unquote_ident() + unquote_literal()

3 files changed, 149 insertions, 3 deletions

diff --git a/python/modules/cquoting.c b/python/modules/cquoting.c
index ff06972b..a929233d 100644
--- a/python/modules/cquoting.c
+++ b/python/modules/cquoting.c
@@ -302,6 +302,114 @@ static PyObject *quote_bytea_raw(PyObject *self, PyObject *args)
 	return common_quote(args, quote_bytea_raw_body);
 }
 
+/* SQL unquote */
+static const char doc_unquote_literal[] =
+"Unquote SQL quoted string.\n\n"
+"C implementation.";
+
+static PyObject *do_sql_ext(unsigned char *src, Py_ssize_t src_len)
+{
+	unsigned char *dst, *src_end = src + src_len;
+	struct Buf buf;
+
+	dst = buf_init(&buf, src_len);
+        if (!dst)
+		return NULL;
+
+        while (src < src_end) {
+		if (*src == '\'') {
+			src++;
+			if (src < src_end && *src == '\'') {
+				*dst++ = *src++;
+				continue;
+			}
+			goto failed;
+		}
+		if (*src != '\\') {
+			*dst++ = *src++;
+			continue;
+		}
+		if (++src >= src_end)
+			goto failed;
+		switch (*src) {
+		case 't': *dst++ = '\t'; src++; break;
+		case 'n': *dst++ = '\n'; src++; break;
+		case 'r': *dst++ = '\r'; src++; break;
+		case 'a': *dst++ = '\a'; src++; break;
+		case 'b': *dst++ = '\b'; src++; break;
+		default:
+			if (*src >= '0' && *src <= '7') {
+				unsigned char c = *src++ - '0';
+				if (src < src_end && *src >= '0' && *src <= '7') {
+					c = (c << 3) | ((*src++) - '0');
+					if (src < src_end && *src >= '0' && *src <= '7')
+						c = (c << 3) | ((*src++) - '0');
+				}
+				*dst++ = c;
+			} else {
+				*dst++ = *src++;
+			}
+		}
+        }
+	return buf_pystr(&buf, 0, dst);
+failed:
+	PyErr_Format(PyExc_ValueError, "Broken exteded SQL string");
+	return NULL;
+}
+
+static PyObject *do_sql_std(unsigned char *src, Py_ssize_t src_len)
+{
+	unsigned char *dst, *src_end = src + src_len;
+	struct Buf buf;
+
+	dst = buf_init(&buf, src_len);
+        if (!dst)
+		return NULL;
+
+        while (src < src_end) {
+		if (*src != '\'') {
+			*dst++ = *src++;
+			continue;
+		}
+		src++;
+		if (src >= src_end || *src != '\'')
+			goto failed;
+		*dst++ = *src++;
+        }
+	return buf_pystr(&buf, 0, dst);
+failed:
+	PyErr_Format(PyExc_ValueError, "Broken standard SQL string");
+	return NULL;
+}
+
+static PyObject *unquote_literal(PyObject *self, PyObject *args)
+{
+	unsigned char *src = NULL;
+        Py_ssize_t src_len = 0;
+	int stdstr = 0;
+	PyObject *value = NULL;
+        if (!PyArg_ParseTuple(args, "O|i", &value, &stdstr))
+                return NULL;
+	if (PyString_AsStringAndSize(value, (char **)&src, &src_len) < 0)
+		return NULL;
+	if (src_len == 4 && strcasecmp((char *)src, "null") == 0) {
+		Py_INCREF(Py_None);
+		return Py_None;
+	}
+	if (src_len < 2 || src[src_len - 1] != '\'')
+		goto badstr;
+	if (src[0] == '\'') {
+		src++; src_len -= 2;
+		return stdstr ? do_sql_std(src, src_len) : do_sql_ext(src, src_len);
+	} else if (src_len > 2 && (src[0] | 0x20) == 'e' && src[1] == '\'') {
+		src += 2; src_len -= 3;
+		return do_sql_ext(src, src_len);
+	}
+badstr:
+	Py_INCREF(value);
+	return value;
+}
+
 /* C unescape */
 static const char doc_unescape[] =
 "Unescape C-style escaped string.\n\n"
@@ -627,6 +735,7 @@ cquoting_methods[] = {
 	{ "unescape", unescape, METH_VARARGS, doc_unescape },
 	{ "db_urlencode", db_urlencode, METH_VARARGS, doc_db_urlencode },
 	{ "db_urldecode", db_urldecode, METH_VARARGS, doc_db_urldecode },
+	{ "unquote_literal", unquote_literal, METH_VARARGS, doc_unquote_literal },
 	{ NULL }
 };
 
diff --git a/python/skytools/_pyquoting.py b/python/skytools/_pyquoting.py
index 28a57577..2897c7f9 100644
--- a/python/skytools/_pyquoting.py
+++ b/python/skytools/_pyquoting.py
@@ -10,6 +10,7 @@ import urllib, re
 __all__ = [
     "quote_literal", "quote_copy", "quote_bytea_raw",
     "db_urlencode", "db_urldecode", "unescape",
+    "unquote_literal",
 ]
 
 # 
@@ -135,7 +136,7 @@ _esc_map = {
     '\\': '\\',
 }
 
-def _sub_unescape(m):
+def _sub_unescape_c(m):
     v = m.group(1)
     if (len(v) == 1) and (v < '0' or v > '7'):
         try:
@@ -149,5 +150,34 @@ def unescape(val):
     """Removes C-style escapes from string.
     Python implementation.
     """
-    return _esc_rc.sub(_sub_unescape, val)
+    return _esc_rc.sub(_sub_unescape_c, val)
+
+_esql_re = r"''|\\([0-7]{1,3}|.)"
+_esql_rc = re.compile(_esc_re)
+def _sub_unescape_sqlext(m):
+    if m.group() == "''":
+        return "'"
+    v = m.group(1)
+    if (len(v) == 1) and (v < '0' or v > '7'):
+        try:
+            return _esc_map[v]
+        except KeyError:
+            return v
+    return chr(int(v, 8))
+
+def unquote_literal(val, stdstr = False):
+    """Unquotes SQL string.
+    Ordinary '' quoted string can be both standard and extended quoted.
+    E'' quoting is always taken as extended quoted.
+    """
+    if val[0] == "'" and val[-1] == "'":
+        if stdstr:
+            return val[1:-1].replace("''", "'")
+        else:
+            return _esql_rc.sub(_sub_unescape_sqlext, val[1:-1])
+    elif len(val) > 2 and val[0] in ('E', 'e') and val[1] == "'" and val[-1] == "'":
+        return _esql_rc.sub(_sub_unescape_sqlext, val[2:-1])
+    elif val.tolower() == "null":
+        return None
+    return val
 
diff --git a/python/skytools/quoting.py b/python/skytools/quoting.py
index 19854a5f..10492b37 100644
--- a/python/skytools/quoting.py
+++ b/python/skytools/quoting.py
@@ -9,7 +9,8 @@ __all__ = [
     "db_urlencode", "db_urldecode", "unescape",
 
     "quote_bytea_literal", "quote_bytea_copy", "quote_statement",
-    "quote_ident", "quote_fqident", "quote_json", "unescape_copy"
+    "quote_ident", "quote_fqident", "quote_json", "unescape_copy",
+    "unquote_ident", "unquote_literal",
 ]
 
 try:
@@ -105,3 +106,9 @@ def unescape_copy(val):
         return None
     return unescape(val)
 
+def unquote_sql_ident(val):
+    """Unquotes possibly quoted SQL identifier."""
+    if val[0] == '"' and val[-1] == '"':
+        return val[1:-1].replace('""', '"')
+    return val
+