Swap http-parse to llhttp (#56)

Replaces the underlying HTTP parser with llhttp as http-parser is
no longer actively maintained. However we still have http-parser
to power our URL parsing function at the moment.

Co-authored-by: Fantix King <fantix.king@gmail.com>

.github/workflows/tests.yml

@@ -41,13 +41,14 @@ jobs:
           __version__\s*=\s*(?:['"])([[:PEP440:]])(?:['"])
 
     - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v1
+      uses: actions/setup-python@v2
       if: steps.release.outputs.version == 0
       with:
         python-version: ${{ matrix.python-version }}
 
     - name: Test
       if: steps.release.outputs.version == 0
       run: |
+        pip install wheel
         pip install -e .[test]
         python setup.py test

.gitignore

@@ -29,3 +29,5 @@ __pycache__/
 /.pytest_cache
 /.mypy_cache
 /.vscode
+.eggs
+.venv

.gitmodules

@@ -1,3 +1,6 @@
 [submodule "vendor/http-parser"]
 	path = vendor/http-parser
 	url = https://github.com/nodejs/http-parser.git
+[submodule "vendor/llhttp"]
+	path = vendor/llhttp
+	url = https://github.com/nodejs/llhttp.git

Makefile

@@ -13,17 +13,16 @@ release: compile test
 	python3 setup.py sdist upload
 
 
-test:
-	python3 setup.py test
+test: compile
+	python3 -m unittest -v
 
 clean:
 	find $(ROOT)/httptools/parser -name '*.c' | xargs rm -f
 	find $(ROOT)/httptools/parser -name '*.html' | xargs rm -f
 
-distclean:
+distclean: clean
 	git --git-dir="$(ROOT)/vendor/http-parser/.git" clean -dfx
-	find $(ROOT)/httptools/parser -name '*.c' | xargs rm -f
-	find $(ROOT)/httptools/parser -name '*.html' | xargs rm -f
+	git --git-dir="$(ROOT)/vendor/llhttp/.git" clean -dfx
 
 
 testinstalled:

README.md

@@ -8,8 +8,11 @@ The package is available on PyPI: `pip install httptools`.
 # APIs
 
 httptools contains two classes `httptools.HttpRequestParser`,
-`httptools.HttpResponseParser` and a function for parsing URLs
-`httptools.parse_url`.  See unittests for examples.
+`httptools.HttpResponseParser` (fulfilled through
+[llhttp](https://github.com/nodejs/llhttp)) and a function for
+parsing URLs `httptools.parse_url` (through
+[http-parse](https://github.com/nodejs/http-parser) for now).
+See unittests for examples.
 
 
 ```python

httptools/parser/__init__.py

@@ -1,4 +1,5 @@
 from .parser import *  # NoQA
 from .errors import *  # NoQA
+from .url_parser import *  # NoQA
 
-__all__ = parser.__all__ + errors.__all__  # NoQA
+__all__ = parser.__all__ + errors.__all__ + url_parser.__all__  # NoQA

httptools/parser/cparser.pxd

@@ -1,139 +1,156 @@
-from libc.stdint cimport uint16_t, uint32_t, uint64_t
+from libc.stdint cimport int32_t, uint8_t, uint16_t, uint64_t
 
 
-cdef extern from "../../vendor/http-parser/http_parser.h":
-    ctypedef int (*http_data_cb) (http_parser*,
+cdef extern from "llhttp.h":
+    struct llhttp__internal_s:
+        int32_t _index
+        void *_span_pos0
+        void *_span_cb0
+        int32_t error
+        const char *reason
+        const char *error_pos
+        void *data
+        void *_current
+        uint64_t content_length
+        uint8_t type
+        uint8_t method
+        uint8_t http_major
+        uint8_t http_minor
+        uint8_t header_state
+        uint16_t flags
+        uint8_t upgrade
+        uint16_t status_code
+        uint8_t finish
+        void *settings
+    ctypedef llhttp__internal_s llhttp__internal_t
+    ctypedef llhttp__internal_t llhttp_t
+
+    ctypedef int (*llhttp_data_cb) (llhttp_t*,
                                   const char *at,
                                   size_t length) except -1
 
-    ctypedef int (*http_cb) (http_parser*) except -1
-
-    struct http_parser:
-        unsigned int type
-        unsigned int flags
-        unsigned int state
-        unsigned int header_state
-        unsigned int index
-
-        uint32_t nread
-        uint64_t content_length
-
-        unsigned short http_major
-        unsigned short http_minor
-        unsigned int status_code
-        unsigned int method
-        unsigned int http_errno
-
-        unsigned int upgrade
-
-        void *data
-
-    struct http_parser_settings:
-        http_cb      on_message_begin
-        http_data_cb on_url
-        http_data_cb on_status
-        http_data_cb on_header_field
-        http_data_cb on_header_value
-        http_cb      on_headers_complete
-        http_data_cb on_body
-        http_cb      on_message_complete
-        http_cb      on_chunk_header
-        http_cb      on_chunk_complete
-
-    enum http_parser_type:
+    ctypedef int (*llhttp_cb) (llhttp_t*) except -1
+
+    struct llhttp_settings_s:
+        llhttp_cb      on_message_begin
+        llhttp_data_cb on_url
+        llhttp_data_cb on_status
+        llhttp_data_cb on_header_field
+        llhttp_data_cb on_header_value
+        llhttp_cb      on_headers_complete
+        llhttp_data_cb on_body
+        llhttp_cb      on_message_complete
+        llhttp_cb      on_chunk_header
+        llhttp_cb      on_chunk_complete
+    ctypedef llhttp_settings_s llhttp_settings_t
+
+    enum llhttp_type:
+        HTTP_BOTH,
         HTTP_REQUEST,
-        HTTP_RESPONSE,
-        HTTP_BOTH
+        HTTP_RESPONSE
+    ctypedef llhttp_type llhttp_type_t
 
-    enum http_errno:
+    enum llhttp_errno:
         HPE_OK,
-        HPE_CB_message_begin,
-        HPE_CB_url,
-        HPE_CB_header_field,
-        HPE_CB_header_value,
-        HPE_CB_headers_complete,
-        HPE_CB_body,
-        HPE_CB_message_complete,
-        HPE_CB_status,
-        HPE_CB_chunk_header,
-        HPE_CB_chunk_complete,
-        HPE_INVALID_EOF_STATE,
-        HPE_HEADER_OVERFLOW,
+        HPE_INTERNAL,
+        HPE_STRICT,
+        HPE_LF_EXPECTED,
+        HPE_UNEXPECTED_CONTENT_LENGTH,
         HPE_CLOSED_CONNECTION,
-        HPE_INVALID_VERSION,
-        HPE_INVALID_STATUS,
         HPE_INVALID_METHOD,
         HPE_INVALID_URL,
-        HPE_INVALID_HOST,
-        HPE_INVALID_PORT,
-        HPE_INVALID_PATH,
-        HPE_INVALID_QUERY_STRING,
-        HPE_INVALID_FRAGMENT,
-        HPE_LF_EXPECTED,
+        HPE_INVALID_CONSTANT,
+        HPE_INVALID_VERSION,
         HPE_INVALID_HEADER_TOKEN,
         HPE_INVALID_CONTENT_LENGTH,
         HPE_INVALID_CHUNK_SIZE,
-        HPE_INVALID_CONSTANT,
-        HPE_INVALID_INTERNAL_STATE,
-        HPE_STRICT,
+        HPE_INVALID_STATUS,
+        HPE_INVALID_EOF_STATE,
+        HPE_INVALID_TRANSFER_ENCODING,
+        HPE_CB_MESSAGE_BEGIN,
+        HPE_CB_HEADERS_COMPLETE,
+        HPE_CB_MESSAGE_COMPLETE,
+        HPE_CB_CHUNK_HEADER,
+        HPE_CB_CHUNK_COMPLETE,
         HPE_PAUSED,
-        HPE_UNKNOWN
+        HPE_PAUSED_UPGRADE,
+        HPE_USER
+    ctypedef llhttp_errno llhttp_errno_t
 
-    enum flags:
-        F_CHUNKED,
+    enum llhttp_flags:
         F_CONNECTION_KEEP_ALIVE,
         F_CONNECTION_CLOSE,
         F_CONNECTION_UPGRADE,
-        F_TRAILING,
+        F_CHUNKED,
         F_UPGRADE,
-        F_SKIPBODY
-
-    enum http_method:
-        DELETE, GET, HEAD, POST, PUT, CONNECT, OPTIONS, TRACE, COPY,
-        LOCK, MKCOL, MOVE, PROPFIND, PROPPATCH, SEARCH, UNLOCK, BIND,
-        REBIND, UNBIND, ACL, REPORT, MKACTIVITY, CHECKOUT, MERGE,
-        MSEARCH, NOTIFY, SUBSCRIBE, UNSUBSCRIBE, PATCH, PURGE, MKCALENDAR,
-        LINK, UNLINK
-
-    void http_parser_init(http_parser *parser, http_parser_type type)
-
-    size_t http_parser_execute(http_parser *parser,
-                               const http_parser_settings *settings,
-                               const char *data,
-                               size_t len)
-
-    int http_should_keep_alive(const http_parser *parser)
-
-    void http_parser_settings_init(http_parser_settings *settings)
-
-    const char *http_errno_name(http_errno err)
-    const char *http_errno_description(http_errno err)
-    const char *http_method_str(http_method m)
-
-    # URL Parser
-
-    enum http_parser_url_fields:
-        UF_SCHEMA   = 0,
-        UF_HOST     = 1,
-        UF_PORT     = 2,
-        UF_PATH     = 3,
-        UF_QUERY    = 4,
-        UF_FRAGMENT = 5,
-        UF_USERINFO = 6,
-        UF_MAX      = 7
-
-    struct http_parser_url_field_data:
-        uint16_t off
-        uint16_t len
-
-    struct http_parser_url:
-        uint16_t field_set
-        uint16_t port
-        http_parser_url_field_data[<int>UF_MAX] field_data
-
-    void http_parser_url_init(http_parser_url *u)
-
-    int http_parser_parse_url(const char *buf,
-                              size_t buflen,
-                              int is_connect,
-                              http_parser_url *u)
+        F_CONTENT_LENGTH,
+        F_SKIPBODY,
+        F_TRAILING,
+        F_LENIENT,
+        F_TRANSFER_ENCODING
+    ctypedef llhttp_flags llhttp_flags_t
+
+    enum llhttp_method:
+        HTTP_DELETE,
+        HTTP_GET,
+        HTTP_HEAD,
+        HTTP_POST,
+        HTTP_PUT,
+        HTTP_CONNECT,
+        HTTP_OPTIONS,
+        HTTP_TRACE,
+        HTTP_COPY,
+        HTTP_LOCK,
+        HTTP_MKCOL,
+        HTTP_MOVE,
+        HTTP_PROPFIND,
+        HTTP_PROPPATCH,
+        HTTP_SEARCH,
+        HTTP_UNLOCK,
+        HTTP_BIND,
+        HTTP_REBIND,
+        HTTP_UNBIND,
+        HTTP_ACL,
+        HTTP_REPORT,
+        HTTP_MKACTIVITY,
+        HTTP_CHECKOUT,
+        HTTP_MERGE,
+        HTTP_MSEARCH,
+        HTTP_NOTIFY,
+        HTTP_SUBSCRIBE,
+        HTTP_UNSUBSCRIBE,
+        HTTP_PATCH,
+        HTTP_PURGE,
+        HTTP_MKCALENDAR,
+        HTTP_LINK,
+        HTTP_UNLINK,
+        HTTP_SOURCE,
+        HTTP_PRI,
+        HTTP_DESCRIBE,
+        HTTP_ANNOUNCE,
+        HTTP_SETUP,
+        HTTP_PLAY,
+        HTTP_PAUSE,
+        HTTP_TEARDOWN,
+        HTTP_GET_PARAMETER,
+        HTTP_SET_PARAMETER,
+        HTTP_REDIRECT,
+        HTTP_RECORD,
+        HTTP_FLUSH
+    ctypedef llhttp_method llhttp_method_t
+
+    void llhttp_init(llhttp_t* parser, llhttp_type_t type, const llhttp_settings_t* settings)
+
+    void llhttp_settings_init(llhttp_settings_t* settings)
+
+    llhttp_errno_t llhttp_execute(llhttp_t* parser, const char* data, size_t len)
+
+    void llhttp_resume_after_upgrade(llhttp_t* parser)
+
+    int llhttp_should_keep_alive(const llhttp_t* parser)
+
+    const char* llhttp_get_error_pos(const llhttp_t* parser)
+    const char* llhttp_get_error_reason(const llhttp_t* parser)
+    const char* llhttp_method_name(llhttp_method_t method)
+
+    void llhttp_set_error_reason(llhttp_t* parser, const char* reason);