wordsplit: rewrite positional parameters implementation

This improves 3e07e3ad

* include/wordsplit.h (ws_paramv,ws_paramc)
(ws_parambuf,ws_paramidx,ws_paramsiz): New fields.
(WRDSO_PARAMV,WRDSO_PARAM_NEGIDX): New options.
(WRDSE_BADPARAM): New error code.
(wordsplit_free_parambuf): New proto.
* src/wordsplit.c (wordsplit_init): Initialize new fields.
(wsplt_assign_var): Fix double-free and memory leak.
(expvar): Expand positional parameters.
(begin_var_p): Add '#'
(wordsplit_free_envbuf): Fix condition.
(wordsplit_free_parambuf): New function.
(wordsplit_free): Call wordsplit_free_parambuf.
(_wordsplit_errstr): New error description.
* tests/wordsplit.at: Update wsp invocations.
Test positional parameters.
* tests/wsp.c: Rewrite.

4 files changed, 704 insertions, 471 deletions

diff --git a/include/wordsplit.h b/include/wordsplit.h
index a175275..d7eb26f 100644
--- a/include/wordsplit.h
+++ b/include/wordsplit.h
@@ -76,4 +76,14 @@ struct wordsplit
   size_t ws_envidx;         /* Index of first free slot */
   size_t ws_envsiz;         /* Size of the ws_envbuf array */
+
+  char const **ws_paramv;   /* [WRDSO_PARAMV] User-supplied positional
+                               parameters */
+  size_t ws_paramc;         /* Number of positional parameters */
+
+        /* Temporary storage for parameters. Works similarly to ws_enbuf.
+         */
+  char **ws_parambuf;
+  size_t ws_paramidx;
+  size_t ws_paramsiz;
   
   int (*ws_getvar) (char **ret, const char *var, size_t len, void *clos);
@@ -100,5 +110,5 @@ struct wordsplit
                                See ws_getvar for a discussion of possible
                                return values. */
-        
+
   const char *ws_input;     /* Input string (the S argument to wordsplit. */  
   size_t ws_len;            /* Length of ws_input. */
@@ -227,4 +237,10 @@ struct wordsplit
 #define WRDSO_NOCMDSPLIT     0x00002000
 
+/* Enable positional parameters */
+#define WRDSO_PARAMV         0x00004000
+/* Enable negative positional indices (${-1} is the last positional
+   parameter) */
+#define WRDSO_PARAM_NEGIDX   0x00008000
+
 #define WRDSO_BSKEEP          WRDSO_BSKEEP_WORD     
 #define WRDSO_OESC            WRDSO_OESC_WORD       
@@ -251,4 +267,5 @@ struct wordsplit
 #define WRDSE_GLOBERR    8
 #define WRDSE_USERERR    9
+#define WRDSE_BADPARAM  10
 
 int wordsplit (const char *s, wordsplit_t *ws, int flags);
@@ -257,4 +274,5 @@ void wordsplit_free (wordsplit_t *ws);
 void wordsplit_free_words (wordsplit_t *ws);
 void wordsplit_free_envbuf (wordsplit_t *ws);
+void wordsplit_free_parambuf (struct wordsplit *ws);
 int wordsplit_get_words (wordsplit_t *ws, size_t *wordc, char ***wordv);
 
diff --git a/src/wordsplit.c b/src/wordsplit.c
index 521a1eb..6a33636 100644
--- a/src/wordsplit.c
+++ b/src/wordsplit.c
@@ -1,4 +1,4 @@
 /* wordsplit - a word splitter
-   Copyright (C) 2009-2018 Sergey Poznyakoff
+   Copyright (C) 2009-2019 Sergey Poznyakoff
 
    This program is free software; you can redistribute it and/or modify it
@@ -28,4 +28,5 @@
 #include <pwd.h>
 #include <glob.h>
+#include <limits.h>
 
 #if ENABLE_NLS
@@ -57,4 +58,7 @@
  ((wsp)->ws_flags & WRDSF_RETURN_DELIMS || ((wsp)->ws_options & WRDSO_MAXWORDS))
 
+#define to_num(c) \
+  (ISDIGIT(c) ? c - '0' : (ISXDIGIT(c) ? toupper(c) - 'A' + 10 : 255 ))
+
 #define ALLOC_INIT 128
 #define ALLOC_INCR 128
@@ -285,4 +289,12 @@ wordsplit_init (struct wordsplit *wsp, const char *input, size_t len,
         }
     }
+
+  if (!(wsp->ws_options & WRDSO_PARAMV))
+    {
+      wsp->ws_paramv = NULL;
+      wsp->ws_paramc = 0;
+    }
+  wsp->ws_paramidx = wsp->ws_paramsiz = 0;
+  wsp->ws_parambuf = NULL;
   
   wsp->ws_endp = 0;
@@ -1037,5 +1049,5 @@ wordsplit_find_env (struct wordsplit *wsp, const char *name, size_t len,
 static int
 wsplt_assign_var (struct wordsplit *wsp, const char *name, size_t namelen,
-                  char *value)
+                  char const *value)
 {
   int n = (wsp->ws_flags & WRDSF_ENV_KV) ? 2 : 1;
@@ -1072,5 +1084,5 @@ wsplt_assign_var (struct wordsplit *wsp, const char *name, size_t namelen,
                       for (; j > 1; j--)
                         free (newenv[j-1]);
-                      free (newenv[j-1]);
+                      free (newenv);
                       return _wsplt_nomem (wsp);
                     }
@@ -1143,4 +1155,68 @@ wsplt_assign_var (struct wordsplit *wsp, const char *name, size_t namelen,
 }
 
+int
+wsplt_assign_param (struct wordsplit *wsp, int param_idx, char *value)
+{
+  char *v;
+  
+  if (param_idx < 0)
+    return WRDSE_BADPARAM;
+  if (param_idx == wsp->ws_paramc)
+    {
+      char **parambuf;
+      if (!wsp->ws_parambuf)
+        {
+          size_t i;
+          
+          parambuf = calloc ((size_t)param_idx + 1, sizeof (parambuf[0]));
+          if (!parambuf)
+            return _wsplt_nomem (wsp);
+
+          for (i = 0; i < wsp->ws_paramc; i++)
+            {
+              parambuf[i] = strdup (wsp->ws_paramv[i]);
+              if (!parambuf[i])
+                {
+                  for (; i > 1; i--)
+                    free (parambuf[i-1]);
+                  free (parambuf);
+                  return _wsplt_nomem (wsp);
+                }
+            }
+          
+          wsp->ws_parambuf = parambuf;
+          wsp->ws_paramidx = param_idx;
+          wsp->ws_paramsiz = param_idx + 1;
+        }
+      else
+        {
+          size_t n = wsp->ws_paramsiz;
+          
+          if ((size_t) -1 / 3 * 2 / sizeof (wsp->ws_parambuf[0]) <= n)
+            return _wsplt_nomem (wsp);
+          n += (n + 1) / 2;
+          parambuf = realloc (wsp->ws_parambuf, n * sizeof (wsp->ws_parambuf[0]));
+          if (!parambuf)
+            return _wsplt_nomem (wsp);
+          wsp->ws_parambuf = parambuf;
+          wsp->ws_paramsiz = n;
+          wsp->ws_parambuf[param_idx] = NULL;
+        }
+
+      wsp->ws_paramv = (const char**) wsp->ws_parambuf;
+      wsp->ws_paramc = param_idx + 1;
+    }
+  else if (param_idx > wsp->ws_paramc)
+    return WRDSE_BADPARAM;
+  
+  v = strdup (value);
+  if (!v)
+    return _wsplt_nomem (wsp);
+    
+  free (wsp->ws_parambuf[param_idx]);
+  wsp->ws_parambuf[param_idx] = v;
+  return WRDSE_OK;
+}
+
 /* Recover from what looked like a variable reference, but turned out
    not to be one. STR points to first character after '$'. */
@@ -1178,4 +1254,6 @@ expvar (struct wordsplit *wsp, const char *str, size_t len,
   int rc;
   struct wordsplit ws;
+  int is_param = 0;
+  long param_idx = 0;
   
   if (ISVARBEG (str[0]))
@@ -1186,14 +1264,37 @@ expvar (struct wordsplit *wsp, const char *str, size_t len,
       *pend = str + i - 1;
     }
-  else if (ISDIGIT (str[0]))
+  else if ((wsp->ws_options & WRDSO_PARAMV) && ISDIGIT (str[0]))
     {
       i = 1;
       *pend = str;
+      is_param = 1;
+      param_idx = to_num (str[0]);
     }
-  else if (str[0] == '{' && (ISVARBEG (str[1]) || ISDIGIT (str[1])))
+  else if ((wsp->ws_options & WRDSO_PARAMV) && str[0] == '#')
+    {
+      char b[16];
+      snprintf (b, sizeof(b), "%d", (int) wsp->ws_paramc);
+      value = strdup (b);
+      if (!value)
+        return _wsplt_nomem (wsp);
+      if (wsnode_new (wsp, &newnode))
+        return 1;
+      wsnode_insert (wsp, newnode, *ptail, 0);
+      *ptail = newnode;
+      newnode->flags = _WSNF_WORD | _WSNF_NOEXPAND | flg;
+      newnode->v.word = value;
+      return 0;
+    }
+  else if (str[0] == '{'
+           && (ISVARBEG (str[1])
+               || (is_param = (((wsp->ws_options & WRDSO_PARAMV)
+                                && ISDIGIT (str[1]))
+                               || ((wsp->ws_options & WRDSO_PARAM_NEGIDX)
+                                   && (str[1] == '-'
+                                       && ISDIGIT (str[2]))))) != 0))
     {
       str++;
       len--;