summaryrefslogtreecommitdiff
path: root/tests/tokens.py
diff options
context:
space:
mode:
authorOwen Jacobson <owen@grimoire.ca>2017-11-14 02:51:27 -0500
committerOwen Jacobson <owen@grimoire.ca>2017-11-14 02:51:27 -0500
commit6ab0fb837e5b1cc40002037e2ed15505f99cbbe3 (patch)
treed7fbc91d82588af74b5f4a931c7ba3344f63ab41 /tests/tokens.py
parent10dad94b84ae0a41b1ff03d6d705732834958efb (diff)
Macro expander.
This includes a fairly complete quasiquote system, and a complete rework of the expander.
Diffstat (limited to 'tests/tokens.py')
-rw-r--r--tests/tokens.py38
1 files changed, 22 insertions, 16 deletions
diff --git a/tests/tokens.py b/tests/tokens.py
index 3eb58b8..0e98494 100644
--- a/tests/tokens.py
+++ b/tests/tokens.py
@@ -1,23 +1,29 @@
from hypothesis.strategies import just, one_of, characters, text, lists, tuples
from hypothesis.strategies import composite, recursive
+from actinide import tokenizer as t
+
# Generators for token families
-# Generates the `(` token.
-def open_parens():
- return just('(')
+# Generates the `(` and ')' tokens.
+def parens():
+ return one_of(just(p) for p in t.parens)
-# Generates the ')' token.
-def close_parens():
- return just(')')
+def quotes():
+ return one_of(
+ just("'"),
+ just('`'),
+ just(','),
+ just(',@'),
+ )
# Generates characters that are legal, unescaped, inside of a string.
def string_bare_characters():
- return characters(blacklist_characters='\\"')
+ return characters(blacklist_characters=t.string_escaped)
# Generates legal string escape sequences.
def string_escaped_characters():
- return one_of(just('"'), just('\\')).map(lambda c: '\\' + c)
+ return one_of(just(c) for c in t.string_escaped).map(lambda c: '\\' + c)
# Generates single-character string representations, including escapes.
def string_characters():
@@ -34,7 +40,7 @@ def strings():
# Generates characters which are legal within a symbol.
def symbol_characters():
- return characters(blacklist_characters=' \t\n();"')
+ return characters(blacklist_characters=t.whitespace + t.parens + t.quotes + t.string_delim + t.comment_delim)
# Generates legal symbols.
def symbols():
@@ -42,11 +48,11 @@ def symbols():
# Generates single whitespace characters.
def whitespace_characters():
- return one_of(just('\n'), just(' '), just('\t'))
+ return one_of(just(c) for c in t.whitespace)
# Generates a single token.
def tokens():
- return one_of(symbols(), strings(), open_parens(), close_parens())
+ return one_of(symbols(), strings(), parens(), quotes())
# Generates a string which may not be empty, but which does not contain a token.
def nontokens():
@@ -78,16 +84,16 @@ def spaced_tokens():
return tuples(intertokens(), strategy)
def unspaced(strategy):
return tuples(one_of(just(''), intertokens()), strategy)
+ def spaced_quotes():
+ return spaced(quotes())
def spaced_symbols():
return spaced(symbols())
def spaced_strings():
return unspaced(strings())
- def spaced_open_parens():
- return unspaced(open_parens())
- def spaced_close_parens():
- return unspaced(close_parens())
+ def spaced_parens():
+ return unspaced(parens())
- return one_of(spaced_symbols(), spaced_strings(), spaced_open_parens(), spaced_close_parens())
+ return one_of(spaced_symbols(), spaced_quotes(), spaced_strings(), spaced_parens())
# Generats a list of pairs as per spaced_token().
def spaced_token_sequences():