| 1 | /*
|
|---|
| 2 | * Url functions
|
|---|
| 3 | *
|
|---|
| 4 | * Copyright 2000 Huw D M Davies for CodeWeavers.
|
|---|
| 5 | */
|
|---|
| 6 |
|
|---|
| 7 | #include <string.h>
|
|---|
| 8 | #include "windef.h"
|
|---|
| 9 | #include "winbase.h"
|
|---|
| 10 | #include "winerror.h"
|
|---|
| 11 | #include "shlwapi.h"
|
|---|
| 12 | #include "debugtools.h"
|
|---|
| 13 |
|
|---|
| 14 | #ifdef __WIN32OS2__
|
|---|
| 15 | #undef FIXME
|
|---|
| 16 | #ifdef DEBUG
|
|---|
| 17 | #define FIXME WriteLog("FIXME %s", __FUNCTION__); WriteLog
|
|---|
| 18 | #else
|
|---|
| 19 | #define FIXME 1 ? (void)0 : (void)((int (*)(char *, ...)) NULL)
|
|---|
| 20 | #endif
|
|---|
| 21 | #endif
|
|---|
| 22 |
|
|---|
| 23 | DEFAULT_DEBUG_CHANNEL(shell);
|
|---|
| 24 |
|
|---|
| 25 | static const unsigned char HashDataLookup[256] = {
|
|---|
| 26 | 0x01, 0x0E, 0x6E, 0x19, 0x61, 0xAE, 0x84, 0x77, 0x8A, 0xAA, 0x7D, 0x76, 0x1B,
|
|---|
| 27 | 0xE9, 0x8C, 0x33, 0x57, 0xC5, 0xB1, 0x6B, 0xEA, 0xA9, 0x38, 0x44, 0x1E, 0x07,
|
|---|
| 28 | 0xAD, 0x49, 0xBC, 0x28, 0x24, 0x41, 0x31, 0xD5, 0x68, 0xBE, 0x39, 0xD3, 0x94,
|
|---|
| 29 | 0xDF, 0x30, 0x73, 0x0F, 0x02, 0x43, 0xBA, 0xD2, 0x1C, 0x0C, 0xB5, 0x67, 0x46,
|
|---|
| 30 | 0x16, 0x3A, 0x4B, 0x4E, 0xB7, 0xA7, 0xEE, 0x9D, 0x7C, 0x93, 0xAC, 0x90, 0xB0,
|
|---|
| 31 | 0xA1, 0x8D, 0x56, 0x3C, 0x42, 0x80, 0x53, 0x9C, 0xF1, 0x4F, 0x2E, 0xA8, 0xC6,
|
|---|
| 32 | 0x29, 0xFE, 0xB2, 0x55, 0xFD, 0xED, 0xFA, 0x9A, 0x85, 0x58, 0x23, 0xCE, 0x5F,
|
|---|
| 33 | 0x74, 0xFC, 0xC0, 0x36, 0xDD, 0x66, 0xDA, 0xFF, 0xF0, 0x52, 0x6A, 0x9E, 0xC9,
|
|---|
| 34 | 0x3D, 0x03, 0x59, 0x09, 0x2A, 0x9B, 0x9F, 0x5D, 0xA6, 0x50, 0x32, 0x22, 0xAF,
|
|---|
| 35 | 0xC3, 0x64, 0x63, 0x1A, 0x96, 0x10, 0x91, 0x04, 0x21, 0x08, 0xBD, 0x79, 0x40,
|
|---|
| 36 | 0x4D, 0x48, 0xD0, 0xF5, 0x82, 0x7A, 0x8F, 0x37, 0x69, 0x86, 0x1D, 0xA4, 0xB9,
|
|---|
| 37 | 0xC2, 0xC1, 0xEF, 0x65, 0xF2, 0x05, 0xAB, 0x7E, 0x0B, 0x4A, 0x3B, 0x89, 0xE4,
|
|---|
| 38 | 0x6C, 0xBF, 0xE8, 0x8B, 0x06, 0x18, 0x51, 0x14, 0x7F, 0x11, 0x5B, 0x5C, 0xFB,
|
|---|
| 39 | 0x97, 0xE1, 0xCF, 0x15, 0x62, 0x71, 0x70, 0x54, 0xE2, 0x12, 0xD6, 0xC7, 0xBB,
|
|---|
| 40 | 0x0D, 0x20, 0x5E, 0xDC, 0xE0, 0xD4, 0xF7, 0xCC, 0xC4, 0x2B, 0xF9, 0xEC, 0x2D,
|
|---|
| 41 | 0xF4, 0x6F, 0xB6, 0x99, 0x88, 0x81, 0x5A, 0xD9, 0xCA, 0x13, 0xA5, 0xE7, 0x47,
|
|---|
| 42 | 0xE6, 0x8E, 0x60, 0xE3, 0x3E, 0xB3, 0xF6, 0x72, 0xA2, 0x35, 0xA0, 0xD7, 0xCD,
|
|---|
| 43 | 0xB4, 0x2F, 0x6D, 0x2C, 0x26, 0x1F, 0x95, 0x87, 0x00, 0xD8, 0x34, 0x3F, 0x17,
|
|---|
| 44 | 0x25, 0x45, 0x27, 0x75, 0x92, 0xB8, 0xA3, 0xC8, 0xDE, 0xEB, 0xF8, 0xF3, 0xDB,
|
|---|
| 45 | 0x0A, 0x98, 0x83, 0x7B, 0xE5, 0xCB, 0x4C, 0x78, 0xD1 };
|
|---|
| 46 |
|
|---|
| 47 | static BOOL URL_NeedEscape(CHAR ch, DWORD dwFlags)
|
|---|
| 48 | {
|
|---|
| 49 |
|
|---|
| 50 | if (isalnum(ch))
|
|---|
| 51 | return FALSE;
|
|---|
| 52 |
|
|---|
| 53 | if(dwFlags & URL_ESCAPE_SPACES_ONLY) {
|
|---|
| 54 | if(ch == ' ')
|
|---|
| 55 | return TRUE;
|
|---|
| 56 | else
|
|---|
| 57 | return FALSE;
|
|---|
| 58 | }
|
|---|
| 59 |
|
|---|
| 60 | if (ch <= 31 || ch >= 127)
|
|---|
| 61 | return TRUE;
|
|---|
| 62 |
|
|---|
| 63 | else {
|
|---|
| 64 | switch (ch) {
|
|---|
| 65 | case ' ':
|
|---|
| 66 | case '<':
|
|---|
| 67 | case '>':
|
|---|
| 68 | case '\"':
|
|---|
| 69 | case '{':
|
|---|
| 70 | case '}':
|
|---|
| 71 | case '|':
|
|---|
| 72 | case '\\':
|
|---|
| 73 | case '^':
|
|---|
| 74 | case ']':
|
|---|
| 75 | case '[':
|
|---|
| 76 | case '`':
|
|---|
| 77 | case '&':
|
|---|
| 78 | return TRUE;
|
|---|
| 79 |
|
|---|
| 80 | default:
|
|---|
| 81 | return FALSE;
|
|---|
| 82 | }
|
|---|
| 83 | }
|
|---|
| 84 | }
|
|---|
| 85 |
|
|---|
| 86 | /*************************************************************************
|
|---|
| 87 | * UrlCanonicalizeA [SHLWAPI]
|
|---|
| 88 | */
|
|---|
| 89 | HRESULT WINAPI UrlCanonicalizeA(LPCSTR pszUrl, LPSTR pszCanonicalized,
|
|---|
| 90 | LPDWORD pcchCanonicalized, DWORD dwFlags)
|
|---|
| 91 | {
|
|---|
| 92 | HRESULT hr = S_OK;
|
|---|
| 93 |
|
|---|
| 94 | LPSTR lpszUrlCpy;
|
|---|
| 95 | INT nLen;
|
|---|
| 96 |
|
|---|
| 97 | TRACE("(%s %p %p 0x%08lx)\n", debugstr_a(pszUrl), pszCanonicalized,
|
|---|
| 98 | pcchCanonicalized, dwFlags);
|
|---|
| 99 |
|
|---|
| 100 | nLen = strlen(pszUrl);
|
|---|
| 101 | lpszUrlCpy = HeapAlloc(GetProcessHeap(), 0, nLen + 1);
|
|---|
| 102 |
|
|---|
| 103 | if (dwFlags & URL_DONT_SIMPLIFY)
|
|---|
| 104 | memcpy(lpszUrlCpy, pszUrl, nLen + 1);
|
|---|
| 105 | else {
|
|---|
| 106 | FIXME("Simplify path\n");
|
|---|
| 107 | memcpy(lpszUrlCpy, pszUrl, nLen + 1);
|
|---|
| 108 | }
|
|---|
| 109 |
|
|---|
| 110 | if(dwFlags & URL_UNESCAPE)
|
|---|
| 111 | UrlUnescapeA(lpszUrlCpy, NULL, NULL, URL_UNESCAPE_INPLACE);
|
|---|
| 112 |
|
|---|
| 113 | if(dwFlags & (URL_ESCAPE_UNSAFE | URL_ESCAPE_SPACES_ONLY)) {
|
|---|
| 114 | DWORD EscapeFlags = dwFlags & (URL_ESCAPE_SPACES_ONLY
|
|---|
| 115 | /* | URL_ESCAPE_PERCENT */);
|
|---|
| 116 | hr = UrlEscapeA(lpszUrlCpy, pszCanonicalized, pcchCanonicalized,
|
|---|
| 117 | EscapeFlags);
|
|---|
| 118 | } else { /* No escapping needed, just copy the string */
|
|---|
| 119 | nLen = strlen(lpszUrlCpy);
|
|---|
| 120 | if(nLen < *pcchCanonicalized)
|
|---|
| 121 | memcpy(pszCanonicalized, lpszUrlCpy, nLen + 1);
|
|---|
| 122 | else {
|
|---|
| 123 | hr = E_POINTER;
|
|---|
| 124 | nLen++;
|
|---|
| 125 | }
|
|---|
| 126 | *pcchCanonicalized = nLen;
|
|---|
| 127 | }
|
|---|
| 128 |
|
|---|
| 129 | HeapFree(GetProcessHeap(), 0, lpszUrlCpy);
|
|---|
| 130 |
|
|---|
| 131 | return hr;
|
|---|
| 132 | }
|
|---|
| 133 |
|
|---|
| 134 | /*************************************************************************
|
|---|
| 135 | * UrlCanonicalizeW [SHLWAPI]
|
|---|
| 136 | */
|
|---|
| 137 | HRESULT WINAPI UrlCanonicalizeW(LPCWSTR pszUrl, LPWSTR pszCanonicalized,
|
|---|
| 138 | LPDWORD pcchCanonicalized, DWORD dwFlags)
|
|---|
| 139 | {
|
|---|
| 140 | FIXME("(%s %p %p 0x%08lx): stub\n",debugstr_w(pszUrl),
|
|---|
| 141 | pszCanonicalized, pcchCanonicalized, dwFlags);
|
|---|
| 142 | return E_NOTIMPL;
|
|---|
| 143 | }
|
|---|
| 144 |
|
|---|
| 145 | /*************************************************************************
|
|---|
| 146 | * UrlEscapeA [SHLWAPI]
|
|---|
| 147 | *
|
|---|
| 148 | * Converts unsafe characters into their escape sequences.
|
|---|
| 149 | *
|
|---|
| 150 | * The converted string is returned in pszEscaped if the buffer size
|
|---|
| 151 | * (which should be supplied in pcchEscaped) is large enough, in this
|
|---|
| 152 | * case the function returns S_OK and pcchEscaped contains the length
|
|---|
| 153 | * of the escaped string. If the buffer is not large enough the
|
|---|
| 154 | * function returns E_POINTER and pcchEscaped contains the required
|
|---|
| 155 | * buffer size (including room for the '\0').
|
|---|
| 156 | *
|
|---|
| 157 | * By default the function stops converting at the first '?' or
|
|---|
| 158 | * '#'. [MSDN says differently]. If URL_ESCAPE_SPACE_ONLY flag is set
|
|---|
| 159 | * then only spaces are converted, but the conversion continues past a
|
|---|
| 160 | * '?' or '#'.
|
|---|
| 161 | *
|
|---|
| 162 | * BUGS:
|
|---|
| 163 | *
|
|---|
| 164 | * None of the URL_ define values are documented, so they were
|
|---|
| 165 | * determined by trial and error. MSDN mentions URL_ESCAPE_PERCENT
|
|---|
| 166 | * but I can't find a value that does this under win2000.
|
|---|
| 167 | * URL_DONT_ESCAPE_EXTRA_INFO appears to be the default which is what
|
|---|
| 168 | * we assume here. URL_ESCAPE_SEGMENT_ONLY is not implemented
|
|---|
| 169 | * (value??). A value of 0x2000 for dwFlags seems to escape
|
|---|
| 170 | * '/'s too - this is neither documented on MSDN nor implemented here.
|
|---|
| 171 | * For character values that are converted see URL_NeedEscape.
|
|---|
| 172 | */
|
|---|
| 173 | HRESULT WINAPI UrlEscapeA(
|
|---|
| 174 | LPCSTR pszUrl,
|
|---|
| 175 | LPSTR pszEscaped,
|
|---|
| 176 | LPDWORD pcchEscaped,
|
|---|
| 177 | DWORD dwFlags)
|
|---|
| 178 | {
|
|---|
| 179 | LPCSTR src;
|
|---|
| 180 | DWORD needed = 0, ret;
|
|---|
| 181 | BOOL stop_escapping = FALSE;
|
|---|
| 182 | char next[3], *dst = pszEscaped;
|
|---|
| 183 | char hex[] = "0123456789ABCDEF";
|
|---|
| 184 | INT len;
|
|---|
| 185 |
|
|---|
| 186 | TRACE("(%s %p %p 0x%08lx)\n", debugstr_a(pszUrl), pszEscaped,
|
|---|
| 187 | pcchEscaped, dwFlags);
|
|---|
| 188 |
|
|---|
| 189 | if(dwFlags & ~URL_ESCAPE_SPACES_ONLY)
|
|---|
| 190 | FIXME("Unimplemented flags: %08lx\n", dwFlags);
|
|---|
| 191 |
|
|---|
| 192 | for(src = pszUrl; *src; src++) {
|
|---|
| 193 | if(!(dwFlags & URL_ESCAPE_SPACES_ONLY) &&
|
|---|
| 194 | (*src == '#' || *src == '?'))
|
|---|
| 195 | stop_escapping = TRUE;
|
|---|
| 196 |
|
|---|
| 197 | if(URL_NeedEscape(*src, dwFlags) && stop_escapping == FALSE) {
|
|---|
| 198 | next[0] = '%';
|
|---|
| 199 | next[1] = hex[(*src >> 4) & 0xf];
|
|---|
| 200 | next[2] = hex[*src & 0xf];
|
|---|
| 201 | len = 3;
|
|---|
| 202 | } else {
|
|---|
| 203 | next[0] = *src;
|
|---|
| 204 | len = 1;
|
|---|
| 205 | }
|
|---|
| 206 |
|
|---|
| 207 | if(needed + len <= *pcchEscaped) {
|
|---|
| 208 | memcpy(dst, next, len);
|
|---|
| 209 | dst += len;
|
|---|
| 210 | }
|
|---|
| 211 | needed += len;
|
|---|
| 212 | }
|
|---|
| 213 |
|
|---|
| 214 | if(needed < *pcchEscaped) {
|
|---|
| 215 | *dst = '\0';
|
|---|
| 216 | ret = S_OK;
|
|---|
| 217 | } else {
|
|---|
| 218 | needed++; /* add one for the '\0' */
|
|---|
| 219 | ret = E_POINTER;
|
|---|
| 220 | }
|
|---|
| 221 | *pcchEscaped = needed;
|
|---|
| 222 | return ret;
|
|---|
| 223 | }
|
|---|
| 224 |
|
|---|
| 225 | /*************************************************************************
|
|---|
| 226 | * UrlEscapeW [SHLWAPI]
|
|---|
| 227 | */
|
|---|
| 228 | HRESULT WINAPI UrlEscapeW(
|
|---|
| 229 | LPCWSTR pszUrl,
|
|---|
| 230 | LPWSTR pszEscaped,
|
|---|
| 231 | LPDWORD pcchEscaped,
|
|---|
| 232 | DWORD dwFlags)
|
|---|
| 233 | {
|
|---|
| 234 | FIXME("(%s %p %p 0x%08lx): stub\n",debugstr_w(pszUrl),
|
|---|
| 235 | pszEscaped, pcchEscaped, dwFlags);
|
|---|
| 236 | return E_NOTIMPL;
|
|---|
| 237 | }
|
|---|
| 238 |
|
|---|
| 239 |
|
|---|
| 240 | /*************************************************************************
|
|---|
| 241 | * UrlUnescapeA [SHLWAPI]
|
|---|
| 242 | *
|
|---|
| 243 | * Converts escape sequences back to ordinary characters.
|
|---|
| 244 | *
|
|---|
| 245 | * If URL_ESCAPE_INPLACE is set in dwFlags then pszUnescaped and
|
|---|
| 246 | * pcchUnescaped are ignored and the converted string is returned in
|
|---|
| 247 | * pszUrl, otherwise the string is returned in pszUnescaped.
|
|---|
| 248 | * pcchUnescaped should contain the size of pszUnescaped on calling
|
|---|
| 249 | * and will contain the length the the returned string on return if
|
|---|
| 250 | * the buffer is big enough else it will contain the buffer size
|
|---|
| 251 | * required (including room for the '\0'). The function returns S_OK
|
|---|
| 252 | * on success or E_POINTER if the buffer is not large enough. If the
|
|---|
| 253 | * URL_DONT_ESCAPE_EXTRA_INFO flag is set then the conversion stops at
|
|---|
| 254 | * the first occurrence of either '?' or '#'.
|
|---|
| 255 | *
|
|---|
| 256 | */
|
|---|
| 257 | HRESULT WINAPI UrlUnescapeA(
|
|---|
| 258 | LPCSTR pszUrl,
|
|---|
| 259 | LPSTR pszUnescaped,
|
|---|
| 260 | LPDWORD pcchUnescaped,
|
|---|
| 261 | DWORD dwFlags)
|
|---|
| 262 | {
|
|---|
| 263 | char *dst, next;
|
|---|
| 264 | LPCSTR src;
|
|---|
| 265 | HRESULT ret;
|
|---|
| 266 | DWORD needed;
|
|---|
| 267 | BOOL stop_unescapping = FALSE;
|
|---|
| 268 |
|
|---|
| 269 | TRACE("(%s, %p, %p, %08lx): stub\n", debugstr_a(pszUrl), pszUnescaped,
|
|---|
| 270 | pcchUnescaped, dwFlags);
|
|---|
| 271 |
|
|---|
| 272 | if(dwFlags & URL_UNESCAPE_INPLACE)
|
|---|
| 273 | dst = (char*)pszUrl;
|
|---|
| 274 | else
|
|---|
| 275 | dst = pszUnescaped;
|
|---|
| 276 |
|
|---|
| 277 | for(src = pszUrl, needed = 0; *src; src++, needed++) {
|
|---|
| 278 | if(dwFlags & URL_DONT_UNESCAPE_EXTRA_INFO &&
|
|---|
| 279 | (*src == '#' || *src == '?')) {
|
|---|
| 280 | stop_unescapping = TRUE;
|
|---|
| 281 | next = *src;
|
|---|
| 282 | } else if(*src == '%' && isxdigit(*(src + 1)) && isxdigit(*(src + 2))
|
|---|
| 283 | && stop_unescapping == FALSE) {
|
|---|
| 284 | INT ih;
|
|---|
| 285 | char buf[3];
|
|---|
| 286 | memcpy(buf, src + 1, 2);
|
|---|
| 287 | buf[2] = '\0';
|
|---|
| 288 | ih = strtol(buf, NULL, 16);
|
|---|
| 289 | next = (CHAR) ih;
|
|---|
| 290 | src += 2; /* Advance to end of escape */
|
|---|
| 291 | } else
|
|---|
| 292 | next = *src;
|
|---|
| 293 |
|
|---|
| 294 | if(dwFlags & URL_UNESCAPE_INPLACE || needed < *pcchUnescaped)
|
|---|
| 295 | *dst++ = next;
|
|---|
| 296 | }
|
|---|
| 297 |
|
|---|
| 298 | if(dwFlags & URL_UNESCAPE_INPLACE || needed < *pcchUnescaped) {
|
|---|
| 299 | *dst = '\0';
|
|---|
| 300 | ret = S_OK;
|
|---|
| 301 | } else {
|
|---|
| 302 | needed++; /* add one for the '\0' */
|
|---|
| 303 | ret = E_POINTER;
|
|---|
| 304 | }
|
|---|
| 305 | if(!(dwFlags & URL_UNESCAPE_INPLACE))
|
|---|
| 306 | *pcchUnescaped = needed;
|
|---|
| 307 |
|
|---|
| 308 | return ret;
|
|---|
| 309 | }
|
|---|
| 310 |
|
|---|
| 311 | /*************************************************************************
|
|---|
| 312 | * UrlUnescapeW [SHLWAPI]
|
|---|
| 313 | */
|
|---|
| 314 | HRESULT WINAPI UrlUnescapeW(
|
|---|
| 315 | LPCWSTR pszUrl,
|
|---|
| 316 | LPWSTR pszUnescaped,
|
|---|
| 317 | LPDWORD pcchUnescaped,
|
|---|
| 318 | DWORD dwFlags)
|
|---|
| 319 | {
|
|---|
| 320 | FIXME("(%s, %p, %p, %08lx): stub\n", debugstr_w(pszUrl), pszUnescaped,
|
|---|
| 321 | pcchUnescaped, dwFlags);
|
|---|
| 322 | return E_NOTIMPL;
|
|---|
| 323 | }
|
|---|
| 324 |
|
|---|
| 325 | /*************************************************************************
|
|---|
| 326 | * HashData [SHLWAPI]
|
|---|
| 327 | *
|
|---|
| 328 | * Hash an input block into a variable sized digest.
|
|---|
| 329 | */
|
|---|
| 330 | BOOL WINAPI HashData(const unsigned char *lpSrc, INT nSrcLen,
|
|---|
| 331 | unsigned char *lpDest, INT nDestLen)
|
|---|
| 332 | {
|
|---|
| 333 | INT srcCount = nSrcLen - 1, destCount = nDestLen - 1;
|
|---|
| 334 |
|
|---|
| 335 | if (IsBadReadPtr(lpSrc, nSrcLen) ||
|
|---|
| 336 | IsBadWritePtr(lpDest, nDestLen))
|
|---|
| 337 | return FALSE;
|
|---|
| 338 |
|
|---|
| 339 | while (destCount >= 0)
|
|---|
| 340 | {
|
|---|
| 341 | lpDest[destCount] = (destCount & 0xff);
|
|---|
| 342 | destCount--;
|
|---|
| 343 | }
|
|---|
| 344 |
|
|---|
| 345 | while (srcCount >= 0)
|
|---|
| 346 | {
|
|---|
| 347 | destCount = nDestLen - 1;
|
|---|
| 348 | while (destCount >= 0)
|
|---|
| 349 | {
|
|---|
| 350 | lpDest[destCount] = HashDataLookup[lpSrc[srcCount] ^ lpDest[destCount]];
|
|---|
| 351 | destCount--;
|
|---|
| 352 | }
|
|---|
| 353 | srcCount--;
|
|---|
| 354 | }
|
|---|
| 355 | return TRUE;
|
|---|
| 356 | }
|
|---|
| 357 |
|
|---|
| 358 | /*************************************************************************
|
|---|
| 359 | * UrlHashA [SHLWAPI]
|
|---|
| 360 | *
|
|---|
| 361 | * Hash an ASCII URL.
|
|---|
| 362 | */
|
|---|
| 363 | HRESULT WINAPI UrlHashA(LPCSTR pszUrl, unsigned char *lpDest, INT nDestLen)
|
|---|
| 364 | {
|
|---|
| 365 | if (IsBadStringPtrA(pszUrl, -1) || IsBadWritePtr(lpDest, nDestLen))
|
|---|
| 366 | return E_INVALIDARG;
|
|---|
| 367 |
|
|---|
| 368 | HashData(pszUrl, strlen(pszUrl), lpDest, nDestLen);
|
|---|
| 369 | return NOERROR;
|
|---|
| 370 | }
|
|---|
| 371 |
|
|---|