Merge branch 'master' of https://github.com/ahupp/python-magic into abi3-wheels

* 'master' of https://github.com/ahupp/python-magic:
  python-magic tests: add test files for elf and json, use to test flags
  Magic.__init__: add kwargs to enable/disable different types of magic detection
  typing stubs: add Magic.__init__ extension & follow_symlinks args

Author: ddelange

magic/__init__.py

@@ -40,7 +40,10 @@ class Magic:
 
     def __init__(self, mime=False, magic_file=None, mime_encoding=False,
                  keep_going=False, uncompress=False, raw=False, extension=False,
-                 follow_symlinks=False):
+                 follow_symlinks=False, check_tar=True, check_soft=True,
+                 check_apptype=True, check_elf=True, check_text=True,
+                 check_cdf=True, check_csv=True, check_encoding=True,
+                 check_json=True, check_simh=True):
         """
         Create a new libmagic wrapper.
 
@@ -69,6 +72,27 @@ def __init__(self, mime=False, magic_file=None, mime_encoding=False,
         if follow_symlinks:
             self.flags |= MAGIC_SYMLINK
 
+        if not check_tar:
+            self.flags |= MAGIC_NO_CHECK_TAR
+        if not check_soft:
+            self.flags |= MAGIC_NO_CHECK_SOFT
+        if not check_apptype:
+            self.flags |= MAGIC_NO_CHECK_APPTYPE
+        if not check_elf:
+            self.flags |= MAGIC_NO_CHECK_ELF
+        if not check_text:
+            self.flags |= MAGIC_NO_CHECK_TEXT
+        if not check_cdf:
+            self.flags |= MAGIC_NO_CHECK_CDF
+        if not check_csv:
+            self.flags |= MAGIC_NO_CHECK_CSV
+        if not check_encoding:
+            self.flags |= MAGIC_NO_CHECK_ENCODING
+        if not check_json:
+            self.flags |= MAGIC_NO_CHECK_JSON
+        if not check_simh:
+            self.flags |= MAGIC_NO_CHECK_SIMH
+
         self.cookie = magic_open(self.flags)
         self.lock = threading.Lock()
 
@@ -416,10 +440,16 @@ def version():
 MAGIC_NO_CHECK_SOFT = 0x004000  # Don't check magic entries
 MAGIC_NO_CHECK_APPTYPE = 0x008000  # Don't check application type
 MAGIC_NO_CHECK_ELF = 0x010000  # Don't check for elf details
-MAGIC_NO_CHECK_ASCII = 0x020000  # Don't check for ascii files
-MAGIC_NO_CHECK_TROFF = 0x040000  # Don't check ascii/troff
-MAGIC_NO_CHECK_FORTRAN = 0x080000  # Don't check ascii/fortran
-MAGIC_NO_CHECK_TOKENS = 0x100000  # Don't check ascii/tokens
+MAGIC_NO_CHECK_TEXT = 0x020000  # Don't check for ascii files
+MAGIC_NO_CHECK_ASCII = 0x020000  # Deprecated alias for MAGIC_NO_CHECK_TEXT
+MAGIC_NO_CHECK_TROFF = 0x040000  # Don't check ascii/troff (deprecated)
+MAGIC_NO_CHECK_FORTRAN = 0x080000  # Don't check ascii/fortran (deprecated)
+MAGIC_NO_CHECK_TOKENS = 0x100000  # Don't check ascii/tokens (deprecated)
+MAGIC_NO_CHECK_CDF = 0x0040000  # Don't check for CDF files
+MAGIC_NO_CHECK_CSV = 0x0080000  # Don't check for CSV files
+MAGIC_NO_CHECK_ENCODING = 0x0200000  # Don't check text encodings
+MAGIC_NO_CHECK_JSON = 0x0400000 # Don't check for JSON files
+MAGIC_NO_CHECK_SIMH = 0x0800000 # Don't check for SIMH tape files
 
 MAGIC_PARAM_INDIR_MAX = 0  # Recursion limit for indirect magic
 MAGIC_PARAM_NAME_MAX = 1  # Use count limit for name/use magic

magic/__init__.pyi

@@ -11,7 +11,7 @@ class Magic:
     flags: int = ...
     cookie: Any = ...
     lock: threading.Lock = ...
-    def __init__(self, mime: bool = ..., magic_file: Optional[Any] = ..., mime_encoding: bool = ..., keep_going: bool = ..., uncompress: bool = ..., raw: bool = ...) -> None: ...
+    def __init__(self, mime: bool = ..., magic_file: Optional[Any] = ..., mime_encoding: bool = ..., keep_going: bool = ..., uncompress: bool = ..., raw: bool = ..., extension: bool = ..., follow_symlinks: bool = ..., check_tar: bool = ..., check_soft: bool = ..., check_apptype: bool = ..., check_elf: bool = ..., check_text: bool = ..., check_encoding: bool = ..., check_json: bool = ..., check_simh: bool = ...) -> None: ...
     def from_buffer(self, buf: Union[bytes, str]) -> Text: ...
     def from_file(self, filename: Union[bytes, str, PathLike]) -> Text: ...
     def from_descriptor(self, fd: int, mime: bool = ...) -> Text: ...
@@ -74,10 +74,16 @@ MAGIC_NO_CHECK_TAR: int
 MAGIC_NO_CHECK_SOFT: int
 MAGIC_NO_CHECK_APPTYPE: int
 MAGIC_NO_CHECK_ELF: int
+MAGIC_NO_CHECK_TEXT: int
 MAGIC_NO_CHECK_ASCII: int
 MAGIC_NO_CHECK_TROFF: int
 MAGIC_NO_CHECK_FORTRAN: int
+MAGIC_NO_CHECK_CDF: int
+MAGIC_NO_CHECK_CSV: int
 MAGIC_NO_CHECK_TOKENS: int
+MAGIC_NO_CHECK_ENCODING: int
+MAGIC_NO_CHECK_JSON: int
+MAGIC_NO_CHECK_SIMH: int
 MAGIC_PARAM_INDIR_MAX: int
 MAGIC_PARAM_NAME_MAX: int
 MAGIC_PARAM_ELF_PHNUM_MAX: int

test/python_magic_test.py

@@ -96,6 +96,10 @@ def test_mime_types(self):
             self.assert_values(
                 m,
                 {
+                    "elf-NetBSD-x86_64-echo": (
+                        "application/x-pie-executable",
+                        "application/x-sharedlib",
+                    ),
                     "magic._pyc_": (
                         "application/octet-stream",
                         "text/x-bytecode.python",
@@ -107,7 +111,9 @@ def test_mime_types(self):
                     "text.txt": "text/plain",
                     b"\xce\xbb".decode("utf-8"): "text/plain",
                     b"\xce\xbb": "text/plain",
+                    "test.json": "application/json",
                 },
+                buf_equals_file=False,
             )
         finally:
             os.unlink(dest)
@@ -119,6 +125,88 @@ def test_descriptions(self):
             self.assert_values(
                 m,
                 {
+                    "elf-NetBSD-x86_64-echo": (
+                        "ELF 64-bit LSB shared object, x86-64, version 1 (SYSV)",
+                        "ELF 64-bit LSB pie executable, x86-64, version 1 (SYSV), dynamically linked, interpreter /libexec/ld.elf_so, for NetBSD 8.0, not stripped",
+                    ),
+                    "magic._pyc_": "python 2.4 byte-compiled",
+                    "test.pdf": (
+                        "PDF document, version 1.2",
+                        "PDF document, version 1.2, 2 pages",
+                        "PDF document, version 1.2, 2 page(s)",
+                    ),
+                    "test.gz": (
+                        'gzip compressed data, was "test", from Unix, last '
+                        "modified: Sun Jun 29 01:32:52 2008",
+                        'gzip compressed data, was "test", last modified'
+                        ": Sun Jun 29 01:32:52 2008, from Unix",
+                        'gzip compressed data, was "test", last modified'
+                        ": Sun Jun 29 01:32:52 2008, from Unix, original size 15",
+                        'gzip compressed data, was "test", '
+                        "last modified: Sun Jun 29 01:32:52 2008, "
+                        "from Unix, original size modulo 2^32 15",
+                        'gzip compressed data, was "test", last modified'
+                        ": Sun Jun 29 01:32:52 2008, from Unix, truncated",
+                    ),
+                    "text.txt": "ASCII text",
+                    "test.snappy.parquet": ("Apache Parquet", "Par archive data"),
+                    "test.json": "JSON text data",
+                },
+                buf_equals_file=False,
+            )
+        finally:
+            del os.environ["TZ"]
+
+    def test_descriptions_no_soft(self):
+        m = magic.Magic(check_soft=False)
+        self.assert_values(
+            m,
+            {
+                "elf-NetBSD-x86_64-echo": (
+                    "data",
+                    "ELF 64-bit LSB pie executable, x86-64, version 1 (SYSV), dynamically linked, interpreter /libexec/ld.elf_so, for NetBSD 8.0, not stripped",
+                ),
+                "magic._pyc_": "data",
+                "test.pdf": "ASCII text",
+                "test.gz": "data",
+                "text.txt": "ASCII text",
+                "test.snappy.parquet": "data",
+                "test.json": "JSON text data",
+            },
+            buf_equals_file=False,
+        )
+
+    def test_descriptions_no_elf(self):
+        m = magic.Magic(check_elf=False)
+        self.assert_values(
+            m,
+            {
+                "elf-NetBSD-x86_64-echo": "ELF 64-bit LSB shared object, x86-64, version 1 (SYSV)",
+            },
+            buf_equals_file=True,
+        )
+
+    def test_descriptions_no_json(self):
+        m = magic.Magic(check_elf=False)
+        self.assert_values(
+            m,
+            {
+                "test.json": "data",
+            },
+            buf_equals_file=True,
+        )
+
+    def test_descriptions_no_json(self):
+        m = magic.Magic(check_json=False)
+        os.environ["TZ"] = "UTC"  # To get last modified date of test.gz in UTC
+        try:
+            self.assert_values(
+                m,
+                {
+                    "elf-NetBSD-x86_64-echo": (
+                        "ELF 64-bit LSB shared object, x86-64, version 1 (SYSV)",
+                        "ELF 64-bit LSB pie executable, x86-64, version 1 (SYSV), dynamically linked, interpreter /libexec/ld.elf_so, for NetBSD 8.0, not stripped",
+                    ),
                     "magic._pyc_": "python 2.4 byte-compiled",
                     "test.pdf": (
                         "PDF document, version 1.2",

test/testdata/elf-NetBSD-x86_64-echo

@@ -0,0 +1,7 @@
+[
+  {
+    "one": 2,
+    "three": null,
+    "four": [5, "six", false]
+  }
+]