icuSources/tools/toolutil/ucbuf.h

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 1998-2016, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 *******************************************************************************
   8 *
   9 * File ucbuf.h
  10 *
  11 * Modification History:
  12 *
  13 *   Date        Name        Description
  14 *   05/10/01    Ram         Creation.
  15 *
  16 * This API reads in files and returns UChars
  17 *******************************************************************************
  18 */
  19
  20 #include "unicode/localpointer.h"
  21 #include "unicode/ucnv.h"
  22 #include "filestrm.h"
  23
  24 #if !UCONFIG_NO_CONVERSION
  25
  26 #ifndef UCBUF_H
  27 #define UCBUF_H 1
  28
  29 typedef struct UCHARBUF UCHARBUF;
  30 /**
  31  * End of file value
  32  */
  33 #define U_EOF 0xFFFFFFFF
  34 /**
  35  * Error value if a sequence cannot be unescaped
  36  */
  37 #define U_ERR 0xFFFFFFFE
  38
  39 typedef struct ULine ULine;
  40
  41 struct  ULine {
  42     UChar     *name;
  43     int32_t   len;
  44 };
  45
  46 /**
  47  * Opens the UCHARBUF with the given file stream and code page for conversion
  48  * @param fileName  Name of the file to open.
  49  * @param codepage  The encoding of the file stream to convert to Unicode.
  50  *                  If *codepoge is NULL on input the API will try to autodetect
  51  *                  popular Unicode encodings
  52  * @param showWarning Flag to print out warnings to STDOUT
  53  * @param buffered  If TRUE performs a buffered read of the input file. If FALSE reads
  54  *                  the whole file into memory and converts it.
  55  * @param err is a pointer to a valid <code>UErrorCode</code> value. If this value
  56  *        indicates a failure on entry, the function will immediately return.
  57  *        On exit the value will indicate the success of the operation.
  58  * @return pointer to the newly opened UCHARBUF
  59  */
  60 U_CAPI UCHARBUF* U_EXPORT2
  61 ucbuf_open(const char* fileName,const char** codepage,UBool showWarning, UBool buffered, UErrorCode* err);
  62
  63 /**
  64  * Gets a UTF-16 code unit at the current position from the converted buffer
  65  * and increments the current position
  66  * @param buf Pointer to UCHARBUF structure
  67  * @param err is a pointer to a valid <code>UErrorCode</code> value. If this value
  68  *        indicates a failure on entry, the function will immediately return.
  69  *        On exit the value will indicate the success of the operation.
  70  */
  71 U_CAPI int32_t U_EXPORT2
  72 ucbuf_getc(UCHARBUF* buf,UErrorCode* err);
  73
  74 /**
  75  * Gets a UTF-32 code point at the current position from the converted buffer
  76  * and increments the current position
  77  * @param buf Pointer to UCHARBUF structure
  78  * @param err is a pointer to a valid <code>UErrorCode</code> value. If this value
  79  *        indicates a failure on entry, the function will immediately return.
  80  *        On exit the value will indicate the success of the operation.
  81  */
  82 U_CAPI int32_t U_EXPORT2
  83 ucbuf_getc32(UCHARBUF* buf,UErrorCode* err);
  84
  85 /**
  86  * Gets a UTF-16 code unit at the current position from the converted buffer after
  87  * unescaping and increments the current position. If the escape sequence is for UTF-32
  88  * code point (\\Uxxxxxxxx) then a UTF-32 codepoint is returned
  89  * @param buf Pointer to UCHARBUF structure
  90  * @param err is a pointer to a valid <code>UErrorCode</code> value. If this value
  91  *        indicates a failure on entry, the function will immediately return.
  92  *        On exit the value will indicate the success of the operation.
  93  */
  94 U_CAPI int32_t U_EXPORT2
  95 ucbuf_getcx32(UCHARBUF* buf,UErrorCode* err);
  96
  97 /**
  98  * Gets a pointer to the current position in the internal buffer and length of the line.
  99  * It imperative to make a copy of the returned buffer before performing operations on it.
 100  * @param buf Pointer to UCHARBUF structure
 101  * @param len Output param to receive the len of the buffer returned till end of the line
 102  * @param err is a pointer to a valid <code>UErrorCode</code> value. If this value
 103  *        indicates a failure on entry, the function will immediately return.
 104  *        On exit the value will indicate the success of the operation.
 105  *        Error: U_TRUNCATED_CHAR_FOUND
 106  * @return Pointer to the internal buffer, NULL if EOF
 107  */
 108 U_CAPI const UChar* U_EXPORT2
 109 ucbuf_readline(UCHARBUF* buf,int32_t* len, UErrorCode* err);
 110
 111
 112 /**
 113  * Resets the buffers and the underlying file stream.
 114  * @param buf Pointer to UCHARBUF structure
 115  * @param err is a pointer to a valid <code>UErrorCode</code> value. If this value
 116  *        indicates a failure on entry, the function will immediately return.
 117  *        On exit the value will indicate the success of the operation.
 118  */
 119 U_CAPI void U_EXPORT2
 120 ucbuf_rewind(UCHARBUF* buf,UErrorCode* err);
 121
 122 /**
 123  * Returns a pointer to the internal converted buffer
 124  * @param buf Pointer to UCHARBUF structure
 125  * @param len Pointer to int32_t to receive the lenth of buffer
 126  * @param err is a pointer to a valid <code>UErrorCode</code> value. If this value
 127  *        indicates a failure on entry, the function will immediately return.
 128  *        On exit the value will indicate the success of the operation.
 129  * @return Pointer to internal UChar buffer
 130  */
 131 U_CAPI const UChar* U_EXPORT2
 132 ucbuf_getBuffer(UCHARBUF* buf,int32_t* len,UErrorCode* err);
 133
 134 /**
 135  * Closes the UCHARBUF structure members and cleans up the malloc'ed memory
 136  * @param buf Pointer to UCHARBUF structure
 137  */
 138 U_CAPI void U_EXPORT2
 139 ucbuf_close(UCHARBUF* buf);
 140
 141 #if U_SHOW_CPLUSPLUS_API
 142
 143 U_NAMESPACE_BEGIN
 144
 145 /**
 146  * \class LocalUCHARBUFPointer
 147  * "Smart pointer" class, closes a UCHARBUF via ucbuf_close().
 148  * For most methods see the LocalPointerBase base class.
 149  *
 150  * @see LocalPointerBase
 151  * @see LocalPointer
 152  */
 153 U_DEFINE_LOCAL_OPEN_POINTER(LocalUCHARBUFPointer, UCHARBUF, ucbuf_close);
 154
 155 U_NAMESPACE_END
 156
 157 #endif
 158
 159 /**
 160  * Rewinds the buffer by one codepoint. Does not rewind over escaped characters.
 161  */
 162 U_CAPI void U_EXPORT2
 163 ucbuf_ungetc(int32_t ungetChar,UCHARBUF* buf);
 164
 165
 166 /**
 167  * Autodetects the encoding of the file stream. Only Unicode charsets are autodectected.
 168  * Some Unicode charsets are stateful and need byte identifiers to be converted also to bring
 169  * the converter to correct state for converting the rest of the stream. So the UConverter parameter
 170  * is necessary.
 171  * If the charset was autodetected, the caller must close both the input FileStream
 172  * and the converter.
 173  *
 174  * @param fileName The file name to be opened and encoding autodected
 175  * @param conv  Output param to receive the opened converter if autodetected; NULL otherwise.
 176  * @param cp Output param to receive the detected encoding
 177  * @param err is a pointer to a valid <code>UErrorCode</code> value. If this value
 178  *        indicates a failure on entry, the function will immediately return.
 179  *        On exit the value will indicate the success of the operation.
 180  * @return The input FileStream if its charset was autodetected; NULL otherwise.
 181  */
 182 U_CAPI FileStream * U_EXPORT2
 183 ucbuf_autodetect(const char* fileName, const char** cp,UConverter** conv,
 184 int32_t* signatureLength, UErrorCode* status);
 185
 186 /**
 187  * Autodetects the encoding of the file stream. Only Unicode charsets are autodectected.
 188  * Some Unicode charsets are stateful and need byte identifiers to be converted also to bring
 189  * the converter to correct state for converting the rest of the stream. So the UConverter parameter
 190  * is necessary.
 191  * If the charset was autodetected, the caller must close the converter.
 192  *
 193  * @param fileStream The file stream whose encoding is to be detected
 194  * @param conv  Output param to receive the opened converter if autodetected; NULL otherwise.
 195  * @param cp Output param to receive the detected encoding
 196  * @param err is a pointer to a valid <code>UErrorCode</code> value. If this value
 197  *        indicates a failure on entry, the function will immediately return.
 198  *        On exit the value will indicate the success of the operation.
 199  * @return Boolean whether the Unicode charset was autodetected.
 200  */
 201
 202 U_CAPI UBool U_EXPORT2
 203 ucbuf_autodetect_fs(FileStream* in, const char** cp, UConverter** conv, int32_t* signatureLength, UErrorCode* status);
 204
 205 /**
 206  * Returns the approximate size in UChars required for converting the file to UChars
 207  */
 208 U_CAPI int32_t U_EXPORT2
 209 ucbuf_size(UCHARBUF* buf);
 210
 211 U_CAPI const char* U_EXPORT2
 212 ucbuf_resolveFileName(const char* inputDir, const char* fileName, char* target, int32_t* len, UErrorCode* status);
 213
 214 #endif
 215 #endif
 216