How to find non duplicate <type:keywords> of a file?

Question

I need a vim shortcut to display all non-duplicate struct names in the file being edited. I have made this C program which does this, however I have to leave the editor and execute the program in bash and return back to editing.

Lets say the file being edited is linux/drivers/net/ethernet/smsc/smsc911x.c. The output should be like following:

 0. smsc911x_data
 1. smsc911x_ops
 2. smsc911x_platform_config
 3. phy_device
 4. mii_bus
 5. net_device
 6. napi_struct
 7. regulator_bulk_data
 8. clk
 9. platform_device
10. sk_buff
11. net_device_stats
12. netdev_hw_addr
13. sockaddr
14. ethtool_drvinfo
15. ethtool_eeprom
16. ethtool_ops
17. net_device_ops
18. resource
19. device_node
20. dev_pm_ops
21. of_device_id
22. platform_driver

EDIT: The Parse() vim routine defined in the answer posted below doesn't prints all the struct names. It's output:

0. clk
1. device_node
2. ethtool_drvinfo
3. ethtool_eeprom
4. mii_bus
5. napi_struct
6. net_device
7. netdev_hw_addr
8. phy_device
9. regulator_bulk_data
10. resource
11. sk_buff
12. smsc911x_data
13. smsc911x_platform_config
14. sockaddr

I guess its because it is not able to handle cases like struct { ... } name;

Answer 1

:e smsc911x.c
:new
:read !grep -o "struct [^ ;,)]\+" # | awk '{print $2}' | sort -u

Explaining:

• grep -o: returns only the match chars (useful for multiple match on one line)
• [^ ;,)]: remove characters non part of struct name
• #: use alternate file (smsc911x.c in this case)

See this gets even more results than your program -once removed that numbered order that I don't know where it comes from-

diff --git 1/your_result 2/my_result
index 0bee24a..90d7ce4 100644
--- 1/your_result
+++ 2/my_result
@@ -1,9 +1,13 @@
 clk
+device
 device_node
 dev_pm_ops
+ethtool_cmd
 ethtool_drvinfo
 ethtool_eeprom
 ethtool_ops
+ethtool_regs
+ifreq
 mii_bus
 napi_struct
 netdev_hw_addr

Save this to a mapping or function if you feel you will use it from time to time

Answer 2

Here you are :

function! Parse()
  " keep only the struct definitions
  v/\v.*(\s|\()struct (\w+).*/d
  " keep only the names of the structs
  %s/\v.*(\s|\()struct (\w+).*/\2/
  " sort and keep non-duplicate lines
  %sort u
  " add line numbers
  %s/^/\=printf('%-d. ', line('.')-1)
endfunction

On your file use it with :

:call Parse()

In the file you provided I didn't match as results as you (but I didn't found any smsc911x_data for example).

If you want detailed explanation about how I did just ask :)

Answer 3

I'd still suggest using the program that you wrote. You can save its output to a scratch buffer and then use tag navigation. Something like (hat-tip to Nobe4):

function! Parse()
    let command = "%!/path/to/find_structs " . shellescape(expand('%'))
    new
    " http://vim.wikia.com/wiki/Vim_buffer_FAQ 
    " (How is a scratch (temporary) buffer created?)
    setlocal buftype=nofile
    setlocal bufhidden=hide
    setlocal noswapfile
    set nobuflisted
    execute command
endfunction