Re: [PATCHES] psql \i handling ~ in specified file name
Zach Irmen wrote: Bruce Momjian wrote: Yes, seems like that will be required. Please use my attached version to make the adjustments. Ok. Adjustments made. All psql commands should be handled. Zach Irmen wrote: And finally, I was wondering if arguments with leading pipes (e.g. |~/file) should also get substituted. Yep, that too. Actually, I found out that popen calls seem to handle the tilde fine as this was already working on my FreeBSD box without substitution. I'm not sure if this is true for all Unix systems in general, but I won't bother putting that in unless it turns out to be needed. Oh, yea, | should work fine because it uses the shell. pipe manual page says: The popen() function ``opens'' a process by creating an IPC connection, forking, and invoking the shell. and The command argument is a pointer to a null-terminated string containing a shell command line. This command is passed to /bin/sh using the -c flag; interpretation, if any, is performed by the shell. so we are find with pipe already. -- Bruce Momjian| http://candle.pha.pa.us [EMAIL PROTECTED] | (610) 359-1001 + If your life is a hard drive, | 13 Roberts Road + Christ can be your backup.| Newtown Square, Pennsylvania 19073 ---(end of broadcast)--- TIP 5: Have you checked our extensive FAQ? http://www.postgresql.org/docs/faqs/FAQ.html
Re: [PATCHES] psql \i handling ~ in specified file name
Patch applied. Thanks.
---
Zach Irmen wrote:
Bruce Momjian wrote:
Yes, seems like that will be required. Please use my attached version
to make the adjustments.
Ok. Adjustments made. All psql commands should be handled.
Zach Irmen wrote:
And finally, I was wondering if arguments with leading pipes
(e.g. |~/file) should also get substituted.
Yep, that too.
Actually, I found out that popen calls seem to handle the tilde fine as
this was already working on my FreeBSD box without
substitution. I'm not sure if this is true for all Unix systems in
general, but I won't bother putting that in unless it turns out
to be needed.
Index: command.c
===
RCS file: /projects/cvsroot/pgsql-server/src/bin/psql/command.c,v
retrieving revision 1.108
diff -c -r1.108 command.c
*** command.c 1 Dec 2003 22:21:54 - 1.108
--- command.c 9 Jan 2004 06:51:55 -
***
*** 413,418
--- 413,419
else
{
fname = scan_option(string, OT_NORMAL, NULL, true);
+ expand_tilde(fname);
status = do_edit(fname, query_buf) ? CMD_NEWEDIT : CMD_ERROR;
free(fname);
}
***
*** 494,500
--- 495,504
if (!fname)
pset.gfname = NULL;
else
+ {
+ expand_tilde(fname);
pset.gfname = xstrdup(fname);
+ }
free(fname);
status = CMD_SEND;
}
***
*** 531,536
--- 535,541
}
else
{
+ expand_tilde(fname);
success = (process_file(fname) == EXIT_SUCCESS);
free(fname);
}
***
*** 561,567
--- 566,575
success = false;
}
else
+ {
+ expand_tilde(opt2);
success = do_lo_export(opt1, opt2);
+ }
}
else if (strcmp(cmd + 3, import) == 0)
***
*** 572,578
--- 580,589
success = false;
}
else
+ {
+ expand_tilde(opt1);
success = do_lo_import(opt1, opt2);
+ }
}
else if (strcmp(cmd + 3, list) == 0)
***
*** 602,607
--- 613,619
{
char *fname = scan_option(string, OT_FILEPIPE, NULL, true);
+ expand_tilde(fname);
success = setQFout(fname);
free(fname);
}
***
*** 653,658
--- 665,671
{
char *fname = scan_option(string, OT_NORMAL, NULL, true);
+ expand_tilde(fname);
success = saveHistory(fname ? fname : /dev/tty);
if (success !quiet fname)
***
*** 771,776
--- 784,790
else
{
fname = scan_option(string, OT_FILEPIPE, NULL, true);
+ expand_tilde(fname);
if (!fname)
{
Index: common.c
===
RCS file: /projects/cvsroot/pgsql-server/src/bin/psql/common.c,v
retrieving revision 1.78
diff -c -r1.78 common.c
*** common.c 29 Nov 2003 19:52:06 - 1.78
--- common.c 9 Jan 2004 06:51:56 -
***
*** 814,816
--- 814,878
else
return PQuser(pset.db);
}
+
+
+ /* expand_tilde
+ *
+ * substitute '~' with HOME or '~username' with username's home dir
+ *
+ */
+ char *
+ expand_tilde(char **filename)
+ {
+ if (!filename || !(*filename))
+ return NULL;
+
+ /* MSDOS uses tilde for short versions of long file names, so skip it. */
+ #ifndef WIN32
+
+ /* try tilde expansion */
+ if (**filename == '~')
+ {
+ char *fn;
+ char *home;
+ charoldp,
+*p;
+ struct passwd *pw;
+
+ fn = *filename;
+ home = NULL;
+
+ p = fn + 1;
+ while (*p != '/' *p != '\0')
+ p++;
+
+ oldp = *p;
+ *p = '\0';
+
+ if (*(fn + 1) == '\0')
+ home = getenv(HOME);
+ else if ((pw = getpwnam(fn + 1)) != NULL)
+ home =
Re: [PATCHES] psql \i handling ~ in specified file name
Zach Irmen wrote: Andrew Dunstan [EMAIL PROTECTED] writes: Zach Irmen said: Can I just ifndef WIN32 and not think about it? I'm not sure how that would work either. If we are going to have a Windows port I don't think we should treat it as a poor cousin. I guess I was thinking more about if it should be done as opposed to how it would be done. On the one hand, I think '~' by itself has no meaning in a normal Windows environment, so why should psql on Windows give it one? The readline library on unix, which can be used by psql, interprets the tilde and is the big reason why psql on unix should interpret the tilde as well. On the other hand however, I can see consistency being important in that giving '~' a meaning in psql should give it the same meaning regardless of platform. As I remember, MSDOS uses the ~ to specify short versions of long file names. I think that is enough for us to say that we are best leaving '~' expansion only for Unix. We already dump COPY in a native Win32 format, so it seems we should handle special characters in a similar native way. I will add a comment to this affect in the source code. -- Bruce Momjian| http://candle.pha.pa.us [EMAIL PROTECTED] | (610) 359-1001 + If your life is a hard drive, | 13 Roberts Road + Christ can be your backup.| Newtown Square, Pennsylvania 19073 ---(end of broadcast)--- TIP 1: subscribe and unsubscribe commands go to [EMAIL PROTECTED]
Re: [PATCHES] psql \i handling ~ in specified file name
Joshua D. Drake wrote: As I remember, MSDOS uses the ~ to specify short versions of long file names. I is not just msdos, it is cmd.exe which exists on (to my knowledge) all versions of windows. For example: Program Files == progra~1 Yes, I meant it is a hold-over from dealing with MS-DOS style 8.3 file names. -- Bruce Momjian| http://candle.pha.pa.us [EMAIL PROTECTED] | (610) 359-1001 + If your life is a hard drive, | 13 Roberts Road + Christ can be your backup.| Newtown Square, Pennsylvania 19073 ---(end of broadcast)--- TIP 8: explain analyze is your friend
Re: [PATCHES] psql \i handling ~ in specified file name
As I remember, MSDOS uses the "~" to specify short versions of long file names. I is not just msdos, it is cmd.exe which exists on (to my knowledge) all versions of windows. For example: Program Files == progra~1 I think that is enough for us to say that we are best leaving '~' expansion only for Unix. We already dump COPY in a native Win32 format, so it seems we should handle special characters in a similar native way. I will add a comment to this affect in the source code. -- Command Prompt, Inc., home of Mammoth PostgreSQL - S/ODBC and S/JDBC Postgresql support, programming shared hosting and dedicated hosting. +1-503-222-2783 - [EMAIL PROTECTED] - http://www.commandprompt.com
Re: [PATCHES] psql \i handling ~ in specified file name
Andrew Dunstan wrote: As I remember, MSDOS uses the ~ to specify short versions of long file names. I is not just msdos, it is cmd.exe which exists on (to my knowledge) all versions of windows. For example: Program Files == progra~1 Yes, I meant it is a hold-over from dealing with MS-DOS style 8.3 file names. This pattern should not cause tilde expansion on any platform, I believe. Only a *leading* ~/ or ~username/ should be expanded. The normal home directory on a (modern) Windows machine is usually C:\Documents and Settings\username, IIRC. Yes, I understand, but on an OS that uses tilde so much inside the file name, do we want special meaning when it is leading the file name? -- Bruce Momjian| http://candle.pha.pa.us [EMAIL PROTECTED] | (610) 359-1001 + If your life is a hard drive, | 13 Roberts Road + Christ can be your backup.| Newtown Square, Pennsylvania 19073 ---(end of broadcast)--- TIP 4: Don't 'kill -9' the postmaster
Re: [PATCHES] psql \i handling ~ in specified file name
Peter Eisentraut wrote:
Zach Irmen wrote:
refers to following item on TODO
Have psql '\i ~/tabtab' actually load files it displays from home
dir
This expansion should also apply to all other commands that take file
names.
Here is a patch that handles ~ in all the file cases. Thanks for the
initial patch.
--
Bruce Momjian| http://candle.pha.pa.us
[EMAIL PROTECTED] | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup.| Newtown Square, Pennsylvania 19073
Index: src/bin/psql/command.c
===
RCS file: /cvsroot/pgsql-server/src/bin/psql/command.c,v
retrieving revision 1.108
diff -c -c -r1.108 command.c
*** src/bin/psql/command.c 1 Dec 2003 22:21:54 - 1.108
--- src/bin/psql/command.c 8 Jan 2004 05:17:35 -
***
*** 65,70
--- 65,72
static bool do_connect(const char *new_dbname, const char *new_user);
static bool do_shell(const char *command);
+ static char *expand_tilde(char **filename);
+
/*--
* HandleSlashCmds:
*
***
*** 413,418
--- 415,421
else
{
fname = scan_option(string, OT_NORMAL, NULL, true);
+ expand_tilde(fname);
status = do_edit(fname, query_buf) ? CMD_NEWEDIT : CMD_ERROR;
free(fname);
}
***
*** 494,500
--- 497,506
if (!fname)
pset.gfname = NULL;
else
+ {
+ expand_tilde(fname);
pset.gfname = xstrdup(fname);
+ }
free(fname);
status = CMD_SEND;
}
***
*** 531,536
--- 537,543
}
else
{
+ expand_tilde(fname);
success = (process_file(fname) == EXIT_SUCCESS);
free(fname);
}
***
*** 602,607
--- 609,615
{
char *fname = scan_option(string, OT_FILEPIPE, NULL, true);
+ expand_tilde(fname);
success = setQFout(fname);
free(fname);
}
***
*** 653,658
--- 661,667
{
char *fname = scan_option(string, OT_NORMAL, NULL, true);
+ expand_tilde(fname);
success = saveHistory(fname ? fname : /dev/tty);
if (success !quiet fname)
***
*** 771,776
--- 780,786
else
{
fname = scan_option(string, OT_FILEPIPE, NULL, true);
+ expand_tilde(fname);
if (!fname)
{
***
*** 1678,1683
--- 1688,1747
}
+ /* expand_tilde
+ *
+ * substitute '~' with HOME or '~username' with username's home dir
+ *
+ */
+ static char *
+ expand_tilde(char **filename)
+ {
+ if (!filename || !(*filename))
+ return NULL;
+
+ #ifndef WIN32
+
+ /* try tilde expansion */
+ if (**filename == '~')
+ {
+ char *fn;
+ char *home;
+ char oldp, *p;
+ struct passwd *pw;
+
+ fn = *filename;
+ home = NULL;
+
+ p = fn+1;
+ while (*p != '/' *p != '\0')
+ p++;
+
+ oldp = *p;
+ *p = '\0';
+
+ if (*(fn+1) == '\0')
+ home = getenv(HOME);
+ else if ((pw = getpwnam(fn+1)) != NULL)
+ home = pw-pw_dir;
+
+ *p = oldp;
+ if (home)
+ {
+ char *newfn;
+
+ newfn = malloc(strlen(home) + strlen(p) + 1);
+ strcpy(newfn,home);
+ strcat(newfn,p);
+
+ free(fn);
+ *filename = newfn;
+ }
+ }
+
+ #endif
+
+ return *filename;
+ }
/*
* process_file
---(end of broadcast)---
TIP 4: Don't 'kill -9' the postmaster
Re: [PATCHES] psql \i handling ~ in specified file name
Bruce Momjian wrote: Here is a patch that handles ~ in all the file cases. Beat me to it. :) I do have a few issues that I was trying to sort out myself regarding this, but I guess now is as good a time as any to ask them here. First off, there should be a check after the malloc to make sure NULL wasn't returned in the expand_tilde function. I missed that one. Secondly, there are a couple of SQL commands (like COPY and LOAD) and psql commands handled outside command.c (like \copy) which also take filenames. I'm guessing that eventually you'll want substitution in those cases as well. So does this mean that the expand_tilde function probably should not be in command.c? Placing it in common.c seems the logical place to make it at least available to all the psql commands (\copy included). And finally, I was wondering if arguments with leading pipes (e.g. |~/file) should also get substituted. Zach Irmen ---(end of broadcast)--- TIP 1: subscribe and unsubscribe commands go to [EMAIL PROTECTED]
Re: [PATCHES] psql \i handling ~ in specified file name
Zach Irmen wrote: refers to following item on TODO Have psql '\i ~/tabtab' actually load files it displays from home dir This expansion should also apply to all other commands that take file names. ---(end of broadcast)--- TIP 3: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly
Re: [PATCHES] psql \i handling ~ in specified file name
Tom Lane [EMAIL PROTECTED] writes:
What happens if getenv(HOME) returns NULL?
Yeah, the strdup fails. I'll take it out to fix that.
You also need to think about Windows
Can I just ifndef WIN32 and not think about it? I'm not sure how that would
work either.
Finally, couldn't we reduce the number
of times strings are strdup'd only to be freed again? I don't think
doing it this way makes the code simpler --- it took me a fair while
to convince myself there was no memory leak.
Ok. I cut it down to one dup within the function. Attempt 2 is below.
--
Zach Irmen
Index: command.c
===
RCS file: /projects/cvsroot/pgsql-server/src/bin/psql/command.c,v
retrieving revision 1.108
diff -c -r1.108 command.c
*** command.c 1 Dec 2003 22:21:54 - 1.108
--- command.c 21 Dec 2003 08:10:50 -
***
*** 65,70
--- 65,72
static bool do_connect(const char *new_dbname, const char *new_user);
static bool do_shell(const char *command);
+ static char *expand_tilde(char **filename);
+
/*--
* HandleSlashCmds:
*
***
*** 531,536
--- 533,539
}
else
{
+ expand_tilde(fname);
success = (process_file(fname) == EXIT_SUCCESS);
free(fname);
}
***
*** 1678,1683
--- 1681,1740
}
+ /* expand_tilde
+ *
+ * substitute '~' with HOME or '~username' with username's home dir
+ *
+ */
+ static char *
+ expand_tilde(char **filename)
+ {
+ if (!filename || !(*filename))
+ return NULL;
+
+ #ifndef WIN32
+
+ /* try tilde expansion */
+ if (**filename == '~')
+ {
+ char *fn;
+ char *home;
+ char oldp, *p;
+ struct passwd *pw;
+
+ fn = *filename;
+ home = NULL;
+
+ p = fn+1;
+ while (*p != '/' *p != '\0')
+ p++;
+
+ oldp = *p;
+ *p = '\0';
+
+ if (*(fn+1) == '\0')
+ home = getenv(HOME);
+ else if ((pw = getpwnam(fn+1)) != NULL)
+ home = pw-pw_dir;
+
+ *p = oldp;
+ if (home)
+ {
+ char *newfn;
+
+ newfn = malloc(strlen(home) + strlen(p) + 1);
+ strcpy(newfn,home);
+ strcat(newfn,p);
+
+ free(fn);
+ *filename = newfn;
+ }
+ }
+
+ #endif
+
+ return *filename;
+ }
/*
* process_file
---(end of broadcast)---
TIP 8: explain analyze is your friend
Re: [PATCHES] psql \i handling ~ in specified file name
Zach Irmen said: Tom Lane [EMAIL PROTECTED] writes: What happens if getenv(HOME) returns NULL? Yeah, the strdup fails. I'll take it out to fix that. You also need to think about Windows Can I just ifndef WIN32 and not think about it? I'm not sure how that would work either. If we are going to have a Windows port I don't think we should treat it as a poor cousin. This information is easily discoverable on MSDN in 5 minutes. The answer is that you need to call NetUserGetInfo at level 11 to get a user's home directory. See http://msdn.microsoft.com/library/en-us/dnucmg/html/ucmgch09.asp and http://msdn.microsoft.com/library/en-us/netmgmt/netmgmt/netusergetinfo.asp (Thinks: using this requires linking with NetAPI32.dll - is that a legal or practical problem for us?) Also, I strongly recommend that the name be canonicalised - see initdb.c for an example. Doing this reduces platform-specific code a lot. Windows is quite able to handle filenames with a forward slash. (Thinks again: when we have WIN32 done we need to trawl through the code to try to abstract out as much as possible, so that people who don't/won't/can't do Windows specific code won't have to bother.) cheers andrew ---(end of broadcast)--- TIP 7: don't forget to increase your free space map settings
Re: [PATCHES] psql \i handling ~ in specified file name
Zach Irmen said: Tom Lane [EMAIL PROTECTED] writes: What happens if getenv(HOME) returns NULL? Yeah, the strdup fails. I'll take it out to fix that. You also need to think about Windows Can I just ifndef WIN32 and not think about it? I'm not sure how that would work either. If we are going to have a Windows port I don't think we should treat it as a poor cousin. This information is easily discoverable on MSDN in 5 minutes. The answer is that you need to call NetUserGetInfo at level 11 to get a user's home directory. See http://msdn.microsoft.com/library/en-us/dnucmg/html/ucmgch09.asp and http://msdn.microsoft.com/library/en-us/netmgmt/netmgmt/netuser getinfo.asp (Thinks: using this requires linking with NetAPI32.dll - is that a legal or practical problem for us?) This is already done by libpq if compiled on Win32. Which means also psql.exe is linked against it on Win32. (see win32.mak in interfaces/libpq). So it shuoldn't be a problem, I think. //Magnus ---(end of broadcast)--- TIP 4: Don't 'kill -9' the postmaster
Re: [PATCHES] psql \i handling ~ in specified file name
Andrew Dunstan [EMAIL PROTECTED] writes: Zach Irmen said: Can I just ifndef WIN32 and not think about it? I'm not sure how that would work either. If we are going to have a Windows port I don't think we should treat it as a poor cousin. I guess I was thinking more about if it should be done as opposed to how it would be done. On the one hand, I think '~' by itself has no meaning in a normal Windows environment, so why should psql on Windows give it one? The readline library on unix, which can be used by psql, interprets the tilde and is the big reason why psql on unix should interpret the tilde as well. On the other hand however, I can see consistency being important in that giving '~' a meaning in psql should give it the same meaning regardless of platform. -- Zach Irmen ---(end of broadcast)--- TIP 2: you can get off all lists at once with the unregister command (send unregister YourEmailAddressHere to [EMAIL PROTECTED])
