tclscan.c

tcl是工具命令语言

    if (totalVars > 0) {
	objs = (Tcl_Obj **) ckalloc(sizeof(Tcl_Obj*) * totalVars);
	for (i = 0; i < totalVars; i++) {
	    objs[i] = NULL;
	}
    }

    string = Tcl_GetStringFromObj(objv[1], NULL);
    baseString = string;

    /*
     * Iterate over the format string filling in the result objects until
     * we reach the end of input, the end of the format string, or there
     * is a mismatch.
     */

    objIndex = 0;
    nconversions = 0;
    while (*format != '\0') {
	format += Tcl_UtfToUniChar(format, &ch);

	flags = 0;

	/*
	 * If we see whitespace in the format, skip whitespace in the string.
	 */

	if (Tcl_UniCharIsSpace(ch)) {
	    offset = Tcl_UtfToUniChar(string, &sch);
	    while (Tcl_UniCharIsSpace(sch)) {
		if (*string == '\0') {
		    goto done;
		}
		string += offset;
		offset = Tcl_UtfToUniChar(string, &sch);
	    }
	    continue;
	}
	    
	if (ch != '%') {
	    literal:
	    if (*string == '\0') {
		underflow = 1;
		goto done;
	    }
	    string += Tcl_UtfToUniChar(string, &sch);
	    if (ch != sch) {
		goto done;
	    }
	    continue;
	}

	format += Tcl_UtfToUniChar(format, &ch);
	if (ch == '%') {
	    goto literal;
	}

	/*
	 * Check for assignment suppression ('*') or an XPG3-style
	 * assignment ('%n$').
	 */

	if (ch == '*') {
	    flags |= SCAN_SUPPRESS;
	    format += Tcl_UtfToUniChar(format, &ch);
	} else if ((ch < 0x80) && isdigit(UCHAR(ch))) { /* INTL: "C" locale. */
	    value = strtoul(format-1, &end, 10); /* INTL: "C" locale. */
	    if (*end == '$') {
		format = end+1;
		format += Tcl_UtfToUniChar(format, &ch);
		objIndex = (int) value - 1;
	    }
	}

	/*
	 * Parse any width specifier.
	 */

	if ((ch < 0x80) && isdigit(UCHAR(ch))) { /* INTL: "C" locale. */
	    width = strtoul(format-1, &format, 10); /* INTL: "C" locale. */
	    format += Tcl_UtfToUniChar(format, &ch);
	} else {
	    width = 0;
	}

	/*
	 * Handle any size specifier.
	 */

	switch (ch) {
	case 'l':
	case 'L':
#ifndef TCL_WIDE_INT_IS_LONG
	    flags |= SCAN_LONGER;
#endif
	    /*
	     * Fall through so we skip to the next character.
	     */
	case 'h':
	    format += Tcl_UtfToUniChar(format, &ch);
	}

	/*
	 * Handle the various field types.
	 */

	switch (ch) {
	    case 'n':
		if (!(flags & SCAN_SUPPRESS)) {
		    objPtr = Tcl_NewIntObj(string - baseString);
		    Tcl_IncrRefCount(objPtr);
		    objs[objIndex++] = objPtr;
		}
		nconversions++;
		continue;

	    case 'd':
		op = 'i';
		base = 10;
		fn = (long (*)())strtol;
#ifndef TCL_WIDE_INT_IS_LONG
		lfn = (Tcl_WideInt (*)())strtoll;
#endif
		break;
	    case 'i':
		op = 'i';
		base = 0;
		fn = (long (*)())strtol;
#ifndef TCL_WIDE_INT_IS_LONG
		lfn = (Tcl_WideInt (*)())strtoll;
#endif
		break;
	    case 'o':
		op = 'i';
		base = 8;
		fn = (long (*)())strtoul;
#ifndef TCL_WIDE_INT_IS_LONG
		lfn = (Tcl_WideInt (*)())strtoull;
#endif
		break;
	    case 'x':
		op = 'i';
		base = 16;
		fn = (long (*)())strtoul;
#ifndef TCL_WIDE_INT_IS_LONG
		lfn = (Tcl_WideInt (*)())strtoull;
#endif
		break;
	    case 'u':
		op = 'i';
		base = 10;
		flags |= SCAN_UNSIGNED;
		fn = (long (*)())strtoul;
#ifndef TCL_WIDE_INT_IS_LONG
		lfn = (Tcl_WideInt (*)())strtoull;
#endif
		break;

	    case 'f':
	    case 'e':
	    case 'g':
		op = 'f';
		break;

	    case 's':
		op = 's';
		break;

	    case 'c':
		op = 'c';
		flags |= SCAN_NOSKIP;
		break;
	    case '[':
		op = '[';
		flags |= SCAN_NOSKIP;
		break;
	}

	/*
	 * At this point, we will need additional characters from the
	 * string to proceed.
	 */

	if (*string == '\0') {
	    underflow = 1;
	    goto done;
	}
	
	/*
	 * Skip any leading whitespace at the beginning of a field unless
	 * the format suppresses this behavior.
	 */

	if (!(flags & SCAN_NOSKIP)) {
	    while (*string != '\0') {
		offset = Tcl_UtfToUniChar(string, &sch);
		if (!Tcl_UniCharIsSpace(sch)) {
		    break;
		}
		string += offset;
	    }
	    if (*string == '\0') {
		underflow = 1;
		goto done;
	    }
	}

	/*
	 * Perform the requested scanning operation.
	 */
	
	switch (op) {
	    case 's':
		/*
		 * Scan a string up to width characters or whitespace.
		 */

		if (width == 0) {
		    width = (size_t) ~0;
		}
		end = string;
		while (*end != '\0') {
		    offset = Tcl_UtfToUniChar(end, &sch);
		    if (Tcl_UniCharIsSpace(sch)) {
			break;
		    }
		    end += offset;
		    if (--width == 0) {
			break;
		    }
		}
		if (!(flags & SCAN_SUPPRESS)) {
		    objPtr = Tcl_NewStringObj(string, end-string);
		    Tcl_IncrRefCount(objPtr);
		    objs[objIndex++] = objPtr;
		}
		string = end;
		break;

	    case '[': {
		CharSet cset;

		if (width == 0) {
		    width = (size_t) ~0;
		}
		end = string;

		format = BuildCharSet(&cset, format);
		while (*end != '\0') {
		    offset = Tcl_UtfToUniChar(end, &sch);
		    if (!CharInSet(&cset, (int)sch)) {
			break;
		    }
		    end += offset;
		    if (--width == 0) {
			break;
		    }
		}
		ReleaseCharSet(&cset);

		if (string == end) {
		    /*
		     * Nothing matched the range, stop processing
		     */
		    goto done;
		}
		if (!(flags & SCAN_SUPPRESS)) {
		    objPtr = Tcl_NewStringObj(string, end-string);
		    Tcl_IncrRefCount(objPtr);
		    objs[objIndex++] = objPtr;
		}
		string = end;
		
		break;
	    }
	    case 'c':
		/*
		 * Scan a single Unicode character.
		 */

		string += Tcl_UtfToUniChar(string, &sch);
		if (!(flags & SCAN_SUPPRESS)) {
		    objPtr = Tcl_NewIntObj((int)sch);
		    Tcl_IncrRefCount(objPtr);
		    objs[objIndex++] = objPtr;
		}
		break;

	    case 'i':
		/*
		 * Scan an unsigned or signed integer.
		 */

		if ((width == 0) || (width > sizeof(buf) - 1)) {
		    width = sizeof(buf) - 1;
		}
		flags |= SCAN_SIGNOK | SCAN_NODIGITS | SCAN_NOZERO;
		for (end = buf; width > 0; width--) {
		    switch (*string) {
			/*
			 * The 0 digit has special meaning at the beginning of
			 * a number.  If we are unsure of the base, it
			 * indicates that we are in base 8 or base 16 (if it is
			 * followed by an 'x').
			 *
			 * 8.1 - 8.3.4 incorrectly handled 0x... base-16
			 * cases for %x by not reading the 0x as the
			 * auto-prelude for base-16. [Bug #495213]
			 */
			case '0':
			    if (base == 0) {
				base = 8;
				flags |= SCAN_XOK;
			    }
			    if (base == 16) {
				flags |= SCAN_XOK;
			    }
			    if (flags & SCAN_NOZERO) {
				flags &= ~(SCAN_SIGNOK | SCAN_NODIGITS
					| SCAN_NOZERO);
			    } else {
				flags &= ~(SCAN_SIGNOK | SCAN_XOK
					| SCAN_NODIGITS);
			    }
			    goto addToInt;

			case '1': case '2': case '3': case '4':
			case '5': case '6': case '7':
			    if (base == 0) {
				base = 10;
			    }
			    flags &= ~(SCAN_SIGNOK | SCAN_XOK | SCAN_NODIGITS);
			    goto addToInt;

			case '8': case '9':
			    if (base == 0) {
				base = 10;
			    }
			    if (base <= 8) {
				break;
			    }
			    flags &= ~(SCAN_SIGNOK | SCAN_XOK | SCAN_NODIGITS);
			    goto addToInt;

			case 'A': case 'B': case 'C':
			case 'D': case 'E': case 'F': 
			case 'a': case 'b': case 'c':
			case 'd': case 'e': case 'f':
			    if (base <= 10) {
				break;
			    }
			    flags &= ~(SCAN_SIGNOK | SCAN_XOK | SCAN_NODIGITS);
			    goto addToInt;

			case '+': case '-':
			    if (flags & SCAN_SIGNOK) {
				flags &= ~SCAN_SIGNOK;
				goto addToInt;
			    }
			    break;

			case 'x': case 'X':
			    if ((flags & SCAN_XOK) && (end == buf+1)) {
				base = 16;
				flags &= ~SCAN_XOK;
				goto addToInt;
			    }
			    break;
		    }

		    /*
		     * We got an illegal character so we are done accumulating.
		     */

		    break;

		    addToInt:
		    /*
		     * Add the character to the temporary buffer.
		     */

		    *end++ = *string++;
		    if (*string == '\0') {
			break;
		    }
		}

		/*
		 * Check to see if we need to back up because we only got a
		 * sign or a trailing x after a 0.
		 */

		if (flags & SCAN_NODIGITS) {
		    if (*string == '\0') {
			underflow = 1;
		    }
		    goto done;
		} else if (end[-1] == 'x' || end[-1] == 'X') {
		    end--;
		    string--;
		}


		/*
		 * Scan the value from the temporary buffer.  If we are
		 * returning a large unsigned value, we have to convert it back
		 * to a string since Tcl only supports signed values.
		 */

		if (!(flags & SCAN_SUPPRESS)) {
		    *end = '\0';
#ifndef TCL_WIDE_INT_IS_LONG
		    if (flags & SCAN_LONGER) {
			wideValue = (Tcl_WideInt) (*lfn)(buf, NULL, base);
			if ((flags & SCAN_UNSIGNED) && (wideValue < 0)) {
			    /* INTL: ISO digit */
			    sprintf(buf, "%" TCL_LL_MODIFIER "u",
				    (Tcl_WideUInt)wideValue);
			    objPtr = Tcl_NewStringObj(buf, -1);
			} else {
			    objPtr = Tcl_NewWideIntObj(wideValue);
			}
		    } else {
#endif /* !TCL_WIDE_INT_IS_LONG */
			value = (long) (*fn)(buf, NULL, base);
			if ((flags & SCAN_UNSIGNED) && (value < 0)) {
			    sprintf(buf, "%lu", value); /* INTL: ISO digit */
			    objPtr = Tcl_NewStringObj(buf, -1);
			} else {
			    if ((unsigned long) value > UINT_MAX) {
				objPtr = Tcl_NewLongObj(value);
			    } else {
				objPtr = Tcl_NewIntObj(value);
			    }
			}
#ifndef TCL_WIDE_INT_IS_LONG
		    }
#endif
		    Tcl_IncrRefCount(objPtr);
		    objs[objIndex++] = objPtr;
		}

		break;

	    case 'f':
		/*
		 * Scan a floating point number
		 */

		if ((width == 0) || (width > sizeof(buf) - 1)) {
		    width = sizeof(buf) - 1;
		}
		flags &= ~SCAN_LONGER;
		flags |= SCAN_SIGNOK | SCAN_NODIGITS | SCAN_PTOK | SCAN_EXPOK;
		for (end = buf; width > 0; width--) {
		    switch (*string) {
			case '0': case '1': case '2': case '3':
			case '4': case '5': case '6': case '7':
			case '8': case '9':
			    flags &= ~(SCAN_SIGNOK | SCAN_NODIGITS);
			    goto addToFloat;
			case '+': case '-':
			    if (flags & SCAN_SIGNOK) {
				flags &= ~SCAN_SIGNOK;
				goto addToFloat;
			    }
			    break;
			case '.':
			    if (flags & SCAN_PTOK) {
				flags &= ~(SCAN_SIGNOK | SCAN_PTOK);
				goto addToFloat;
			    }
			    break;
			case 'e': case 'E':
			    /*
			     * An exponent is not allowed until there has
			     * been at least one digit.
			     */

			    if ((flags & (SCAN_NODIGITS | SCAN_EXPOK))
				    == SCAN_EXPOK) {
				flags = (flags & ~(SCAN_EXPOK|SCAN_PTOK))
				    | SCAN_SIGNOK | SCAN_NODIGITS;
				goto addToFloat;
			    }
			    break;
		    }

		    /*
		     * We got an illegal character so we are done accumulating.
		     */

		    break;

		    addToFloat:
		    /*
		     * Add the character to the temporary buffer.
		     */

		    *end++ = *string++;
		    if (*string == '\0') {
			break;
		    }
		}

		/*
		 * Check to see if we need to back up because we saw a
		 * trailing 'e' or sign.
		 */

		if (flags & SCAN_NODIGITS) {
		    if (flags & SCAN_EXPOK) {
			/*
			 * There were no digits at all so scanning has
			 * failed and we are done.
			 */
			if (*string == '\0') {
			    underflow = 1;
			}
			goto done;
		    }

		    /*
		     * We got a bad exponent ('e' and maybe a sign).
		     */

		    end--;
		    string--;
		    if (*end != 'e' && *end != 'E') {
			end--;
			string--;
		    }
		}

		/*
		 * Scan the value from the temporary buffer.
		 */

		if (!(flags & SCAN_SUPPRESS)) {
		    double dvalue;
		    *end = '\0';
		    dvalue = strtod(buf, NULL);
		    objPtr = Tcl_NewDoubleObj(dvalue);
		    Tcl_IncrRefCount(objPtr);
		    objs[objIndex++] = objPtr;
		}
		break;
	}
	nconversions++;
    }

    done:
    result = 0;
    code = TCL_OK;

    if (numVars) {
	/*
	 * In this case, variables were specified (classic scan)
	 */
	for (i = 0; i < totalVars; i++) {
	    if (objs[i] != NULL) {
		result++;
		if (Tcl_ObjSetVar2(interp, objv[i+3], NULL,
			objs[i], 0) == NULL) {
		    Tcl_AppendStringsToObj(Tcl_GetObjResult(interp),
			    "couldn't set variable \"",
			    Tcl_GetString(objv[i+3]), "\"", (char *) NULL);
		    code = TCL_ERROR;
		}
		Tcl_DecrRefCount(objs[i]);
	    }
	}
    } else {
	/*
	 * Here no vars were specified, we want a list returned (inline scan)
	 */
	objPtr = Tcl_NewObj();
	for (i = 0; i < totalVars; i++) {
	    if (objs[i] != NULL) {
		Tcl_ListObjAppendElement(NULL, objPtr, objs[i]);
		Tcl_DecrRefCount(objs[i]);
	    } else {
		/*
		 * More %-specifiers than matching chars, so we
		 * just spit out empty strings for these
		 */
		Tcl_ListObjAppendElement(NULL, objPtr, Tcl_NewObj());
	    }
	}
    }
    if (objs != NULL) {
	ckfree((char*) objs);
    }
    if (code == TCL_OK) {
	if (underflow && (nconversions == 0)) {
	    if (numVars) {
		objPtr = Tcl_NewIntObj(-1);
	    } else {
		if (objPtr) {
		    Tcl_SetListObj(objPtr, 0, NULL);
		} else {
		    objPtr = Tcl_NewObj();
		}
	    }
	} else if (numVars) {
	    objPtr = Tcl_NewIntObj(result);
	}
	Tcl_SetObjResult(interp, objPtr);
    }
    return code;
}