MIME type and character set extraction

The zope.mimetype.typegetter module provides a selection of MIME type extractors (implementations of zope.mimetype.interfaces.IMimeTypeGetter) and charset extractors (implementations of zope.mimetype.interfaces.ICharsetGetter). These may be used to determine what the MIME type and character set for uploaded data should be.

These two interfaces represent the site policy regarding interpreting upload data in the face of missing or inaccurate input.

Let’s go ahead and import the module:

>>> from zope.mimetype import typegetter

MIME types

There are a number of interesting MIME-type extractors:

mimeTypeGetter()
A minimal extractor that never attempts to guess.
mimeTypeGuesser()
An extractor that tries to guess the content type based on the name and data if the input contains no content type information.
smartMimeTypeGuesser()
An extractor that checks the content for a variety of constructs to try and refine the results of the mimeTypeGuesser(). This is able to do things like check for XHTML that’s labelled as HTML in upload data.

mimeTypeGetter()

We’ll start with the simplest, which does no content-based guessing at all, but uses the information provided by the browser directly. If the browser did not provide any content-type information, or if it cannot be parsed, the extractor simply asserts a “safe” MIME type of application/octet-stream. (The rationale for selecting this type is that since there’s really nothing productive that can be done with it other than download it, it’s impossible to mis-interpret the data.)

When there’s no information at all about the content, the extractor returns None:

>>> print(typegetter.mimeTypeGetter())
None

Providing only the upload filename or data, or both, still produces None, since no guessing is being done:

>>> print(typegetter.mimeTypeGetter(name="file.html"))
None

>>> print(typegetter.mimeTypeGetter(data=b"<html>...</html>"))
None

>>> print(typegetter.mimeTypeGetter(
...     name="file.html", data=b"<html>...</html>"))
None

If a content type header is available for the input, that is used since that represents explicit input from outside the application server. The major and minor parts of the content type are extracted and returned as a single string:

>>> typegetter.mimeTypeGetter(content_type="text/plain")
'text/plain'

>>> typegetter.mimeTypeGetter(content_type="text/plain; charset=utf-8")
'text/plain'

If the content-type information is provided but malformed (not in conformance with RFC 2822), it is ignored, since the intent cannot be reliably guessed:

>>> print(typegetter.mimeTypeGetter(content_type="foo bar"))
None

This combines with ignoring the other values that may be provided as expected:

>>> print(typegetter.mimeTypeGetter(
...     name="file.html", data=b"<html>...</html>", content_type="foo bar"))
None

mimeTypeGuesser()

A more elaborate extractor that tries to work around completely missing information can be found as the mimeTypeGuesser() function. This function will only guess if there is no usable content type information in the input. This extractor can be thought of as having the following pseudo-code:

def mimeTypeGuesser(name=None, data=None, content_type=None):
    type = mimeTypeGetter(name=name, data=data, content_type=content_type)
    if type is None:
        type = guess the content type
    return type

Let’s see how this affects the results we saw earlier. When there’s no input to use, we still get None:

>>> print(typegetter.mimeTypeGuesser())
None

Providing only the upload filename or data, or both, now produces a non-None guess for common content types:

>>> typegetter.mimeTypeGuesser(name="file.html")
'text/html'

>>> typegetter.mimeTypeGuesser(data=b"<html>...</html>")
'text/html'

>>> typegetter.mimeTypeGuesser(name="file.html", data=b"<html>...</html>")
'text/html'

Note that if the filename and data provided separately produce different MIME types, the result of providing both will be one of those types, but which is unspecified:

>>> mt_1 = typegetter.mimeTypeGuesser(name="file.html")
>>> mt_1
'text/html'

>>> mt_2 = typegetter.mimeTypeGuesser(data=b"<?xml version='1.0'?>...")
>>> mt_2
'text/xml'

>>> mt = typegetter.mimeTypeGuesser(
...     data=b"<?xml version='1.0'?>...", name="file.html")
>>> mt in (mt_1, mt_2)
True

If a content type header is available for the input, that is used in the same way as for the mimeTypeGetter() function:

>>> typegetter.mimeTypeGuesser(content_type="text/plain")
'text/plain'

>>> typegetter.mimeTypeGuesser(content_type="text/plain; charset=utf-8")
'text/plain'

If the content-type information is provided but malformed, it is ignored:

>>> print(typegetter.mimeTypeGetter(content_type="foo bar"))
None

When combined with values for the filename or content data, those are still used to provide reasonable guesses for the content type:

>>> typegetter.mimeTypeGuesser(name="file.html", content_type="foo bar")
'text/html'

>>> typegetter.mimeTypeGuesser(
...     data=b"<html>...</html>", content_type="foo bar")
'text/html'

Information from a parsable content-type is still used even if a guess from the data or filename would provide a different or more-refined result:

>>> typegetter.mimeTypeGuesser(
...     data=b"GIF89a...", content_type="application/octet-stream")
'application/octet-stream'

smartMimeTypeGuesser()

The smartMimeTypeGuesser() function applies more knowledge to the process of determining the MIME-type to use. Essentially, it takes the result of the mimeTypeGuesser() function and attempts to refine the content-type based on various heuristics.

We still see the basic behavior that no input produces None:

>>> print(typegetter.smartMimeTypeGuesser())
None

An unparsable content-type is still ignored:

>>> print(typegetter.smartMimeTypeGuesser(content_type="foo bar"))
None

The interpretation of uploaded data will be different in at least some interesting cases. For instance, the mimeTypeGuesser() function provides these results for some XHTML input data:

>>> typegetter.mimeTypeGuesser(
...     data=b"<?xml version='1.0' encoding='utf-8'?><html>...</html>",
...     name="file.html")
'text/html'

The smart extractor is able to refine this into more usable data:

>>> typegetter.smartMimeTypeGuesser(
...     data=b"<?xml version='1.0' encoding='utf-8'?>...",
...     name="file.html")
'application/xhtml+xml'

In this case, the smart extractor has refined the information determined from the filename using information from the uploaded data. The specific approach taken by the extractor is not part of the interface, however.

charsetGetter()

If you’re interested in the character set of textual data, you can use the charsetGetter function (which can also be registered as the ICharsetGetter utility):

The simplest case is when the character set is already specified in the content type.

>>> typegetter.charsetGetter(content_type='text/plain; charset=mambo-42')
'mambo-42'

Note that the charset name is lowercased, because all the default ICharset and ICharsetCodec utilities are registered for lowercase names.

>>> typegetter.charsetGetter(content_type='text/plain; charset=UTF-8')
'utf-8'

If it isn’t, charsetGetter can try to guess by looking at actual data

>>> typegetter.charsetGetter(content_type='text/plain', data=b'just text')
'ascii'
>>> typegetter.charsetGetter(content_type='text/plain', data=b'\xe2\x98\xba')
'utf-8'
>>> import codecs
>>> typegetter.charsetGetter(data=codecs.BOM_UTF16_BE + b'\x12\x34')
'utf-16be'
>>> typegetter.charsetGetter(data=codecs.BOM_UTF16_LE + b'\x12\x34')
'utf-16le'

If the character set cannot be determined, charsetGetter returns None.

>>> typegetter.charsetGetter(content_type='text/plain', data=b'\xff')
>>> typegetter.charsetGetter()