string: Fix split_ident() to handle idents with '_' and/or digits

elprans · elprans · commit 320e5b1faee4 · 2011-03-16T09:16:06.000-04:00
string.split_ident() used ad-hoc logic to determine if an identifier
must be quoted, effectively prohibiting identifiers with underscores or
digits, which are both considered to be safe by PostgreSQL.
diff --git a/postgresql/string.py b/postgresql/string.py
@@ -197,9 +197,9 @@ def split_ident(text, sep = ',', quote = '"', maxsplit = -1):
 				)
 			else:
 				nr.append(x[1:-1].replace('""', '"'))
-		elif not x.isalpha():
+		elif needs_quoting(x):
 			raise ValueError(
-				"non-alpha characters in unquoted identifier", x
+				"non-ident characters in unquoted identifier", x
 			)
 		else:
 			# postgres implies a lower, so to stay consistent
diff --git a/postgresql/test/test_string.py b/postgresql/test/test_string.py
@@ -23,6 +23,9 @@
 	('"schema.base"', ['schema.base'], '"schema.base"'),
 	('schEm&Aring;."base"', ['schem&aring;', 'base'], 'schem&aring;.base'),
 	('scheMa."base"', ['schema', 'base'], 'schema.base'),
+	('sche_ma.base', ['sche_ma', 'base'], 'sche_ma.base'),
+	('_schema.base', ['_schema', 'base'], '_schema.base'),
+	('a000.b111', ['a000', 'b111'], 'a000.b111'),
 	('" schema"."base"', [' schema', 'base'], '" schema".base'),
 	('" schema"."ba se"', [' schema', 'ba se'], '" schema"."ba se"'),
 	('" ""schema"."ba""se"', [' "schema', 'ba"se'], '" ""schema"."ba""se"'),

Original file line number	Diff line number	Diff line change
`@@ -197,9 +197,9 @@ def split_ident(text, sep = ',', quote = '"', maxsplit = -1):`
`197`	`197`	`)`
`198`	`198`	`else:`
`199`	`199`	`nr.append(x[1:-1].replace('""', '"'))`
`200`		`- elif not x.isalpha():`
	`200`	`+ elif needs_quoting(x):`
`201`	`201`	`raise ValueError(`
`202`		`- "non-alpha characters in unquoted identifier", x`
	`202`	`+ "non-ident characters in unquoted identifier", x`
`203`	`203`	`)`
`204`	`204`	`else:`
`205`	`205`	`# postgres implies a lower, so to stay consistent`