-D or --dump-header option combined with shell redirection. This approach extracts only the headers from the server's reply, allowing you to parse or store the specific MIME type without capturing the entire response body.
Understanding Header Extraction
When you run a standardcurl command, it typically outputs the response body to your terminal or saves it to a file specified by -o. However, the HTTP headers, which include metadata like Content-Type, are handled separately. To isolate these headers, you need to direct them to a different destination than the content itself. The most reliable method across Windows, Linux, and macOS is to use the --dump-header flag (or its shorthand -D).
This flag tells curl to write the received HTTP response headers into a specified file. By default, if you do not specify an output file for the body, curl prints it to stdout. If you want to save both the headers and the body, but keep them in separate files, you must explicitly define the output for each. This separation is crucial when you only care about the metadata, such as verifying that a resource is indeed an image or a JSON document, rather than downloading the potentially large payload.
Step-by-Step Implementation on Windows
On Windows, usingcurl.exe requires careful handling of file paths and redirection operators. The process involves two main steps: dumping the headers to one file and optionally suppressing or redirecting the body.
First, construct your command with the -D flag pointing to your desired header file. For example:
curl.exe -D headers.txt https://example.comContent-Type, you can redirect stdout to NUL:
curl.exe -D headers.txt -o NUL https://example.com-o NUL part ensures that no data is written to a visible file, saving disk space and time. The headers.txt file will now contain lines like HTTP/1.1 200 OK followed by Content-Type: application/json; charset=utf-8. You can then use standard Windows tools like findstr to extract just the content type line if needed. Note that on older versions of Windows, ensure you are using the native curl.exe included in Windows 10 version 1803 and later, or Git Bash, to avoid compatibility issues with URL encoding.
Implementation on Linux and macOS
Linux and macOS users benefit from powerful shell piping capabilities, making the extraction even more streamlined. The syntax forcurl remains identical, but the post-processing differs due to the availability of tools like grep and awk.
To save the headers to a file while discarding the body, use:
curl -D response_headers.txt -o /dev/null https://example.com/dev/null acts as the black hole for the response body, similar to NUL on Windows. Once the headers are saved, you can immediately extract the Content-Type value. A common pattern is to pipe the header dump directly into grep:
curl -s -D - https://example.com | grep -i "content-type"-s flag suppresses the progress meter, and -D - sends the headers to stdout instead of a file, allowing the pipe to work. This method is highly efficient for scripting and automation tasks where you need to make decisions based on the media type returned by the server. For more complex scenarios involving return codes, you might also want to check how to handle exit statuses, as detailed in our guide on setting return codes from CURL.
Parsing the Content-Type Value
Having the raw header text is often not enough; you usually need the specific string value, such asimage/png. On Unix-like systems, you can combine curl with sed or awk to clean up the output. For instance, to get just the value after the colon:
curl -s -D - https://example.com | grep -i "content-type" | cut -d':' -f2-Select-String and regex patterns to achieve the same result without relying on external utilities.
It is important to note that some servers may send multiple Content-Type headers or none at all if the request fails. Always validate the existence of the header before attempting to parse it. Additionally, if you are dealing with authentication or cookies in your requests, remember that copying cURL commands from browser devtools may omit certain security details, so manual verification is recommended. See our article on DevTools copying behavior for more context.
Frequently Asked Questions
How do I extract just the MIME type string?
You can use shell tools like grep and cut to filter the header file. After saving headers with -D, run grep -i "content-type" on the file, then use cut to remove the header name and colon, leaving only the value.
Does curl save headers automatically if I don't use -D?
No, curl does not save headers to a file by default. They are either printed to the terminal along with the body or suppressed. You must explicitly use the -D or --dump-header flag to direct headers to a specific file path.
What if the server returns a 404 error? Will I still get the Content-Type?
Yes, curl downloads headers regardless of the HTTP status code. Even for error responses like 404 or 500, the server usually includes a Content-Type header indicating whether the error page is HTML, JSON, or plain text.
Can I save headers and body to the same file?
While possible, it is not recommended for parsing. Using -D writes headers, and -o writes the body. If you omit -o, the body goes to stdout. Combining them into one stream makes it difficult to distinguish where headers end and content begins programmatically.
Is there a way to do this without saving to a temporary file?
Yes, you can pipe the output directly. Use -D - to send headers to stdout, then pipe (|) them into grep or other text processing tools. This avoids creating intermediate files on disk, which is ideal for quick checks in scripts.