Re: [PATCH v2] credential/libsecret: support password_expiry_utc

From: Junio C Hamano <gitster@pobox.com>
Date: May 4, 2023, 17:42 UTC
Message-ID: <xmqqjzxoj8mm.fsf@gitster.g>
In-Reply-To: <pull.1469.v2.git.git.1679729764851.gitgitgadget@gmail.com>

"M Hickford via GitGitGadget" <gitgitgadget@gmail.com> writes:

Show 33 quoted lines

> From: M Hickford <mirth.hickford@gmail.com>
>
> Signed-off-by: M Hickford <mirth.hickford@gmail.com>
> ---
> diff --git a/contrib/credential/libsecret/git-credential-libsecret.c b/contrib/credential/libsecret/git-credential-libsecret.c
> index 2c5d76d789f..3f2b530db79 100644
> --- a/contrib/credential/libsecret/git-credential-libsecret.c
> +++ b/contrib/credential/libsecret/git-credential-libsecret.c
> @@ -39,6 +39,7 @@ struct credential {
>  	char *path;
>  	char *username;
>  	char *password;
> +	char *password_expiry_utc;
>  };
>  
>  #define CREDENTIAL_INIT { 0 }
> @@ -54,6 +55,20 @@ struct credential_operation {
>  
>  /* ----------------- Secret Service functions ----------------- */
>  
> +static const SecretSchema schema = {
> +	"org.git.Password",
> +	SECRET_SCHEMA_NONE,
> +	{
> +		{  "user", SECRET_SCHEMA_ATTRIBUTE_STRING },
> +		{  "object", SECRET_SCHEMA_ATTRIBUTE_STRING },
> +		{  "protocol", SECRET_SCHEMA_ATTRIBUTE_STRING },
> +		{  "port", SECRET_SCHEMA_ATTRIBUTE_INTEGER },
> +		{  "server", SECRET_SCHEMA_ATTRIBUTE_STRING },
> +		{  "password_expiry_utc", SECRET_SCHEMA_ATTRIBUTE_INTEGER },
> +		{  NULL, 0 },
> +	}
> +};

We used to use the bog-standard "COMPAT_NETWORK" but now we are adding an extra element, and that makes it necessary to define our own? OK.

Show 23 quoted lines

>  static char *make_label(struct credential *c)
>  {
>  	if (c->port)
> @@ -78,6 +93,9 @@ static GHashTable *make_attr_list(struct credential *c)
>  		g_hash_table_insert(al, "port", g_strdup_printf("%hu", c->port));
>  	if (c->path)
>  		g_hash_table_insert(al, "object", g_strdup(c->path));
> +	if (c->password_expiry_utc)
> +		g_hash_table_insert(al, "password_expiry_utc",
> +			g_strdup(c->password_expiry_utc));
>  
>  	return al;
>  }
> @@ -101,9 +119,11 @@ static int keyring_get(struct credential *c)
>  
>  	attributes = make_attr_list(c);
>  	items = secret_service_search_sync(service,
> -					   SECRET_SCHEMA_COMPAT_NETWORK,
> +					   &schema,
>  					   attributes,
> -					   SECRET_SEARCH_LOAD_SECRETS | SECRET_SEARCH_UNLOCK,
> +					   SECRET_SEARCH_LOAD_SECRETS | SECRET_SEARCH_UNLOCK |
> +					   // for backwards compatibility

No // comments please.

> +					   SECRET_SCHEMA_DONT_MATCH_NAME,

SECRET_SCHEMA_DONT_MATCH_NAME does not seem to be listed as one of the flags to be used with secret_service_search_sync(),

    https://www.manpagez.com/html/libsecret-1/libsecret-1-0.18.6/SecretService.php#secret-service-search-sync

and the only reference to it I found was as a flag to be placed in the schema.

    https://www.manpagez.com/html/libsecret-1/libsecret-1-0.18.6/migrating-schemas.php
    https://www.manpagez.com/html/libsecret-1/libsecret-1-0.18.6/libsecret-SecretSchema.php

But I'll take your word for it.

I found nothing unexpected or surprising in the rest of the patch to this file. They all looked just a fallout of having to store and retrieve one extra item from the database together with many other things we already store and retrieve. Cleanly written.

Show 45 quoted lines

> diff --git a/t/lib-credential.sh b/t/lib-credential.sh
> index 5ea8bc9f1dc..9ebf7eeae48 100644
> --- a/t/lib-credential.sh
> +++ b/t/lib-credential.sh
> @@ -43,6 +43,7 @@ helper_test_clean() {
>  	reject $1 https example.com store-user
>  	reject $1 https example.com user1
>  	reject $1 https example.com user2
> +	reject $1 https example.com user3
>  	reject $1 http path.tld user
>  	reject $1 https timeout.tld user
>  	reject $1 https sso.tld
> @@ -298,6 +299,35 @@ helper_test_timeout() {
>  	'
>  }
>  
> +helper_test_password_expiry_utc() {
> +	HELPER=$1
> +
> +	test_expect_success "helper ($HELPER) stores password_expiry_utc" '
> +		check approve $HELPER <<-\EOF
> +		protocol=https
> +		host=example.com
> +		username=user3
> +		password=pass
> +		password_expiry_utc=9999999999
> +		EOF
> +	'
>
> +	test_expect_success "helper ($HELPER) gets password_expiry_utc" '
> +		check fill $HELPER <<-\EOF
> +		protocol=https
> +		host=example.com
> +		username=user3
> +		--
> +		protocol=https
> +		host=example.com
> +		username=user3
> +		password=pass
> +		password_expiry_utc=9999999999
> +		--
> +		EOF
> +	'
> +}
> +

Is any random number usable for this test, or is there some constraints (like, "it cannot be too small to be a timestamp in the past, because the entry will be expired immediately")? If there is some constraint like that, is it a good idea to also test that (like, "make sure an entry with expiry already happened is rejected")?

Thanks.