我的gets_ws功能

< snip> - Richard Heathfield Usenet是一个奇怪的地方 - dmr 29/7/1999 http://www.cpax.org.uk 电子邮件：rjh在上面的域名（但显然放弃了www） 让我们为maxlen设置size_t，不是吗？我们不想去/那里/再次！ 我确实有一个size_t，但后来我在 函数并且不希望我的lint因为比较size_t和 int而惹我生气，或者其他东西。 （回想起来，这可能是一个 的案例，其中一个明确的演员是值得的。） 我会解决它。 这里有一个使用getc而不是fgetc的情况。因为标准的getc被给予 显式许可，不止一次地评估它的参数 当作为宏实现时（fgetc是* not *给出的许可），它可能 可能让实现者更快地使getc。可能不是很多b $ b，但是要记住这一点。 直觉表明getc在stdin工作获得 的方式（并且put，putc在stdout上工作）。对于维护程序员，恕我直言，这可能会让人感到困惑。 其他人已经注意到这里没有检查EOF。 和/那是我发布的原因这对clc。因为我的所有目的，我已经亲自验证了输入，我从来没有在测试中发现 。 而不是ungetcing ch并返回0，你可以简单地将 跟随while循环包含在''else''块中。 是的，那会更清楚，不是吗。 如果maxlen为0怎么办？ i< -1会失败，因为我最初为0.我确实检查了这个条件。 然而，它会设置为buff [0]到''\0''，即使我'' m告诉 buff有0，而不是1个字节。我相信另一个断言（）在这里是 。 buff [i ++] = ch;在语义上等同于上面两行，并且 足够简单和惯用，它不会让任何维护 代码的人感到困惑。 Okey doke。谢谢！ - Andrew Poelstra< http://www.wpsoftware.net/projects/> 要给我发电子邮件，请使用apoelstra。在上述领域。 你们人讨厌数学。 - 詹姆斯哈里斯 int gets_ws (char *buff, int maxlen, int sc, FILE *fh); This function reads up to maxlen characters from fh, stopping when itencounters whitespace, sc, or EOF. If EOF is encountered, the functionreturns EOF. If buff is NULL, read characters are discarded. Otherwise, they areput into buff. The function returns the number of characters read or0 in the case of a NULL buffer. I use it in my XML parser to read a tag name:gets_ws (buff, max, ''>'', fh);and if it reads max characters (by returning max), I simply deletethe remaining characters:gets_ws (NULL, 0, ''>'', fh); Here is the code: gets_ws.c:/* Reads from stream up until first whitespace or sc character. ** If given a non-NULL buffer, it will fill that. Otherwise, it ** will simply go through the characters and throw them away. ** Returns number of characters read or EOF on failure. */static int gets_ws (char *buff, size_t maxlen, int sc, FILE *fh){int i = 0;int ch = fgetc (fh); assert (fh != NULL);if (buff == NULL){while (!isspace (ch) && ch != sc)ch = fgetc (fh); ungetc (ch);return 0;} while (i < (maxlen - 1) && !isspace (ch) && ch != sc){if (ch == EOF)return EOF; buff[i] = ch;++i; ch = fgetc (fh);}ungetc (ch); return i;} Comments or suggestions are welcome. --Andrew Poelstra <http://www.wpsoftware.net/projects/>To email me, use "apoelstra" at the above domain."You people hate mathematics." -- James Harris 解决方案 <snip><snip> This "discard loop" should also have a check for EOF, shouldn''t it? /Michael BrennanLet''s have a size_t for maxlen, shall we? We don''t want to go /there/ again! There is a case for using getc rather than fgetc here. Because getc is givenexplicit licence by the Standard to evaluate its argument more than oncewhen implemented as a macro (a licence that fgetc is *not* given), it maybe possible for the implementor to make getc a bit quicker. Probably notmuch in it, but it''s something to bear in mind. Someone else has already noted the lack of a check for EOF here. Instead of ungetc-ing ch and returning 0 here, you could simply enclose thefollowing while loop in an ''else'' block. What if maxlen is 0? buff[i++] = ch; is semantically equivalent to the above two lines, and issufficiently simple and idiomatic that it won''t confuse anyone maintainingthe code. <snip> --Richard Heathfield"Usenet is a strange place" - dmr 29/7/1999 http://www.cpax.org.ukemail: rjh at above domain (but drop the www, obviously) Let''s have a size_t for maxlen, shall we? We don''t want to go /there/ again! I did have a size_t, but then I was returning -1 in one version of thefunction and didn''t want my lint to nab me for comparing a size_t andint, or something to that effect. (In retrospect, that may have been acase where an explicit cast was merited.) I''ll fix it. There is a case for using getc rather than fgetc here. Because getc is givenexplicit licence by the Standard to evaluate its argument more than oncewhen implemented as a macro (a licence that fgetc is *not* given), it maybe possible for the implementor to make getc a bit quicker. Probably notmuch in it, but it''s something to bear in mind. Intuition suggests that getc works from stdin in the same way that getsdoes (and puts, putc work on stdout). It could potentially be confusingfor a maintenance programmer, IMHO. Someone else has already noted the lack of a check for EOF here. And /that/ is why I posted this to clc. Since for all my purposes, Ihave personally validated the input, I never would have spotted thatin testing. Instead of ungetc-ing ch and returning 0 here, you could simply enclose thefollowing while loop in an ''else'' block. Yes, that would be more clear, wouldn''t it. What if maxlen is 0? i < -1 will fail, as i is initially 0. I did check that condition.However, it will then set to buff[0] to ''\0'', even though I''m toldthat buff has 0, not 1 bytes. I believe another assert() is in orderhere. buff[i++] = ch; is semantically equivalent to the above two lines, and issufficiently simple and idiomatic that it won''t confuse anyone maintainingthe code. Okey doke. Thanks! --Andrew Poelstra <http://www.wpsoftware.net/projects/>To email me, use "apoelstra" at the above domain."You people hate mathematics." -- James Harris 这篇关于我的gets_ws功能的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持！