Re: [PATCH] strbuf: add strbuf_add_real_path()
On 02/27, René Scharfe wrote: > Am 27.02.2017 um 19:22 schrieb Brandon Williams: > >On 02/25, René Scharfe wrote: > >>+void strbuf_add_real_path(struct strbuf *sb, const char *path) > >>+{ > >>+ if (sb->len) { > >>+ struct strbuf resolved = STRBUF_INIT; > >>+ strbuf_realpath(&resolved, path, 1); > >>+ strbuf_addbuf(sb, &resolved); > >>+ strbuf_release(&resolved); > >>+ } else > >>+ strbuf_realpath(sb, path, 1); > > > >I know its not required but I would have braces on the 'else' branch > >since they were needed on the 'if' branch. But that's up to you and > >your style :) > > Personally I'd actually prefer them as well, but the project's style > has traditionally been to avoid braces on such trailing single-line > branches to save lines. The CodingGuidelines for this topic have > been clarified recently, though, and seem to require them now. > Interesting. > > René Having the project's guidelines align with your own preference makes things a bit easier! -- Brandon Williams
Re: [PATCH] strbuf: add strbuf_add_real_path()
Am 27.02.2017 um 19:22 schrieb Brandon Williams: On 02/25, René Scharfe wrote: +void strbuf_add_real_path(struct strbuf *sb, const char *path) +{ + if (sb->len) { + struct strbuf resolved = STRBUF_INIT; + strbuf_realpath(&resolved, path, 1); + strbuf_addbuf(sb, &resolved); + strbuf_release(&resolved); + } else + strbuf_realpath(sb, path, 1); I know its not required but I would have braces on the 'else' branch since they were needed on the 'if' branch. But that's up to you and your style :) Personally I'd actually prefer them as well, but the project's style has traditionally been to avoid braces on such trailing single-line branches to save lines. The CodingGuidelines for this topic have been clarified recently, though, and seem to require them now. Interesting. René
Re: [PATCH] strbuf: add strbuf_add_real_path()
On 02/25, René Scharfe wrote: > Add a function for appending the canonized absolute pathname of a given > path to a strbuf. It keeps the existing contents intact, as expected of > a function of the strbuf_add() family, while avoiding copying the result > if the given strbuf is empty. It's more consistent with the rest of the > strbuf API than strbuf_realpath(), which it's wrapping. > > Also add a semantic patch demonstrating its intended usage and apply it > to the current tree. Using strbuf_add_real_path() instead of calling > strbuf_addstr() and real_path() avoids an extra copy to a static buffer. > Seems like a reasonable thing to do. When I wrote strbuf_realpath() I think I looked at the strbuf_getcwd() function for what it did (since it handled paths) and it simply uses the provided buffer disregarding what is already stored in it. > Signed-off-by: Rene Scharfe > --- > contrib/coccinelle/strbuf.cocci | 6 ++ > setup.c | 2 +- > strbuf.c| 11 +++ > strbuf.h| 14 ++ > 4 files changed, 32 insertions(+), 1 deletion(-) > > diff --git a/contrib/coccinelle/strbuf.cocci b/contrib/coccinelle/strbuf.cocci > index 63995f22ff..1d580e49b0 100644 > --- a/contrib/coccinelle/strbuf.cocci > +++ b/contrib/coccinelle/strbuf.cocci > @@ -38,3 +38,9 @@ expression E1, E2, E3; > @@ > - strbuf_addstr(E1, find_unique_abbrev(E2, E3)); > + strbuf_add_unique_abbrev(E1, E2, E3); > + > +@@ > +expression E1, E2; > +@@ > +- strbuf_addstr(E1, real_path(E2)); > ++ strbuf_add_real_path(E1, E2); > diff --git a/setup.c b/setup.c > index 967f289f1e..f14cbcd338 100644 > --- a/setup.c > +++ b/setup.c > @@ -254,7 +254,7 @@ int get_common_dir_noenv(struct strbuf *sb, const char > *gitdir) > if (!is_absolute_path(data.buf)) > strbuf_addf(&path, "%s/", gitdir); > strbuf_addbuf(&path, &data); > - strbuf_addstr(sb, real_path(path.buf)); > + strbuf_add_real_path(sb, path.buf); > ret = 1; > } else { > strbuf_addstr(sb, gitdir); > diff --git a/strbuf.c b/strbuf.cq > index 8fec6579f7..ace58e7367 100644 > --- a/strbuf.c > +++ b/strbuf.c > @@ -707,6 +707,17 @@ void strbuf_add_absolute_path(struct strbuf *sb, const > char *path) > strbuf_addstr(sb, path); > } > > +void strbuf_add_real_path(struct strbuf *sb, const char *path) > +{ > + if (sb->len) { > + struct strbuf resolved = STRBUF_INIT; > + strbuf_realpath(&resolved, path, 1); > + strbuf_addbuf(sb, &resolved); > + strbuf_release(&resolved); > + } else > + strbuf_realpath(sb, path, 1); I know its not required but I would have braces on the 'else' branch since they were needed on the 'if' branch. But that's up to you and your style :) > +} > + > int printf_ln(const char *fmt, ...) > { > int ret; > diff --git a/strbuf.h b/strbuf.h > index cf1b5409e7..cf8e4bf532 100644 > --- a/strbuf.h > +++ b/strbuf.h > @@ -441,6 +441,20 @@ extern int strbuf_getcwd(struct strbuf *sb); > */ > extern void strbuf_add_absolute_path(struct strbuf *sb, const char *path); > > +/** > + * Canonize `path` (make it absolute, resolve symlinks, remove extra > + * slashes) and append it to `sb`. Die with an informative error > + * message if there is a problem. > + * > + * The directory part of `path` (i.e., everything up to the last > + * dir_sep) must denote a valid, existing directory, but the last > + * component need not exist. > + * > + * Callers that don't mind links should use the more lightweight > + * strbuf_add_absolute_path() instead. > + */ > +extern void strbuf_add_real_path(struct strbuf *sb, const char *path); > + > > /** > * Normalize in-place the path contained in the strbuf. See > -- > 2.12.0 > -- Brandon Williams
Re: [PATCH] strbuf: add strbuf_add_real_path()
On Sat, Feb 25, 2017 at 05:00:33PM +0100, René Scharfe wrote: > Add a function for appending the canonized absolute pathname of a given > path to a strbuf. It keeps the existing contents intact, as expected of > a function of the strbuf_add() family, while avoiding copying the result > if the given strbuf is empty. It's more consistent with the rest of the > strbuf API than strbuf_realpath(), which it's wrapping. > > Also add a semantic patch demonstrating its intended usage and apply it > to the current tree. Using strbuf_add_real_path() instead of calling > strbuf_addstr() and real_path() avoids an extra copy to a static buffer. It's also re-entrant, which real_path() is not. > +void strbuf_add_real_path(struct strbuf *sb, const char *path) > +{ > + if (sb->len) { > + struct strbuf resolved = STRBUF_INIT; > + strbuf_realpath(&resolved, path, 1); > + strbuf_addbuf(sb, &resolved); > + strbuf_release(&resolved); > + } else > + strbuf_realpath(sb, path, 1); > +} The wrapping here seems a little backwards. If strbuf_add_real_path() were the inner one, then we would not need this extra allocation. I know that the reasons are historical, but I don't think it would be impossible to teach the realpath code to do it. OTOH, it may not be worth the effort. It's not like strbuf_realpath() doesn't allocate secondary strbufs for its work already, so dropping one more is probably not that exciting. And certainly think your patch is an incremental improvement. Out of curiosity, I took a stab at the patch, which is below. Only lightly tested by me, and it does make the logic a bit more complicated to read, as you have to adjust for the original "base" in several places. diff --git a/abspath.c b/abspath.c index 2f0c26e0e..286072f48 100644 --- a/abspath.c +++ b/abspath.c @@ -12,9 +12,9 @@ int is_directory(const char *path) } /* removes the last path component from 'path' except if 'path' is root */ -static void strip_last_component(struct strbuf *path) +static void strip_last_component(struct strbuf *path, size_t base) { - size_t offset = offset_1st_component(path->buf); + size_t offset = base + offset_1st_component(path->buf + base); size_t len = path->len; /* Find start of the last component */ @@ -49,14 +49,15 @@ static void get_next_component(struct strbuf *next, struct strbuf *remaining) } /* copies root part from remaining to resolved, canonicalizing it on the way */ -static void get_root_part(struct strbuf *resolved, struct strbuf *remaining) +static void get_root_part(struct strbuf *resolved, struct strbuf *remaining, + size_t base) { int offset = offset_1st_component(remaining->buf); - strbuf_reset(resolved); + strbuf_setlen(resolved, base); strbuf_add(resolved, remaining->buf, offset); #ifdef GIT_WINDOWS_NATIVE - convert_slashes(resolved->buf); + convert_slashes(resolved->buf + base); #endif strbuf_remove(remaining, 0, offset); } @@ -78,8 +79,8 @@ static void get_root_part(struct strbuf *resolved, struct strbuf *remaining) * informative error message if there is a problem. Otherwise, return * NULL on errors (without generating any output). */ -char *strbuf_realpath(struct strbuf *resolved, const char *path, - int die_on_error) +char *strbuf_add_real_path(struct strbuf *resolved, const char *path, + int die_on_error) { struct strbuf remaining = STRBUF_INIT; struct strbuf next = STRBUF_INIT; @@ -87,6 +88,7 @@ char *strbuf_realpath(struct strbuf *resolved, const char *path, char *retval = NULL; int num_symlinks = 0; struct stat st; + size_t base = resolved->len; if (!*path) { if (die_on_error) @@ -96,9 +98,9 @@ char *strbuf_realpath(struct strbuf *resolved, const char *path, } strbuf_addstr(&remaining, path); - get_root_part(resolved, &remaining); + get_root_part(resolved, &remaining, base); - if (!resolved->len) { + if (resolved->len == base) { /* relative path; can use CWD as the initial resolved path */ if (strbuf_getcwd(resolved)) { if (die_on_error) @@ -118,7 +120,7 @@ char *strbuf_realpath(struct strbuf *resolved, const char *path, continue; /* '.' component */ } else if (next.len == 2 && !strcmp(next.buf, "..")) { /* '..' component; strip the last path component */ - strip_last_component(resolved); + strip_last_component(resolved, base); continue; } @@ -127,12 +129,12 @@ char *strbuf_realpath(struct strbuf *resolved, const char *path, strbuf_addch(resolved, '/'); strbuf_addbuf(resolved, &next); -
[PATCH] strbuf: add strbuf_add_real_path()
Add a function for appending the canonized absolute pathname of a given path to a strbuf. It keeps the existing contents intact, as expected of a function of the strbuf_add() family, while avoiding copying the result if the given strbuf is empty. It's more consistent with the rest of the strbuf API than strbuf_realpath(), which it's wrapping. Also add a semantic patch demonstrating its intended usage and apply it to the current tree. Using strbuf_add_real_path() instead of calling strbuf_addstr() and real_path() avoids an extra copy to a static buffer. Signed-off-by: Rene Scharfe --- contrib/coccinelle/strbuf.cocci | 6 ++ setup.c | 2 +- strbuf.c| 11 +++ strbuf.h| 14 ++ 4 files changed, 32 insertions(+), 1 deletion(-) diff --git a/contrib/coccinelle/strbuf.cocci b/contrib/coccinelle/strbuf.cocci index 63995f22ff..1d580e49b0 100644 --- a/contrib/coccinelle/strbuf.cocci +++ b/contrib/coccinelle/strbuf.cocci @@ -38,3 +38,9 @@ expression E1, E2, E3; @@ - strbuf_addstr(E1, find_unique_abbrev(E2, E3)); + strbuf_add_unique_abbrev(E1, E2, E3); + +@@ +expression E1, E2; +@@ +- strbuf_addstr(E1, real_path(E2)); ++ strbuf_add_real_path(E1, E2); diff --git a/setup.c b/setup.c index 967f289f1e..f14cbcd338 100644 --- a/setup.c +++ b/setup.c @@ -254,7 +254,7 @@ int get_common_dir_noenv(struct strbuf *sb, const char *gitdir) if (!is_absolute_path(data.buf)) strbuf_addf(&path, "%s/", gitdir); strbuf_addbuf(&path, &data); - strbuf_addstr(sb, real_path(path.buf)); + strbuf_add_real_path(sb, path.buf); ret = 1; } else { strbuf_addstr(sb, gitdir); diff --git a/strbuf.c b/strbuf.cq index 8fec6579f7..ace58e7367 100644 --- a/strbuf.c +++ b/strbuf.c @@ -707,6 +707,17 @@ void strbuf_add_absolute_path(struct strbuf *sb, const char *path) strbuf_addstr(sb, path); } +void strbuf_add_real_path(struct strbuf *sb, const char *path) +{ + if (sb->len) { + struct strbuf resolved = STRBUF_INIT; + strbuf_realpath(&resolved, path, 1); + strbuf_addbuf(sb, &resolved); + strbuf_release(&resolved); + } else + strbuf_realpath(sb, path, 1); +} + int printf_ln(const char *fmt, ...) { int ret; diff --git a/strbuf.h b/strbuf.h index cf1b5409e7..cf8e4bf532 100644 --- a/strbuf.h +++ b/strbuf.h @@ -441,6 +441,20 @@ extern int strbuf_getcwd(struct strbuf *sb); */ extern void strbuf_add_absolute_path(struct strbuf *sb, const char *path); +/** + * Canonize `path` (make it absolute, resolve symlinks, remove extra + * slashes) and append it to `sb`. Die with an informative error + * message if there is a problem. + * + * The directory part of `path` (i.e., everything up to the last + * dir_sep) must denote a valid, existing directory, but the last + * component need not exist. + * + * Callers that don't mind links should use the more lightweight + * strbuf_add_absolute_path() instead. + */ +extern void strbuf_add_real_path(struct strbuf *sb, const char *path); + /** * Normalize in-place the path contained in the strbuf. See -- 2.12.0