batch-file - 需要从 1 行很长的文本文件中替换 13 个空格

Question

我有一个文件 (1.8 Mb)，其中有 1 行(很长的)文本。该行上的值通常由 13 个空格分隔。我想做的是用管道 | 替换这 13 个空格分隔符，以便我可以使用 SSIS 处理此文本文件。

到目前为止，我没有成功地使用批处理文件以编程方式处理此文件。

我尝试使用从另一篇 SO 帖子中获得的以下代码。

    @echo off 
REM create empty file:
break>R1.txt
setlocal enabledelayedexpansion
REM prevent empty lines by adding line numbers (find /v /n "")
REM parse the file, taking the second token (*, %%b) with delimiters
REM ] (to eliminate line numbers) and space (to eliminate leading spaces)
for /f "tokens=1,* delims=] " %%a in ('find /v /n "" ^<PXZP_SND_XZ01_GFT10553.dat') do (
  call :sub1 "%%b"
  REM write the string without quotes:
  REM removing the qoutes from the string would make the special chars poisonous again
  >>PXZP_SND_XZ01_GFT10553.dat echo(!s:"=!
)

REM Show the written file:
type PXZP_SND_XZ01_GFT10553.dat 
goto :eof

:sub1
set S=%*
REM do 13 times (adapt to your Needs):
for /l %%i in (1,1,13) do (
  REM replace "space qoute" with "quote" (= removing the last space
  set S=!S: "=|!
)
goto :eof

有人可以帮我吗？我的文本文件示例:

96859471/971 AAAA HAWAII               96860471/971 BBBB HAWAII               96861471/971 CCCC HAWAII               96863471/971 DDDD HAWAII               

Answer 1

for /F loop无法处理超过 8190 个字符的行。但是，有一种方法可以读取行数较长的文件:使用 set /P 在一个循环中，连同 input redirection < ; set /P最多读取 1023 个字符，除非遇到换行符或文件末尾；自 set /P 以来，对同一个打开(输入重定向)文件句柄多次执行它允许读取 1023 个字符部分的非常长的行不重置文件指针。

另一个挑战是返回(回显)很长的行，这对于 echo command 是不可能的再次因为大约 8190 个字符的行限制(适用于命令行和变量内容)。同样，按 block 处理也有帮助:首先，获取文件结束字符(EOF，ASCII 0x1A)；然后取一个文本/字符串部分，附加一个 EOF 并使用 echo 将结果写入临时文件(附加一个换行符)，连同 output redirection > ;接下来使用 copy 将文件复制到自身上, 但以 ASCII 文本模式读取它以丢弃 EOF 和之后的所有内容(因此先前由 echo 附加的换行符)并以二进制模式写入它以获得结果数据的精确副本;最后，使用 type 输入文件内容.

以下脚本使用了这些技术(请参阅代码中的所有解释性 rem 注释):

@echo off
setlocal EnableExtensions DisableDelayedexpansion

rem // Define constants here:
set "_INPUT=.\PXZP_SND_XZ01_GFT10553.dat" & rem // (this is the input file)
set "_OUTPUT=.\R1.txt" & rem // (set to `con` to display the result on the console)
set "_TEMPF=%TEMP%\%~n0_%RANDOM%.tmp" & rem // (specifies a temporary file)
set "_SEARCH=     " & rem // (this is the string to be found)
set "_REPLAC=|"     & rem // (this is the replacement string)
set "_LTRIM=#"      & rem // (set to something to left-trim sub-strings)
(set _LF=^
%= blank line =%
) & rem // (this block stores a new-line character in a variable)
rem // This stores an end-of-file character in a variable:
for /F %%E in ('forfiles /P "%~dp0." /M "%~nx0" /C "cmd /C echo 0x1A"') do set "_EOF=%%E"

rem /* The input file is going to be processed in a sub-routine,
rem    which accesses the file content via input redirection `<`: */
< "%_INPUT%" > "%_OUTPUT%" call :PROCESS

endlocal
exit /B


:PROCESS
    rem // Reset variables that store a partial string to be processed and a separator:
    set "PART=" & set "SEP="
    setlocal EnableDelayedExpansion
:READ
    rem /* At this point 1023 characters are read from the input file at most, until
    rem    a line-break or the end of the file is encountered:*/
    set "NEW=" & set /P NEW=""
    rem // The read characters are appended to a string buffer that will be processed:
    set "PART=!PART!!NEW!"
    rem /* Skip processing when the string buffer is empty, which is the case when the end
    rem    of the file has already been reached: */
:LOOP
    if defined PART (
        rem /* Make the search string accessible as a `for` meta-variable reference in
        rem    to not have to use normal (immediate) `%`-expansion, which could cause
        rem    trouble with some special characters under some circumstances: */
        for /F delims^=^ eol^= %%K in ("!_SEARCH!") do (
            rem /* Try to split the string buffer at the first search string and store the
            rem    portion at the right, using sub-string substitution: */
            set "RIGHT=!PART:*%%K=!"
            rem /* Check whether the split was successful, hence whether a search string
            rem    even occurred in the string buffer; if not, jump back and read more
            rem    characters; otherwise (when the end of the file was reached) clear the
            rem    right portion and continue processing: */
            if "!RIGHT!"=="!PART!" if not defined NEW (set "RIGHT=") else goto :READ
            rem /* Clear the variable that will receive the portion left to the first
            rem    occurrence of the search string in the string buffer; then replace each
            rem    occurrence in the string buffer by a new-line character: */
            set "LEFT=" & set ^"PART=!PART:%%K=^%_LF%%_LF%!^"
            rem /* Iterate over all lines of the altered string buffer, which is now a
            rem    multi-line string, then get the first line, which constitutes the
            rem    portion at the left of the first search string; the (first) line is
            rem    preceded by an `_` just for it not to appear blank, because `for /F`
            rem    skips over empty lines; this character is removed later: */
            for /F delims^=^ eol^= %%L in (^"_!PART!^") do (
                rem // Execute the loop body only for the first iteration:
                if not defined LEFT (
                    rem /* Store the (augmented) left portion with delayed expansion
                    rem    disabled in order not to get trouble with `!` in the string: */
                    setlocal DisableDelayedExpansion & set "LEFT=%%L"
                    rem // Enable delayed expansion to be able to safely echo the string:
                    setlocal EnableDelayedExpansion
                    rem /* Write to a temporary file the output string, which consists of
                    rem    a replacement string (except for the very first time), the left
                    rem    portion with the preceding `_` removed and an end-of-file
                    rem    character; a line-break is automatically appended by `echo`: */
                    > "!_TEMPF!" echo(!SEP!!LEFT:~1!%_EOF%
                    rem /* Copy the temporary file onto itself, but remove the end-of-file
                    rem    character and everything after, then type the file content;
                    rem    this is a safe way of echoing a string without a line-break: */
                    > nul copy /Y /A "!_TEMPF!" + nul "!_TEMPF!" /B & type "!_TEMPF!"
                    rem /* Restore the environment present at the beginning of the loop
                    rem    body, then ensure the left portion not to appear empty: */
                    endlocal & endlocal & set "LEFT=_"
                )
            )
            rem // If specified, left-trim the right portion, so remove leading spaces:
            if defined _LTRIM (
                for /F "tokens=* eol=  delims= " %%T in ("!RIGHT!_") do (
                    for /F delims^=^ eol^= %%S in (^""!NEW!"^") do (
                        endlocal & set "NEW=%%~S" & set "RIGHT=%%T"
                    )
                    setlocal EnableDelayedExpansion & set "RIGHT=!RIGHT:~,-1!"
                )
            )
            rem // Set the replacement string now to skip it only for the first output:
            set "SEP=!_REPLAC!"
            rem /* Move the right portion into the string buffer; if there is still some
            rem    amount of text left, jump back to find more occurrences of the search
            rem    string; if not, jump back and read more characters, unless the end of
            rem    the file has already been reached: */
            set "PART=!RIGHT!" & if defined PART (
                if defined NEW if "!PART:~1024!"=="" goto :READ
                goto :LOOP
            ) else if defined NEW goto :READ
        )
    )
    endlocal
    rem // Clean up the temporary file:
    del "%_TEMPF%"
    exit /B

存在以下限制:

• 两个连续搜索字符串之间的字符串部分(= 5 × SPACE 在上述方法中)必须短于大约 8190 个字符；
• 搜索字符串不能为空，不能以!开头, * , ~并且不得包含 = ;
• 替换字符串不能包含! ;