Content Security Policy Level 3 Editor’s Draft, 22 June 2018

This section is not normative.        

This document defines Content Security Policy (CSP), a tool which developers can use to lock down their applications in various ways, mitigating the risk of content injection vulnerabilities such as cross-site scripting, and reducing the privilege with which their applications execute.

CSP is not intended as a first line of defense against content injection vulnerabilities. Instead, CSP is best used as defense-in-depth. It reduces the harm that a malicious injection can cause, but it is not a replacement for careful input validation and output encoding.

This document is an iteration on Content Security Policy Level 2, with the goal of more clearly explaining the interactions between CSP, HTML, and Fetch on the one hand, and providing clear hooks for modular extensibility on the other. Ideally, this will form a stable core upon which we can build new functionality.

1.1. Examples        

1.1.1. Control Execution        

Content-Security-Policy: script-src https://cdn.example.com/scripts/; object-src 'none'

1.2. Goals        

Content Security Policy aims to do to a few related things:

1. Mitigate the risk of content-injection attacks by giving developers fairly granular control over

• The resources which can be requested (and subsequently embedded or executed) on behalf of a specific Document or Worker

• The execution of inline script

• Dynamic code execution (via eval() and similar constructs)

• The application of inline style

1.3. Changes from Level 2        

This document describes an evolution of the Content Security Policy Level 2 specification [CSP2]. The following is a high-level overview of the changes:

1. The specification has been rewritten from the ground up in terms of the [FETCH] specification, which should make it simpler to integrate CSP’s requirements and restrictions with other specifications (and with Service Workers in particular).

2. The child-src model has been substantially altered:

1. The frame-src directive, which was deprecated in CSP Level 2, has been undeprecated, but continues to defer to child-src if not present (which defers to default-src in turn).

2. A worker-src directive has been added, deferring to script-src if not present (which likewise defers to default-src in turn).

3. child-src is now deprecated.

4. Dedicated workers now always inherit their creator’s policy.

3. The URL matching algorithm now treats insecure schemes and ports as matching their secure variants. That is, the source expression http://example.com:80 will match both http://example.com:80and https://example.com:443.

   Likewise, 'self' now matches https: and wss: variants of the page’s origin, even on pages whose scheme is http.

4. Violation reports generated from inline script or style will now report "inline" as the blocked resource. Likewise, blocked eval() execution will report "eval" as the blocked resource.

5. The manifest-src directive has been added.

6. The report-uri directive is deprecated in favor of the new report-to directive, which relies on [REPORTING] as infrastructure.

7. The 'strict-dynamic' source expression will now allow script which executes on a page to load more script via non-"parser-inserted" script elements. Details are in §8.2 Usage of "'strict-dynamic'".

8. The 'unsafe-hashes' source expression will now allow event handlers, style attributes and javascript: navigation targets to match hashes. Details in §8.3 Usage of "'unsafe-hashes'".

9. The source expression matching has been changed to require explicit presence of any non-network scheme, rather than local scheme, unless that non-network scheme is the same as the scheme of protected resource, as described in §6.6.1.6 Does url match expression in origin with redirect count?.

10. Hash-based source expressions may now match external scripts if the script element that triggers the request specifies a set of integrity metadata which is listed in the current policy. Details in §8.4 Allowing external JavaScript via hashes.

11. The disown-opener directive ensures that a resource can’t be opened in such a way as to give another browsing context control over its contents.

     disown-opener is a work in progress. <https://github.com/w3c/webappsec-csp/issues/194>                

12. The navigate-to directive gives a resource control over the endpoints to which it can initiate navigation.

13. Reports generated for inline violations will contain a sample attribute if the relevant directive contains the 'report-sample' expression.

2. Framework        

2.1. Infrastructure        

This document uses ABNF grammar to specify syntax, as defined in [RFC5234]. It also relies on the #ruleABNF extension defined in Section 7 of [RFC7230].

This document depends on the Infra Standard for a number of foundational concepts used in its algorithms and prose [INFRA].

2.2. Policies        

A policy defines allowed and restricted behaviors, and may be applied to a Window, WorkerGlobalScope, or WorkletGlobalScope as described in §4.2.2 Initialize a global object’s CSP list.

Each policy has an associated directive set, which is an ordered set of directives that define the policy’s implications when applied.

Each policy has an associated disposition, which is either "enforce" or "report".

Each policy has an associated source, which is either "header" or "meta".

Multiple policies can be applied to a single resource, and are collected into a list of policies known as a CSP list.

A CSP list contains a header-delivered Content Security Policy if it contains a policy whose source is "header".

A serialized CSP is an ASCII string consisting of a semicolon-delimited series of serialized directives, adhering to the following ABNF grammar [RFC5234]:

serialized-policy= serialized-directive *( OWS ";" [ OWS serialized-directive ] )
                    ; OWS is defined in section 3.2.3 of RFC 7230

A serialized CSP list is an ASCII string consisting of a comma-delimited series of serialized CSPs, adhering to the following ABNF grammar [RFC5234]:

serialized-policy-list= 1#serialized-policy
                    ; The '#' rule is defined in section 7 of RFC 7230

2.2.1. Parse a serialized CSP        

To parse a serialized CSP, given a serialized CSP (serialized), a source (source), and a disposition(disposition), execute the following steps.

This algorithm returns a Content Security Policy object. If serialized could not be parsed, the object’s directive set will be empty.

1. Let policy be a new policy with an empty directive set, a source of source, and a disposition of disposition.

2. For each token returned by strictly splitting serialized on the U+003B SEMICOLON character (;):

1. Strip leading and trailing ASCII whitespace from token.

2. If token is an empty string, continue.

3. Let directive name be the result of collecting a sequence of code points from token which are not ASCII whitespace.

4. If policy’s directive set contains a directive whose name is directive name, continue.

   In this case, the user agent SHOULD notify developers that a duplicate directive was ignored. A console warning might be appropriate, for example.

5. Let directive value be the result of splitting token on ASCII whitespace.

6. Let directive be a new directive whose name is directive name, and value is directive value.

7. Append directive to policy’s directive set.

3. Return policy.

2.2.2. Parse a serialized CSP list        

To parse a serialized CSP list, given a serialized CSP list (list), a source (source), and a disposition(disposition), execute the following steps.

This algorithm returns a list of Content Security Policy objects. If list cannot be parsed, the returned list will be empty.

1. Let policies be an empty list.

2. For each token returned by splitting list on commas:

1. Let policy be the result of parsing token, with a source of source, and disposition of disposition.

2. If policy’s directive set is empty, continue.

3. Append policy to policies.

3. Return policies.

2.3. Directives        

Each policy contains an ordered set of directives (its directive set), each of which controls a specific behavior. The directives defined in this document are described in detail in §6 Content Security Policy Directives.

Each directive is a name / value pair. The name is a non-empty string, and the value is a set of non-empty strings. The value MAY be empty.

A serialized directive is an ASCII string, consisting of one or more whitespace-delimited tokens, and adhering to the following ABNF [RFC5234]:

serialized-directive= directive-name [ RWS directive-value ]directive-name= 1*( ALPHA / DIGIT / "-" )directive-value= *( %x09 / %x20-%x2B / %x2D-%x3A / %x3C-%7E )
                       ; Directive values may contain whitespace and VCHAR characters,
                       ; excluding ";" and ","

; RWS is defined in section 3.2.3 of RFC7230. ALPHA, DIGIT, and
; VCHAR are defined in Appendix B.1 of RFC 5234.

Directives have a number of associated algorithms:

1. A pre-request check, which takes a request and a policy as an argument, and is executed during §4.1.3 Should request be blocked by Content Security Policy?. This algorithm returns "Allowed" unless otherwise specified.

2. A post-request check, which takes a request, a response, and a policy as arguments, and is executed during §4.1.4 Should response to request be blocked by Content Security Policy?. This algorithm returns "Allowed" unless otherwise specified.

3. A response check, which takes a request, a response, and a policy as arguments, and is executed during §4.1.4 Should response to request be blocked by Content Security Policy?. This algorithm returns "Allowed" unless otherwise specified.

4. An inline check, which takes an Element a type string, and a source string as arguments, and is executed during §4.2.4 Should element’s inline type behavior be blocked by Content Security Policy? and during §4.2.5 Should navigation request of type from source in target be blocked by Content Security Policy? for javascript: requests. This algorithm returns "Allowed" unless otherwise specified.

5. An initialization, which takes a Document or global object, a response, and a policy as arguments. This algorithm is executed during §4.2.1 Initialize a Document's CSP list, and has no effect unless otherwise specified.

6. A pre-navigation check, which takes a request, a navigation type string ("form-submission" or "other"), two browsing contexts, and a policy as arguments, and is executed during §4.2.5 Should navigation request of type from source in target be blocked by Content Security Policy?. It returns "Allowed" unless otherwise specified.

7. A navigation response check, which takes a request, a navigation type string ("form-submission" or "other"), a response, two browsing contexts, a check type string ("source" or "response"), and a policyas arguments, and is executed during §4.2.6 Should navigation response to navigation request of type from source in target be blocked by Content Security Policy?. It returns "Allowed" unless otherwise specified.

2.3.1. Source Lists        

Many directives' values consist of source lists: sets of strings which identify content that can be fetched and potentially embedded or executed. Each string represents one of the following types of source expression:

1. Keywords such as 'none' and 'self' (which match nothing and the current URL’s origin, respectively)

2. Serialized URLs such as https://example.com/path/to/file.js (which matches a specific file) or https://example.com/ (which matches everything on that origin)

3. Schemes such as https: (which matches any resource having the specified scheme)

4. Hosts such as example.com (which matches any resource on the host, regardless of scheme) or *.example.com (which matches any resource on the host’s subdomains (and any of its subdomains' subdomains, and so on))

5. Nonces such as 'nonce-ch4hvvbHDpv7xCSvXCs3BrNggHdTzxUA' (which can match specific elements on a page)

6. Digests such as 'sha256-abcd...' (which can match specific elements on a page)

A serialized source list is an ASCII string, consisting of a whitespace-delimited series of source expressions, adhering to the following ABNF grammar [RFC5234]:

serialized-source-list= ( source-expression *( RWS source-expression ) ) / "'none'"source-expression= scheme-source / host-source / keyword-source
                         / nonce-source / hash-source

; Schemes: "https:" / "custom-scheme:" / "another.custom-scheme:"scheme-source= scheme-part ":"

; Hosts: "example.com" / "*.example.com" / "https://*.example.com:12/path/to/file.js"host-source= [ scheme-part "://" ] host-part [ ":" port-part ] [ path-part ]scheme-part= scheme
              ; scheme is defined in section 3.1 of RFC 3986.host-part= "*" / [ "*." ] 1*host-char *( "." 1*host-char )host-char= ALPHA / DIGIT / "-"port-part= 1*DIGIT / "*"path-part= path-absolute
              ; path-absolute is defined in section 3.3 of RFC 3986.

; Keywords:keyword-source= "'self'" / "'unsafe-inline'" / "'unsafe-eval'"
                 / "'strict-dynamic'" / "'unsafe-hashes'" /
                 / "'report-sample'" / "'unsafe-allow-redirects'"

ISSUE: Bikeshed unsafe-allow-redirects.

; Nonces: 'nonce-[nonce goes here]'nonce-source= "'nonce-" base64-value "'"base64-value= 1*( ALPHA / DIGIT / "+" / "/" / "-" / "_" )*2( "=" )

; Digests: 'sha256-[digest goes here]'hash-source= "'" hash-algorithm "-" base64-value "'"hash-algorithm= "sha256" / "sha384" / "sha512"

The host-char production intentionally contains only ASCII characters; internationalized domain names cannot be entered directly as part of a serialized CSP, but instead MUST be Punycode-encoded [RFC3492]. For example, the domain üüüüüü.de MUST be represented as xn--tdaaaaaa.de.

Note: Though IP address do match the grammar above, only 127.0.0.1 will actually match a URL when used in a source expression (see §6.6.1.5 Does url match source list in origin with redirect count? for details). The security properties of IP addresses are suspect, and authors ought to prefer hostnames whenever possible.

Note: The base64-value grammar allows both base64 and base64url encoding. These encodings are treated as equivalant when processing hash-source values. Nonces, however, are strict string matches: we use the base64-value grammar to limit the characters available, and reduce the complexity for the server-side operator (encodings, etc), but the user agent doesn’t actually care about any underlying value, nor does it do any decoding of the nonce-source value.

2.4. Violations        

A violation represents an action or resource which goes against the set of policy objects associated with a global object.

Each violation has a global object, which is the global object whose policy has been violated.

Each violation has a url which is its global object’s URL.

Each violation has a status which is a non-negative integer representing the HTTP status code of the resource for which the global object was instantiated.

Each violation has a resource, which is either null, "inline", "eval", or a URL. It represents the resource which violated the policy.

Each violation has a referrer, which is either null, or a URL. It represents the referrer of the resource whose policy was violated.

Each violation has a policy, which is the policy that has been violated.

Each violation has a disposition, which is the disposition of the policy that has been violated.

Each violation has an effective directive which is a non-empty string representing the directive whose enforcement caused the violation.

Each violation has a source file, which is either null or a URL.

Each violation has a line number, which is a non-negative integer.

Each violation has a column number, which is a non-negative integer.

Each violation has a element, which is either null or an element.

Each violation has a sample, which is a string. It is the empty string unless otherwise specified.

Note: A violation’s sample will be populated with the first 40 characters of an inline script, event handler, or style that caused an violation. Violations which stem from an external file will not include a sample in the violation report.

2.4.1. Create a violation object for global, policy, and directive        

Given a global object (global), a policy (policy), and a string (directive), the following algorithm creates a new violation object, and populates it with an initial set of data:

1. Let violation be a new violation whose global object is global, policy is policy, effective directive is directive, and resource is null.

2. If the user agent is currently executing script, and can extract a source file’s URL, line number, and column number from the global, set violation’s source file, line number, and column numberaccordingly.

    Is this kind of thing specified anywhere? I didn’t see anything that looked useful in [ECMA262].

   Note: User agents need to ensure that the source file is the URL requested by the page, pre-redirects. If that’s not possible, user agents need to strip the URL down to an origin to avoid unintentional leakage.

3. If global is a Window object, set violation’s referrer to global’s document's referrer.

4. Set violation’s status to the HTTP status code for the resource associated with violation’s global object.

    How, exactly, do we get the status code? We don’t actually store it anywhere.

5. Return violation.

2.4.2. Create a violation object for request, policy, and directive        

Given a request (request), a policy (policy), and a string (directive), the following algorithm creates a new violation object, and populates it with an initial set of data:

1. Let violation be the result of executing §2.4.1 Create a violation object for global, policy, and directive on request’s client’s global object, policy, and directive.

2. Set violation’s resource to request’s url.

   Note: We use request’s url, and not its current url, as the latter might contain information about redirect targets to which the page MUST NOT be given access.

3. Return violation.

3. Policy Delivery        

A server MAY declare a policy for a particular resource representation via an HTTP response header field whose value is a serialized CSP. This mechanism is defined in detail in §3.1 The Content-Security-Policy HTTP Response Header Field and §3.2 The Content-Security-Policy-Report-Only HTTP Response Header Field, and the integration with Fetch and HTML is described in §4.1 Integration with Fetch and §4.2 Integration with HTML.

A policy may also be declared inline in an HTML document via a meta element’s http-equiv attribute, as described in §3.3 The <meta> element.

3.1. The Content-Security-Policy HTTP Response Header Field        

The Content-Security-Policy HTTP response header field is the preferred mechanism for delivering a policy from a server to a client. The header’s value is represented by the following ABNF [RFC5234]:

Content-Security-Policy = 1#serialized-policy

Content-Security-Policy: script-src 'self';
                         report-to csp-reporting-endpoint

A server MAY send different Content-Security-Policy header field values with different representationsof the same resource.

A server SHOULD NOT send more than one HTTP response header field named "Content-Security-Policy" with a given resource representation.

When the user agent receives a Content-Security-Policy header field, it MUST parse and enforce each serialized CSP it contains as described in §4.1 Integration with Fetch, §4.2 Integration with HTML.

3.2. The Content-Security-Policy-Report-Only HTTP Response Header Field        

The Content-Security-Policy-Report-Only HTTP response header field allows web developers to experiment with policies by monitoring (but not enforcing) their effects. The header’s value is represented by the following ABNF [RFC5234]:

Content-Security-Policy-Report-Only = 1#serialized-policy

This header field allows developers to piece together their security policy in an iterative fashion, deploying a report-only policy based on their best estimate of how their site behaves, watching for violation reports, and then moving to an enforced policy once they’ve gained confidence in that behavior.

Content-Security-Policy-Report-Only: script-src 'self';
                                     report-to csp-reporting-endpoint

A server MAY send different Content-Security-Policy-Report-Only header field values with different representations of the same resource.

A server SHOULD NOT send more than one HTTP response header field named "Content-Security-Policy-Report-Only" with a given resource representation.

When the user agent receives a Content-Security-Policy-Report-Only header field, it MUST parse and monitor each serialized CSP it contains as described in §4.1 Integration with Fetch and §4.2 Integration with HTML.

Note: The Content-Security-Policy-Report-Only header is not supported inside a meta element.

3.3. The <meta> element        

A Document may deliver a policy via one or more HTML meta elements whose http-equiv attributes are an ASCII case-insensitive match for the string "Content-Security-Policy". For example:

<meta http-equiv="Content-Security-Policy" content="script-src 'self'">

Implementation details can be found in HTML’s Content Security Policy state http-equiv processing instructions [HTML].

Note: The Content-Security-Policy-Report-Only header is not supported inside a meta element. Neither are the report-uri, frame-ancestors, and sandbox directives.

Authors are strongly encouraged to place meta elements as early in the document as possible, because policies in meta elements are not applied to content which precedes them. In particular, note that resources fetched or prefetched using the Link HTTP response header field, and resources fetched or prefetched using link and script elements which precede a meta-delivered policy will not be blocked.

Note: A policy specified via a meta element will be enforced along with any other policies active for the protected resource, regardless of where they’re specified. The general impact of enforcing multiple policies is described in §8.1 The effect of multiple policies.

Note: Modifications to the content attribute of a meta element after the element has been parsed will be ignored.

4. Integrations        

This section is non-normative.        

This document defines a set of algorithms which are used in other specifications in order to implement the functionality. These integrations are outlined here for clarity, but those external documents are the normative references which ought to be consulted for detailed information.

4.1. Integration with Fetch        

A number of directives control resource loading in one way or another. This specification provides algorithms which allow Fetch to make decisions about whether or not a particular request should be blocked or allowed, and about whether a particular response should be replaced with a network error.

1. §4.1.3 Should request be blocked by Content Security Policy? is called as part of step #5 of its Main Fetch algorithm. This allows directives' pre-request checks to be executed against each requestbefore it hits the network, and against each redirect that a request might go through on its way to reaching a resource.

2. §4.1.4 Should response to request be blocked by Content Security Policy? is called as part of step #13 of its Main Fetch algorithm. This allows directives' post-request checks and response checks to be executed on the response delivered from the network or from a Service Worker.

A policy is generally enforced upon a global object, but the user agent needs to parse any policy delivered via an HTTP response header field before any global object is created in order to handle directives that require knowledge of a response’s details. To that end:

1. A response has an associated CSP list which contains any policy objects delivered in the response’s header list.

2. §4.1.1 Set response’s CSP list is called in the HTTP fetch and HTTP-network fetch algorithms.

   Note: These two calls should ensure that a response’s CSP list is set, regardless of how the response is created. If we hit the network (via HTTP-network fetch, then we parse the policy before we handle the Set-Cookie header. If we get a response from a Service Worker (via HTTP fetch, we’ll process its CSP list before handing the response back to our caller.

4.1.1. Set response’s CSP list        

Given a response (response), this algorithm evaluates its header list for serialized CSP values, and populates its CSP list accordingly:

1. Set response’s CSP list to the empty list.

2. Let policies be the result of parsing the result of extracting header list values given Content-Security-Policy and response’s header list, with a source of "header", and a disposition of "enforce".

3. Append to policies the result of parsing the result of extracting header list values given Content-Security-Policy-Report-Only and response’s header list, with a source of "header", and a dispositionof "report".

4. For each policy in policies:

1. Insert policy into response’s CSP list.

4.1.2. Report Content Security Policy violations for request        

Given a request (request), this algorithm reports violations based on client’s "report only" policies.

1. Let CSP list be request’s client’s global object’s CSP list.

2. For each policy in CSP list:

1. If policy’s disposition is "enforce", then skip to the next policy.

2. Let violates be the result of executing §6.6.1.1 Does request violate policy? on request and policy.

3. If violates is not "Does Not Violate", then execute §5.3 Report a violation on the result of executing §2.4.2 Create a violation object for request, policy, and directive on request, policy, and violates.

4.1.3. Should request be blocked by Content Security Policy?        

Given a request (request), this algorithm returns Blocked or Allowed and reports violations based on request’s client’s Content Security Policy.

1. Let CSP list be request’s client’s global object’s CSP list.

2. Let result be "Allowed".

3. For each policy in CSP list:

1. Execute §5.3 Report a violation on the result of executing §2.4.2 Create a violation object for request, policy, and directive on request, policy, and violates.

2. Set result to "Blocked".

3. If policy’s disposition is "report", then skip to the next policy.

4. Let violates be the result of executing §6.6.1.1 Does request violate policy? on request and policy.

5. If violates is not "Does Not Violate", then:

4. Return result.

4.1.4. Should response to request be blocked by Content Security Policy?        

Given a response (response) and a request (request), this algorithm returns Blocked or Allowed, and reports violations based on request’s client’s Content Security Policy.

1. Let CSP list be request’s client’s global object’s CSP list.

2. Let result be "Allowed".

3. For each policy in CSP list:

   Note: This portion of the check verifies that the page can load the response. That is, that a Service Worker hasn’t substituted a file which would violate the page’s CSP.

1. If the result of executing directive’s post-request check is "Blocked", then:

2. Execute §5.3 Report a violation on the result of executing §2.4.2 Create a violation object for request, policy, and directive on request, policy, and directive.

3. If policy’s disposition is "enforce", then set result to "Blocked".

4. For each directive in policy:

4. For each policy in response’s CSP list:

   Note: This portion of the check allows policies delivered with the response to determine whether the response is allowed to be delivered.

1. If the result of executing directive’s response check on request, response, and policy is "Blocked", then:

2. Execute §5.3 Report a violation on the result of executing §2.4.2 Create a violation object for request, policy, and directive on request, policy, and directive.

3. If policy’s disposition is "enforce", then set result to "Blocked".

4. For each directive in policy:

5. Return result.

4.2. Integration with HTML        

1. The Document, WorkerGlobalScope, and WorkletGlobalScope objects have a CSP list, which holds all the policy objects which are active for a given context. This list is empty unless otherwise specified, and is populated via the §4.2.2 Initialize a global object’s CSP list algorithm.

    This concept is missing from W3C’s Workers. <https://github.com/w3c/html/issues/187>                

2. A global object’s CSP list is the result of executing §4.2.3 Retrieve the CSP list of an object with the global object as the object.

3. A policy is enforced or monitored for a global object by inserting it into the global object’s CSP list.

4. §4.2.2 Initialize a global object’s CSP list is called during the initializing a new Document object and run a worker algorithms in order to bind a set of policy objects associated with a response to a newly created Document, WorkerGlobalScope or WorkletGlobalScope.

5. §4.2.4 Should element’s inline type behavior be blocked by Content Security Policy? is called during the prepare a script and update a style block algorithms in order to determine whether or not an inline script or style block is allowed to execute/render.

6. §4.2.4 Should element’s inline type behavior be blocked by Content Security Policy? is called during handling of inline event handlers (like onclick) and inline style attributes in order to determine whether or not they ought to be allowed to execute/render.

7. policy is enforced during processing of the meta element’s http-equiv.

8. A Document's embedding document is the Document through which the Document's browsing context is nested.

9. HTML populates each request’s cryptographic nonce metadata and parser metadata with relevant data from the elements responsible for resource loading.

    Stylesheet loading is not yet integrated with Fetch in W3C’s HTML. <https://github.com/whatwg/html/issues/198>                

    Stylesheet loading is not yet integrated with Fetch in WHATWG’s HTML. <https://github.com/whatwg/html/issues/968>                

10. §6.2.1.1 Is base allowed for document? is called during base's set the frozen base URL algorithm to ensure that the href attribute’s value is valid.

11. §6.2.2.2 Should plugin element be blocked a priori by Content Security Policy?: is called during the processing of object, embed, and applet elements to determine whether they may trigger a fetch.

    Note: Fetched plugin resources are handled in §4.1.4 Should response to request be blocked by Content Security Policy?.

     This hook is missing from W3C’s HTML. <https://github.com/w3c/html/issues/547>                

12. §4.2.5 Should navigation request of type from source in target be blocked by Content Security Policy? is called during the process a navigate fetch algorithm, and §4.2.6 Should navigation response to navigation request of type from source in target be blocked by Content Security Policy?is called during the process a navigate response algorithm to apply directive’s navigation checks, as well as inline checks for navigations to javascript: URLs.

     W3C’s HTML is not based on Fetch, and does not have a process a navigate responsealgorithm into which to hook. <https://github.com/w3c/html/issues/548>                

4.2.1. Initialize a Document's CSP list        

Given a Document (document), and a response (response), the user agent performs the following steps in order to initialize document’s CSP list:

1. If response’s url’s scheme is a local scheme:

   Note: local scheme includes about:, and this algorithm will therefore copy the embedding document’s policies for an iframe srcdoc Document.

   Note: We do all this to ensure that a page cannot bypass its policy by embedding a frame or popping up a new window containing content it controls (blob: resources, or document.write()).

1. For each policy in doc’s CSP list:

2. Insert a copy of policy into document’s CSP list.

3. Let documents be an empty list.

4. If document has an embedding document (embedding), then add embedding to documents.

5. If document has an opener browsing context, then add its active document to documents.

6. For each doc in documents:

2. For each policy in response’s CSP list, insert policy into document’s CSP list.

3. For each policy in document’s CSP list:

1. Execute directive’s initialization algorithm on document and response.

2. For each directive in policy:

4.2.2. Initialize a global object’s CSP list        

Given a global object (global), and a response (response), the user agent performs the following steps in order to initialize global’s CSP list:

1. If response’s url’s scheme is a local scheme, or if global is a DedicatedWorkerGlobalScope:

   Note: local scheme includes about:, and this algorithm will therefore copy the embedding document’s policies for an iframe srcdoc Document.

1. For each policy in owner’s CSP list:

2. Insert a copy of policy into global’s CSP list.

3. Let owners be an empty list.

4. Add each of the items in global’s owner set to owners.

5. For each owner in owners:

2. If global is a SharedWorkerGlobalScope or ServiceWorkerGlobalScope:

1. For each policy in response’s CSP list, insert policy into global’s CSP list.

3. If global is a WorkletGlobalScope:

1. Insert a copy of policy into global’s CSP list.

2. Let owner be global’s owner document.

3. For each policy in owner’s CSP list:

4.2.3. Retrieve the CSP list of an object        

To obtain object’s CSP list:

1. If object is a Document return object’s CSP list.

2. If object is a Window return object’s associated Document’s CSP list.

3. If object is a WorkerGlobalScope, return object’s CSP list.

4. If object is a WorkletGlobalScope, return object’s CSP list.

5. Return null.

4.2.4. Should element’s inline type behavior be blocked by Content Security Policy?        

Given an Element (element), a string (type), and a string (source) this algorithm returns "Allowed" if the element is allowed to have inline definition of a particular type of behavior (script execution, style application, event handlers, etc.), and "Blocked" otherwise:

Note: The valid values for type are "script", "script attribute", "style", and "style attribute".

1. Assert: element is not null.

2. Let result be "Allowed".

3. For each policy in element’s Document's global object’s CSP list:

1. If directive’s inline check returns "Allowed" when executed upon element, type, and source, skip to the next directive.

2. Otherwise, let violation be the result of executing §2.4.1 Create a violation object for global, policy, and directive on the current settings object’s global object, policy, and directive’s name.

3. Set violation’s resource to "inline".

4. Set violation’s element to element.

5. If directive’s value contains the expression "'report-sample'", then set violation’s sample to the substring of source containing its first 40 characters.

6. Execute §5.3 Report a violation on violation.

7. If policy’s disposition is "enforce", then set result to "Blocked".

8. For each directive in policy’s directive set:

4. Return result.

4.2.5. Should navigation request of type from source in target be blocked by Content Security Policy?        

Given a request (navigation request), a string (type, either "form-submission" or "other"), and two browsing contexts (source and target), this algorithm return "Blocked" if the active policy blocks the navigation, and "Allowed" otherwise:

1. Let result be "Allowed".

2. For each policy in source’s active document’s CSP list:

1. If directive’s pre-navigation check returns "Allowed" when executed upon navigation request, type, source, target, and policy skip to the next directive.

2. Otherwise, let violation be the result of executing §2.4.1 Create a violation object for global, policy, and directive on source’s relevant global object, policy, and directive’s name.

3. Set violation’s resource to navigation request’s URL.

4. Execute §5.3 Report a violation on violation.

5. If policy’s disposition is "enforce", then set result to "Blocked".

6. For each directive in policy:

3. If result is "Allowed", and if navigation request’s url’s scheme is javascript:

1. For each directive in policy:

2. If directive’s inline check returns "Allowed" when executed upon null, "navigation" and navigation request’s url, skip to the next directive.

3. Otherwise, let violation be the result of executing §2.4.1 Create a violation object for global, policy, and directive on source’s relevant global object, policy, and directive’s name.

4. Set violation’s resource to navigation request’s URL.

5. Execute §5.3 Report a violation on violation.

6. If policy’s disposition is "enforce", then set result to "Blocked".

7. For each policy in source’s active document’s CSP List:

4. Return result.

4.2.6. Should navigation response to navigation request of type from source in target be blocked by Content Security Policy?        

Given a request (navigation request), a string (type, either "form-submission" or "other"), a responsenavigation response, and two browsing contexts (source and target), this algorithm returns "Blocked" if the active policy blocks the navigation, and "Allowed" otherwise:

1. Let result be "Allowed".

2. For each policy in navigation response’s CSP list:

   Note: Some directives (like frame-ancestors) allow a response’s Content Security Policy to act on the navigation.

1. If directive’s navigation response check returns "Allowed" when executed upon navigation request, type, navigation response, source, target, "response", and policy skip to the next directive.

2. Otherwise, let violation be the result of executing §2.4.1 Create a violation object for global, policy, and directive on null, policy, and directive’s name.

   Note: We use null for the global object, as no global exists: we haven’t processed the navigation to create a Document yet.

3. Set violation’s resource to navigation response’s URL.

4. Execute §5.3 Report a violation on violation.

5. If policy’s disposition is "enforce", then set result to "Blocked".

6. For each directive in policy:

3. For each policy in source’s active document’s CSP List:

   Note: Some directives in the source context (like navigate-to) need the response before acting on the navigation.

1. If directive’s navigation response check returns "Allowed" when executed upon navigation request, type, navigation response, source, target, "source", and policy skip to the next directive.

2. Otherwise, let violation be the result of executing §2.4.1 Create a violation object for global, policy, and directive on source’s relevant global object, policy, and directive’s name.

3. Set violation’s resource to navigation request’s URL.

4. Execute §5.3 Report a violation on violation.

5. If policy’s disposition is "enforce", then set result to "Blocked".

6. For each directive in policy:

4. Return result.

4.3. Integration with ECMAScript        

ECMAScript defines a HostEnsureCanCompileStrings() abstract operation which allows the host environment to block the compilation of strings into ECMAScript code. This document defines an implementation of that abstract operation thich examines the relevant CSP list to determine whether such compilation ought to be blocked.

4.3.1. EnsureCSPDoesNotBlockStringCompilation(callerRealm, calleeRealm, source)        

Given two realms (callerRealm and calleeRealm), and a string (source), this algorithm returns normally if string compilation is allowed, and throws an "EvalError" if not:

1. Let globals be a list containing callerRealm’s global object and calleeRealm’s global object.

2. For each global in globals:

1. Let source-list be null.

2. If policy contains a directive whose name is "script-src", then set source-list to that directive's value.

   Otherwise if policy contains a directive whose name is "default-src", then set source-list to that directive’s value.

3. If source-list is not null, and does not contain a source expression which is an ASCII case-insensitive match for the string "'unsafe-eval'", then:

4. Let violation be the result of executing §2.4.1 Create a violation object for global, policy, and directive on global, policy, and "script-src".

5. Set violation’s resource to "inline".

6. If source-list contains the expression "'report-sample'", then set violation’s sample to the substring of source containing its first 40 characters.

7. Execute §5.3 Report a violation on violation.

8. If policy’s disposition is "enforce", then set result to "Blocked".

9. Let result be "Allowed".

10. For each policy in global’s CSP list:

11. If result is "Blocked", throw an EvalError exception.

 HostEnsureCanCompileStrings() does not include the string which is going to be compiled as a parameter. We’ll also need to update HTML to pipe that value through to CSP. <https://github.com/tc39/ecma262/issues/938>        

5. Reporting        

When one or more of a policy’s directives is violated, a violation report may be generated and sent out to a reporting endpoint associated with the policy.

5.1. Violation DOM Events        

enum SecurityPolicyViolationEventDisposition{"enforce","report"};

[Constructor(DOMString type, optional SecurityPolicyViolationEventInit eventInitDict)]
interface SecurityPolicyViolationEvent: Event {
    readonly    attribute USVString      documentURI;
    readonly    attribute USVString      referrer;
    readonly    attribute USVString      blockedURI;
    readonly    attribute DOMString      violatedDirective;
    readonly    attribute DOMString      effectiveDirective;
    readonly    attribute DOMString      originalPolicy;
    readonly    attribute USVString      sourceFile;
    readonly    attribute DOMString      sample;
    readonly    attribute SecurityPolicyViolationEventDisposition      disposition;
    readonly    attribute unsigned short statusCode;
    readonly    attribute unsigned long  lineNumber;
    readonly    attribute unsigned long  columnNumber;
};

dictionary SecurityPolicyViolationEventInit: EventInit {
    required USVString      documentURI;
             USVString      referrer= "";
             USVString      blockedURI= "";
    required DOMString      violatedDirective;
    required DOMString      effectiveDirective;
    required DOMString      originalPolicy;
             USVString      sourceFile= "";
             DOMString      sample= "";
    required SecurityPolicyViolationEventDisposition disposition;
    required unsigned short statusCode;
             unsigned long  lineNumber= 0;
             unsigned long  columnNumber= 0;
};

5.2. Obtain the deprecated serialization of violation        

Given a violation (violation), this algorithm returns a JSON text string representation of the violation, suitable for submission to a reporting endpoint associated with the deprecated report-uri directive.

1. Let object be a new JavaScript object with properties initialized as follows:

• "document-uri"

• The result of executing the URL serializer on violation’s url, with the exclude fragment flag set.

• "referrer"

• The result of executing the URL serializer on violation’s referrer, with the exclude fragment flag set.

• "blocked-uri"

• The result of executing the URL serializer on violation’s resource, with the exclude fragmentflag set.

• "effective-directive"

• violation’s effective directive

• "violated-directive"

• violation’s effective directive

• "original-policy"

• The serialization of violation’s policy

• "disposition"

• The disposition of violation’s policy

• "status-code"

• violation’s status

• "script-sample"

• violation’s sample

  Note: The name script-sample was chosen for compatibility with an earlier iteration of this feature which has shipped in Firefox since its initial implementation of CSP. Despite the name, this field will contain samples for non-script violations, like stylesheets. The data contained in a SecurityPolicyViolationEvent object, and in reports generated via the new report-to directive, is named in a more encompassing fashion: sample.

1. Set object’s "source-file" property to the result of executing the URL serializer on violation’s source file, with the exclude fragment flag set.

2. Set object’s "line-number" property to violation’s line number.

3. Set object’s "column-number" property to violation’s column number.

5.3. Report a violation        

Given a violation (violation), this algorithm reports it to the endpoint specified in violation’s policy, and fires a SecurityPolicyViolationEvent at violation’s element, or at violation’s global object as described below:

1. Let global be violation’s global object.

2. Let target be violation’s element.

3. Queue a task to run the following steps:

   Note: We "queue a task" here to ensure that the event targeting and dispatch happens after JavaScript completes execution of the task responsible for a given violation (which might manipulate the DOM).

• data

• violation

• type

• "CSP"

• endpoint group

• group

• settings

• settings object

• method

• "POST"

• url

• violation’s url

• origin

• violation’s global object’s relevant settings object’s origin

• window

• "no-window"

• client

• violation’s global object’s relevant settings object

• destination

• "report"

• initiator

• ""

• credentials mode

• "same-origin"

• keepalive flag

• "true"

• header list

• A header list containing a single header whose name is "Content-Type", and value is "application/csp-report"

• body

• The result of executing §5.2 Obtain the deprecated serialization of violation on violation

• redirect mode

• "error"

• documentURI

• The result of executing the URL serializer on violation’s url, with the exclude fragment flag set.

• referrer

• The result of executing the URL serializer on violation’s referrer, with the exclude fragmentflag set.

• blockedURI

• The result of executing the URL serializer on violation’s resource, with the exclude fragment flag set.

• effectiveDirective

• violation’s effective directive

• violatedDirective

• violation’s effective directive

• originalPolicy

• The serialization of violation’s policy

• disposition

• violation’s disposition

• sourceFile

• The result of executing the URL serializer on violation’s source file, with the exclude fragment flag set if the violation’s source file it not null and the empty string otherwise.

• statusCode

• violation’s status

• lineNumber

• violation’s line number

• columnNumber

• violation’s column number

• sample

• violation’s sample

• bubbles

• true

• composed

• true

1. Let group be directive’s value.

2. Let settings object be violation’s global object’s relevant settings object.

3. Execute [REPORTING]'s Queue data as type for endpoint group on settings algorithm with the following arguments:

4. If violation’s policy’s directive set contains a directive named "report-to", skip the remaining substeps.

5. For each token returned by splitting a string on ASCII whitespace with directive’s value as the input.

6. Let endpoint be the result of executing the URL parser with token as the input, and violation’s url as the base URL.

7. If endpoint is not a valid URL, skip the remaining substeps.

8. Let request be a new request, initialized as follows:

   Note: request’s mode defaults to "no-cors"; the response is ignored entirely.

9. Fetch request. The result will be ignored.

10. Set target be violation’s global object.

11. If target is a Window, set target to target’s associated Document.

12. If target is not null, and global is a Window, and target’s shadow-including root is not global’s associated Document, set target to null.

    Note: This ensures that we fire events only at elements connected to violation’s policy’s Document. If a violation is caused by an element which isn’t connected to that document, we’ll fire the event at the document rather than the element in order to ensure that the violation is visible to the document’s listeners.

13. If target is null:

14. Fire an event named securitypolicyviolation that uses the SecurityPolicyViolationEventinterface at target with its attributes initialized as follows:

    Note: Both effectiveDirective and violatedDirective are the same value. This is intentional to maintain backwards compatibility.

    Note: We set the composed attribute, which means that this event can be captured on its way into, and will bubble its way out of a shadow tree. target, et al will be automagically scoped correctly for the main tree.

15. If violation’s policy’s directive set contains a directive named "report-uri" (directive):

    Note: All of this should be considered deprecated. It sends a single request per violation, which simply isn’t scalable. As soon as this behavior can be removed from user agents, it will be.

    Note: report-uri only takes effect if report-to is not present. That is, the latter overrides the former, allowing for backwards compatibility with browsers that don’t support the new mechanism.

16. If violation’s policy’s directive set contains a directive named "report-to" (directive):

6. Content Security Policy Directives        

This specification defines a number of types of directives which allow developers to control certain aspects of their sites' behavior. This document defines directives which govern resource fetching (in §6.1 Fetch Directives), directives which govern the state of a document (in §6.2 Document Directives), directives which govern aspects of navigation (in §6.3 Navigation Directives), and directives which govern reporting (in §6.4 Reporting Directives). These form the core of Content Security Policy; other directives are defined in a modular fashion in ancillary documents (see §6.5 Directives Defined in Other Documents for examples).

To mitigate the risk of cross-site scripting attacks, web developers SHOULD include directives that regulate sources of script and plugins. They can do so by including:

• Both the script-src and object-src directives, or

• a default-src directive

In either case, developers SHOULD NOT include either 'unsafe-inline', or data: as valid sources in their policies. Both enable XSS attacks by allowing code to be included directly in the document itself; they are best avoided completely.

6.1. Fetch Directives        

Fetch directives control the locations from which certain resource types may be loaded. For instance, script-src allows developers to allow trusted sources of script to execute on a page, while font-srccontrols the sources of web fonts.

6.1.1. child-src        

The child-src directive is deprecated. Authors who wish to regulate nested browsing contexts and workers SHOULD use the frame-src and worker-src directives, respectively.

The child-src directive governs the creation of nested browsing contexts (e.g. iframe and framenavigations) and Worker execution contexts. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "child-src"
directive-value = serialized-source-list

This directive controls requests which will populate a frame or a worker. More formally, requests falling into one of the following categories:

• destination is "document", and whose target browsing context is a nested browsing context (e.g. requests which will populate an iframe or frame element)

• destination is either "serviceworker", "sharedworker", or "worker" (which are fed to the run a workeralgorithm for ServiceWorker, SharedWorker, and Worker, respectively).

Content-Security-Policy: child-src https://example.com/

Fetches for the following code will all return network errors, as the URLs provided do not match child-src's source list:

<iframe src="https://example.org"></iframe>
<script>
  var blockedWorker = new Worker("data:application/javascript,...");
</script>

6.1.1.1. child-src Pre-request check        

This directive’s pre-request check is as follows:

Given a request (request) and a policy (policy):

1. Let name be the result of executing §6.6.1.11 Get the effective directive for request on request.

2. If name is not frame-src, return "Allowed".

3. If policy contains a directive whose name is name, return "Allowed"

4. Return the result of executing the pre-request check for the directive whose name is name on request and policy, using this directive’s value for the comparison.

6.1.1.2. child-src Post-request check        

This directive’s post-request check is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Let name be the result of executing §6.6.1.11 Get the effective directive for request on request.

2. If name is not frame-src, return "Allowed".

3. If policy contains a directive whose name is name, return "Allowed"

4. Return the result of executing the post-request check for the directive whose name is name on request, response, and policy, using this directive’s value for the comparison.

6.1.2. connect-src        

The connect-src directive restricts the URLs which can be loaded using script interfaces. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "connect-src"
directive-value = serialized-source-list

This directive controls requests which transmit or receive data from other origins. This includes APIs like fetch(), [XHR], [EVENTSOURCE], [BEACON], and a's ping. This directive also controls WebSocket [WEBSOCKETS] connections, though those aren’t technically part of Fetch.

The connect-src directive allows you to ensure that these and similar sorts of connections are only opened to origins you trust. Sending a policy that defines a list of source expressions for this directive is straightforward. For example, to limit connections to only https://example.com, send the following header:

Content-Security-Policy: connect-src https://example.com/

Fetches for the following code will all return network errors, as the URLs provided do not match connect-src's source list:

<a ping="https://example.org">...
<script>
  var xhr = new XMLHttpRequest();
  xhr.open('GET', 'https://example.org/');
  xhr.send();

  var ws = new WebSocket("https://example.org/");

  var es = new EventSource("https://example.org/");

  navigator.sendBeacon("https://example.org/", { ... });
</script>

6.1.2.1. connect-src Pre-request check        

This directive’s pre-request check is as follows:

Given a request (request) and a policy (policy):

1. Assert: policy is unused.

2. If request’s initiator is "fetch" or its destination is "":

1. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.2.2. connect-src Post-request check        

This directive’s post-request check is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Assert: policy is unused.

2. If request’s initiator is "fetch" or its destination is "":

1. If the result of executing §6.6.1.4 Does response to request match source list? on response, request, and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.3. default-src        

The default-src directive serves as a fallback for the other fetch directives. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "default-src"
directive-value = serialized-source-list

If a default-src directive is present in a policy, its value will be used as the policy’s default source list. That is, given default-src 'none'; script-src 'self', script requests will use 'self' as the source list to match against. Other requests will use 'none'. This is spelled out in more detail in the §4.1.3 Should request be blocked by Content Security Policy? and §4.1.4 Should response to request be blocked by Content Security Policy? algorithms.

Content-Security-Policy: default-src 'self'

will have the same behavior as the following header:

Content-Security-Policy: connect-src 'self';
                         font-src 'self';
                         frame-src 'self';
                         img-src 'self';
                         manifest-src 'self';
                         media-src 'self';
                         prefetch-src 'self';
                         object-src 'self';
                         script-src 'self';
                         style-src 'self';
                         worker-src 'self'

That is, when default-src is set, every fetch directive that isn’t explicitly set will fall back to the value default-src specifies.

Content-Security-Policy: default-src 'self'; script-src https://example.com

will have the same behavior as the following header:

Content-Security-Policy: connect-src 'self';
                         font-src 'self';
                         frame-src 'self';
                         img-src 'self';
                         manifest-src 'self';
                         media-src 'self';
                         prefetch-src 'self';
                         object-src 'self';
                         script-src https://example.com;
                         style-src 'self';
                         worker-src 'self'

Given this behavior, one good way to build a policy for a site would be to begin with a default-src of 'none', and to build up a policy from there which allowed only those resource types which are necessary for the particular page the policy will apply to.

6.1.3.1. default-src Pre-request check        

This directive’s pre-request check is as follows:

Given a request (request) and a policy (policy):

1. Let name be the result of executing §6.6.1.11 Get the effective directive for request on request.

2. If name is null, return "Allowed".

3. If policy contains a directive whose name is name, return "Allowed".

4. If name is "frame-src", and policy contains a directive whose name is "child-src", return "Allowed".

5. If name is "worker-src", and policy contains a directive whose name is "script-src", return "Allowed".

6. Otherwise, return the result of executing the pre-request check for the directive whose name is name on request and policy, using this directive’s value for the comparison.

6.1.3.2. default-src Post-request check        

This directive’s post-request check is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Let name be the result of executing §6.6.1.11 Get the effective directive for request on request.

2. If name is null, return "Allowed".

3. If policy contains a directive whose name is name, return "Allowed".

4. If name is "frame-src", and policy contains a directive whose name is "child-src", return "Allowed".

5. If name is "worker-src", and policy contains a directive whose name is "script-src", return "Allowed".

6. Otherwise, return the result of executing the post-request check for the directive whose name is name on request, response, and policy, using this directive’s value for the comparison.

6.1.4. font-src        

The font-src directive restricts the URLs from which font resources may be loaded. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "font-src"
directive-value = serialized-source-list

Content-Security-Policy: font-src https://example.com/

Fetches for the following code will return a network errors, as the URL provided do not match font-src's source list:

<style>
  @font-face {
    font-family: "Example Font";
    src: url("https://example.org/font");
  }
  body {
    font-family: "Example Font";
  }
</style>

6.1.4.1. font-src Pre-request check        

This directive’s pre-request check is as follows:

Given a request (request) and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is "font":

1. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.4.2. font-src Post-request check        

This directive’s post-request check is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is "font":

1. If the result of executing §6.6.1.4 Does response to request match source list? on response, request, and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.5. frame-src        

The frame-src directive restricts the URLs which may be loaded into nested browsing contexts. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "frame-src"
directive-value = serialized-source-list

Content-Security-Policy: frame-src https://example.com/

Fetches for the following code will return a network errors, as the URL provided do not match frame-src's source list:

<iframe src="https://example.org/">
</iframe>

6.1.5.1. frame-src Pre-request check        

This directive’s pre-request check is as follows:

Given a request (request) and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is "document" and target browsing context is a nested browsing context:

1. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.5.2. frame-src Post-request check        

This directive’s post-request check is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is "document" and target browsing context is a nested browsing context:

1. If the result of executing §6.6.1.4 Does response to request match source list? on response, request, and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.6. img-src        

The img-src directive restricts the URLs from which image resources may be loaded. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "img-src"
directive-value = serialized-source-list

This directive controls requests which load images. More formally, this includes requests whose destination is "image" [FETCH].

Content-Security-Policy: img-src https://example.com/

Fetches for the following code will return a network errors, as the URL provided do not match img-src's source list:

<img src="https://example.org/img">

6.1.6.1. img-src Pre-request check        

This directive’s pre-request check is as follows:

Given a request (request) and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is "image":

1. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.6.2. img-src Post-request check        

This directive’s post-request check is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is "image":

1. If the result of executing §6.6.1.4 Does response to request match source list? on response, request, and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.7. manifest-src        

The manifest-src directive restricts the URLs from which application manifests may be loaded [APPMANIFEST]. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "manifest-src"
directive-value = serialized-source-list

Content-Security-Policy: manifest-src https://example.com/

Fetches for the following code will return a network errors, as the URL provided do not match manifest-src's source list:

<link rel="manifest" href="https://example.org/manifest">

6.1.7.1. manifest-src Pre-request check        

This directive’s pre-request check is as follows:

Given a request (request) and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is "manifest":

1. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.7.2. manifest-src Post-request check        

This directive’s post-request check is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is "manifest":

1. If the result of executing §6.6.1.4 Does response to request match source list? on response, request, and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.8. media-src        

The media-src directive restricts the URLs from which video, audio, and associated text track resources may be loaded. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "media-src"
directive-value = serialized-source-list

Content-Security-Policy: media-src https://example.com/

Fetches for the following code will return a network errors, as the URL provided do not match media-src's source list:

<audio src="https://example.org/audio"></audio>
<video src="https://example.org/video">
    <track kind="subtitles" src="https://example.org/subtitles">
</video>

6.1.8.1. media-src Pre-request check        

This directive’s pre-request check is as follows:

Given a request (request) and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is one of "audio", "video", or "track":

1. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.8.2. media-src Post-request check        

This directive’s post-request check is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is one of "audio", "video", or "track":

1. If the result of executing §6.6.1.4 Does response to request match source list? on response, request, and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.9. prefetch-src        

The prefetch-src directive restricts the URLs from which resources may be prefetched or prerendered. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "prefetch-src"
directive-value = serialized-source-list

Content-Security-Policy: prefetch-src https://example.com/

Fetches for the following code will return network errors, as the URLs provided do not match prefetch-src's source list:

<link rel="prefetch" src="https://example.org/"></link>
<link rel="prerender" src="https://example.org/"></link>

6.1.9.1. prefetch-src Pre-request check        

This directive’s pre-request check is as follows:

Given a request (request) and a policy (policy):

1. Assert: policy is unused.

2. If request’s initiator is "prefetch" or "prerender":

1. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.9.2. prefetch-src Post-request check        

This directive’s post-request check is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Assert: policy is unused.

2. If request’s initiator is "prefetch" or "prerender":

1. If the result of executing §6.6.1.4 Does response to request match source list? on response, request, and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.10. object-src        

The object-src directive restricts the URLs from which plugin content may be loaded. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "object-src"
directive-value = serialized-source-list

Content-Security-Policy: object-src https://example.com/

Fetches for the following code will return a network errors, as the URL provided do not match object-src's source list:

<embed src="https://example.org/flash"></embed>
<object data="https://example.org/flash"></object>
<applet archive="https://example.org/flash"></applet>

If plugin content is loaded without an associated URL (perhaps an object element lacks a data attribute, but loads some default plugin based on the specified type), it MUST be blocked if object-src's value is 'none', but will otherwise be allowed.

Note: The object-src directive acts upon any request made on behalf of an object, embed, or appletelement. This includes requests which would populate the nested browsing context generated by the former two (also including navigations). This is true even when the data is semantically equivalent to content which would otherwise be restricted by another directive, such as an object element with a text/html MIME type.

Note: When a plugin resource is navigated to directly (that is, as a plugin document in the top-level browsing context or a nested browsing context, and not as an embedded subresource via embed, object, or applet), any policy delivered along with that resource will be applied to the plugin document. This means, for instance, that developers can prevent the execution of arbitrary resources as plugin content by delivering the policy object-src 'none' along with a response. Given plugins' power (and the sometimes-interesting security model presented by Flash and others), this could mitigate the risk of attack vectors like Rosetta Flash.

6.1.10.1. object-src Pre-request check        

This directive’s pre-request check is as follows:

Given a request (request) and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is "object" or "embed":

1. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.10.2. object-src Post-request check        

This directive’s post-request check is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is "object" or "embed":

1. If the result of executing §6.6.1.4 Does response to request match source list? on response, request, and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.11. script-src        

The script-src directive restricts the locations from which scripts may be executed. This includes not only URLs loaded directly into script elements, but also things like inline script blocks and XSLT stylesheets [XSLT] which can trigger script execution. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "script-src"
directive-value = serialized-source-list

The script-src directive governs five things:

1. Script requests MUST pass through §4.1.3 Should request be blocked by Content Security Policy?.

2. Script responses MUST pass through §4.1.4 Should response to request be blocked by Content Security Policy?.

3. Inline script blocks MUST pass through §4.2.4 Should element’s inline type behavior be blocked by Content Security Policy?. Their behavior will be blocked unless every policy allows inline script, either implicitly by not specifying a script-src (or default-src) directive, or explicitly, by specifying "unsafe-inline", a nonce-source or a hash-source that matches the inline block.

4. The following JavaScript execution sinks are gated on the "unsafe-eval" source expression:

   Note: If a user agent implements non-standard sinks like setImmediate() or execScript(), they SHOULD also be gated on "unsafe-eval".

• eval()

• Function()

• setTimeout() with an initial argument which is not callable.

• setInterval() with an initial argument which is not callable.

6.1.11.1. script-src Pre-request check        

This directive’s pre-request check is as follows:

Given a request (request) and a policy (policy):

1. If the result of executing §6.6.1.11 Get the effective directive for request on request is "worker-src", and policy contains a directive whose name is "worker-src", return "Allowed".

   Note: If worker-src is present, we’ll defer to it when handling worker requests.

2. If request’s destination is script-like:

1. If the request’s parser metadata is "parser-inserted", return "Blocked".

   Otherwise, return "Allowed".

   Note: "'strict-dynamic'" is explained in more detail in §8.2 Usage of "'strict-dynamic'".

2. Let integrity sources be the result of executing the algorithm defined in Subresource Integrity §parse-metadata on request’s integrity metadata. [SRI]

3. If integrity sources is "no metadata" or an empty set, skip the remaining substeps.

4. Let bypass due to integrity match be true.

5. For each source in integrity sources:

6. If bypass due to integrity match is true, return "Allowed".

7. If this directive’s value does not contain a source expression whose hash-algorithm is a case-sensitive match for source’s hash-algo component, and whose base64-value is a case-sensitive match for source’s base64-value, then set bypass due to integrity match to false.

8. If the result of executing §6.6.1.2 Does nonce match source list? on request’s cryptographic nonce metadata and this directive’s value is "Matches", return "Allowed".

9. Let integrity expressions be the set of source expressions in this directive’s value that match the hash-source grammar.

10. If integrity expressions is not empty:

    Note: Here, we verify only that the request contains a set of integrity metadata which is a subset of the hash-source source expressions specified by this directive. We rely on the browser’s enforcement of Subresource Integrity [SRI] to block non-matching resources upon response.

11. If this directive’s value contains a source expression that is an ASCII case-insensitive match for the "'strict-dynamic'" keyword-source:

12. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.11.2. script-src Post-request check        

This directive’s post-request check is as follows:

Given a request (request), a response (response), and a policy (policy):

1. If the result of executing §6.6.1.11 Get the effective directive for request on request is "worker-src", and policy contains a directive whose name is "worker-src", return "Allowed".

   Note: If worker-src is present, we’ll defer to it when handling worker requests.

2. If request’s destination is script-like:

1. If the result of executing §6.6.1.2 Does nonce match source list? on request’s cryptographic nonce metadata and this directive’s value is "Matches", return "Allowed".

2. If this directive’s value contains "'strict-dynamic'", and request’s parser metadata is not "parser-inserted", return "Allowed".

3. If the result of executing §6.6.1.4 Does response to request match source list? on response, request, and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.11.3. script-src Inline Check        

This directive’s inline check algorithm is as follows:

Given an Element (element), a string (type), and a string (source):

1. If type is "script", "script attribute" or "navigation":

1. Assert: element is not null or type is "navigation".

2. If the result of executing §6.6.2.3 Does element match source list for type and source? on element, this directive’s value, type, and source, is "Does Not Match", return "Blocked".

2. Return "Allowed".

6.1.12. style-src        

The style-src directive restricts the locations from which style may be applied to a Document. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "style-src"
directive-value = serialized-source-list

The style-src directive governs several things:

1. Style requests MUST pass through §4.1.3 Should request be blocked by Content Security Policy?. This includes:

1. Stylesheet requests originating from a link element.

2. Stylesheet requests originating from the @import rule.

3. Stylesheet requests originating from a Link HTTP response header field [RFC8288].

2. Responses to style requests MUST pass through §4.1.4 Should response to request be blocked by Content Security Policy?.

3. Inline style blocks MUST pass through §4.2.4 Should element’s inline type behavior be blocked by Content Security Policy?. The styles will be blocked unless every policy allows inline style, either implicitly by not specifying a style-src (or default-src) directive, or explicitly, by specifying "unsafe-inline", a nonce-source or a hash-source that matches the inline block.

4. The following CSS algorithms are gated on the unsafe-eval source expression:

   This would include, for example, all invocations of CSSOM’s various cssText setters and insertRulemethods [CSSOM] [HTML].

    This needs to be better explained. <https://github.com/w3c/webappsec-csp/issues/212>                

1. insert a CSS rule

2. parse a CSS rule,

3. parse a CSS declaration block

4. parse a group of selectors

6.1.12.1. style-src Pre-request Check        

This directive’s pre-request check is as follows:

Given a request (request) and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is "style":

1. If the result of executing §6.6.1.2 Does nonce match source list? on request’s cryptographic nonce metadata and this directive’s value is "Matches", return "Allowed".

2. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.12.2. style-src Post-request Check        

This directive’s post-request check is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is "style":

1. If the result of executing §6.6.1.2 Does nonce match source list? on request’s cryptographic nonce metadata and this directive’s value is "Matches", return "Allowed".

2. If the result of executing §6.6.1.4 Does response to request match source list? on response, request, and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.12.3. style-src Inline Check        

This directive’s inline check algorithm is as follows:

Given an Element (element), a string (type), and a string (source):

1. If type is "style" or "style attribute":

1. If the result of executing §6.6.2.3 Does element match source list for type and source? on element, this directive’s value, type, and source, is "Does Not Match", return "Blocked".

2. Return "Allowed".

This directive’s initialization algorithm is as follows:

 Do something interesting to the execution context in order to lock down interesting CSSOM algorithms. I don’t think CSSOM gives us any hooks here, so let’s work with them to put something reasonable together.

6.1.13. worker-src        

The worker-src directive restricts the URLs which may be loaded as a Worker, SharedWorker, or ServiceWorker. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "worker-src"
directive-value = serialized-source-list

Content-Security-Policy: worker-src https://example.com/

Fetches for the following code will return a network errors, as the URL provided do not match worker-src's source list:

<script>
  var blockedWorker = new Worker("data:application/javascript,...");
  blockedWorker = new SharedWorker("https://example.org/");
  navigator.serviceWorker.register('https://example.org/sw.js');
</script>

6.1.13.1. worker-src Pre-request Check        

This directive’s pre-request check is as follows:

Given a request (request) and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is one of "serviceworker", "sharedworker", or "worker":

4. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.1.13.2. worker-src Post-request Check        

This directive’s post-request check is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is one of "serviceworker", "sharedworker", or "worker":

1. If the result of executing §6.6.1.4 Does response to request match source list? on response, request, and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.2. Document Directives        

The following directives govern the properties of a document or worker environment to which a policy applies.

6.2.1. base-uri        

The base-uri directive restricts the URLs which can be used in a Document's base element. The syntax for the directive’s name and value is described by the following ABNF:

directive-name  = "base-uri"
directive-value = serialized-source-list

The following algorithm is called during HTML’s set the frozen base url algorithm in order to monitor and enforce this directive:

6.2.1.1. Is base allowed for document?        

Given a URL (base), and a Document (document), this algorithm returns "Allowed" if base may be used as the value of a base element’s href attribute, and "Blocked" otherwise:

1. For each policy in document’s global object’s csp list:

   Note: We compare against the fallback base URL in order to deal correctly with things like an iframe srcdoc Document which has been sandboxed into an opaque origin.

1. Let violation be the result of executing §2.4.1 Create a violation object for global, policy, and directive on document’s global object, policy, and "base-uri".

2. Set violation’s resource to "inline".

3. Execute §5.3 Report a violation on violation.

4. If policy’s disposition is "enforce", return "Blocked".

5. Let source list be null.

6. If a directive whose name is "base-uri" is present in policy’s directive set, set source list to that directive’s value.

7. If source list is null, skip to the next policy.

8. If the result of executing §6.6.1.5 Does url match source list in origin with redirect count? on base, source list, document’s fallback base URL’s origin, and 0 is "Does Not Match":

2. Return "Allowed".

6.2.2. plugin-types        

The plugin-types directive restricts the set of plugins that can be embedded into a document by limiting the types of resources which can be loaded. The directive’s syntax is described by the following ABNF grammar:

directive-name  = "plugin-types"
directive-value = media-type-list

media-type-list= media-type *( RWS media-type )media-type= type "/" subtype
; type and subtype are defined in RFC 2045

If a plugin-types directive is present, instantiation of an embed or object element will fail if any of the following conditions hold:

1. The element does not explicitly declare a valid MIME type via a type attribute.

2. The declared type does not match one of the items in the directive’s value.

3. The fetched resource does not match the declared type.

Content-Security-Policy: plugin-types application/pdf

Fetches for the following code will all return network errors:

<!-- No 'type' declaration -->
<object data="https://example.com/flash"></object>

<!-- Non-matching 'type' declaration -->
<object data="https://example.com/flash" type="application/x-shockwave-flash"></object>

<!-- Non-matching resource -->
<object data="https://example.com/flash" type="application/pdf"></object>

If the page allowed Flash content by sending the following header:

Content-Security-Policy: plugin-types application/x-shockwave-flash

Then the second item above would load successfully:

<!-- Matching 'type' declaration and resource -->
<object data="https://example.com/flash" type="application/x-shockwave-flash"></object>

6.2.2.1. plugin-types Post-Request Check        

This directive’s post-request check algorithm is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Assert: policy is unused.

2. If request’s destination is either "object" or "embed":

1. Let type be the result of extracting a MIME type from response’s header list.

2. If type is not an ASCII case-insensitive match for any item in this directive’s value, return "Blocked".

3. Return "Allowed".

6.2.2.2. Should plugin element be blocked a priori by Content Security Policy?:

Given an Element (plugin element), this algorithm returns "Blocked" or "Allowed" based on the element’s type attribute and the policy applied to its document:

1. For each policy in plugin element’s node document’s CSP list:

1. Let type be "application/x-java-applet" if plugin element is an applet element, or plugin element’s type attribute’s value if present, or "null" otherwise.

2. Return "Blocked" if any of the following are true:

3. type is null.

4. type is not a valid MIME type.

5. type is not an ASCII case-insensitive match for any item in directive’s value.

6. If policy contains a directive (directive) whose name is plugin-types:

2. Return "Allowed".

6.2.3. sandbox        

The sandbox directive specifies an HTML sandbox policy which the user agent will apply to a resource, just as though it had been included in an iframe with a sandbox property.

The directive’s syntax is described by the following ABNF grammar, with the additional requirement that each token value MUST be one of the keywords defined by HTML specification as allowed values for the iframe sandbox attribute [HTML].

directive-name  = "sandbox"
directive-value = "" / token *( RWS token )

This directive has no reporting requirements; it will be ignored entirely when delivered in a Content-Security-Policy-Report-Only header, or within a meta element.

6.2.3.1. sandbox Response Check        

This directive’s response check algorithm is as follows:

Given a request (request), a response (response), and a policy (policy):

1. Assert: response is unused.

2. If policy’s disposition is not "enforce", then return "Allowed".

3. If request’s destination is one of "serviceworker", "sharedworker", or "worker":

1. If the result of the Parse a sandboxing directive algorithm using this directive’s value as the input contains either the sandboxed scripts browsing context flag or the sandboxed origin browsing context flag flags, return "Blocked".

   Note: This will need to change if we allow Workers to be sandboxed into unique origins, which seems like a pretty reasonable thing to do.

4. Return "Allowed".

6.2.3.2. sandbox Initialization        

This directive’s initialization algorithm is responsible for adjusting a Document's forced sandboxing flag set according to the sandbox values present in its policies, as follows:

Given a Document or global object (context), a response (response), and a policy (policy):

1. Assert: response is unused.

2. If policy’s disposition is not "enforce", or context is not a Document, then abort this algorithm.

   Note: This will need to change if we allow Workers to be sandboxed, which seems like a pretty reasonable thing to do.

3. Parse a sandboxing directive using this directive’s value as the input, and context’s forced sandboxing flag set as the output.

directive-name  = "disown-opener"
directive-value = ""

6.2.4.1. disown-opener Initialization        

This directive’s initialization algorithm is as follows:

Given a Document or global object (context), a response (response), and a policy (policy):

1. Assert: response and policy are unused.

2. If context’s responsible browsing context has an opener browsing context, disown its opener.

 What should this do in an iframe? Anything?

6.3. Navigation Directives        

6.3.1. form-action        

The form-action directive restricts the URLs which can be used as the target of a form submissions from a given context. The directive’s syntax is described by the following ABNF grammar:

directive-name  = "form-action"
directive-value = serialized-source-list

6.3.1.1. form-action Pre-Navigation Check        

Given a request (request), a string navigation type ("form-submission" or "other"), two browsing contexts(source and target), and a policy (policy) this algorithm returns "Blocked" if one or more of the ancestors of target violate the frame-ancestors directive delivered with the response, and "Allowed" otherwise. This constitutes the form-action' directive’s pre-navigation check:

1. Assert: source, target, and policy are unused in this algorithm, as form-action is concerned only with details of the outgoing request.

2. If navigation type is "form-submission":

1. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

3. Return "Allowed".

6.3.2. frame-ancestors        

The frame-ancestors directive restricts the URLs which can embed the resource using frame, iframe, object, embed, or applet element. Resources can use this directive to avoid many UI Redressing [UISECURITY] attacks, by avoiding the risk of being embedded into potentially hostile contexts.

The directive’s syntax is described by the following ABNF grammar:

directive-name  = "frame-ancestors"
directive-value = ancestor-source-list

ancestor-source-list= ( ancestor-source *( RWS ancestor-source) ) / "'none'"ancestor-source= scheme-source / host-source / "'self'"

The frame-ancestors directive MUST be ignored when contained in a policy declared via a meta element.

Note: The frame-ancestors directive’s syntax is similar to a source list, but frame-ancestors will not fall back to the default-src directive’s value if one is specified. That is, a policy that declares default-src 'none' will still allow the resource to be embedded by anyone.

6.3.2.1. frame-ancestors Navigation Response Check        

Given a request (request), a string navigation type ("form-submission" or "other"), a response (navigation response) two browsing contexts (source and target), a string check type ("source" or "response"), and a policy (policy) this algorithm returns "Blocked" if one or more of the ancestors of target violate the frame-ancestors directive delivered with the response, and "Allowed" otherwise. This constitutes the frame-ancestors directive’s navigation response check:

1. Assert: request, navigation response, navigation type, source, and policy are unused in this algorithm, as frame-ancestors is concerned only with navigation response’s frame-ancestorsdirective.

2. If check type is "source", return "Allowed".

   Note: The 'frame-ancestors' directive is relevant only to the target browsing context and it has no impact on the source browsing context.

3. If target is not a nested browsing context, return "Allowed".

4. Let current be target.

5. While current has a parent browsing context (parent):

1. Set current to parent.

2. Let origin be the result of executing the URL parser on the ASCII serialization of parent’s active document’s relevant settings object’s origin.

3. If §6.6.1.5 Does url match source list in origin with redirect count? returns Does Not Match when executed upon origin, this directive’s value, navigation response’s url’s origin, and 0, return "Blocked".

6. Return "Allowed".

6.3.2.2. Relation to X-Frame-Options        

This directive is similar to the X-Frame-Options header that several user agents have implemented. The 'none' source expression is roughly equivalent to that header’s DENY, 'self' to SAMEORIGIN, and so on. The major difference is that many user agents implement SAMEORIGIN such that it only matches against the top-level document’s location, while the frame-ancestors directive checks against each ancestor. If _any_ ancestor doesn’t match, the load is cancelled. [RFC7034]        

In order to allow backwards-compatible deployment, the frame-ancestors directive _obsoletes_ the X-Frame-Options header. If a resource is delivered with an policy that includes a directive named frame-ancestors and whose disposition is "enforce", then the X-Frame-Options header MUST be ignored.

 Spell this out in more detail as part of defining X-Frame-Options integration with the process a navigate response algorithm. <https://github.com/whatwg/html/issues/1230>        

6.3.3. navigate-to        

The navigate-to directive restricts the URLs to which a document can initiate navigations by any means (a, form, window.location, window.open, etc.). This is an enforcement on what navigations this documentinitiates not on what this document is allowed to navigate to. If the form-action directive is present, the navigate-to directive will not act on navigations that are form submissions.

Content-Security-Policy: navigate-to example.com

A document target has the following Content-Security-Policy:

Content-Security-Policy: navigate-to not-example.com

If the initiator attempts to navigate the target to example.com, the navigation is allowed by the navigate-to directive.

If the initiator attempts to navigate the target to not-example.com, the navigation is blocked by the navigate-to directive.

The directive’s syntax is described by the following ABNF grammar:

directive-name  = "navigate-to"
directive-value = serialized-source-list

6.3.3.1. navigate-to Pre-Navigation Check        

Given a request (request), a string navigation type ("form-submission" or "other"), two browsing contexts(source and target), and a policy (policy), this algorithm returns "Blocked" if the navigation violates the navigate-to directive’s constraints, and "Allowed" otherwise. This constitutes the navigate-to' directive’s pre-navigation check:

1. Assert: source and target are unused as 'navigate-to' is concerned with the details of the request.

2. If navigation type is "form-submission" and policy contains a directive named "form-action", return "Allowed".

3. If this directive’s value contains a source expression that is an ASCII case-insensitive match for the "'unsafe-allow-redirects'" keyword-source, return "Allowed".

   Note: If the 'unsafe-allow-redirects' flag is present we have to wait for the response and take into account the response’s status in §6.3.3.2 navigate-to Navigation Response Check.

4. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

5. Return "Allowed".

6.3.3.2. navigate-to Navigation Response Check        

Given a request (request), a string navigation type ("form-submission" or "other"), a response (navigation response) two browsing contexts (source and target), a string check type ("source" or "response"), and a policy (policy), this algorithm returns "Blocked" if the navigation violates the navigate-to directive’s constraints, and "Allowed" otherwise. This constitutes the navigate-to directive’s navigation response check:

1. Assert: source, and target are unused.

2. If check type is "response", return "Allowed".

   Note: The 'navigate-to' directive is relevant only to the source browsing context and it has no impact on the target browsing context.

3. If navigation type is "form-submission" and policy contains a directive named "form-action", return "Allowed".

4. If this directive’s value does not contain a source expression that is an ASCII case-insensitivematch for the "'unsafe-allow-redirects'" keyword-source, return "Allowed".

   Note: If the 'unsafe-allow-redirects' flag is not present we have already checked the navigation in §6.3.3.1 navigate-to Pre-Navigation Check.

5. If navigation response’s status is a redirect status, return "Allowed".

6. If the result of executing §6.6.1.3 Does request match source list? on request and this directive’s value is "Does Not Match", return "Blocked".

7. Return "Allowed".

6.4. Reporting Directives        

Various algorithms in this document hook into the reporting process by constructing a violation object via §2.4.2 Create a violation object for request, policy, and directive or §2.4.1 Create a violation object for global, policy, and directive, and passing that object to §5.3 Report a violation to deliver the report.

6.4.1. report-uri        

Content-Security-Policy: ...; report-uri https://endpoint.com; report-to groupname

The report-uri directive defines a set of endpoints to which violation reports will be sent when particular behaviors are prevented.

directive-name  = "report-uri"
directive-value = uri-reference *( RWS uri-reference )

; The uri-reference grammar is defined in Section 4.1 of RFC 3986.

The directive has no effect in and of itself, but only gains meaning in combination with other directives.

6.4.2. report-to        

The report-to directive defines a reporting group to which violation reports ought to be sent [REPORTING]. The directive’s behavior is defined in §5.3 Report a violation. The directive’s name and value are described by the following ABNF:

directive-name  = "report-to"
directive-value = token

6.5. Directives Defined in Other Documents        

This document defines a core set of directives, and sets up a framework for modular extension by other specifications. At the time this document was produced, the following stable documents extend CSP:

• [MIX] defines block-all-mixed-content

• [UPGRADE-INSECURE-REQUESTS] defines upgrade-insecure-requests

• [SRI] defines require-sri-for

Extensions to CSP MUST register themselves via the process outlined in [RFC7762]. In particular, note the criteria discussed in Section 4.2 of that document.

New directives SHOULD use the pre-request check, post-request check, response check, and initialization hooks in order to integrate themselves into Fetch and HTML.

6.6. Matching Algorithms        

6.6.1. URL Matching        

6.6.1.1. Does request violate policy?        

Given a request (request) and a policy (policy), this algorithm returns the violated directive if the request violates the policy, and "Does Not Violate" otherwise.

1. Let violates be "Does Not Violate".

2. For each directive in policy:

1. Let result be the result of executing directive’s pre-request check on request and policy.

2. If result is "Blocked", then let violates be directive.

3. Return violates.

6.6.1.2. Does nonce match source list?        

Given a request’s cryptographic nonce metadata (nonce) and a source list (source list), this algorithm returns "Matches" if the nonce matches one or more source expressions in the list, and "Does Not Match" otherwise:

1. Assert: source list is not null.

2. If nonce is the empty string, return "Does Not Match".

3. For each expression in source list:

1. If expression matches the nonce-source grammar, and nonce is a case-sensitive match for expression’s base64-value part, return "Matches".

4. Return "Does Not Match".

6.6.1.3. Does request match source list?        

Given a request (request), and a source list (source list), this algorithm returns the result of executing §6.6.1.5 Does url match source list in origin with redirect count? on request’s current url, source list, request’s origin, and request’s redirect count.

Note: This is generally used in directives' pre-request check algorithms to verify that a given requestis reasonable.

6.6.1.4. Does response to request match source list?        

Given a request (request), and a source list (source list), this algorithm returns the result of executing §6.6.1.5 Does url match source list in origin with redirect count? on response’s url, source list, request’s origin, and request’s redirect count.

Note: This is generally used in directives' post-request check algorithms to verify that a given response is reasonable.

6.6.1.5. Does url match source list in origin with redirect count?        

Given a URL (url), a source list (source list), an origin (origin), and a number (redirect count), this algorithm returns "Matches" if the URL matches one or more source expressions in source list, or "Does Not Match" otherwise:

1. Assert: source list is not null.

2. If source list is an empty list, return "Does Not Match".

3. If source list contains a single item which is an ASCII case-insensitive match for the string "'none'", return "Does Not Match".

   Note: An empty source list (that is, a directive without a value: script-src, as opposed to script-src host1) is equivalent to a source list containing 'none', and will not match any URL.

4. For each expression in source list:

1. If §6.6.1.6 Does url match expression in origin with redirect count? returns "Matches" when executed upon url, expression, origin, and redirect count, return "Matches".

5. Return "Does Not Match".

6.6.1.6. Does url match expression in origin with redirect count?        

Given a URL (url), a source expression (expression), an origin (origin), and a number (redirect count), this algorithm returns "Matches" if url matches expression, and "Does Not Match" otherwise.

Note: origin is the origin of the resource relative to which the expression should be resolved. "'self'", for instance, will have distinct meaning depending on that bit of context.

1. If expression is the string "*", return "Matches" if one or more of the following conditions is met:

   Note: This logic means that in order to allow a resource from a non-network scheme, it has to be either explicitly specified (e.g. default-src * data: custom-scheme-1: custom-scheme-2:), or the protected resource must be loaded from the same scheme.

1. url’s scheme is a network scheme.

2. url’s scheme is the same as origin’s scheme.

2. If expression matches the scheme-source or host-source grammar:

1. If expression has a scheme-part, and it does not scheme-part match url’s scheme, return "Does Not Match".

2. If expression matches the scheme-source grammar, return "Matches".

3. If expression matches the host-source grammar:

1. Let path be the resulting of joining url’s path on the U+002F SOLIDUS character (/).

2. If expression’s path-part does not path-part match path, return "Does Not Match".

3. If url’s host is null, return "Does Not Match".

4. If expression does not have a scheme-part, and origin’s scheme does not scheme-part matchurl’s scheme, return "Does Not Match".

   Note: As with scheme-part above, we allow schemeless host-source expressions to be upgraded from insecure schemes to secure schemes.

5. If expression’s host-part does not host-part match url’s host, return "Does Not Match".

6. Let port-part be expression’s port-part if present, and null otherwise.

7. If port-part does not port-part match url’s port and url’s scheme, return "Does Not Match".

8. If expression contains a non-empty path-part, and redirect count is 0, then:

9. Return "Matches".

4. If expression is an ASCII case-insensitive match for "'self'", return "Matches" if one or more of the following conditions is met:

   Note: Like the scheme-part logic above, the "'self'" matching algorithm allows upgrades to secure schemes when it is safe to do so. We limit these upgrades to endpoints running on the default port for a particular scheme or a port that matches the origin of the protected resource, as this seems sufficient to deal with upgrades that can be reasonably expected to succeed.

1. url’s scheme is "https" or "wss"

2. origin’s scheme is "http"

3. origin is the same as url’s origin

4. origin’s host is the same as url’s host, origin’s port and url’s port are either the same or the default ports for their respective schemes, and one or more of the following conditions is met:

5. Return "Does Not Match".

6.6.1.7. scheme-part matching        

An ASCII string scheme-part matches another ASCII string if a CSP source expression that contained the first as a scheme-part could potentially match a URL containing the latter as a scheme. For example, we say that "http" scheme-part matches "https".

Note: The matching relation is asymmetric. For example, the source expressions https: and https://example.com/ do not match the URL http://example.com/. We always allow a secure upgrade from an explicitly insecure expression. script-src http: is treated as equivalent to script-src http: https:, script-src http://example.com to script-src http://example.com https://example.com, and connect-src ws: to connect-src ws: wss:.

More formally, two ASCII strings (A and B) are said to scheme-part match if the following algorithm returns "Matches":

1. If one of the following is true, return "Matches":

1. A is an ASCII case-insensitive match for B.

2. A is an ASCII case-insensitive match for "http", and B is an ASCII case-insensitive match for "https".

3. A is an ASCII case-insensitive match for "ws", and B is an ASCII case-insensitive match for "wss", "http", or "https".

4. A is an ASCII case-insensitive match for "wss", and B is an ASCII case-insensitive match for "https".

2. Return "Does Not Match".

6.6.1.8. host-part matching        

An ASCII string host-part matches another ASCII string if a CSP source expression that contained the first as a host-part could potentially match a URL containing the latter as a host. For example, we say that "www.example.com" host-part matches "www.example.com".

More formally, two ASCII strings (A and B) are said to host-part match if the following algorithm returns "Matches":

Note: The matching relation is asymmetric. That is, A matching B does not mean that B will match A. For example, *.example.com host-part matches www.example.com, but www.example.com does not host-part match *.example.com.

1. If the first character of A is an U+002A ASTERISK character (*):

1. Let remaining be the result of removing the leading ("*") from A.

2. If remaining (including the leading U+002E FULL STOP character (.)) is an ASCII case-insensitive match for the rightmost characters of B, then return "Matches". Otherwise, return "Does Not Match".

2. If A is not an ASCII case-insensitive match for B, return "Does Not Match".

3. If A matches the IPv4address rule from [RFC3986], and is not "127.0.0.1"; or if A is an IPv6 address, return "Does Not Match".

   Note: A future version of this specification may allow literal IPv6 and IPv4 addresses, depending on usage and demand. Given the weak security properties of IP addresses in relation to named hosts, however, authors are encouraged to prefer the latter whenever possible.

4. Return "Matches".

6.6.1.9. port-part matching        

An ASCII string (port A) port-part matches two other ASCII strings (port B and scheme B) if a CSP source expression that contained the first as a port-part could potentially match a URL containing the latter as port and scheme. For example, "80" port-part matches matches "80"/"http".

1. If port A is empty:

1. If port B is the default port for scheme B, return "Matches". Otherwise, return "Does Not Match".

2. If port A is equal to "*", return "Matches".

3. If port A is a case-sensitive match for port B, return "Matches".

4. If port B is empty:

1. If port A is the default port for scheme B, return "Matches". Otherwise, return "Does not Match".

5. Return "Does Not Match".

6.6.1.10. path-part matching        

An ASCII string (path A) path-part matches another ASCII string (path B) if a CSP source expression that contained the first as a path-part could potentially match a URL containing the latter as a path. For example, we say that "/subdirectory/" path-part matches "/subdirectory/file".

Note: The matching relation is asymmetric. That is, path A matching path B does not mean that path B will match path A.

1. If path A is empty, return "Matches".

2. If path A consists of one character that is equal to the U+002F SOLIDUS character (/) and path B is empty, return "Matches".

3. Let exact match be false if the final character of path A is the U+002F SOLIDUS character (/), and true otherwise.

4. Let path list A and path list B be the result of strictly splitting path A and path B respectively on the U+002F SOLIDUS character (/).

5. If path list A has more items than path list B, return "Does Not Match".

6. If exact match is true, and path list A does not have the same number of items as path list B, return "Does Not Match".

7. If exact match is false:

1. Assert: the final item in path list A is the empty string.

2. Remove the final item from path list A.

8. For each piece A in path list A:

1. Let piece B be the next item in path list B.

2. Percent decode piece A.

3. Percent decode piece B.

4. If piece A is not a case-sensitive match for piece B, return "Does Not Match".

9. Return "Matches".

6.6.1.11. Get the effective directive for request        

Each fetch directive controls a specific destination of request. Given a request (request), the following algorithm returns either null or the name of the request’s effective directive:

1. Switch on request’s destination, and execute the associated steps:

• ""

• 
  

• "manifest"

• 
  

• "object"

• "embed"

• 
  

• "prefetch"

• "prerender"

• 
  

• "document"

• 
  

• "audio"

• "track"

• "video"

• 
  

• "font"

• 
  

• "image"

• 
  

• "style"

• 
  

• "script"

• "xslt"

• 
  

• "sharedworker"

• "worker"

• 
  

1. Return worker-src.

2. Return script-src.

3. Return style-src.

4. Return img-src.

5. Return font-src.

6. Return media-src.

7. If the request’s target browsing context is a nested browsing context, return frame-src.

8. Return prefetch-src.

9. Return object-src.

10. Return manifest-src.

11. If the request’s initiator is the empty string, return connect-src.

6.6.2. Element Matching Algorithms        

6.6.2.1. Is element nonceable?        

Given an Element (element), this algorithm returns "Nonceable" if a nonce-source expression can match the element (as discussed in §7.2 Nonce Stealing), and "Not Nonceable" if such expressions should not be applied.

1. If element does not have an attribute named "nonce", return "Not Nonceable".

2. If element is a script element, then for each attribute in element:

1. If attribute’s name is an ASCII case-insensitive match for the string "<script" or the string "<style", return "Not Nonceable".

2. If attribute’s value contains an ASCII case-insensitive match the string "<script" or the string "<style", return "Not Nonceable".

3. If element had a duplicate-attribute parse error during tokenization, return "Not Nonceable".

    We need some sort of hook in HTML to record this error if we’re planning on using it here. <https://github.com/whatwg/html/issues/3257>                

4. Return "Nonceable".

 This processing is meant to mitigate the risk of dangling markup attacks that steal the nonce from an existing element in order to load injected script. It is fairly expensive, however, as it requires that we walk through all attributes and their values in order to determine whether the script should execute. Here, we try to minimize the impact by doing this check only for script elements when a nonce is present, but we should probably consider this algorithm as "at risk" until we know its impact. <https://github.com/w3c/webappsec-csp/issues/98>        

6.6.2.2. Does a source list allow all inline behavior for type?        

A source list allows all inline behavior of a given type if it contains the keyword-source expression 'unsafe-inline', and does not override that expression as described in the following algorithm:

Given a source list (list) and a string (type), the following algorithm returns "Allows" if all inline content of a given type is allowed and "Does Not Allow" otherwise.

1. Let allow all inline be false.

2. For each expression in list:

1. If expression matches the nonce-source or hash-source grammar, return "Does Not Allow".

2. If type is "script", "script attribute" or "navigation" and expression matches the keyword-source "'strict-dynamic'", return "Does Not Allow".

   Note: 'strict-dynamic' only applies to scripts, not other resource types. Usage is explained in more detail in §8.2 Usage of "'strict-dynamic'".

3. If expression is an ASCII case-insensitive match for the keyword-source "'unsafe-inline'", set allow all inline to true.

3. If allow all inline is true, return "Allows". Otherwise, return "Does Not Allow".

'unsafe-inline' http://a.com http://b.com
'unsafe-inline'

Source lists that do not allow all inline behavior due to the presence of nonces and/or hashes, or absence of 'unsafe-inline':

'sha512-321cba' 'nonce-abc'
http://example.com 'unsafe-inline' 'nonce-abc'

Source lists that do not allow all inline behavior when type is 'script' or 'script attribute' due to the presence of 'strict-dynamic', but allow all inline behavior otherwise:

'unsafe-inline' 'strict-dynamic'
http://example.com 'strict-dynamic' 'unsafe-inline'

6.6.2.3. Does element match source list for type and source?        

Given an Element (element), a source list (list), a string (type), and a string (source), this algorithm returns "Matches" or "Does Not Match".

Note: source will be interpreted with the encoding of the page in which it is embedded. See the integration points in §4.2 Integration with HTML for more detail.

1. If §6.6.2.2 Does a source list allow all inline behavior for type? returns "Allows" given list and type, return "Matches".

2. If type is "script" or "style", and §6.6.2.1 Is element nonceable? returns "Nonceable" when executed upon element:

   Note: Nonces only apply to inline script and inline style, not to attributes of either element or to javascript: navigations.

1. If expression matches the nonce-source grammar, and element has a nonce attribute whose value is a case-sensitive match for expression’s base64-value part, return "Matches".

2. For each expression in list:

3. Let unsafe-hashes flag be false.

4. For each expression in list:

1. If expression is an ASCII case-insensitive match for the keyword-source "'unsafe-hashes'", set unsafe-hashes flag to true. Break out of the loop.

5. If type is "script" or "style", or unsafe-hashes flag is true:

   Note: Hashes apply to inline script and inline style. If the "'unsafe-hashes'" source expression is present, they will also apply to event handlers, style attributes and javascript: navigations.

1. If expression matches the hash-source grammar:

2. Let actual be the result of base64 encoding the result of applying algorithm to source.

3. Let expected be expression’s base64-value part, with all '-' characters replaced with '+', and all '_' characters replaced with '/'.

   Note: This replacement normalizes hashes expressed in base64url encoding into base64 encoding for matching.

4. If actual is a case-sensitive match for expected, return "Matches".

5. Let algorithm be null.

6. If expression’s hash-algorithm part is an ASCII case-insensitive match for "sha256", set algorithm to SHA-256.

7. If expression’s hash-algorithm part is an ASCII case-insensitive match for "sha384", set algorithm to SHA-384.

8. If expression’s hash-algorithm part is an ASCII case-insensitive match for "sha512", set algorithm to SHA-512.

9. If algorithm is not null:

10. For each expression in list:

6. Return "Does Not Match".

7. Security and Privacy Considerations        

7.1. Nonce Reuse        

Nonces override the other restrictions present in the directive in which they’re delivered. It is critical, then, that they remain unguessable, as bypassing a resource’s policy is otherwise trivial.

If a server delivers a nonce-source expression as part of a policy, the server MUST generate a unique value each time it transmits a policy. The generated value SHOULD be at least 128 bits long (before encoding), and SHOULD be generated via a cryptographically secure random number generator in order to ensure that the value is difficult for an attacker to predict.

Note: Using a nonce to allow inline script or style is less secure than not using a nonce, as nonces override the restrictions in the directive in which they are present. An attacker who can gain access to the nonce can execute whatever script they like, whenever they like. That said, nonces provide a substantial improvement over 'unsafe-inline' when layering a content security policy on top of old code. When considering 'unsafe-inline', authors are encouraged to consider nonces (or hashes) instead.

7.2. Nonce Stealing        

Dangling markup attacks such as those discussed in [FILEDESCRIPTOR-2015] can be used to repurpose a page’s legitimate nonces for injections. For example, given an injection point before a scriptelement:

<p>Hello, [INJECTION POINT]</p>
<script nonce=abc src=/good.js></script>

If an attacker injects the string "<script src='https://evil.com/evil.js' ", then the browser will receive the following:

<p>Hello, <script src='https://evil.com/evil.js' </p>
<script nonce=abc src=/good.js></script>

It will then parse that code, ending up with a script element with a src attribute pointing to a malicious payload, an attribute named </p>, an attribute named "<script", a nonce attribute, and a second src attribute which is helpfully discarded as duplicate by the parser.

The §6.6.2.1 Is element nonceable? algorithm attempts to mitigate this specific attack by walking through script or style element attributes, looking for the string "<script" or "<style" in their names or values.

7.3. Nonce Retargeting        

Nonces bypass host-source expressions, enabling developers to load code from any origin. This, generally, is fine, and desirable from the developer’s perspective. However, if an attacker can inject a base element, then an otherwise safe page can be subverted when relative URLs are resolved. That is, on https://example.com/ the following code will load https://example.com/good.js:

<script nonce=abc src=/good.js></script>

However, the following will load https://evil.com/good.js:

<base href="https://evil.com">
<script nonce=abc src=/good.js></script>

To mitigate this risk, it is advisable to set an explicit base element on every page, or to limit the ability of an attacker to inject their own base element by setting a base-uri directive in your page’s policy. For example, base-uri 'none'.

7.4. CSS Parsing        

The style-src directive restricts the locations from which the protected resource can load styles. However, if the user agent uses a lax CSS parsing algorithm, an attacker might be able to trick the user agent into accepting malicious "stylesheets" hosted by an otherwise trustworthy origin.

These attacks are similar to the CSS cross-origin data leakage attack described by Chris Evans in 2009 [CSS-ABUSE]. User agents SHOULD defend against both attacks using the same mechanism: stricter CSS parsing rules for style sheets with improper MIME types.

7.5. Violation Reports        

The violation reporting mechanism in this document has been designed to mitigate the risk that a malicious web site could use violation reports to probe the behavior of other servers. For example, consider a malicious web site that allows https://example.com as a source of images. If the malicious site attempts to load https://example.com/login as an image, and the example.com server redirects to an identity provider (e.g. identityprovider.example.net), CSP will block the request. If violation reports contained the full blocked URL, the violation report might contain sensitive information contained in the redirected URL, such as session identifiers or purported identities. For this reason, the user agent includes only the URL of the original request, not the redirect target.

Note also that violation reports should be considered attacker-controlled data. Developers who wish to collect violation reports in a dashboard or similar service should be careful to properly escape their content before rendering it (and should probably themselves use CSP to further mitigate the risk of injection). This is especially true for the "script-sample" property of violation reports, and the sampleproperty of SecurityPolicyViolationEvent, which are both completely attacker-controlled strings.

7.6. Paths and Redirects        

To avoid leaking path information cross-origin (as discussed in Egor Homakov’s Using Content-Security-Policy for Evil), the matching algorithm ignores the path component of a source expression if the resource being loaded is the result of a redirect. For example, given a page with an active policy of img-src example.com example.org/path:

• Directly loading https://example.org/not-path would fail, as it doesn’t match the policy.

• Directly loading https://example.com/redirector would pass, as it matches example.com.

• Assuming that https://example.com/redirector delivered a redirect response pointing to https://example.org/not-path, the load would succeed, as the initial URL matches example.com, and the redirect target matches example.org/path if we ignore its path component.

This restriction reduces the granularity of a document’s policy when redirects are in play, a necessary compromise to avoid brute-forced information leaks of this type.

The relatively long thread "Remove paths from CSP?" from public-webappsec@w3.org has more detailed discussion around alternate proposals.

7.7. Secure Upgrades        

To mitigate one variant of history-scanning attacks like Yan Zhu’s Sniffly, CSP will not allow pages to lock themselves into insecure URLs via policies like script-src http://example.com. As described in §6.6.1.7 scheme-part matching, the scheme portion of a source expression will always allow upgrading to a secure variant.

7.8. CSP Inheriting to avoid bypasses        

As described in §4.2.1 Initialize a Document's CSP list and §4.2.2 Initialize a global object’s CSP list, documents loaded from local schemes will inherit a copy of the policies in the CSP list of the embedding document or opener browsing context. The goal is to ensure that a page can’t bypass its policy by embedding a frame or opening a new window containing content that is entirely under its control (srcdoc documents, blob: or data: URLs, about:blank documents that can be manipulated via document.write(), etc).

<iframe srcdoc="<script>alert(1);</script>"></iframe>

Note that we create a copy of the CSP list which means that the new Document's CSP list is a snapshot of the relevant policies at its creation time. Modifications in the CSP list of the new Document won’t affect the embedding document or opener browsing context’s CSP list or vice-versa.

<iframe srcdoc='<meta http-equiv="Content-Security-Policy" content="img-src example.com;">
                   <img src="not-example.com/image">'></iframe>

<img src="not-example.com/image">

8. Authoring Considerations        

8.1. The effect of multiple policies        

This section is not normative.        

The above sections note that when multiple policies are present, each must be enforced or reported, according to its type. An example will help clarify how that ought to work in practice. The behavior of an XMLHttpRequest might seem unclear given a site that, for whatever reason, delivered the following HTTP headers:

Content-Security-Policy: default-src 'self' http://example.com http://example.net;
                         connect-src 'none';
Content-Security-Policy: connect-src http://example.com/;
                         script-src http://example.com/

Is a connection to example.com allowed or not? The short answer is that the connection is not allowed. Enforcing both policies means that a potential connection would have to pass through both unscathed. Even though the second policy would allow this connection, the first policy contains connect-src 'none', so its enforcement blocks the connection. The impact is that adding additional policies to the list of policies to enforce can only further restrict the capabilities of the protected resource.

To demonstrate that further, consider a script tag on this page. The first policy would lock scripts down to 'self', http://example.com and http://example.net via the default-src directive. The second, however, would only allow script from http://example.com/. Script will only load if it meets both policy’s criteria: in this case, the only origin that can match is http://example.com, as both policies allow it.

8.2. Usage of "'strict-dynamic'"        

Host- and path-based policies are tough to get right, especially on sprawling origins like CDNs. The solutions to Cure53’s H5SC Minichallenge 3: "Sh*t, it’s CSP!" [H5SC3] are good examples of the kinds of bypasses which such policies can enable, and though CSP is capable of mitigating these bypasses via exhaustive declaration of specific resources, those lists end up being brittle, awkward, and difficult to implement and maintain.

The "'strict-dynamic'" source expression aims to make Content Security Policy simpler to deploy for existing applications who have a high degree of confidence in the scripts they load directly, but low confidence in their ability to provide a reasonable list of resources to load up front.

If present in a script-src or default-src directive, it has two main effects:

1. host-source and scheme-source expressions, as well as the "'unsafe-inline'" and "'self' keyword-sources will be ignored when loading script.

   hash-source and nonce-source expressions will be honored.

2. Script requests which are triggered by non-"parser-inserted" script elements are allowed.

The first change allows you to deploy "'strict-dynamic' in a backwards compatible way, without requiring user-agent sniffing: the policy 'unsafe-inline' https: 'nonce-abcdefg' 'strict-dynamic' will act like 'unsafe-inline' https: in browsers that support CSP1, https: 'nonce-DhcnhD3khTMePgXwdayK9BsMqXjhguVV' in browsers that support CSP2, and 'nonce-DhcnhD3khTMePgXwdayK9BsMqXjhguVV' 'strict-dynamic' in browsers that support CSP3.

The second allows scripts which are given access to the page via nonces or hashes to bring in their dependencies without adding them explicitly to the page’s policy.

Content-Security-Policy: script-src 'nonce-DhcnhD3khTMePgXwdayK9BsMqXjhguVV' 'strict-dynamic'

And serves the following HTML with that policy active:

...
<script src="https://cdn.example.com/script.js" nonce="DhcnhD3khTMePgXwdayK9BsMqXjhguVV" ></script>
...

This will generate a request for https://cdn.example.com/script.js, which will not be blocked because of the matching nonce attribute.

If script.js contains the following code:

var s = document.createElement('script');
s.src = 'https://othercdn.not-example.net/dependency.js';
document.head.appendChild(s);

document.write('<scr' + 'ipt src="/sadness.js"></scr' + 'ipt>');

dependency.js will load, as the script element created by createElement() is not "parser-inserted".

sadness.js will not load, however, as document.write() produces script elements which are "parser-inserted".

<button id="action" onclick="doSubmit()">

Content-Security-Policy:  script-src 'unsafe-hashes' 'sha256-jzgBGA4UWFFmpOBq0JpdsySukE1FrEN5bUpoK8Z29fY='

Content-Security-Policy: script-src 'sha256-abc123' 'sha512-321cba'

<script integrity="sha256-abc123" ...></script>
<script integrity="sha512-321cba" ...></script>
<script integrity="sha256-abc123 sha512-321cba" ...></script>

<script integrity="sha384-xyz789" ...></script>
<script integrity="sha384-xyz789 sha512-321cba" ...></script>
<script integrity="sha256-abc123 sha384-xyz789 sha512-321cba" ...></script>

<script integrity="sha256-abc123 sha1024-abcd" ...></script>
<script integrity="sha512-321cba entirely-invalid" ...></script>
<script integrity="sha256-abc123 not-a-hash-at-all sha512-321cba" ...></script>

9. Implementation Considerations        

9.1. Vendor-specific Extensions and Addons        

Policy enforced on a resource SHOULD NOT interfere with the operation of user-agent features like addons, extensions, or bookmarklets. These kinds of features generally advance the user’s priority over page authors, as espoused in [HTML-DESIGN].

Moreover, applying CSP to these kinds of features produces a substantial amount of noise in violation reports, significantly reducing their value to developers.

Chrome, for example, excludes the chrome-extension: scheme from CSP checks, and does some work to ensure that extension-driven injections are allowed, regardless of a page’s policy.

10. IANA Considerations        

10.1. Directive Registry        

The Content Security Policy Directive registry should be updated with the following directives and references [RFC7762]:

• base-uri

• This document (see §6.2.1 base-uri)

• child-src

• This document (see §6.1.1 child-src)

• connect-src

• This document (see §6.1.2 connect-src)

• default-src

• This document (see §6.1.3 default-src)

• disown-opener

• This document (see §6.2.4 disown-opener)

• font-src

• This document (see §6.1.4 font-src)

• form-action

• This document (see §6.3.1 form-action)

• frame-ancestors

• This document (see §6.3.2 frame-ancestors)

• frame-src

• This document (see §6.1.5 frame-src)

• img-src

• This document (see §6.1.6 img-src)

• manifest-src

• This document (see §6.1.7 manifest-src)

• media-src

• This document (see §6.1.8 media-src)

• object-src

• This document (see §6.1.10 object-src)

• plugin-types

• This document (see §6.2.2 plugin-types)

• report-uri

• This document (see §6.4.1 report-uri)

• report-to

• This document (see §6.4.2 report-to)

• sandbox

• This document (see §6.2.3 sandbox)

• script-src

• This document (see §6.1.11 script-src)

• style-src

• This document (see §6.1.12 style-src)

• worker-src

• This document (see §6.1.13 worker-src)

10.2. Headers        

The permanent message header field registry should be updated with the following registrations: [RFC3864]        

10.2.1. Content-Security-Policy        

• Header field name

• Content-Security-Policy

• Applicable protocol

• http

• Status

• standard

• Author/Change controller

• W3C

• Specification document

• This specification (See §3.1 The Content-Security-Policy HTTP Response Header Field)

10.2.2. Content-Security-Policy-Report-Only        

• Header field name

• Content-Security-Policy-Report-Only

• Applicable protocol

• http

• Status

• standard

• Author/Change controller

• W3C

• Specification document

• This specification (See §3.2 The Content-Security-Policy-Report-Only HTTP Response Header Field)

11. Acknowledgements        

Lots of people are awesome. For instance:

• Mario and all of Cure53.

• Artur Janc, Michele Spagnuolo, Lukas Weichselbaum, Jochen Eisinger, and the rest of Google’s CSP Cabal.