New utility: nmem_strsplit_escape

diff --git a/include/yaz/nmem.h b/include/yaz/nmem.h
index c12ddc0..192f362 100644 (file)
--- a/include/yaz/nmem.h
+++ b/include/yaz/nmem.h
@@ -134,6 +134,21 @@ YAZ_EXPORT void nmem_strsplitx(NMEM nmem, const char *delim,
 YAZ_EXPORT void nmem_strsplit_blank(NMEM nmem, const char *dstr,
                                     char ***darray, int *num);
 
+
+/** \brief allocates sub strings out of string using certain delimitors
+    \param nmem NMEM handle
+    \param delim delimitor chars (splits on each char in there) 
+    \param dstr string to be split
+    \param darray result string array for each sub string
+    \param num number of result strings
+    \param collapse 1=collapse multiple delims to one; 0=no collapse
+    \param escape_char != 0, an escape char (could be \)
+*/
+YAZ_EXPORT void nmem_strsplit_escape(NMEM nmem, const char *delim,
+                                     const char *dstr,
+                                     char ***darray, int *num, int collapse,
+                                     int escape_char);
+
 /** \brief allocates and sets integer for NMEM
     \param nmem NMEM handle
     \param v integer value

diff --git a/src/nmemsdup.c b/src/nmemsdup.c
index f697e73..fcfb074 100644 (file)
--- a/src/nmemsdup.c
+++ b/src/nmemsdup.c
@@ -67,40 +67,19 @@ void nmem_strsplit(NMEM nmem, const char *delim, const char *dstr,
 void nmem_strsplitx(NMEM nmem, const char *delim, const char *dstr,
                     char ***darray, int *num, int collapse)
 {
-    const char *cp = dstr;
-    *num = 0;
+    nmem_strsplit_escape(nmem, delim, dstr, darray, num, collapse, 0);
+}
 
+void nmem_strsplit_escape(NMEM nmem, const char *delim, const char *dstr,
+                          char ***darray, int *num, int collapse,
+                          int escape_char)
+{
+    *darray = 0;
+    /* two passes over the input string.. */
     while (1)
     {
-        if (collapse)
-        {
-            if (!*cp)
-                break;
-            while (*cp && strchr(delim, *cp))
-                cp++;
-            if (!*cp)
-                break;
-            while (*cp && !strchr(delim, *cp))
-                cp++;
-            (*num)++;
-        }
-        else
-        {
-            (*num)++;
-            while (*cp && !strchr(delim, *cp))
-                cp++;
-            if (!*cp)
-                break;
-            cp++;
-        }
-    }
-    if (!*num)
-        *darray = 0;
-    else
-    {
         size_t i = 0;
-        *darray = (char **) nmem_malloc(nmem, *num * sizeof(**darray));
-        cp = dstr;
+        const char *cp = dstr;
         while (1)
         {
             const char *cp0;
@@ -108,26 +87,44 @@ void nmem_strsplitx(NMEM nmem, const char *delim, const char *dstr,
             {
                 if (!*cp)
                     break;
-                while (*cp && strchr(delim, *cp))
+                while (*cp && strchr(delim, *cp) && *cp != escape_char)
                     cp++;
                 if (!*cp)
                     break;
                 cp0 = cp;
                 while (*cp && !strchr(delim, *cp))
+                {
+                    if (*cp == escape_char)
+                        cp++;
                     cp++;
-                (*darray)[i++] = nmem_strdupn(nmem, cp0, cp - cp0);
+                }
+                if (*darray)
+                    (*darray)[i] = nmem_strdupn(nmem, cp0, cp - cp0);
+                i++;
             }
             else
             {
                 cp0 = cp;
                 while (*cp && !strchr(delim, *cp))
+                {
+                    if (*cp == escape_char)
+                        cp++;
                     cp++;
-                (*darray)[i++] = nmem_strdupn(nmem, cp0, cp - cp0);
+                }
+                if (*darray)
+                    (*darray)[i] = nmem_strdupn(nmem, cp0, cp - cp0);
+                i++;
                 if (!*cp)
                     break;
                 cp++;
             }
         }
+        *num = i;
+        if (!*num)
+            break; /* no items, so stop, *darray=0 already */
+        else if (*darray)
+            break; /* second pass, stop */
+        *darray = (char **) nmem_malloc(nmem, *num * sizeof(**darray));
     }
 }
 

diff --git a/test/test_nmem.c b/test/test_nmem.c
index 027b823..cc34f27 100644 (file)
--- a/test/test_nmem.c
+++ b/test/test_nmem.c
@@ -72,8 +72,16 @@ void tst_nmem_strsplit(void)
     YAZ_CHECK(num > 0 && !strcmp(array[0], "a"));
     YAZ_CHECK(num > 1 && !strcmp(array[1], "b"));
     YAZ_CHECK(num > 2 && !strcmp(array[2], "cd"));
+
     nmem_strsplitx(nmem, ",", ",a,b,,cd", &array, &num, 0);
+    YAZ_CHECK(num == 5);
+    YAZ_CHECK(num > 0 && !strcmp(array[0], ""));
+    YAZ_CHECK(num > 1 && !strcmp(array[1], "a"));
+    YAZ_CHECK(num > 2 && !strcmp(array[2], "b"));
+    YAZ_CHECK(num > 3 && !strcmp(array[3], ""));
+    YAZ_CHECK(num > 4 && !strcmp(array[4], "cd"));
 
+    nmem_strsplit_escape(nmem, ",", ",a,b,,cd", &array, &num, 0, '\\');
     YAZ_CHECK(num == 5);
     YAZ_CHECK(num > 0 && !strcmp(array[0], ""));
     YAZ_CHECK(num > 1 && !strcmp(array[1], "a"));
@@ -81,6 +89,24 @@ void tst_nmem_strsplit(void)
     YAZ_CHECK(num > 3 && !strcmp(array[3], ""));
     YAZ_CHECK(num > 4 && !strcmp(array[4], "cd"));
 
+    nmem_strsplit_escape(nmem, ",", ",a,b\\,,cd", &array, &num, 0, '\\');
+    YAZ_CHECK(num == 4);
+    YAZ_CHECK(num > 0 && !strcmp(array[0], ""));
+    YAZ_CHECK(num > 1 && !strcmp(array[1], "a"));
+    YAZ_CHECK(num > 2 && !strcmp(array[2], "b\\,"));
+    YAZ_CHECK(num > 3 && !strcmp(array[3], "cd"));
+
+    nmem_strsplit_escape(nmem, ",", "\\,a,b\\,,cd", &array, &num, 0, '\\');
+    YAZ_CHECK(num == 3);
+    YAZ_CHECK(num > 0 && !strcmp(array[0], "\\,a"));
+    YAZ_CHECK(num > 1 && !strcmp(array[1], "b\\,"));
+    YAZ_CHECK(num > 2 && !strcmp(array[2], "cd"));
+
+    nmem_strsplit_escape(nmem, ",", "\\,a,b\\,\\,cd", &array, &num, 0, '\\');
+    YAZ_CHECK(num == 2);
+    YAZ_CHECK(num > 0 && !strcmp(array[0], "\\,a"));
+    YAZ_CHECK(num > 1 && !strcmp(array[1], "b\\,\\,cd"));
+
     nmem_destroy(nmem);
 }