case = 3;
 544                 break;
 545             }
 546             if (*relative.ap2 == L'/') {
 547                 /* case where pszRelative is root to location */
 548                 process_case = 4;
 549                 break;
 550             }
 551             /* case where scheme is followed by document path */
 552             process_case = 5;
 553             break;
 554         }
 555         if ((*relative.ap2 == L'/') && (*(relative.ap2+1) == L'/')) {
 556             /* case where pszRelative replaces scheme, location,
 557              * and following and handles PLUGGABLE
 558              */
 559             process_case = 2;
 560             break;
 561         }
 562         process_case = 1;
 563         break;
 564     } while(FALSE); /* a litte trick to allow easy exit from nested if's */
 565
 566
 567     ret = S_OK;
 568     switch (process_case) {
 569
 570     case 1:  /*
 571               * Return pszRelative appended to what ever is in pszCombined,
 572               * (which may the string "file:///"
 573               */
 574         len = strlenW(mrelative) + strlenW(preliminary);
 575         if (len+1 > *pcchCombined) {
 576             *pcchCombined = len;
 577             ret = E_POINTER;
 578             break;
 579         }
 580         strcatW(preliminary, mrelative);
 581         break;
 582
 583     case 2:  /*
 584               * Same as case 1, but if URL_PLUGGABLE_PROTOCOL was specified
 585               * and pszRelative starts with "//", then append a "/"
 586               */
 587         len = strlenW(mrelative) + 1;
 588         if (len+1 > *pcchCombined) {
 589             *pcchCombined = len;
 590             ret = E_POINTER;
 591             break;
 592         }
 593         strcpyW(preliminary, mrelative);
 594         if (!(dwFlags & URL_PLUGGABLE_PROTOCOL) &&
 595             URL_JustLocation(relative.ap2))
 596             strcatW(preliminary, single_slash);
 597         break;
 598
 599     case 3:  /*
 600               * Return the pszBase scheme with pszRelative. Basicly
 601               * keeps the scheme and replaces the domain and following.
 602               */
 603         len = base.sizep1 + 1 + relative.sizep2 + 1;
 604         if (len+1 > *pcchCombined) {
 605             *pcchCombined = len;
 606             ret = E_POINTER;
 607             break;
 608         }
 609         strncpyW(preliminary, base.ap1, base.sizep1 + 1);
 610         work = preliminary + base.sizep1 + 1;
 611         strcpyW(work, relative.ap2);
 612         if (!(dwFlags & URL_PLUGGABLE_PROTOCOL) &&
 613             URL_JustLocation(relative.ap2))
 614             strcatW(work, single_slash);
 615         break;
 616
 617     case 4:  /*
 618               * Return the pszBase scheme and location but everything
 619               * after the location is pszRelative. (Replace document
 620               * from root on.)
 621               */
 622         len = base.sizep1 + 1 + sizeloc + relative.sizep2 + 1;
 623         if (len+1 > *pcchCombined) {
 624             *pcchCombined = len;
 625             ret = E_POINTER;
 626             break;
 627         }
 628         strncpyW(preliminary, base.ap1, base.sizep1+1+sizeloc);
 629         work = preliminary + base.sizep1 + 1 + sizeloc;
 630         if (dwFlags & URL_PLUGGABLE_PROTOCOL)
 631             *(work++) = L'/';
 632         strcpyW(work, relative.ap2);
 633         break;
 634
 635     case 5:  /*
 636               * Return the pszBase without its document (if any) and
 637               * append pszRelative after its scheme.
 638               */
 639         len = base.sizep1 + 1 + base.sizep2 + relative.sizep2;
 640         if (len+1 > *pcchCombined) {
 641             *pcchCombined = len;
 642             ret = E_POINTER;
 643             break;
 644         }
 645         strncpyW(preliminary, base.ap1, base.sizep1+1+base.sizep2);
 646         work = preliminary + base.sizep1+1+base.sizep2 - 1;
 647         if (*work++ != L'/')
 648             *(work++) = L'/';
 649         strcpyW(work, relative.ap2);
 650         break;
 651
 652     default:
 653         FIXME("How did we get here????? process_case=%ld\n", process_case);
 654         ret = E_INVALIDARG;
 655     }
 656
 657     if (ret == S_OK) {
 658         /*
 659          * Now that the combining is done, process the escape options if
 660          * necessary, otherwise just copy the string.
 661          */
 662         myflags = dwFlags & (URL_ESCAPE_PERCENT |
 663                              URL_ESCAPE_SPACES_ONLY |
 664                              URL_DONT_ESCAPE_EXTRA_INFO |
 665                              URL_ESCAPE_SEGMENT_ONLY);
 666         if (myflags)
 667             ret = UrlEscapeW(preliminary, pszCombined,
 668                              pcchCombined, myflags);
 669         else {
 670             len = (strlenW(preliminary) + 1) * sizeof(WCHAR);
 671             memcpy(pszCombined, preliminary, len);
 672             *pcchCombined = strlenW(preliminary);
 673         }
 674         TRACE("return-%ld len=%ld, %s\n",
 675               process_case, *pcchCombined, debugstr_w(pszCombined));
 676     }
 677     HeapFree(GetProcessHeap(), 0, preliminary);
 678     return ret;
 679 }
 680
 681 /*************************************************************************
 682  *      UrlEscapeA      [SHLWAPI.@]
 683  *
 684  * Converts unsafe characters into their escape sequences.
 685  *
 686  * The converted string is returned in pszEscaped if the buffer size
 687  * (which should be supplied in pcchEscaped) is large enough, in this
 688  * case the function returns S_OK and pcchEscaped contains the length
 689  * of the escaped string.  If the buffer is not large enough the
 690  * function returns E_POINTER and pcchEscaped contains the required
 691  * buffer size (including room for the '\0').
 692  *
 693  * By default the function stops converting at the first '?' or
 694  * '#'. [MSDN says differently].  If URL_ESCAPE_SPACE_ONLY flag is set
 695  * then only spaces are converted, but the conversion continues past a
 696  * '?' or '#'.
 697  *
 698  * BUGS:
 699  *  Have now implemented the following flags:
 700  *     URL_ESCAPE_SPACES_ONLY
 701  *     URL_DONT_ESCAPE_EXTRA_INFO
 702  *     URL_ESCAPE_SEGMENT_ONLY
 703  *     URL_ESCAPE_PERCENT
 704  *  Initial testing seems to indicate that this is now working like
 705  *  native shlwapi version 5. Note that these functions did not work
 706  *  well (or at all) in shlwapi version 4.
 707  *
 708  */
 709 HRESULT WINAPI UrlEscapeA(
 710         LPCSTR pszUrl,
 711         LPSTR pszEscaped,
 712         LPDWORD pcchEscaped,
 713         DWORD dwFlags)
 714 {
 715     LPCSTR src;
 716     DWORD needed = 0, ret;
 717     BOOL stop_escaping = FALSE;
 718     char next[3], *dst = pszEscaped;
 719     char hex[] = "0123456789ABCDEF";
 720     INT len;
 721
 722     TRACE("(%s %p %p 0x%08lx)\n", debugstr_a(pszUrl), pszEscaped,
 723           pcchEscaped, dwFlags);
 724
 725     if(dwFlags & ~(URL_ESCAPE_SPACES_ONLY |
 726                    URL_ESCAPE_SEGMENT_ONLY |
 727                    URL_DONT_ESCAPE_EXTRA_INFO |
 728                    URL_ESCAPE_PERCENT))
 729         FIXME("Unimplemented flags: %08lx\n", dwFlags);
 730
 731     /* fix up flags */
 732     if (dwFlags & URL_ESCAPE_SPACES_ONLY)
 733         /* if SPACES_ONLY specified, reset the other controls */
 734         dwFlags &= ~(URL_DONT_ESCAPE_EXTRA_INFO |
 735                      URL_ESCAPE_PERCENT |
 736                      URL_ESCAPE_SEGMENT_ONLY);
 737
 738     else
 739         /* if SPACES_ONLY *not* specified the assume DONT_ESCAPE_EXTRA_INFO */
 740         dwFlags |= URL_DONT_ESCAPE_EXTRA_INFO;
 741
 742     for(src = pszUrl; *src; src++) {
 743         if(!(dwFlags & URL_ESCAPE_SEGMENT_ONLY) &&
 744            (dwFlags & URL_DONT_ESCAPE_EXTRA_INFO) &&
 745            (*src == '#' || *src == '?'))
 746             stop_escaping = TRUE;
 747
 748         if(URL_NeedEscapeA(*src, dwFlags) && stop_escaping == FALSE) {
 749             /* TRACE("escaping %c\n", *src); */
 750             next[0] = '%';
 751             next[1] = hex[(*src >> 4) & 0xf];
 752             next[2] = hex[*src & 0xf];
 753             len = 3;
 754         } else {
 755             /* TRACE("passing %c\n", *src); */
 756             next[0] = *src;
 757             len = 1;
 758         }
 759
 760         if(needed + len <= *pcchEscaped) {
 761             memcpy(dst, next, len);
 762             dst += len;
 763         }
 764         needed += len;
 765     }
 766
 767     if(needed < *pcchEscaped) {
 768         *dst = '\0';
 769         ret = S_OK;
 770     } else {
 771         needed++; /* add one for the '\0' */
 772         ret = E_POINTER;
 773     }
 774     *pcchEscaped = needed;
 775     return ret;
 776 }
 777
 778 /*************************************************************************
 779  *      UrlEscapeW      [SHLWAPI.@]
 780  *
 781  * See UrlEscapeA for list of assumptions, bugs, and FIXMEs
 782  */
 783 HRESULT WINAPI UrlEscapeW(
 784         LPCWSTR pszUrl,
 785         LPWSTR pszEscaped,
 786         LPDWORD pcchEscaped,
 787         DWORD dwFlags)
 788 {
 789     LPCWSTR src;
 790     DWORD needed = 0, ret;
 791     BOOL stop_escaping = FALSE;
 792     WCHAR next[5], *dst = pszEscaped;
 793     CHAR hex[] = "0123456789ABCDEF";
 794     INT len;
 795
 796     TRACE("(%s %p %p 0x%08lx)\n", debugstr_w(pszUrl), pszEscaped,
 797           pcchEscaped, dwFlags);
 798
 799     if(dwFlags & ~(URL_ESCAPE_SPACES_ONLY |
 800                    URL_ESCAPE_SEGMENT_ONLY |
 801                    URL_DONT_ESCAPE_EXTRA_INFO |
 802                    URL_ESCAPE_PERCENT))
 803         FIXME("Unimplemented flags: %08lx\n", dwFlags);
 804
 805     /* fix up flags */
 806     if (dwFlags & URL_ESCAPE_SPACES_ONLY)
 807         /* if SPACES_ONLY specified, reset the other controls */
 808         dwFlags &= ~(URL_DONT_ESCAPE_EXTRA_INFO |
 809                      URL_ESCAPE_PERCENT |
 810                      URL_ESCAPE_SEGMENT_ONLY);
 811
 812     else
 813         /* if SPACES_ONLY *not* specified the assume DONT_ESCAPE_EXTRA_INFO */
 814         dwFlags |= URL_DONT_ESCAPE_EXTRA_INFO;
 815
 816     for(src = pszUrl; *src; src++) {
 817         /*
 818          * if(!(dwFlags & URL_ESCAPE_SPACES_ONLY) &&
 819          *   (*src == L'#' || *src == L'?'))
 820          *    stop_escaping = TRUE;
 821          */
 822         if(!(dwFlags & URL_ESCAPE_SEGMENT_ONLY) &&
 823            (dwFlags & URL_DONT_ESCAPE_EXTRA_INFO) &&
 824            (*src == L'#' || *src == L'?'))
 825             stop_escaping = TRUE;
 826
 827         if(URL_NeedEscapeW(*src, dwFlags) && stop_escaping == FALSE) {
 828             /* TRACE("escaping %c\n", *src); */
 829             next[0] = L'%';
 830             /*
 831              * I would have assumed that the W form would escape
 832              * the character with 4 hex digits (or even 8),
 833              * however, experiments show that native shlwapi escapes
 834              * with only 2 hex digits.
 835              *   next[1] = hex[(*src >> 12) & 0xf];
 836              *   next[2] = hex[(*src >> 8) & 0xf];
 837              *   next[3] = hex[(*src >> 4) & 0xf];
 838              *   next[4] = hex[*src & 0xf];
 839              *   len = 5;
 840              */
 841             next[1] = hex[(*src >> 4) & 0xf];
 842             next[2] = hex[*src & 0xf];
 843             len = 3;
 844         } else {
 845             /* TRACE("passing %c\n", *src); */
 846             next[0] = *src;
 847             len = 1;
 848         }
 849
 850         if(needed + len <= *pcchEscaped) {
 851             memcpy(dst, next, len*sizeof(WCHAR));
 852             dst += len;
 853         }
 854         needed += len;
 855     }
 856
 857     if(needed < *pcchEscaped) {
 858         *dst = L'\0';
 859         ret = S_OK;
 860     } else {
 861         needed++; /* add one for the '\0' */
 862         ret = E_POINTER;
 863     }
 864     *pcchEscaped = needed;
 865     return ret;
 866 }
 867
 868
 869 /*************************************************************************
 870  *      UrlUnescapeA    [SHLWAPI.@]
 871  *
 872  * Converts escape sequences back to ordinary characters.
 873  *
 874  * If URL_ESCAPE_INPLACE is set in dwFlags then pszUnescaped and
 875  * pcchUnescaped are ignored and the converted string is returned in
 876  * pszUrl, otherwise the string is returned in pszUnescaped.
 877  * pcchUnescaped should contain the size of pszUnescaped on calling
 878  * and will contain the length the the returned string on return if
 879  * the buffer is big enough else it will contain the buffer size
 880  * required (including room for the '\0').  The function returns S_OK
 881  * on success or E_POINTER if the buffer is not large enough.  If the
 882  * URL_DONT_ESCAPE_EXTRA_INFO flag is set then the conversion stops at
 883  * the first occurrence of either '?' or '#'.
 884  *
 885  */
 886 HRESULT WINAPI UrlUnescapeA(
 887         LPCSTR pszUrl,
 888         LPSTR pszUnescaped,
 889         LPDWORD pcchUnescaped,
 890         DWORD dwFlags)
 891 {
 892     char *dst, next;
 893     LPCSTR src;
 894     HRESULT ret;
 895     DWORD needed;
 896     BOOL stop_unescaping = FALSE;
 897
 898     TRACE("(%s, %p, %p, 0x%08lx)\n", debugstr_a(pszUrl), pszUnescaped,
 899           pcchUnescaped, dwFlags);
 900
 901     if(dwFlags & URL_UNESCAPE_INPLACE)
 902         dst = (char*)pszUrl;
 903     else
 904         dst = pszUnescaped;
 905
 906     for(src = pszUrl, needed = 0; *src; src++, needed++) {
 907         if(dwFlags & URL_DONT_UNESCAPE_EXTRA_INFO &&
 908            (*src == '#' || *src == '?')) {
 909             stop_unescaping = TRUE;
 910             next = *src;
 911         } else if(*src == '%' && isxdigit(*(src + 1)) && isxdigit(*(src + 2))
 912                   && stop_unescaping == FALSE) {
 913             INT ih;
 914             char buf[3];
 915             memcpy(buf, src + 1, 2);
 916             buf[2] = '\0';
 917             ih = strtol(buf, NULL, 16);
 918             next = (CHAR) ih;
 919             src += 2; /* Advance to end of escape */
 920         } else
 921             next = *src;
 922
 923         if(dwFlags & URL_UNESCAPE_INPLACE || needed < *pcchUnescaped)
 924             *dst++ = next;
 925     }
 926
 927     if(dwFlags & URL_UNESCAPE_INPLACE || needed < *pcchUnescaped) {
 928         *dst = '\0';
 929         ret = S_OK;
 930     } else {
 931         needed++; /* add one for the '\0' */
 932         ret = E_POINTER;
 933     }
 934     if(!(dwFlags & URL_UNESCAPE_INPLACE))
 935         *pcchUnescaped = needed;
 936
 937     if (ret == S_OK) {
 938         TRACE("result %s\n", (dwFlags & URL_UNESCAPE_INPLACE) ?
 939               debugstr_a(pszUrl) : debugstr_a(pszUnescaped));
 940     }
 941
 942     return ret;
 943 }
 944
 945 /*************************************************************************
 946  *      UrlUnescapeW    [SHLWAPI.@]
 947  *
 948  * See UrlUnescapeA for list of assumptions, bugs, and FIXMEs
 949  */
 950 HRESULT WINAPI UrlUnescapeW(
 951         LPCWSTR pszUrl,
 952         LPWSTR pszUnescaped,
 953         LPDWORD pcchUnescaped,
 954         DWORD dwFlags)
 955 {
 956     WCHAR *dst, next;
 957     LPCWSTR src;
 958     HRESULT ret;
 959     DWORD needed;
 960     BOOL stop_unescaping = FALSE;
 961
 962     TRACE("(%s, %p, %p, 0x%08lx)\n", debugstr_w(pszUrl), pszUnescaped,
 963           pcchUnescaped, dwFlags);
 964
 965     if(dwFlags & URL_UNESCAPE_INPLACE)
 966         dst = (WCHAR*)pszUrl;
 967     else
 968         dst = pszUnescaped;
 969
 970     for(src = pszUrl, needed = 0; *src; src++, needed++) {
 971         if(dwFlags & URL_DONT_UNESCAPE_EXTRA_INFO &&
 972            (*src == L'#' || *src == L'?')) {
 973             stop_unescaping = TRUE;
 974             next = *src;
 975         } else if(*src == L'%' && isxdigitW(*(src + 1)) && isxdigitW(*(src + 2))
 976                   && stop_unescaping == FALSE) {
 977             INT ih;
 978             WCHAR buf[3];
 979             memcpy(buf, src + 1, 2*sizeof(WCHAR));
 980             buf[2] = L'\0';
 981             ih = StrToIntW(buf);
 982             next = (WCHAR) ih;
 983             src += 2; /* Advance to end of escape */
 984         } else
 985             next = *src;
 986
 987         if(dwFlags & URL_UNESCAPE_INPLACE || needed < *pcchUnescaped)
 988             *dst++ = next;
 989     }
 990
 991     if(dwFlags & URL_UNESCAPE_INPLACE || needed < *pcchUnescaped) {
 992         *dst = L'\0';
 993         ret = S_OK;
 994     } else {
 995         needed++; /* add one for the '\0' */
 996         ret = E_POINTER;
 997     }
 998     if(!(dwFlags & URL_UNESCAPE_INPLACE))
 999         *pcchUnescaped = needed;
1000
1001     if (ret == S_OK) {
1002         TRACE("result %s\n", (dwFlags & URL_UNESCAPE_INPLACE) ?
1003               debugstr_w(pszUrl) : debugstr_w(pszUnescaped));
1004     }
1005
1006     return ret;
1007 }
1008
1009 /*************************************************************************
1010  *      UrlGetLocationA         [SHLWAPI.@]
1011  *
1012  * Bugs/Features:
1013  *  MSDN (as of 2001-11-01) says that:
1014  *         "The location is the segment of the URL starting with a ?
1015  *          or # character."
1016  *     Neither V4 nor V5 of shlwapi.dll implement the '?' and always return
1017  *     a NULL.
1018  *  MSDN further states that:
1019  *         "If a file URL has a query string, ther returned string
1020  *          the query string."
1021  *     In all test cases if the scheme starts with "fi" then a NULL is
1022  *     returned. V5 gives the following results:
1023  *       NULL     file://aa/b/cd#hohoh
1024  *       #hohoh   http://aa/b/cd#hohoh
1025  *       NULL     fi://aa/b/cd#hohoh
1026  *       #hohoh   ff://aa/b/cd#hohoh
1027  */
1028 LPCSTR WINAPI UrlGetLocationA(
1029         LPCSTR pszUrl)
1030 {
1031     UNKNOWN_SHLWAPI_1 base;
1032     DWORD res1;
1033
1034     base.size = 24;
1035     res1 = SHLWAPI_1(pszUrl, &base);
1036     if (res1) return NULL;  /* invalid scheme */
1037
1038     /* if scheme is file: then never return pointer */
1039     if (strncmp(base.ap1, "file", min(4,base.sizep1)) == 0) return NULL;
1040
1041     /* Look for '#' and return its addr */
1042     return strchr(base.ap2, '#');
1043 }
1044
1045 /*************************************************************************
1046  *      UrlGetLocationW         [SHLWAPI.@]
1047  *
1048  * See UrlGetLocationA for list of assumptions, bugs, and FIXMEs
1049  */
1050 LPCWSTR WINAPI UrlGetLocationW(
1051         LPCWSTR pszUrl)
1052 {
1053     UNKNOWN_SHLWAPI_2 base;
1054     DWORD res1;
1055
1056     base.size = 24;
1057     res1 = SHLWAPI_2(pszUrl, &base);
1058     if (res1) return NULL;  /* invalid scheme */
1059
1060     /* if scheme is file: then never return pointer */
1061     if (strncmpW(base.ap1, fileW, min(4,base.sizep1)) == 0) return NULL;
1062
1063     /* Look for '#' and return its addr */
1064     return strchrW(base.ap2, L'#');
1065 }
1066
1067 /*************************************************************************
1068  *      UrlCompareA     [SHLWAPI.@]
1069  */
1070 INT WINAPI UrlCompareA(
1071         LPCSTR pszUrl1,
1072         LPCSTR pszUrl2,
1073         BOOL fIgnoreSlash)
1074 {
1075     INT ret, len, len1, len2;
1076
1077     if (!fIgnoreSlash)
1078         return strcmp(pszUrl1, pszUrl2);
1079     len1 = strlen(pszUrl1);
1080     if (pszUrl1[len1-1] == L'/') len1--;
1081     len2 = strlen(pszUrl2);
1082     if (pszUrl2[len2-1] == L'/') len2--;
1083     if (len1 == len2)
1084         return strncmp(pszUrl1, pszUrl2, len1);
1085     len = min(len1, len2);
1086     ret = strncmp(pszUrl1, pszUrl2, len);
1087     if (ret) return ret;
1088     if (len1 > len2) return 1;
1089     return -1;
1090 }
1091
1092 /*************************************************************************
1093  *      UrlCompareW     [SHLWAPI.@]
1094  */
1095 INT WINAPI UrlCompareW(
1096         LPCWSTR pszUrl1,
1097         LPCWSTR pszUrl2,
1098         BOOL fIgnoreSlash)
1099 {
1100     INT ret, len, len1, len2;
1101
1102     if (!fIgnoreSlash)
1103         return strcmpW(pszUrl1, pszUrl2);
1104     len1 = strlenW(pszUrl1);
1105     if (pszUrl1[len1-1] == L'/') len1--;
1106     len2 = strlenW(pszUrl2);
1107     if (pszUrl2[len2-1] == L'/') len2--;
1108     if (len1 == len2)
1109         return strncmpW(pszUrl1, pszUrl2, len1);
1110     len = min(len1, len2);
1111     ret = strncmpW(pszUrl1, pszUrl2, len);
1112     if (ret) return ret;
1113     if (len1 > len2) return 1;
1114     return -1;
1115 }
1116
1117 /*************************************************************************
1118  *      HashData        [SHLWAPI.@]
1119  *
1120  * Hash an input block into a variable sized digest.
1121  */
1122 BOOL WINAPI HashData(const unsigned char *lpSrc, INT nSrcLen,
1123                      unsigned char *lpDest, INT nDestLen)
1124 {
1125   INT srcCount = nSrcLen - 1, destCount = nDestLen - 1;
1126
1127   if (IsBadReadPtr(lpSrc, nSrcLen) ||
1128       IsBadWritePtr(lpDest, nDestLen))
1129     return FALSE;
1130
1131   while (destCount >= 0)
1132   {
1133     lpDest[destCount] = (destCount & 0xff);
1134     destCount--;
1135   }
1136
1137   while (srcCount >= 0)
1138   {
1139     destCount = nDestLen - 1;
1140     while (destCount >= 0)
1141     {
1142       lpDest[destCount] = HashDataLookup[lpSrc[srcCount] ^ lpDest[destCount]];
1143       destCount--;
1144     }
1145     srcCount--;
1146   }
1147   return TRUE;
1148 }
1149
1150 /*************************************************************************
1151  *      UrlHashA        [SHLWAPI.@]
1152  *
1153  * Hash an ASCII URL.
1154  */
1155 HRESULT WINAPI UrlHashA(LPCSTR pszUrl, unsigned char *lpDest, INT nDestLen)
1156 {
1157   if (IsBadStringPtrA(pszUrl, -1) || IsBadWritePtr(lpDest, nDestLen))
1158     return E_INVALIDARG;
1159
1160   HashData(pszUrl, strlen(pszUrl), lpDest, nDestLen);
1161   return NOERROR;
1162 }
1163
1164 /*************************************************************************
1165  *      UrlApplySchemeA [SHLWAPI.@]
1166  */
1167 HRESULT WINAPI UrlApplySchemeA(LPCSTR pszIn, LPSTR pszOut, LPDWORD pcchOut, DWORD dwFlags)
1168 {
1169     LPWSTR in, out;
1170     DWORD ret, len, len2;
1171
1172     TRACE("(in %s, out size %ld, flags %08lx) using W version\n",
1173           debugstr_a(pszIn), *pcchOut, dwFlags);
1174
1175     in = (LPWSTR) HeapAlloc(GetProcessHeap(), 0,
1176                               (2*INTERNET_MAX_URL_LENGTH) * sizeof(WCHAR));
1177     out = in + INTERNET_MAX_URL_LENGTH;
1178
1179     MultiByteToWideChar(0, 0, pszIn, -1, in, INTERNET_MAX_URL_LENGTH);
1180     len = INTERNET_MAX_URL_LENGTH;
1181
1182     ret = UrlApplySchemeW(in, out, &len, dwFlags);
1183     if ((ret != S_OK) && (ret != S_FALSE)) {
1184         HeapFree(GetProcessHeap(), 0, in);
1185         return ret;
1186     }
1187
1188     len2 = WideCharToMultiByte(0, 0, out, len+1, 0, 0, 0, 0);
1189     if (len2 > *pcchOut) {
1190         *pcchOut = len2;
1191         HeapFree(GetProcessHeap(), 0, in);
1192         return E_POINTER;
1193     }
1194     WideCharToMultiByte(0, 0, out, len+1, pszOut, *pcchOut, 0, 0);
1195     *pcchOut = len2;
1196     HeapFree(GetProcessHeap(), 0, in);
1197     return ret;
1198 }
1199
1200 HRESULT URL_GuessScheme(LPCWSTR pszIn, LPWSTR pszOut, LPDWORD pcchOut)
1201 {
1202     HKEY newkey;
1203     BOOL j;
1204     INT index, i;
1205     DWORD value_len, data_len, dwType;
1206     WCHAR reg_path[MAX_PATH];
1207     WCHAR value[MAX_PATH], data[MAX_PATH];
1208     WCHAR Wxx, Wyy;
1209
1210     MultiByteToWideChar(0, 0,
1211               "Software\\Microsoft\\Windows\\CurrentVersion\\URL\\Prefixes",
1212                         -1, reg_path, MAX_PATH);
1213     RegOpenKeyExW(HKEY_LOCAL_MACHINE, reg_path, 0, 1, &newkey);
1214     index = 0;
1215     while(value_len = data_len = MAX_PATH,
1216           RegEnumValueW(newkey, index, value, &value_len,
1217                         0, &dwType, (LPVOID)data, &data_len) == 0) {
1218         TRACE("guess %d %s is %s\n",
1219               index, debugstr_w(value), debugstr_w(data));
1220
1221         j = FALSE;
1222         for(i=0; i<value_len; i++) {
1223             Wxx = pszIn[i];
1224             Wyy = value[i];
1225             /* remember that TRUE is not-equal */
1226             j = ChrCmpIW(Wxx, Wyy);
1227             if (j) break;
1228         }
1229         if ((i == value_len) && !j) {
1230             if (strlenW(data) + strlenW(pszIn) + 1 > *pcchOut) {
1231                 *pcchOut = strlenW(data) + strlenW(pszIn) + 1;
1232                 RegCloseKey(newkey);
1233                 return E_POINTER;
1234             }
1235             strcpyW(pszOut, data);
1236             strcatW(pszOut, pszIn);
1237             *pcchOut = strlenW(pszOut);
1238             TRACE("matched and set to %s\n", debugstr_w(pszOut));
1239             RegCloseKey(newkey);
1240             return S_OK;
1241         }
1242         index++;
1243     }
1244     RegCloseKey(newkey);
1245     return -1;
1246 }
1247
1248 HRESULT URL_ApplyDefault(LPCWSTR pszIn, LPWSTR pszOut, LPDWORD pcchOut)
1249 {
1250     HKEY newkey;
1251     DWORD data_len, dwType;
1252     WCHAR reg_path[MAX_PATH];
1253     WCHAR value[MAX_PATH], data[MAX_PATH];
1254
1255     /* get and prepend default */
1256     MultiByteToWideChar(0, 0,
1257          "Software\\Microsoft\\Windows\\CurrentVersion\\URL\\DefaultPrefix",
1258                         -1, reg_path, MAX_PATH);
1259     RegOpenKeyExW(HKEY_LOCAL_MACHINE, reg_path, 0, 1, &newkey);
1260     data_len = MAX_PATH;
1261     value[0] = L'@';
1262     value[1] = L'\0';
1263     RegQueryValueExW(newkey, value, 0, &dwType, (LPBYTE)data, &data_len);
1264     RegCloseKey(newkey);
1265     if (strlenW(data) + strlenW(pszIn) + 1 > *pcchOut) {
1266         *pcchOut = strlenW(data) + strlenW(pszIn) + 1;
1267         return E_POINTER;
1268     }
1269     strcpyW(pszOut, data);
1270     strcatW(pszOut, pszIn);
1271     *pcchOut = strlenW(pszOut);
1272     TRACE("used default %s\n", debugstr_w(pszOut));
1273     return S_OK;
1274 }
1275
1276 /*************************************************************************
1277  *      UrlApplySchemeW [SHLWAPI.@]
1278  */
1279 HRESULT WINAPI UrlApplySchemeW(LPCWSTR pszIn, LPWSTR pszOut, LPDWORD pcchOut, DWORD dwFlags)
1280 {
1281     UNKNOWN_SHLWAPI_2 in_scheme;
1282     DWORD res1;
1283     HRESULT ret;
1284
1285     TRACE("(in %s, out size %ld, flags %08lx)\n",
1286           debugstr_w(pszIn), *pcchOut, dwFlags);
1287
1288     if (dwFlags & URL_APPLY_GUESSFILE) {
1289         FIXME("(%s %p %p(%ld) 0x%08lx): stub URL_APPLY_GUESSFILE not implemented\n",
1290               debugstr_w(pszIn), pszOut, pcchOut, *pcchOut, dwFlags);
1291         strcpyW(pszOut, pszIn);
1292         *pcchOut = strlenW(pszOut);
1293         return S_FALSE;
1294     }
1295
1296     in_scheme.size = 24;
1297     /* See if the base has a scheme */
1298     res1 = SHLWAPI_2(pszIn, &in_scheme);
1299     if (res1) {
1300         /* no scheme in input, need to see if we need to guess */
1301         if (dwFlags & URL_APPLY_GUESSSCHEME) {
1302             if ((ret = URL_GuessScheme(pszIn, pszOut, pcchOut)) != -1)
1303                 return ret;
1304         }
1305     }
1306     else {
1307         /* we have a scheme, see if valid (known scheme) */
1308         if (in_scheme.fcncde) {
1309             /* have valid scheme, so just copy and exit */
1310             if (strlenW(pszIn) + 1 > *pcchOut) {
1311                 *pcchOut = strlenW(pszIn) + 1;
1312                 return E_POINTER;
1313             }
1314             strcpyW(pszOut, pszIn);
1315             *pcchOut = strlenW(pszOut);
1316             TRACE("valid scheme, returing copy\n");
1317             return S_OK;
1318         }
1319     }
1320
1321     /* If we are here, then either invalid scheme,
1322      * or no scheme and can't/failed guess.
1323      */
1324     if ( ( ((res1 == 0) && (dwFlags & URL_APPLY_FORCEAPPLY)) ||
1325            ((res1 != 0)) ) &&
1326          (dwFlags & URL_APPLY_DEFAULT)) {
1327         /* find and apply default scheme */
1328         return URL_ApplyDefault(pszIn, pszOut, pcchOut);
1329     }
1330
1331     /* just copy and give proper return code */
1332     if (strlenW(pszIn) + 1 > *pcchOut) {
1333         *pcchOut = strlenW(pszIn) + 1;
1334         return E_POINTER;
1335     }
1336     strcpyW(pszOut, pszIn);
1337     *pcchOut = strlenW(pszOut);
1338     TRACE("returing copy, left alone\n");
1339     return S_FALSE;
1340 }
1341
1342 /*************************************************************************
1343  *      UrlIsA          [SHLWAPI.@]
1344  */
1345 BOOL WINAPI UrlIsA(LPCSTR pszUrl, URLIS Urlis)
1346 {
1347     UNKNOWN_SHLWAPI_1 base;
1348     DWORD res1;
1349
1350     switch (Urlis) {
1351
1352     case URLIS_OPAQUE:
1353         base.size = 24;
1354         res1 = SHLWAPI_1(pszUrl, &base);
1355         if (res1) return FALSE;  /* invalid scheme */
1356         if ((*base.ap2 == '/') && (*(base.ap2+1) == '/'))
1357             /* has scheme followed by 2 '/' */
1358             return FALSE;
1359         return TRUE;
1360
1361     case URLIS_URL:
1362     case URLIS_NOHISTORY:
1363     case URLIS_FILEURL:
1364     case URLIS_APPLIABLE:
1365     case URLIS_DIRECTORY:
1366     case URLIS_HASQUERY:
1367     default:
1368         FIXME("(%s %d): stub\n", debugstr_a(pszUrl), Urlis);
1369     }
1370     return FALSE;
1371 }
1372
1373 /*************************************************************************
1374  *      UrlIsW          [SHLWAPI.@]
1375  */
1376 BOOL WINAPI UrlIsW(LPCWSTR pszUrl, URLIS Urlis)
1377 {
1378     UNKNOWN_SHLWAPI_2 base;
1379     DWORD res1;
1380
1381     switch (Urlis) {
1382
1383     case URLIS_OPAQUE:
1384         base.size = 24;
1385         res1 = SHLWAPI_2(pszUrl, &base);
1386         if (res1) return FALSE;  /* invalid scheme */
1387         if ((*base.ap2 == L'/') && (*(base.ap2+1) == L'/'))
1388             /* has scheme followed by 2 '/' */
1389             return FALSE;
1390         return TRUE;
1391
1392     case URLIS_URL:
1393     case URLIS_NOHISTORY:
1394     case URLIS_FILEURL:
1395     case URLIS_APPLIABLE:
1396     case URLIS_DIRECTORY:
1397     case URLIS_HASQUERY:
1398     default:
1399         FIXME("(%s %d): stub\n", debugstr_w(pszUrl), Urlis);
1400     }
1401     return FALSE;
1402 }
1403
1404 /*************************************************************************
1405  *      UrlIsNoHistoryA         [SHLWAPI.@]
1406  */
1407 BOOL WINAPI UrlIsNoHistoryA(LPCSTR pszUrl)
1408 {
1409     return UrlIsA(pszUrl, URLIS_NOHISTORY);
1410 }
1411
1412 /*************************************************************************
1413  *      UrlIsNoHistoryW         [SHLWAPI.@]
1414  */
1415 BOOL WINAPI UrlIsNoHistoryW(LPCWSTR pszUrl)
1416 {
1417     return UrlIsW(pszUrl, URLIS_NOHISTORY);
1418 }
1419
1420 /*************************************************************************
1421  *      UrlIsOpaqueA    [SHLWAPI.@]
1422  */
1423 BOOL WINAPI UrlIsOpaqueA(LPCSTR pszUrl)
1424 {
1425     return UrlIsA(pszUrl, URLIS_OPAQUE);
1426 }
1427
1428 /*************************************************************************
1429  *      UrlIsOpaqueW    [SHLWAPI.@]
1430  */
1431 BOOL WINAPI UrlIsOpaqueW(LPCWSTR pszUrl)
1432 {
1433     return UrlIsW(pszUrl, URLIS_OPAQUE);
1434 }
1435
1436 /*************************************************************************
1437  *  Scans for characters of type "type" and when not matching found,
1438  *  returns pointer to it and length in size.
1439  *
1440  * Characters tested based on RFC 1738
1441  */
1442 LPCWSTR  URL_ScanID(LPCWSTR start, LPDWORD size, WINE_URL_SCAN_TYPE type)
1443 {
1444     static DWORD alwayszero = 0;
1445     BOOL cont = TRUE;
1446
1447     *size = 0;
1448
1449     switch(type){
1450
1451     case SCHEME:
1452         while (cont) {
1453             if ( (islowerW(*start) && isalphaW(*start)) ||
1454                  isdigitW(*start) ||
1455                  (*start == L'+') ||
1456                  (*start == L'-') ||
1457                  (*start == L'.')) {
1458                 start++;
1459                 (*size)++;
1460             }
1461             else
1462                 cont = FALSE;
1463         }
1464         break;
1465
1466     case USERPASS:
1467         while (cont) {
1468             if ( isalphaW(*start) ||
1469                  isdigitW(*start) ||
1470                  /* user/password only characters */
1471                  (*start == L';') ||
1472                  (*start == L'?') ||
1473                  (*start == L'&') ||
1474                  (*start == L'=') ||
1475                  /* *extra* characters */
1476                  (*start == L'!') ||
1477                  (*start == L'*') ||
1478                  (*start == L'\'') ||
1479                  (*start == L'(') ||
1480                  (*start == L')') ||
1481                  (*start == L',') ||
1482                  /* *safe* characters */
1483                  (*start == L'$') ||
1484                  (*start == L'_') ||
1485                  (*start == L'+') ||
1486                  (*start == L'-') ||
1487                  (*start == L'.')) {
1488                 start++;
1489                 (*size)++;
1490             } else if (*start == L'%') {
1491                 if (isxdigitW(*(start+1)) &&
1492                     isxdigitW(*(start+2))) {
1493                     start += 3;
1494                     *size += 3;
1495                 } else
1496                     cont = FALSE;
1497             } else
1498                 cont = FALSE;
1499         }
1500         break;
1501
1502     case PORT:
1503         while (cont) {
1504             if (isdigitW(*start)) {
1505                 start++;
1506                 (*size)++;
1507             }
1508             else
1509                 cont = FALSE;
1510         }
1511         break;
1512
1513     case HOST:
1514         while (cont) {
1515             if (isalnumW(*start) ||
1516                 (*start == L'-') ||
1517                 (*start == L'.') ) {
1518                 start++;
1519                 (*size)++;
1520             }
1521             else
1522                 cont = FALSE;
1523         }
1524         break;
1525     default:
1526         FIXME("unknown type %d\n", type);
1527         return (LPWSTR)&alwayszero;
1528     }
1529     /* TRACE("scanned %ld characters next char %p<%c>\n",
1530      *size, start, *start); */
1531     return start;
1532 }
1533
1534 /*************************************************************************
1535  *  Attempt to parse URL into pieces.
1536  */
1537 LONG URL_ParseUrl(LPCWSTR pszUrl, WINE_PARSE_URL *pl)
1538 {
1539     LPCWSTR work;
1540
1541     memset(pl, 0, sizeof(WINE_PARSE_URL));
1542     pl->pScheme = pszUrl;
1543     work = URL_ScanID(pl->pScheme, &pl->szScheme, SCHEME);
1544     if (!*work || (*work != L':')) goto ERROR;
1545     work++;
1546     if ((*work != L'/') || (*(work+1) != L'/')) goto ERROR;
1547     pl->pUserName = work + 2;
1548     work = URL_ScanID(pl->pUserName, &pl->szUserName, USERPASS);
1549     if (*work == L':' ) {
1550         /* parse password */
1551         work++;
1552         pl->pPassword = work;
1553         work = URL_ScanID(pl->pPassword, &pl->szPassword, USERPASS);
1554         if (*work != L'@') {
1555             /* what we just parsed must be the hostname and port
1556              * so reset pointers and clear then let it parse */
1557             pl->szUserName = pl->szPassword = 0;
1558             work = pl->pUserName - 1;
1559             pl->pUserName = pl->pPassword = 0;
1560         }
1561     } else if (*work == L'@') {
1562         /* no password */
1563         pl->szPassword = 0;
1564         pl->pPassword = 0;
1565     } else if (!*work || (*work == L'/') || (*work == L'.')) {
1566         /* what was parsed was hostname, so reset pointers and let it parse */
1567         pl->szUserName = pl->szPassword = 0;
1568         work = pl->pUserName - 1;
1569         pl->pUserName = pl->pPassword = 0;
1570     } else goto ERROR;
1571
1572     /* now start parsing hostname or hostnumber */
1573     work++;
1574     pl->pHostName = work;
1575     work = URL_ScanID(pl->pHostName, &pl->szHostName, HOST);
1576     if (*work == L':') {
1577         /* parse port */
1578         work++;
1579         pl->pPort = work;
1580         work = URL_ScanID(pl->pPort, &pl->szPort, PORT);
1581     }
1582     if (*work == L'/') {
1583         /* see if query string */
1584         pl->pQuery = strchrW(work, L'?');
1585         if (pl->pQuery) pl->szQuery = strlenW(pl->pQuery);
1586     }
1587     TRACE("parse successful: scheme=%p(%ld), user=%p(%ld), pass=%p(%ld), host=%p(%ld), port=%p(%ld), query=%p(%ld)\n",
1588           pl->pScheme, pl->szScheme,
1589           pl->pUserName, pl->szUserName,
1590           pl->pPassword, pl->szPassword,
1591           pl->pHostName, pl->szHostName,
1592           pl->pPort, pl->szPort,
1593           pl->pQuery, pl->szQuery);
1594     return S_OK;
1595   ERROR:
1596     FIXME("failed to parse %s\n", debugstr_w(pszUrl));
1597     return E_INVALIDARG;
1598 }
1599
1600 /*************************************************************************
1601  *      UrlGetPartA     [SHLWAPI.@]
1602  */
1603 HRESULT WINAPI UrlGetPartA(LPCSTR pszIn, LPSTR pszOut, LPDWORD pcchOut,
1604                            DWORD dwPart, DWORD dwFlags)
1605 {
1606     LPWSTR in, out;
1607     DWORD ret, len, len2;
1608
1609     in = (LPWSTR) HeapAlloc(GetProcessHeap(), 0,
1610                               (2*INTERNET_MAX_URL_LENGTH) * sizeof(WCHAR));
1611     out = in + INTERNET_MAX_URL_LENGTH;
1612
1613     MultiByteToWideChar(0, 0, pszIn, -1, in, INTERNET_MAX_URL_LENGTH);
1614
1615     len = INTERNET_MAX_URL_LENGTH;
1616     ret = UrlGetPartW(in, out, &len, dwPart, dwFlags);
1617
1618     if (ret != S_OK) {
1619         HeapFree(GetProcessHeap(), 0, in);
1620         return ret;
1621     }
1622
1623     len2 = WideCharToMultiByte(0, 0, out, len, 0, 0, 0, 0);
1624     if (len2 > *pcchOut) {
1625         *pcchOut = len2;
1626         HeapFree(GetProcessHeap(), 0, in);
1627         return E_POINTER;
1628     }
1629     WideCharToMultiByte(0, 0, out, len+1, pszOut, *pcchOut, 0, 0);
1630     *pcchOut = len2;
1631     HeapFree(GetProcessHeap(), 0, in);
1632     return S_OK;
1633 }
1634
1635 /*************************************************************************
1636  *      UrlGetPartW     [SHLWAPI.@]
1637  */
1638 HRESULT WINAPI UrlGetPartW(LPCWSTR pszIn, LPWSTR pszOut, LPDWORD pcchOut,
1639                            DWORD dwPart, DWORD dwFlags)
1640 {
1641     WINE_PARSE_URL pl;
1642     HRESULT ret;
1643     DWORD size, schsize;
1644     LPCWSTR addr, schaddr;
1645     LPWSTR work;
1646
1647     TRACE("(%s %p %p(%ld) %08lx %08lx)\n",
1648           debugstr_w(pszIn), pszOut, pcchOut, *pcchOut, dwPart, dwFlags);
1649
1650     ret = URL_ParseUrl(pszIn, &pl);
1651     if (!ret) {
1652         schaddr = pl.pScheme;
1653         schsize = pl.szScheme;
1654
1655         switch (dwPart) {
1656         case URL_PART_SCHEME:
1657             if (!pl.szScheme) return E_INVALIDARG;
1658             addr = pl.pScheme;
1659             size = pl.szScheme;
1660             break;
1661
1662         case URL_PART_HOSTNAME:
1663             if (!pl.szHostName) return E_INVALIDARG;
1664             addr = pl.pHostName;
1665             size = pl.szHostName;
1666             break;
1667
1668         case URL_PART_USERNAME:
1669             if (!pl.szUserName) return E_INVALIDARG;
1670             addr = pl.pUserName;
1671             size = pl.szUserName;
1672             break;
1673
1674         case URL_PART_PASSWORD:
1675             if (!pl.szPassword) return E_INVALIDARG;
1676             addr = pl.pPassword;
1677             size = pl.szPassword;
1678             break;
1679
1680         case URL_PART_PORT:
1681             if (!pl.szPort) return E_INVALIDARG;
1682             addr = pl.pPort;
1683             size = pl.szPort;
1684             break;
1685
1686         case URL_PART_QUERY:
1687             if (!pl.szQuery) return E_INVALIDARG;
1688             addr = pl.pQuery;
1689             size = pl.szQuery;
1690             break;
1691
1692         default:
1693             return E_INVALIDARG;
1694         }
1695
1696         if (dwFlags == URL_PARTFLAG_KEEPSCHEME) {
1697             if (*pcchOut < size + schsize + 2) {
1698                 *pcchOut = size + schsize + 2;
1699                 return E_POINTER;
1700             }
1701             strncpyW(pszOut, schaddr, schsize);
1702             work = pszOut + schsize;
1703             *work = L':';
1704             strncpyW(work+1, addr, size);
1705             *pcchOut = size + schsize + 1;
1706             work += (size + 1);
1707             *work = L'\0';
1708         }
1709         else {
1710             if (*pcchOut < size + 1) {*pcchOut = size+1; return E_POINTER;}
1711             strncpyW(pszOut, addr, size);
1712             *pcchOut = size;
1713             work = pszOut + size;
1714             *work = L'\0';
1715         }
1716         TRACE("len=%ld %s\n", *pcchOut, debugstr_w(pszOut));
1717     }
1718     return ret;
1719 }