Re: gEDA-user: pcb: parameter escaping for commands
On Sat, Nov 22, 2008 at 01:49:57PM -0500, DJ Delorie wrote:
Is the supported syntax described anywhere?
Nope. You get to write it! :-)
As long as it can do everything in pcb-menu.res it will be fine.
That's what I wanted to hear (the latter part at least). g
Patch attached. I tried to follow the current coding style, including
usage of libc string functions (DJB-style string buffer functions would
have made it easier). If the style is fine with you, I can create a bug
report and attach it.
It seems to work fine for me, but I could not find any test suite.
Two important changes:
1. Action chaining isn't supported anymore. Wasn't documented, isn't
used by anything AFAIK (resource parser does its own splitting).
2. The sequence \_ will now be parsed as _, so if PCB did use that
to create negation lines, it will be broken until gsch2pcb and possibly
other tools are adapted. Special-casing that sequence is possible, but
IMO ugly.
The patch is only for actions - haven't looked at footprint etc. formats
yet.
PS: How do I display the pin names (besides Generate object report on
a pin)? Neither View - Pins/Vias show Name/Number nor Window -
Pinout show any effect (regardless of selection).
CU Sascha
--
http://sascha.silbe.org/
http://www.infra-silbe.de/
Index: src/hid/common/actions.c
===
RCS file: /cvsroot/pcb/pcb/src/hid/common/actions.c,v
retrieving revision 1.12
diff -u -d -r1.12 actions.c
--- src/hid/common/actions.c 5 Jan 2008 05:37:08 - 1.12
+++ src/hid/common/actions.c 23 Nov 2008 20:56:14 -
@@ -219,123 +219,152 @@
return a-trigger_cb (argc, argv, x, y);
}
+static char *
+strip (const char *s)
+{
+ const char *sp = s;
+ unsigned int len;
+ char *res;
+
+ /* skip leading spaces and tabs */
+ while (*sp isspace ((int) *sp))
+sp++;
+
+ len = strlen(sp);
+
+ /* skip trailing spaces and tabs */
+ while (len isspace((int) sp[len-1]))
+len--;
+
+ if (!len)
+ return ;
+
+ res = (char *) malloc(len+1); // ENOMEM not checked
+ if (!res)
+return NULL;
+
+ strncpy(res, sp, len);
+ res[len] = '\0';
+
+ return res;
+}
+
+static char *
+my_strndup(const char *s, size_t n)
+{
+ char *res = (char *) malloc(n+1);
+ strncpy(res, s, n);
+ res[n] = '\0';
+ return res;
+}
+
+/* returns copy of next parameter, updating sp to point behind separator */
+static char *
+get_next_param(char **sp)
+{
+ char *sep, *res;
+
+ sep = strpbrk(*sp, \\,));
+
+ if (!sep)
+sep = *sp + strlen(*sp);
+
+ if (*sep == '\\' *(sep+1))
+{
+ /* recursively process remaining part, concatenate
+ escaping is considered to be rare, so we don't need to optimize it */
+ size_t curlen = sep-*sp;
+ char escaped = sep[1];
+ char *rem;
+ sep += 2;
+ rem = get_next_param(sep);
+ res = (char *) malloc(curlen+1+strlen(rem)+1);
+ strncpy(res, *sp, curlen);
+ res[curlen] = escaped;
+ strcpy(res+curlen+1, rem);
+ free(rem);
+ *sp = sep;
+ return res;
+}
+
+ /* ignore trailing whitespace */
+ while ((sep *sp) isspace((int) *(sep-1)))
+sep--;
+
+ /* return copy of parameter */
+ res = my_strndup(*sp, sep-*sp);
+ *sp = (*sep) ? sep+1 : sep;
+ return res;
+}
+
int
hid_parse_actions (const char *rstr,
- int (*function) (const char *, int, char **))
+ int (*function) (const char *, int, char **))
{
+ char *str, *sp, *aname;
+ int num = 0, max = 0;
char **list = NULL;
- int max = 0;
- int num;
- char *str = NULL;
- char *sp, *aname, *sp2;
- int maybe_empty = 0;
- int retcode = 0;
+ char *paren_pos;
+ int retcode;
if (function == NULL)
function = hid_actionv;
- /*fprintf(stderr, invoke: `%s'\n, rstr); */
-
- sp = str = strdup (rstr);
+ aname = sp = str = strip (rstr);
+ if (!sp || !*sp)
+return 0;
-another:
- num = 0;
- /* eat leading spaces and tabs */
- while (*sp isspace ((int) *sp))
-sp++;
-
- if (!*sp)
+ /* action names don't need to be escaped so strchr suffices */
+ paren_pos = strchr(sp, '(');
+ if (paren_pos)
{
- retcode = 0;
- goto cleanup;
-}
+ /* NUL-terminate action name */
+ *paren_pos = '\0';
+ sp = paren_pos+1;
+
+ /* scan for parameters */
+ while (*sp *sp != ')')
+{
+ char *param;
+
+ /* ignore leading whitespace */
+ while (*sp isspace ((int) *sp))
+sp++;
+
+ if (!*sp || (*sp == ')'))
+break;
+
+ /* get_next_param updates sp */
+ param = get_next_param(sp);
- aname = sp;
+ /* ensure list is big enough */
+ if (num = max)
+{
+ max += 10;
+ if (list)
+list = (char **) realloc (list, max * sizeof (char *));
+ else
+list = (char **) malloc (max * sizeof (char *));
+}
- /* search for the
Re: gEDA-user: pcb: parameter escaping for commands
1. Action chaining isn't supported anymore. Wasn't documented, isn't=20 used by anything AFAIK (resource parser does its own splitting). Er, what was this? PS: How do I display the pin names (besides Generate object report on=20 a pin)? Neither View - Pins/Vias show Name/Number nor Window -=20 Pinout show any effect (regardless of selection). That's the way. Most pins have the same string for name and number until you run the script from gsch2pcb to change them. ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: pcb: parameter escaping for commands
On Sun, Nov 23, 2008 at 04:14:46PM -0500, DJ Delorie wrote: 1. Action chaining isn't supported anymore. Wasn't documented, isn't=20 used by anything AFAIK (resource parser does its own splitting). Er, what was this? The old parser supported constructs like Display(Name) Display(Value), invoking both functions in sequence. The new one stops after the first ). In the resource file, those are two different entries anyway, so it doesn't matter there. PS: How do I display the pin names (besides Generate object report on=20 a pin)? Neither View - Pins/Vias show Name/Number nor Window -=20 Pinout show any effect (regardless of selection). That's the way. Most pins have the same string for name and number until you run the script from gsch2pcb to change them. Well, it doesn't display any number or name (for pins). By luck, I now found out that both work only if the pointer is above the currently selected element - which is impossible if I use the menu. CU Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ signature.asc Description: Digital signature ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: pcb: parameter escaping for commands
On Sun, 2008-11-23 at 22:12 +0100, Sascha Silbe wrote: Two important changes: 1. Action chaining isn't supported anymore. Wasn't documented, isn't used by anything AFAIK (resource parser does its own splitting). Please don't break this. I already spend long enough finding the obscure bug and fixing it which broke re-entrancy with the old code. This _is_ encountered in xgsch2pcb, where one action loads another action script via an action. -- Peter Clifton Electrical Engineering Division, Engineering Department, University of Cambridge, 9, JJ Thomson Avenue, Cambridge CB3 0FA Tel: +44 (0)7729 980173 - (No signal in the lab!) ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: pcb: parameter escaping for commands
On Mon, 2008-11-24 at 00:51 +, Peter Clifton wrote: On Sun, 2008-11-23 at 22:12 +0100, Sascha Silbe wrote: Two important changes: 1. Action chaining isn't supported anymore. Wasn't documented, isn't used by anything AFAIK (resource parser does its own splitting). Please don't break this. I already spend long enough finding the obscure bug and fixing it which broke re-entrancy with the old code. This _is_ encountered in xgsch2pcb, where one action loads another action script via an action. Given your previous explanation, I guess I misunderstood what you meant by action chaining. What I was refering to was a bug with re-entrancy in the parser which caused breakage. -- Peter Clifton Electrical Engineering Division, Engineering Department, University of Cambridge, 9, JJ Thomson Avenue, Cambridge CB3 0FA Tel: +44 (0)7729 980173 - (No signal in the lab!) ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: pcb: parameter escaping for commands
On Mon, Nov 24, 2008 at 12:51:33AM +, Peter Clifton wrote: 1. Action chaining isn't supported anymore. Wasn't documented, isn't used by anything AFAIK (resource parser does its own splitting). Please don't break this. I already spend long enough finding the obscure bug and fixing it which broke re-entrancy with the old code. This _is_ encountered in xgsch2pcb, where one action loads another action script via an action. I think we're talking about two different things here. The new parser doesn't handle several actions passed as a single string (e.g. Display(Name) Display(Value)). Having an action (lets say SetAll(Clearance,0.1)) parse and execute another action (lets say SetElement(U1,Clearance,0.1)) should work perfectly well. Sorry if my imprecise description caused confusion. CU Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ signature.asc Description: Digital signature ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: pcb: parameter escaping for commands
The old parser supported constructs like Display(Name) Display(Value), invoking both functions in sequence. The new one stops after the first ). I suppose the parser can return a pointer to the leftovers, and we can have a new function that continues parsing until the string is used up. I suspect the command line --action-string and --action-script may need this type of functionality. Well, it doesn't display any number or name (for pins). By luck, I now found out that both work only if the pointer is above the currently selected element - which is impossible if I use the menu. To turn *on* the display, press the 'd' key over the element you're interested in. The menu option changes *which* string is displayed, not *if* it's displayed. Window-Pinout is supposed to ask you to click on an element, file a bug (with which HID you're using) if it doesn't. Shift-D over an element does work, though. ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
gEDA-user: pcb: parameter escaping for commands (was: Re: pcb/gsch2pcb: problem with special characters in pin names)
On Fri, Nov 14, 2008 at 12:55:30PM -0500, DJ Delorie wrote: pcb currently has no escape mechanism for that. It would be added to src/hid/common/actions.c in hid_parse_actions() if you want to work on it. Looks like it's going to need a full rewrite of that function to implement escaping. Is the supported syntax described anywhere? hid_parse_actions is able to parse multiple invocations with parameters, but an invocation without parameters must be the last one; empty parentheses supply a single, empty parameter (instead of no parameters like an invocation without parentheses does) etc. I'd rather do the reimplementation based on a syntax definition than trying to mimic the old function as much as possible. CU Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ signature.asc Description: Digital signature ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: pcb: parameter escaping for commands (was: Re: pcb/gsch2pcb: problem with special characters in pin names)
Is the supported syntax described anywhere? Nope. You get to write it! :-) As long as it can do everything in pcb-menu.res it will be fine. ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
