okbob / pspg Goto Github PK

Would be nice if it could detect CSV structure when running as: pspg < file.csv

If used with \?, the screen is blank (except for the Quit 0..4 C.Freeze) at the bottom. If I press PageDown it's still blank, but when I press it again, I get to see the last 2/3 or 3/4 of the output. I also get a mostly blank screen when using \d on some tables with many columns and constraints/triggers (for example, it only shows the first line of a textsearch trigger, but it does show a blank inverse video, cursor line, that can be moved with the arrow keys).

The 0.4 tag appears to output 0.3devel when --version is passed.

Additionally, the spec file lists the version as 1.0.0, which seems weird.

The attached file (from \dx+ postgis) causes pspg to consume 100% CPU: ./pspg --less-status-bar < dx-postgis.txt. It works without --less-status-bar.
dx-postgis.txt

The width of some chars is badly identified - 丂

Downloaded 0.6 version.
My linux OS is CentOS 5.11.

I ran configure, no errors there.
Then I ran make and get the following error:

gcc src/pspg.c -o pspg -g -O2   -D_GNU_SOURCE   -lncursesw
/tmp/ccIdvgCz.o: In function `main':
/home/user/Src/pspg-0.6/src/pspg.c:2403: undefined reference to `set_escdelay'
collect2: ld returned 1 exit status
make: *** [pspg] Error 1

An initial google search pointed to ncurses.
I have ncurses version 5.5-24 installed.

Yes, I know my OS is old but it's what I have to deal with.

hi there, great project!

However, when I tried to compile:

git clone https://github.com/okbob/pspg.git postgres-pager
cd postgres-pager
make

I got errors about missing files:

pager.c:1:20: fatal error: curses.h: No such file or directory
/usr/bin/ld: cannot find -lncursesw

Some research suggested

sudo apt-get install build-essential libreadline-dev libncurses5-dev libncursesw5-dev

Which somewhat helps, but now I get

gcc -lncursesw pager.c -o pager
/tmp/ccih2jFq.o: In function `initialize_color_pairs':
pager.c:(.text+0x419): undefined reference to `use_default_colors'
pager.c:(.text+0x42d): undefined reference to `init_pair'
pager.c:(.text+0x441): undefined reference to `init_pair'
pager.c:(.text+0x455): undefined reference to `init_pair'
pager.c:(.text+0x469): undefined reference to `init_pair'
pager.c:(.text+0x47d): undefined reference to `init_pair'
/tmp/ccih2jFq.o:pager.c:(.text+0x491): more undefined references to `init_pair' follow
/tmp/ccih2jFq.o: In function `initialize_color_pairs':
pager.c:(.text+0x4af): undefined reference to `assume_default_colors'
pager.c:(.text+0x4c3): undefined reference to `init_pair'
... ... ... ...
pager.c:(.text+0x2a3b): undefined reference to `wrefresh'
pager.c:(.text+0x2aff): undefined reference to `endwin'
collect2: error: ld returned 1 exit status
Makefile:4: recipe for target 'compile' failed
make: *** [compile] Error 1

Any suggestions?

I was able to build on macOS Sierra by linking against ncurses, instead of ncursesw.

I don't know how to make this a conditional in the Makefile.

Hi!

I should very much like to ask to add history in searches, like "less" does: one types '/', then can use keys up and down. Thanks for the great product!

/dk

I always get "Segmentation fault" when trying to show SQL function source code using \sf.

unpaged result
postgres=# select pid, usename from pg_stat_activity limit 0;
pid │ usename
─────┼─────────
(0 rows)

Time: 1.419 ms
postgres=# select pid, usename from pg_stat_activity limit 1;
pid │ usename
───────┼─────────
22436 │ rms
(1 row)

paged result
postgres=# select pid, usename from pg_stat_activity limit 0;
─────┼─────────
─────┼─────────
(0 rows)
Time: 1.002 ms
postgres=# select pid, usename from pg_stat_activity limit 1;
pid │ usename
───────┼─────────
22436 │ rms
(1 row)

The zero row paged result does not show column names, just the UTF8 border characters. The column names are present in the textual results, and presumably ought to be shown.

backward searching

I had a long-running query that I stopped, using Ctrl-C.

It seems, even though I can't find any processes running pspg that the terminal session is totally bunk. Even typing reset doesn't fix it: the cursor is stuck in column 0.

bottom part should be independent (drawing, scrolling) on main data part.

It seems that pspg is not capable of printing some special characters such as "á" which is printed as "M-CM-!". The symbol is properly shown using "less".

LANG is set to ca_ES.utf8 and psql encoding to utf8.

Example output:

┌────────┬─────────────────────┬───────┬────────┐
│ Schema │        Name         │ Type  │ Owner  │
├────────┼─────────────────────┼───────┼────────┤
│ public │ FROM                │ table │ ilmari │
│ public │ JOIN                │ table │ ilmari │
│ public │ arraytest           │ table │ ilmari │
│ public │ bars                │ table │ ilmari │
└────────┴─────────────────────┴───────┴────────┘

In less, if you press / and type a string to search for, you can later press / and then Enter and it will search for the same string again. Pspg should behave this way as well.

It would be nice to be able to horizontally scroll with your mouse. Chrome uses Shift+MouseWheel to do this, so I'd suggest that to be used.

Version : Linux 1ffdea63e0b2 4.13.0-36-generic #40-Ubuntu

gcc -O3 src/pspg.c src/print.c src/unicode.c src/themes.c -o pspg -g -O2   -D_GNU_SOURCE -D_DEFAULT_SOURCE -I/usr/include/ncursesw  -lncursesw -ltinfo
src/pspg.c:157:1: error: conflicting types for 'strcasestr'
 strcasestr(const char *s1, const char *s2)
 ^~~~~~~~~~
In file included from src/pspg.c:19:0:
/usr/include/string.h:368:14: note: previous declaration of 'strcasestr' was here
 extern char *strcasestr (const char *__haystack, const char *__needle)
              ^~~~~~~~~~
Makefile:10: recipe for target 'pspg' failed
make: *** [pspg] Error 1

Grepping the source code revealed that the strcasestr function is never used, but:

• I am not a real C user
• I do not want to break things I do not understand

Locally, removing the declaration made the build successful

Hi,
I installed this, but then of course nothing happened. It would be great to mention setting PAGER is needed to use this (or maybe there is something else I don't know about ?)

Cheers
S

\sf functionname results in a segmentation fault

Like less...

Hi,

i found problem with wide column (output of public key), can be tested with:
--not working (empty screen)
SELECT repeat('A','1000');
--works
SELECT 'a', repeat('A','1000');
--not working
SELECT repeat('A','1000'), 'a';

but maybe caused by some local environment settings...

David

Output gets mangled when --quit-if-one-screen it's in effect:

$ mysql
[…]
Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.

user@host/(none)> select now();
+---------------------+
| now()               |
+---------------------+
| 2018-01-28 23:59:24 |
+---------------------+
1 row in set (0.00 sec)

user@host/(none)> ^DBye
$ mysql --pager='pspg --quit-if-one-screen'
[…]
Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.

user@host/(none)> select now();
+---------------------+| now()               |+---------------------+| 2018-01-28 23:59:46 |+---------------------+1 row in set (0.00 sec)

user@host/(none)> select now()\G
*************************** 1. row ***************************now(): 2018-01-28 23:59:481 row in set (0.00 sec)

user@host/(none)> ^DBye

Thanks

Something similar to examples for less or vim/vimpager below:
less
s /tmp/filename
vimpager
:w /tmp/filename

Hi,
it would be nice if pspg would highlight search results, both in the current line where the cursor is, but also in other lines, so it's immediately obvious if there are more hits on screen (with possibly more hits to the left or right, but assuming the hits are in one column only, that won't be a problem in practise). less does this by default, see -g and -G, but I'm not sure it needs to be configurable.
Thanks!

Just an idea of feature request: since pspg is aware of columns in psql output, it could as well allow to:

1. hide columns from the view - especially usefull for SELECT * queries for wide tables, since there is no "SELECT all_except_specified_columns",
2. reorder columns - also saves typing of all the column names instead of "*", since PostgreSQL itself doesn't allow to change default order and on-disk structure might be subject of on-purpose alignment,
3. ellipse specified columns,
4. show expanded (psql -x) version of the record when selected and (e.g.) ENTER pressed.
5. provide maximum column width as a command line argument.

User interface:
c - enter column management mode, cursor goes to top-line header, hidden columns are displayed dimmed
left/right arrows - selecting column
h - toggle "hide"
</> - shrink/expand column
m - mark column for "move" operation, left/right arrows to position, enter to accept
c/enter - exit column management mode

Hi,
I'm wondering if the top and bottom status lines couldn't be merged into one line at the bottom. The current top line is right-aligned, and the bottom line is left-aligned, so the content wouldn't have to change much (except when -f is used). This would save one line on the screen and would also match more closely the look of "less".
Thanks!

Version 0.7.3 crashes on -f option (open file), previous versions works fine.

Hi! I've created brew formula for your awesome project but there is one small thing I don't know how to handle.

Brew is a package manager for OS X like portage in gentoo or ports in freebsd.
They have a lot of acceptance criteria and one of them is having a test of application for running after compiling.
For small programs they usually check for --help or --version, but this is not considered as a good practice. Better to have some 'real' cases of application usage.
Also, those tests should be non-interactive so CI can run them

So, I've started to write a formula but it's turned out that I can't write any type of test for your app.
I've checked how they usually test other pagers, i.e vimpager: https://github.com/Homebrew/homebrew-core/blob/93a8200c17986038585a140cce4a9bc93e4a3f86/Formula/vimpager.rb

but this approach is not fit for your pager because it bring up interactive interface even if there is not enough content for paging.
Also, you don't have --version or --help.

So, i've created this #27

I have never write any line of C, but I hope there is nothing wrong with this few lines.
If you accept PR, please, do a release so I can update brew formula with new version (0.3?)

Thank you and sorry if my english is not good )

Thanks for pspg. I've just started using it and it looks promising.

I was reading through the man page and also the recommended .psqlrc configs and noticed "-X" in the configuration. I don't know what alternate screen the documentation is referring to. I've looked at the source code and can't make sense of it either.

Would you consider expanding the documentation a bit with regards to "alternate screen" ?

Thanks again!

-m

PSPG works beautifully for me except for the "Save content to file" command. After entering a filename, it returns the message "Cannot write to [filename]" and creates a zero-byte file at the specified path.

Actually, this has been the case for me for all builds since 2017/09 (I just git-pulled and recompiled it again today).

I'm invoking PSPG in my .psqlrc like so:

\setenv PAGER 'pspg -s 11 -X --no-mouse --no-sound'

...and running it on OSX 10.12.16 with psql 9.4.0.

Any ideas? Thanks!

PAGER="./pspg -s 11 -g -i -F" psql -c "SELECT (CASE WHEN random()<0.5 then 'ó' else '-' END),md5(random()::text) FROM generate_series(1,80)"

generates following output (note the misalignments):

 ó    |973263148f6c1da850c5b82f44bedd
 -    | 98d546af5b4f08861a8e0291b3117c22
 ó    |3e7a73ef2ae53719d7a5eaa88b6f41
 ó    |9a48e5cb0b378cbc9c72721c983adb
 -    | 0d5c73047d42b16a3ab7f09a845f1fe2
 -    | 087f7bc9a684617fc9c5522c0ee79fef
 -    | 81db99d85a7c63dc447b5675a0774bbb
 ó    |f00ab7b5e450eb314d8b9f1e4875db
 -    | 2ca1b394e873b1168c0e4eac829ac7ca

This happens with 8-bit (non unicode) locale (ISO8859-2 in my case).

Edit: it's better visible with constant value in 2nd column ("abcdef"):

 -    | abcdef
 ó    |cdef
 -    | abcdef
 ó    |cdef
 -    | abcdef
 -    | abcdef
 ó    |cdef

pspg segfaults when pressing G for this input:

 id 
----
 0
 0
  
 0
 0
 0
 0
 0
 0
 0
(10 rows)

Used ncurses version is 6.1.

I saw you released a new version with resizing implemented, so I got excited and tried

git clone https://github.com/okbob/pspg.git postgres-pager
cd postgres-pager
sudo apt-get install build-essential libreadline-dev libncurses5-dev libncursesw5-dev
gcc pager.c -o pager -L. -lncursesw
./pager -d

as per #1, but got

pager.c: In function ‘main’:
pager.c:1641:26: error: ‘BUTTON5_PRESSED’ undeclared (first use in this function)
       if (event.bstate & BUTTON5_PRESSED)
                          ^
pager.c:1641:26: note: each undeclared identifier is reported only once for each function it appears in

instead. I'm using commit e0a0aba17 here. Any suggestions?

https://github.com/okbob/pspg/blob/0.9.0/src/pspg.c#L31

When I click on some row in ./pspg --less-status-bar < pg_class.txt, the cursor bar goes to the row above the one I clicked.

jq (https://stedolan.github.io/jq/) is a great tool for pretty-printing json. Is it possible to add support
for piping a json column into jq from pspg?

generic piping support with the "|" char would be awesome

It's common in more/less to page forward with the space key. Please implement it, too.

To support non-color terminals, I would suggest to evaluate the 'b' command line switch (black-withe style):

Line 2414: if(!has_colors()) -> if(!has_colors() && style != 0)

When running psql with su:

su postgres -c "psql test"

I get the following error:

Line style is unicode.
Border style is 2.
Pager is used for long output.
Timing is on.
Null display is "¤".
psql (10.1)
Type "help" for help.

test=# select * from test;
cannot to reopen stdin: No such device or address
Time: 14.278 ms
test=#

entering a invalid parameter
Usage: ./pspg [-b] [-s n] [-c n] [-f file] [-X]
is printed on the screen.

What are this for parameters and what are they good for. I couldn't find anything in the README.md

More key commands that many less users will naturally reach for:

• ^E scroll window down one line, keeping cursor in place
• ^Y scroll window up one line, keeping cursor in place
• ^D forward a half window
• ^U backward a half window

select * from pg_stat_activity
PgDn

When I tried to use '\s' in psql (with PAGER set to pspg) I got no output. Further investigation indicated that pspg was exiting with a 'Segmentation fault: 11'.
I tracked this down to pspg.c line 876 (in master):
desc->headline_size = strlen(desc->headline);
A fix was not obvious to me.

# select * from newline;
  t  
─────
 foo↵
 bar
(1 Zeile)
# \pset pager always

... then do the select again, and search for /foo:

Segmentation fault (core dumped)

Core was generated by `pspg -b --less-status-bar --ignore-case'.
Program terminated with signal SIGSEGV, Segmentation fault.
(gdb) bt
#0  __strstr_sse2_unaligned () at ../sysdeps/x86_64/multiarch/strstr-sse2-unaligned.S:40
#1  0x0000557cecc9fedb in window_fill (window_identifier=3, srcy=<optimized out>, srcx=0, cursor_row=0, desc=0x7ffe2ed3f740, 
    scrdesc=0x7ffe2ed3f200, opts=0x7ffe2ed3f1d4) at src/print.c:136
#2  0x0000557cecc9b540 in main (argc=<optimized out>, argv=<optimized out>) at src/pspg.c:1571

Please let me know if you need more details to reproduce. Thanks!

# ./configure
checking for gcc... gcc
checking whether the C compiler works... yes
checking for C compiler default output file name... a.out
checking for suffix of executables...
checking whether we are cross compiling... no
checking for suffix of object files... o
checking whether we are using the GNU C compiler... yes
checking whether gcc accepts -g... yes
checking for gcc option to accept ISO C89... none needed
checking how to run the C preprocessor... gcc -E
checking for grep that handles long lines and -e... /bin/grep
checking for egrep... /bin/grep -E
checking for ANSI C header files... yes
checking for gcc... (cached) gcc
checking whether we are using the GNU C compiler... (cached) yes
checking whether gcc accepts -g... (cached) yes
checking for gcc option to accept ISO C89... (cached) none needed
checking for pkg-config... /usr/bin/pkg-config
checking pkg-config is at least version 0.9.0... yes
checking for ncursesw via pkg-config... yes
checking for working ncursesw/curses.h... yes
checking for working ncursesw.h... no
checking for working ncurses.h... yes
configure: creating ./config.status
config.status: creating config.make
# make
gcc -O3 src/pspg.c src/print.c src/unicode.c src/themes.c -o pspg -g -O2      -lncursesw
src/pspg.c: In function ‘get_code’:
src/pspg.c:1388: warning: passing argument 1 of ‘tigetstr’ discards qualifiers from pointer target type
/usr/include/curses.h:792: note: expected ‘char *’ but argument is of type ‘const char *’
#

What to do?

The problem is here:

It's nice to be able to mark text for copy. On my linux workstation, mouse control only selects the active row which is not a good tradeoff on balance; it would be nice to have command switch to let O/S handle all mouse activity...thanks

Cloning the newest sources (7ed50cb), I sadly get pager.c:20:26: fatal error: ncurses/term.h: No such file or directory:

$ sudo apt-get update
$ sudo apt-get install build-essential libreadline-dev libncurses5-dev libncursesw5-dev
build-essential is already the newest version (12.1ubuntu2).
libncurses5-dev is already the newest version (6.0+20160213-1ubuntu1).
libncursesw5-dev is already the newest version (6.0+20160213-1ubuntu1).
libreadline-dev is already the newest version (6.3-8ubuntu2).

$ gcc pager.c -o pager -L. -lncursesw                                                  
pager.c:20:26: fatal error: ncurses/term.h: No such file or directory
compilation terminated.

Update Looks like I can comment out line 20, which compiles...