From f8604f5c7658d1bc051267ed36c0b12e67350da1 Mon Sep 17 00:00:00 2001
From: Dan Hirsch <thequux@upstandinghackers.com>
Date: Fri, 1 Nov 2013 18:01:44 -0400
Subject: [PATCH] Add CFFI python bindings
---
src/bindings/python/hammer.py | 244 ++++++++++++++++++++++++++++++++++
1 file changed, 244 insertions(+)
create mode 100644 src/bindings/python/hammer.py
diff --git a/src/bindings/python/hammer.py b/src/bindings/python/hammer.py
new file mode 100644
index 00000000..d3d1e3d6
--- /dev/null
+++ b/src/bindings/python/hammer.py
@@ -0,0 +1,244 @@
+from cffi import FFI
+
+ffi = FFI()
+
+## Types
+ffi.cdef("typedef struct HAllocator_ HAllocator;")
+ffi.cdef("typedef struct HArena_ HArena;")
+ffi.cdef("typedef int bool;")
+ffi.cdef("typedef struct HParseState_ HParseState;")
+ffi.cdef("""
+typedef enum HParserBackend_ {
+ PB_MIN = 0,
+ PB_PACKRAT = 0, // PB_MIN is always the default.
+ PB_REGULAR,
+ PB_LLk,
+ PB_LALR,
+ PB_GLR
+// TODO: support PB_MAX
+} HParserBackend;
+""")
+ffi.cdef("""
+typedef enum HTokenType_ {
+ // Before you change the explicit values of these, think of the poor bindings ;_;
+ TT_NONE = 1,
+ TT_BYTES = 2,
+ TT_SINT = 4,
+ TT_UINT = 8,
+ TT_SEQUENCE = 16,
+ TT_RESERVED_1, // reserved for backend-specific internal use
+ TT_ERR = 32,
+ TT_USER = 64,
+ TT_MAX
+} HTokenType;
+""")
+ffi.cdef("""
+typedef struct HCountedArray_ {
+ size_t capacity;
+ size_t used;
+ HArena * arena;
+ struct HParsedToken_ **elements;
+} HCountedArray;
+""")
+ffi.cdef("""
+typedef struct HBytes_ {
+ const uint8_t *token;
+ size_t len;
+} HBytes;
+""")
+ffi.cdef("""
+typedef struct HParsedToken_ {
+ HTokenType token_type;
+ union {
+ HBytes bytes;
+ int64_t sint;
+ uint64_t uint;
+ double dbl;
+ float flt;
+ HCountedArray *seq; // a sequence of HParsedToken's
+ void *user;
+ };
+ size_t index;
+ char bit_offset;
+} HParsedToken;
+""")
+ffi.cdef("""
+typedef struct HParseResult_ {
+ const HParsedToken *ast;
+ long long bit_length;
+ HArena * arena;
+} HParseResult;
+""")
+
+ffi.cdef("""typedef HParsedToken* (*HAction)(const HParseResult *p);""")
+ffi.cdef("""typedef bool (*HPredicate)(HParseResult *p);""")
+ffi.cdef("""
+typedef struct HCFChoice_ HCFChoice;
+typedef struct HRVMProg_ HRVMProg;
+typedef struct HParserVtable_ HParserVtable;
+""")
+
+ffi.cdef("typedef struct HParser_ HParser;")
+ffi.cdef("""
+typedef struct HParserTestcase_ {
+ unsigned char* input;
+ size_t length;
+ char* output_unambiguous;
+} HParserTestcase;
+
+typedef struct HCaseResult_ {
+ bool success;
+ union {
+ const char* actual_results; // on failure, filled in with the results of h_write_result_unamb
+ size_t parse_time; // on success, filled in with time for a single parse, in nsec
+ };
+} HCaseResult;
+
+typedef struct HBackendResults_ {
+ HParserBackend backend;
+ bool compile_success;
+ size_t n_testcases;
+ size_t failed_testcases; // actually a count...
+ HCaseResult *cases;
+} HBackendResults;
+
+typedef struct HBenchmarkResults_ {
+ size_t len;
+ HBackendResults *results;
+} HBenchmarkResults;
+""")
+
+## The following section was generated by
+## $ perl ../desugar-header.pl <../../hammer.h |sed -e 's/.*/ffi.cdef("&")/'
+ffi.cdef("HParseResult* h_parse(const HParser* parser, const uint8_t* input, size_t length);")
+ffi.cdef("HParseResult* h_parse__m(HAllocator* mm__, const HParser* parser, const uint8_t* input, size_t length);")
+ffi.cdef("HParser* h_token(const uint8_t *str, const size_t len);")
+ffi.cdef("HParser* h_token__m(HAllocator* mm__, const uint8_t *str, const size_t len);")
+ffi.cdef("HParser* h_ch(const uint8_t c);")
+ffi.cdef("HParser* h_ch__m(HAllocator* mm__, const uint8_t c);")
+ffi.cdef("HParser* h_ch_range(const uint8_t lower, const uint8_t upper);")
+ffi.cdef("HParser* h_ch_range__m(HAllocator* mm__, const uint8_t lower, const uint8_t upper);")
+ffi.cdef("HParser* h_int_range(const HParser *p, const int64_t lower, const int64_t upper);")
+ffi.cdef("HParser* h_int_range__m(HAllocator* mm__, const HParser *p, const int64_t lower, const int64_t upper);")
+ffi.cdef("HParser* h_bits(size_t len, bool sign);")
+ffi.cdef("HParser* h_bits__m(HAllocator* mm__, size_t len, bool sign);")
+ffi.cdef("HParser* h_int64(void);")
+ffi.cdef("HParser* h_int64__m(HAllocator* mm__);")
+ffi.cdef("HParser* h_int32(void);")
+ffi.cdef("HParser* h_int32__m(HAllocator* mm__);")
+ffi.cdef("HParser* h_int16(void);")
+ffi.cdef("HParser* h_int16__m(HAllocator* mm__);")
+ffi.cdef("HParser* h_int8(void);")
+ffi.cdef("HParser* h_int8__m(HAllocator* mm__);")
+ffi.cdef("HParser* h_uint64(void);")
+ffi.cdef("HParser* h_uint64__m(HAllocator* mm__);")
+ffi.cdef("HParser* h_uint32(void);")
+ffi.cdef("HParser* h_uint32__m(HAllocator* mm__);")
+ffi.cdef("HParser* h_uint16(void);")
+ffi.cdef("HParser* h_uint16__m(HAllocator* mm__);")
+ffi.cdef("HParser* h_uint8(void);")
+ffi.cdef("HParser* h_uint8__m(HAllocator* mm__);")
+ffi.cdef("HParser* h_whitespace(const HParser* p);")
+ffi.cdef("HParser* h_whitespace__m(HAllocator* mm__, const HParser* p);")
+ffi.cdef("HParser* h_left(const HParser* p, const HParser* q);")
+ffi.cdef("HParser* h_left__m(HAllocator* mm__, const HParser* p, const HParser* q);")
+ffi.cdef("HParser* h_right(const HParser* p, const HParser* q);")
+ffi.cdef("HParser* h_right__m(HAllocator* mm__, const HParser* p, const HParser* q);")
+ffi.cdef("HParser* h_middle(const HParser* p, const HParser* x, const HParser* q);")
+ffi.cdef("HParser* h_middle__m(HAllocator* mm__, const HParser* p, const HParser* x, const HParser* q);")
+ffi.cdef("HParser* h_action(const HParser* p, const HAction a);")
+ffi.cdef("HParser* h_action__m(HAllocator* mm__, const HParser* p, const HAction a);")
+ffi.cdef("HParser* h_in(const uint8_t *charset, size_t length);")
+ffi.cdef("HParser* h_in__m(HAllocator* mm__, const uint8_t *charset, size_t length);")
+ffi.cdef("HParser* h_not_in(const uint8_t *charset, size_t length);")
+ffi.cdef("HParser* h_not_in__m(HAllocator* mm__, const uint8_t *charset, size_t length);")
+ffi.cdef("HParser* h_end_p(void);")
+ffi.cdef("HParser* h_end_p__m(HAllocator* mm__);")
+ffi.cdef("HParser* h_nothing_p(void);")
+ffi.cdef("HParser* h_nothing_p__m(HAllocator* mm__);")
+ffi.cdef("HParser* h_sequence(HParser* p, ...);")
+ffi.cdef("HParser* h_sequence__m(HAllocator *mm__, HParser* p, ...);")
+ffi.cdef("HParser* h_sequence__a(void* args);")
+ffi.cdef("HParser* h_sequence__ma(HAllocator* mm__, void* args);")
+ffi.cdef("HParser* h_choice(HParser* p, ...);")
+ffi.cdef("HParser* h_choice__m(HAllocator *mm__, HParser* p, ...);")
+ffi.cdef("HParser* h_choice__a(void* args);")
+ffi.cdef("HParser* h_choice__ma(HAllocator* mm__, void* args);")
+ffi.cdef("HParser* h_butnot(const HParser* p1, const HParser* p2);")
+ffi.cdef("HParser* h_butnot__m(HAllocator* mm__, const HParser* p1, const HParser* p2);")
+ffi.cdef("HParser* h_difference(const HParser* p1, const HParser* p2);")
+ffi.cdef("HParser* h_difference__m(HAllocator* mm__, const HParser* p1, const HParser* p2);")
+ffi.cdef("HParser* h_xor(const HParser* p1, const HParser* p2);")
+ffi.cdef("HParser* h_xor__m(HAllocator* mm__, const HParser* p1, const HParser* p2);")
+ffi.cdef("HParser* h_many(const HParser* p);")
+ffi.cdef("HParser* h_many__m(HAllocator* mm__, const HParser* p);")
+ffi.cdef("HParser* h_many1(const HParser* p);")
+ffi.cdef("HParser* h_many1__m(HAllocator* mm__, const HParser* p);")
+ffi.cdef("HParser* h_repeat_n(const HParser* p, const size_t n);")
+ffi.cdef("HParser* h_repeat_n__m(HAllocator* mm__, const HParser* p, const size_t n);")
+ffi.cdef("HParser* h_optional(const HParser* p);")
+ffi.cdef("HParser* h_optional__m(HAllocator* mm__, const HParser* p);")
+ffi.cdef("HParser* h_ignore(const HParser* p);")
+ffi.cdef("HParser* h_ignore__m(HAllocator* mm__, const HParser* p);")
+ffi.cdef("HParser* h_sepBy(const HParser* p, const HParser* sep);")
+ffi.cdef("HParser* h_sepBy__m(HAllocator* mm__, const HParser* p, const HParser* sep);")
+ffi.cdef("HParser* h_sepBy1(const HParser* p, const HParser* sep);")
+ffi.cdef("HParser* h_sepBy1__m(HAllocator* mm__, const HParser* p, const HParser* sep);")
+ffi.cdef("HParser* h_epsilon_p(void);")
+ffi.cdef("HParser* h_epsilon_p__m(HAllocator* mm__);")
+ffi.cdef("HParser* h_length_value(const HParser* length, const HParser* value);")
+ffi.cdef("HParser* h_length_value__m(HAllocator* mm__, const HParser* length, const HParser* value);")
+ffi.cdef("HParser* h_attr_bool(const HParser* p, HPredicate pred);")
+ffi.cdef("HParser* h_attr_bool__m(HAllocator* mm__, const HParser* p, HPredicate pred);")
+ffi.cdef("HParser* h_and(const HParser* p);")
+ffi.cdef("HParser* h_and__m(HAllocator* mm__, const HParser* p);")
+ffi.cdef("HParser* h_not(const HParser* p);")
+ffi.cdef("HParser* h_not__m(HAllocator* mm__, const HParser* p);")
+ffi.cdef("HParser* h_indirect(void);")
+ffi.cdef("HParser* h_indirect__m(HAllocator* mm__);")
+ffi.cdef("void h_bind_indirect(HParser* indirect, const HParser* inner);")
+ffi.cdef("void h_bind_indirect__m(HAllocator* mm__, HParser* indirect, const HParser* inner);")
+ffi.cdef("void h_parse_result_free(HParseResult *result);")
+ffi.cdef("void h_parse_result_free__m(HAllocator* mm__, HParseResult *result);")
+ffi.cdef("void h_pprint(FILE* stream, const HParsedToken* tok, int indent, int delta);")
+ffi.cdef("int h_compile(HParser* parser, HParserBackend backend, const void* params);")
+ffi.cdef("int h_compile__m(HAllocator* mm__, HParser* parser, HParserBackend backend, const void* params);")
+ffi.cdef("HBenchmarkResults * h_benchmark(HParser* parser, HParserTestcase* testcases);")
+ffi.cdef("HBenchmarkResults * h_benchmark__m(HAllocator* mm__, HParser* parser, HParserTestcase* testcases);")
+
+lib = ffi.verify("#include <hammer/hammer.h>",
+ libraries=['hammer'])
+
+
+# Quick test
+def fromCobj(cobj):
+ # TODO: Free the toplevel parser
+ tt = cobj.token_type
+ if cobj.token_type == lib.TT_BYTES:
+ return ffi.buffer(cobj.bytes.token, cobj.bytes.len)[:]
+ elif cobj.token_type == lib.TT_ERR:
+ # I have no idea what this is for
+ pass
+ elif cobj.token_type == lib.TT_NONE:
+ return None
+ elif cobj.token_type == lib.TT_SEQUENCE:
+ return [fromCobj(cobj.seq.elements[i])
+ for i in range(cobj.seq.used)]
+ elif cobj.token_type == lib.TT_SINT:
+ return cobj.sint
+ elif cobj.token_type == lib.TT_UINT:
+ return cobj.uint
+
+def fromParseResult(cobj):
+ ret = fromCobj(cobj.ast)
+ lib.h_parse_result_free(cobj)
+ return ret
+
+def run_test():
+ p_test = lib.h_sepBy1(lib.h_choice(lib.h_ch(ord('1')),
+ lib.h_ch(ord('2')),
+ lib.h_ch(ord('3')),
+ ffi.NULL),
+ lib.h_ch(ord(',')))
+ return fromParseResult(lib.h_parse(p_test, "1,2,3", 5))
+
--
GitLab