If I have a syntax such as:

aa { 
  bb, auth="show", cb();
  cc { 
    dd, auth="super", cb();
  }
  ee, cb();
} 

How exactly is it enforced that someone with super/show or any authorization can run certain commands?

cli> show list1 ?
value1
data1
string1
cli> show list1 s  <Enter>

cv_string_get(cvec_i(cvv, 0)); this method returns "show list1 s" instead of "show list1 string1" in the above case
Should be able to retrieve the entire string

Consider the following pattern.

 typedef system-id {
   type string {
     pattern
       '^[0-9A-Fa-f]{4}\.[0-9A-Fa-f]{4}\.[0-9A-Fa-f]{4}';
   }
 }

It has only ^ at the beginning.

clixon> set system-id 1111.1111.1111
CLI syntax error: "set system-id 1111.1111.1111": regexp match fail: 1111.1111.1111 does not match ^[0-9A-Fa-f]{4}.[0-9A-Fa-f]{4}.[0-9A-Fa-f]{4}

Even though it is a valid regular expression match. Was working in 4.3 version.

Workaround is either both ^ $ should be given or without both of them.

Given the commands tree

Aa {
  Bb, cb();
}

I would like to able to enter commands without matching case. Currently, entering the command aa bb results in an "Unknown command" error. I would like to be able to enter the command aa bb and have it execute successfully.

On certain compilers I get an warning that says that cligen_custom.h should be in quote rather than angle brackets since it is in the compilation directory. Attached is a very small patch to fix this.

patch_cligen_config.h.in.txt

Hi there,

We have been using cligen for a while and it seems to work nicely when accessed over SSH. However on a new device and directly connecting via a console/serial port we are seeing a weird issue regarding garbage characters being showing up when pasting multiple lines at once.

Say I copy paste these 5 commands/lines at once:

just a dummy test command 1
just a dummy test command 2
just a dummy test command 3
just a dummy test command 4
just a dummy test command 5

This is how they are received on the CLI launched via Serial port:

test# just a dummy test command 1
CLI syntax error: Unknown keyword: "just" in "just a dummy test command 1"
test# just a dummy test command 2���������5�
CLI syntax error: Unknown keyword: "just" in "just a dummy test command 2"
test# +jR

i.e. the lines after the 1st one are interpreted as garbage characters.

while on a CLI launched via SSH they are parsed fine:

test# just a dummy test command 1
CLI syntax error: Unknown keyword: "just" in "just a dummy test command 1"
test# just a dummy test command 2
CLI syntax error: Unknown keyword: "just" in "just a dummy test command 2"
test# just a dummy test command 3
CLI syntax error: Unknown keyword: "just" in "just a dummy test command 3"
test# just a dummy test command 4
CLI syntax error: Unknown keyword: "just" in "just a dummy test command 4"
test# just a dummy test command 5
CLI syntax error: Unknown keyword: "just" in "just a dummy test command 5"
test#

With some trial & error we found that commenting this line i.e.

    new_termios.c_iflag |= (IGNBRK|IGNPAR);

resolves the issue on the serial port. Setting either IGNBRK or IGNPAR or both of them brings the issue back but if both are left unset this garbage character issue does not happen.

I do not have much idea about terminal settings & termios. Do you know why we need these bits to be set and any repercussions of keeping these bits unset?

Just for reference these are the default terminal settings of serial port i.e. the content of old_termios:

c_iflag		ICRNL|IUTF8|IXOFF|IXON
c_oflag		ONLCR|OPOST
c_cflag		CLOCAL|CREAD|CSIZE|HUPCL|B38400
c_lflag		ECHO|ECHOCTL|ECHOE|ECHOK|ECHOKE|ICANON|ISIG
c_cc		3,28,127,21,4,0,1,0,17,19,26,0,18,15,23,22,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
c_ispeed	38400
c_ospeed	38400

It would be helpful if there was the ability to transform or input strings - for example if I enter a password and I want to hash it before passing it down to lower levels. Ideally, the transform
function would be user definable.

A possible syntax would be an extension of the string type in the clispec file:
<var:string transform:userTransformFunction()>

CLIgen has several options that can be set programmatically using the API here:
https://github.com/clicon/cligen/blob/master/cligen_handle.h
Most of them are not documented in the cligen-tutorial pdf.
The easiest would be to to add them there.
Thanks for pointing that out.

Originally posted by @olofhagsand in #22 (comment)

I currently have this tree structure in my .cli file:

aa {
    bb, cb(); {
        (
            [cc]
            [dd]
        ), cb();
    }
}

When I execute the command aa bb, the callback cb() is execute twice, however it is only executed once for the commands aa bb cc, aa bb dd, etc. Is this expected behaviour?

So I have the following syntax:

aa<aa:string choice:bb|cc|dd|ee>;

Is there a way for me to hide an option (for example dd) when the user hits tab?

To reproduce the issue follow below steps:-

set gl_scrolling_mode to 0 in cligen to enable multiline commands.
now type type a command such that it fits in a single line.(eg: aa bb cc)
Then type a command which spans across multiple lines.(eg: any command that occupies atleast 2 lines on the cli terminal)

Then use the up arrow key twice and you will see data is no longer legible.

Ex: consider a command that takes integer as keyword value
cli> set value

cli>set value ""              <------------Exists CLI shell

Below diff solves the problem

supervisor@deeksha_march9_spine>srv2:~/development/cligen $ git diff
diff --git a/cligen_match.c b/cligen_match.c
index 2318610..6cc4af7 100644
--- a/cligen_match.c
+++ b/cligen_match.c
@@ -155,6 +155,9 @@ match_object(cligen_handle h,
 {
   int    match = 0;
   size_t len = 0;
+  enum cv_type t;
+
+  t = co->co_vtype;
 
   if (str)
       len = strlen(str);
@@ -190,8 +193,13 @@ match_object(cligen_handle h,
       }
     break;
   case CO_VARIABLE:
-      if (str == NULL || len==0)
-          match++;
+      if (str == NULL || len==0) {
+           if ((t>= CGV_INT8) && (t<=CGV_DEC64)) {
+               *reason = cligen_reason("'%s' is not a number", str);
+           } else {
+               match++;
+           }
+      }
       else
           if ((match = match_variable(h, co, str, reason)) < 0)
               return -1;
supervisor@deeksha_march9_spine>srv2:~/development/cligen $

.cli file:

show("show entries") {
           table("Display table entries") <table_name:string> ("Table name") {
		          column("Column name") @ {
				     <col_name:string cust_col_expansion()>  {
				     value("Column value") <col_val:string>("Search value"), custom_callback();
				     }
                }
           }
}

I have a requirement where i want to get certain entries from a table based on <col_name> and <co_val>
I have a custom expansion function cust_col_expansion that adds keywords col1 , col2 and col3.

cli> show table list1 column <PRESS ?>
<col_name>  column name
col1                 column 1
col2                 column 2
col3                 column 3
cli> show table list1 column col1 value <PRESS ?>
<col_val>      Enter value to be searched
cli> show table list1 column col1 value attr1            <------- scenario 1 
cli> show table list1 column col1 value val1 col2 value attr2    <-------- scenario 2 

In scenario 1: the generated parse tree will have 1 'col_name' node with value 'col1' and one 'col_val' node with value attr1
In scenario 2 : the generated parse tree will have 2 'col_name' nodes with value 'col1' and 'col2' and 2 'col_val' nodes with value attr1 and attr2

so my question is: is there any way we can dynamically change the node names 'col_name' and 'col_val' using some custom expansions so that the generated parse tree in the backend has unique nodes as referring these nodes using unique names can be useful in recursive case

(Request from [email protected])
Consider the following multi-type statements:

  (<peer:ipv4addr>("IPv4 address") | <peer:string>("Peer group name");

or

  {
    <peer:int32>("IPv4 address");
    <peer:string length[5:10]>("Peer group name");
   }

In both cases, CLIgen will produce only a single help text., even though there are two different instances (and uses) of peer:

  cli> ?
  <peer>                IPv4 address

The proposal is to enhance the help-texts so that all different help-texts (not only one) is shown. Such as:

  cli> ?
  <peer>                Peer group name
  <peer>                IPv4 address

Note that this will also apply in case of differing help texts of the same type.

When accessing the CLI using a serial port, the content wraps before the edge of the screen:

localhost.localdomain tnsr(config)# auth u
ser dsfjlkdsadfmdslkf
fhtosprjdemsofjtneyfq

where this was typed as a single line. notice that it wraps after 21 characters on the second and succeeding lines.

this does not happen when accessing the CLI via ssh.

the problem seems to be in cligen_handle.c:cligen_gwinsz() - a serial console seems to report a screen dimension of 0 x 0. in cligen_handle.c:cligen_terminal_length_set() any line length is set to a minimum of 21 characters.

in the case where the terminal dimensions are not available (0 x 0), it should disable all the special scrolling behavior and allow the terminal to deal with over length lines. or something like that.

Lets say I have the following syntax

aa {
  bb, hide {
     cc, cb();
  }
}

Right now you are unable to hide the command aa bb (you are able to tab it).

Is there a syntax to specify a valid command regardless of the order the command is entered?

For example, I want the following commands to be valid:

cli> aa bb cc dd
cli> aa bb dd cc
cli> aa cc bb dd
cli> aa cc dd bb
cli> aa dd cc bb
cli> aa dd bb cc

Right now I believe you would have to explicitly state each command in the syntax like so:

aa { 
  bb { 
    cc dd;
    dd cc;
  }
  cc { 
    bb dd;
    dd bb;
  }
  dd { 
    bb cc;
    cc bb;
  }
}

It is fine implementing this when there are only a few commands which you want like this. (i.e here we have 3 keyword, so only 3! (6) to consider)

However, if you have lots of commands like this. then the amount of permutations that you need to consider grows quickly!

Any solution to this?

Request that you move some of the structures which are not publicly exposed - cbuf, and cvec come to mind - to separate private header files which are not installed.

in our use case, we have added functions that are not of general interest and rather than maintain them as diffs to the cligen core, we could simply create a new file and avoid conflicts and errors.

Is there anyway currently to determine if a cg_obj is an optional? I am currently implementing a command to print the full command tree and would like to avoid printing all permutations for optionals. For example, with the tree

aa {
    bb {
        [
           [cc]
           [dd]
        ]
        ee;
    }
}

I would like to get the output

aa bb cc
aa bb dd
aa bb ee

but am currently getting

aa bb ee
aa bb cc ee
aa bb cc dd ee
aa bb dd ee
aa bb dd cc ee

If there is some way I could determine that cc and dd are optionals it would make this much easier.

Hello,
I am trying to write a cli using your cligen module. I am facing a issue.
Below is my syntax:-

aa
{
[bb]
[cc]
[dd];
}

now on cli prompt output:-
cli>aa bb cc dd --- valid output
cli>aa cc dd bb --- not a valid output

I am wondering why i am not able to give the second command.
Is there a restriction on order of parameters at the same level. If yes, I would like to have an order independent command processing. Can we have a syntax which will help in in achieving my goal.

when i do
cli> aa cc ?
it should show both
bb
dd

as help ouptut.

Thanks and Regards,
Deepak Sharma

For a terminal command with variable, I would love the ability to enter multiple values for the variable. For example, implementing the bash command ls with the cli format
ls("list files")[<ls-var:string>], cb();
I would like to be able to enter commands with multiple variables (eg. ls -l -a, ls -a dirname, etc.) without having to enter the variables in quotes (ls "-l -a") to simulate typical bash syntax. Is this something that would be feasible? I looked into using sets for this but it doesn't seem to fit this use case

Choice with a space is not parsed by Cligen code.

Example:

1. choice ("Choice example") <string choice:unknown|NoSpace|with\ space>, fn_cb(); -->not working
2. choice ("Choice example") <string choice:unknown|NoSpace|”with space”>, fn_cb(); -->not working

As per yang RFC, space is a valid character in choice type.

Please note: Auto-complete work for first case, but when you press enter, it will say invalid token.

size_t diff;

On some platforms size_t type may be defined as:

typedef undefined int size_t

and block which allocates memory may not happen.

cbuf_append function in cligen_buf.c

Issue is seen when redundant keywords are present in the referenced tree at the same hierarchy

consider tree syntax :
file:op.cli
treename= "op";
show {
word1;
word2;
}

file : test.cli
show {
word3;
word4;
}
@op;

cli: op> show <PRESS ?>
word1
word2
cli :op>

cli: cfg><PRESS ?>
show
show                    <------------------ duplicate keywords
cli: cfg> show <PRESS ?>     <-------------------no suggestion are seen

Expected behaviour :

cli: cfg> show <PRESS ?>
word1
word2
word3
word4
cli: cfg>

please make cligen capable of handling unicode

When a user provides a valid incomplete command, is it possible for cligen to put the user into the scope/level of the syntax structure of that command rather than returning the "Unkown command" error message ?

For example, given the following syntax....
aa {
bb;
cc;
dd {
ee;
}
}

The user types the command : "aa\n" . Cligen would then not return "Unknown Command", but allows the user to perform the commands defined inside the "aa" block at the next prompt. The prompt could contain the current scope/level of the command.

cli>aa
cli(aa)>dd
cli(aa/dd)>ee

Thanks,
Marek Gimza

Escaping the comment character (which clixon sets to '#') is useful to enter strings that include the comment character. Specifically, passwords might contain all sort of special characters.

Currently the low level cli_trim() routine handles comments removal. So implementing escaping of the comment character would require moving it to higher level syntax aware code.

Hi olofhagsand:

I used latest cligen code(2018-01-27 7925db3)with following cli file,

snmp set <$cfg_type%g:string choice:sysname> ,cb();     
                                                    
snmp <$test:string choice:hello>,cb();                  
snmp set <$priv_cfg_type%g:string choice:priv> ,cb();   
                                                    
snmp set <$access_cfg_type%g:string choice:access>,cb();

Hello,

What would be the best way to implement output modifiers, e.g. show something | include bla to filter out the output of the given command?
Other example, but probably different implementation: show configuration | display json, which would totally change the output (produce a JSON output of the configuration in that case).

Thanks,
Sylvain

Hello!

From the documentation I see that cligen provides ability to hook a translation function. Using that one can convert the plain text into an encrypted / obfuscated text to be stored in the db I think. How to go about it in RESTCONF? Does the clixon backend have some ability to tweak the data before it get's stored in the candidate / running db itself? Or is there a way for the restconf application to covert the transform the entered plain text password into an encrypted one ?

Thanks!

EDIT : Wrong repo ! Raised the issue on clicon/clixon#127. Sorry for the confusion!

Given the .cli syntax

aa<aa-var:string>), cb();

I would like to be able to type the command aa without a variable, as well as aa <variable>. Is the possible to achieve currently?

Move the cligen repo to the clicon project at https://github.com/clicon.
The reason is that cligen and clixon are closely related and would benefit from being in the same project.
It would be less confusing.

Hello,

I'm trying to develop a CLI using cligen, and I would like to have some hidden (non-documented) commands. I have read in the tutorial that it's possible using the keyword "hide". However I have not managed to make it work when the command contains variables (parameters).

The cligen_tutorial example contains a simple hidden command. It is possible to reproduce the issue if the definition of the command in the file "tutorial.cli" is changed as shown below:

#secret,hide,secret("message");
secret("Help text") param:string,hide,secret("message");

Then the parameter is hide but not the command.

Am I doing something wrong?

Thanks and Regards,
Sergio

Given a "hierarchical" cligen set command structure as follows:

  a @{
    b;
    c; @{
      d;
      e;
    }
  }

It is possible to issue command: a c b and a c d b, but query and completion does not work properly.
For example, the following lacks the "b" option:

> a c ?
  <cr>
  d                    
  e    

expected behavior:

> a c ?
  <cr>
  b        <-------------
  d                    
  e    

Given the following syntax:

aa @{
    bb<int32>;
    cc;
    dd;
}

If i type in the following, I see the following:

cli > aa bb <TAB>
 cc                    dd                    <int32>

This should instead produce:

cli > aa bb <TAB>
 <int32>

Another bug I have found is associated with the following syntax:

aa @{
    bb;
    cc;
    dd;
}
aa ff;

If I type in the following and then hit tab, I see:

cli> aa ff <TAB>
 bb                    cc                    dd  

In the case I should not see anything when hitting tab. It should look like this:

cli> aa ff <TAB>
cli> aa ff 

when I press the delete button, the ~ sign is printed instead of deleting the character

Example clispec:

value;
value2;

Run:

> ./cligen_file -f foo.cli
cli> value?
  value
cli>

Expected: value2 as a separate option
This error was introduced in CLIgen 5.4

Expansion plugin defined in .cli gets called twice for each expansion.

I have also debugged and found that the below code snippet might the causing the issue.

static int                                                                      
cli_tab_hook(cligen_handle h,                                                   
         int          *cursorp)                                                 
{                                                                               
    int retval = -1;                                                            
    int completed = 0;                                                          
                                                                                
    if (cli_try_complete(h, cursorp, &completed) < 0)                           
    goto done;                                                                  
    if (!completed){                                                            
    if (cli_show_help_commands(h, cligen_buf(h), cligen_tabmode(h) & CLIGEN_TABMODE_COLUMNS) < 0)
        goto done;                                                              
    }                                                                           
    retval = 0;                                                                 
 done:                                                                          
    return retval;                                                              
}

In the above code snippet cli_try_complete as well as cli_show_help_commands both call the expansion function.
In the above code completed variable is not at all modified by cli_try_complete API.
As per my understanding completed should be set to 1 if we get the full expansion.

As per discussion over chat with Olof cli_try_complete indeed has issues as completed is never set.

step 1 : execute a command in cli

supervisor@deeksha_jan28: op> show config set interface 

step 2 : press ctrl+r and type "set" and then press right arrow

supervisor@deeksha_jan28: op> show config [Cset interface

you can see that '[c' get added before 'set'

Performance slow down, when string size is large. Since invoke 'strlen' too many times.

for (i=0;i<strlen(str);i++)
if (str[i] != '\')
str[j++] = str[i];
for (;j<strlen(str);j++)
str[j] = '\0';

May can replace by below:

size_t len = strlen(str);
for (i=0;i<len;i++)
if (str[i] != '\')
str[j++] = str[i];
for (;j<len;j++)
str[j] = '\0';

In CLIgen runtime, comments are characters on a line after the special "comment" character which is default '#'.
Example:
# This is a comment
cmd a#this is a comment
You can set the comment character to \0 in which no comments are allowed.

However, thus may be somewhat too basic/raw. In bash for example (https://linux.die.net/man/1/sh), comments

In a non-interactive shell, or an interactive shell in which the interactive_comments option to the shopt builtin is enabled (see SHELL BUILTIN COMMANDS below), a word beginning with # causes that word and all remaining characters on that line to be ignored. An interactive shell without the interactive_comments option enabled does not allow comments. The interactive_comments option is on by default in interactive shells.

CLIgen does not have the concept of "interactive".

The proposal is to introduce a word beginning with # causes that word and all remaining characters on that line to be ignored. This is different from the current behavior where all occurrences of # are used as comment.

Need to add the possibility to use anchors (^ and $) around patterns (regexp) which is part of POSIX.
Now https://github.com/clicon/cligen/blob/master/cligen_regex.c#L91 always wraps any regexp with ^(...)$

I want to be able to display a multiword string when hitting tab using the show attribute. I have tried to use the following syntax:

aa<int32 show:"A number">

But I get an error stating complaining about the "

When I try to use the following

aa<int32 show:A\ number>

When I tab it see the following:

cli> aa 
<A\ number>

Any way to use show so that I see the following:

cli> aa 
<A number>

Hi,

I have not seen any issue if input string is small but seen unexpected behavior when taking a input string >= 1024

getting below bt

#0 __GI_raise (sig=sig@entry=6)
at /usr/src/debug/glibc/2.26-r0/git/sysdeps/unix/sysv/linux/raise.c:51
#1 0x00000034fca355e5 in __GI_abort ()
at /usr/src/debug/glibc/2.26-r0/git/stdlib/abort.c:90
#2 0x00000034fca74bde in __libc_message (action=action@entry=do_abort,
fmt=fmt@entry=0x34fcb744d3 "%s\n") aaaaaaaa
at /usr/src/debug/glibc/2.26-r0/git/sysdeps/posix/libc_fatal.c:181 aaaaaaaa
#3 0x00000034fca7abaa in malloc_printerr (
str=str@entry=0x34fcb72628 "corrupted size vs. prev_size")
at /usr/src/debug/glibc/2.26-r0/git/malloc/malloc.c:5368
#4 0x00000034fca7bcc4 in _int_free (av=av@entry=0x34fcda9c40 <main_arena>,
p=, have_lock=have_lock@entry=1)
at /usr/src/debug/glibc/2.26-r0/git/malloc/malloc.c:4302
#5 0x00000034fca7e222 in _int_realloc (av=av@entry=0x34fcda9c40 <main_arena>,
oldp=oldp@entry=0x1163c40, oldsize=oldsize@entry=1056, nb=nb@entry=80)
at /usr/src/debug/glibc/2.26-r0/git/malloc/malloc.c:4666
#6 0x00000034fca8082b in __GI___libc_realloc (oldmem=0x1163c50, bytes=72)
at /usr/src/debug/glibc/2.26-r0/git/malloc/malloc.c:3231
#7 0x000000350760837d in pt_realloc (pt=0x729548) at cligen_gen.c:315
#8 0x0000003507609d4c in co_insert (pt=0x729548, co1=0x7209d0)
at cligen_gen.c:1010
#9 0x0000003507617755 in pt_expand_treeref (h=0x66c8b0, coprev=0x0,

is any restriction on input length ?

Thank

To reproduce

git clone https://github.com/olofhagsand/cligen.git
mkdir build
cd build
../cligen/configure
make

I stumbled onto an interesting bug related to the use of asserts() and the -DNDEBUG compiler flag. I have been experimenting with using CMake to build Cligen (you can find that attempt here: https://github.com/theonemcdonald/cligen/tree/cmake).

The CMake default CFLAGS for each build type are as follows:

CMAKE_C_FLAGS_DEBUG: -g
CMAKE_C_FLAGS_RELEASE: -O3 -DNDEBUG
CMAKE_C_FLAGS_RELWITHDEBINFO: -O2 -g -DNDEBUG
CMAKE_C_FLAGS_MINSIZEREL: -Os -DNDEBUG

The failure case can be readily observed by compiling Cligen with -DNDEBUG and running the cligen_hello test application. In this case, typing hello<tab><tab><tab>... causes Cligen to just keep repeating the hello keyword:

building without -DNDEBUG yields the correct behavior:

The issue appears to be related to an assert() call having undesired side-effects.

After consulting with Olof, we determined this to be a likely culprit that should be explored further:

cligen_match.c:239:18: warning: variable 'levels' is uninitialized when used here [-Wuninitialized]
    if (level >= levels)
                 ^~~~~~
cligen_match.c:236:15: note: initialize the variable 'levels' to silence this warning
    int levels;
              ^
               = 0

Hi,

I am trying to parse following sample input file and cligen code hangs at the time of parsing. I enabled debug flag and screen is flooded with following message.
cgy_list_push

The cligen file I am using is as below:

`#CLIgen input file
treename="main";
@config_command;

treename="config_command";
show("Retrieve yang object data") @show, show_cb();

treename="container_inventory_list_show";
list(""), fn_cb();{
("list key name"), fn_cb();
{
[aa ("des")]
[ab ("des")]
[ac ("des")]
[ad ("des")]
[ae ("des")]
[af ("des")]
[ag ("des")]
[ah ("des")]
[ai ("des")]
[aj ("des")]
[ak ("des")]
[al ("des")]
[am ("des")]
[an ("des")]
[ao ("des")]
[ap ("des")]
[aq ("des")]
[ar ("des")]
[as ("des")]
[at ("des")]
[au ("des")]
[av ("des")]
[aw ("des")]
[ax ("des")]
[ay ("des")]
[az ("des")]
[ba ("des")]
[bb ("des")]
[bc ("des")]
[bd ("des")]
[be ("des")]
[bf ("des")], fn_cb();
}
}

treename="container_inventory_show";
inventory(""), fn_cb();
{
[@container_inventory_list_show], fn_cb();
}

treename="container_show";
container(""), fn_cb();
{
<string length[1:32]>("The name of the container."), fn_cb();
{
[@container_inventory_show], fn_cb();
}
}

treename="show";
@container_show;
`

• I tried to narrow down the problem and come up with following observations: If i remove some parameters from container_inventory_list_show tree (say if I remove ax to be), the code works perfectly fine. After adding new parameter one by one in same tree, the code become slower. And after adding few more parameters, the cligen code hang in parsing.
• The issue is not in bison because I tried with different bison versions (2.6, 3.0.4, 3.2 etc) and issue still there.

This is from the last code on master pulled today 2019-03-19

brett@brett-Precision-7740:~/git/cligen$ ./cligen_tutorial -f tutorial.cli 
Syntax:
aa bb, cb();{
      ca("help ca") <int32:int32>, letters();{
            dd, letters();
            ee, letters();
         }
      cb("help cb"), letters();{
         dd, letters();
         ee, letters();
      }
      ff, cb();
   }
access-list("ACL") permit("permit") <src:ipv4addr show:"A.B.C.D">("Source address") <dst:ipv4addr show:"A.B.C.D">("Destination address");
add @sub, add();
change{
   prompt <new:string>("New prompt"), setprompt();
   tree, changetree(0 : (null) = tree1
);
}
del @sub, del();
ex <a:int32 show:"number">("A number");
hello("Greet the world") world, hello(0 : (null) = Hello World!
);
increment <var:string translate:incstr()>, cb();
interface <ifname:string interface("")>("interface name"), cb();
ip("The IP protocol"){
   tcp("The TCP protocol") <port:int32>("Port number"), cb(0 : (null) = tcp
);
   udp("The UDP protocol") <addr:ipv4addr>("IPv4 address"), cb(0 : (null) = udp
);
}
quit("quit the CLI"), quit();
recurse @tutorial;
secret, hide, secret(0 : (null) = message
);
values{
   aa, cb();
   <int32:int32>, cb();
   <int64:int64>, cb();
   <string:string>, cb();
}
Syntax:
x y, fn(0 : (null) = a
);
Syntax:
change tree, changetree(0 : (null) = tutorial
);
z x;
cli> interface eth0
CLI syntax error in: "interface eth0": Unknown command
cli> 

The interface function is being called to provide the list of options, but then the command is not recognized nor is the callback function cb called.

Does not seem to be working regardless of if you provide fraction digits, (always set to 0 for whatever reason).