threads / discuss / 23608

"git log -- SHA-1" - how to get/simulate this functionality?

Subject: "git log -- SHA-1" - how to get/simulate this functionality?

## tl;dr

7 messages between Apr 26, 2010 and Apr 26, 2010.

replies: 6people: 2as markdown or json

Eugene Sajine· Apr 26, 2010, 17:54 UTC · lore
Hi,

I'm wondering if there is a way to get some repo information if the only thing i know is SHA-1 (no full path to the file) The current way seems to be a bit cumbersome

git log --no-abbrev --raw --all | grep SHA-1
parse for path(s) and then git log -- path...
I would want something like git log -- SHA-1
Any ideas?

Thanks, Eugene

Eugene Sajine· Apr 26, 2010, 17:56 UTC · re: Eugene Sajine · lore

Re: "git log -- SHA-1" - how to get/simulate this functionality?

On Mon, Apr 26, 2010 at 1:54 PM, Eugene Sajine <euguess@gmail.com> wrote:
> Hi,
>
> I'm wondering if there is a way to get some repo information if the
> only thing i know is SHA-1 (no full path to the file)

I mean of course the SHA-1 of the blob, i.e. the result of git hash-object some_file

Thanks, Eugene

Alex Riesen· Apr 26, 2010, 18:08 UTC · re: Eugene Sajine · lore

Re: "git log -- SHA-1" - how to get/simulate this functionality?

On Mon, Apr 26, 2010 at 19:56, Eugene Sajine <euguess@gmail.com> wrote:
Show 8 quoted lines
> On Mon, Apr 26, 2010 at 1:54 PM, Eugene Sajine <euguess@gmail.com> wrote:
>> Hi,
>>
>> I'm wondering if there is a way to get some repo information if the
>> only thing i know is SHA-1 (no full path to the file)
>
> I mean of course the SHA-1 of the blob, i.e. the result of git
> hash-object some_file

Try 'git cat-file -t <sha>'. It prints what object type is under the hash. Than you can use either 'git cat-file <type> <sha>' (the most precise version, which shall fail if anything is wrong), 'git cat-file -p <sha>' or plain 'git show <sha>' (which always works, unless repo's broken).

Eugene Sajine· Apr 26, 2010, 18:22 UTC · re: Alex Riesen · lore

Re: "git log -- SHA-1" - how to get/simulate this functionality?

On Mon, Apr 26, 2010 at 2:08 PM, Alex Riesen <raa.lkml@gmail.com> wrote:
Show 15 quoted lines
> On Mon, Apr 26, 2010 at 19:56, Eugene Sajine <euguess@gmail.com> wrote:
>> On Mon, Apr 26, 2010 at 1:54 PM, Eugene Sajine <euguess@gmail.com> wrote:
>>> Hi,
>>>
>>> I'm wondering if there is a way to get some repo information if the
>>> only thing i know is SHA-1 (no full path to the file)
>>
>> I mean of course the SHA-1 of the blob, i.e. the result of git
>> hash-object some_file
>
> Try 'git cat-file -t <sha>'. It prints what object type is under the hash.
> Than you can use either 'git cat-file <type> <sha>' (the most precise version,
> which shall fail if anything is wrong), 'git cat-file -p <sha>' or plain
> 'git show <sha>' (which always works, unless repo's broken).
>

Well thank you for that, but that gives me what i already know - the content. I need to "reverse engineer" the history by having only the content and its SHA-1 in order to indentify the last commitID and get all relevant info (author, date etc. )

This is all about avoiding using keyword expansion (thanks CVS for bad habbits)

Thanks, Eugene

Alex Riesen· Apr 26, 2010, 18:57 UTC · re: Eugene Sajine · lore

Re: "git log -- SHA-1" - how to get/simulate this functionality?

On Mon, Apr 26, 2010 at 20:22, Eugene Sajine <euguess@gmail.com> wrote:
Show 21 quoted lines
> On Mon, Apr 26, 2010 at 2:08 PM, Alex Riesen <raa.lkml@gmail.com> wrote:
>> On Mon, Apr 26, 2010 at 19:56, Eugene Sajine <euguess@gmail.com> wrote:
>>> On Mon, Apr 26, 2010 at 1:54 PM, Eugene Sajine <euguess@gmail.com> wrote:
>>>> Hi,
>>>>
>>>> I'm wondering if there is a way to get some repo information if the
>>>> only thing i know is SHA-1 (no full path to the file)
>>>
>>> I mean of course the SHA-1 of the blob, i.e. the result of git
>>> hash-object some_file
>>
>> Try 'git cat-file -t <sha>'. It prints what object type is under the hash.
>> Than you can use either 'git cat-file <type> <sha>' (the most precise version,
>> which shall fail if anything is wrong), 'git cat-file -p <sha>' or plain
>> 'git show <sha>' (which always works, unless repo's broken).
>>
>
> Well thank you for that, but that gives me what i already know - the content.
> I need to "reverse engineer" the history by having only the content
> and its SHA-1 in order to indentify the last commitID and get all
> relevant info (author, date etc. )

Hmm. In that case, your using of git log look like a good way to get the information. If you wanted an output format which is better suited for automated parsing, you can look at "git rev-list --pretty=format:...", let it output only commit and tree hashes, and grep the output of "git ls-tree" for your hash...

Show 5 quoted lines
> This is all about avoiding using keyword expansion (thanks CVS for bad habbits)
>
> Thanks,
> Eugene
>
Alex Riesen· Apr 26, 2010, 19:37 UTC · re: Alex Riesen · lore

Re: "git log -- SHA-1" - how to get/simulate this functionality?

On Mon, Apr 26, 2010 at 20:57, Alex Riesen <raa.lkml@gmail.com> wrote:
Show 11 quoted lines
> On Mon, Apr 26, 2010 at 20:22, Eugene Sajine <euguess@gmail.com> wrote:
>> Well thank you for that, but that gives me what i already know - the content.
>> I need to "reverse engineer" the history by having only the content
>> and its SHA-1 in order to indentify the last commitID and get all
>> relevant info (author, date etc. )
>
> Hmm. In that case, your using of git log look like a good way to get the
> information. If you wanted an output format which is better suited for
> automated parsing, you can look at "git rev-list --pretty=format:...",
> let it output only commit and tree hashes, and grep the output of
> "git ls-tree" for your hash...

Actually, "git log --no-abbrev --raw --oneline" looks very simple to parse. By the way, you may get ambiguous results: any file any time could have had the given hash. IOW, by looking at the whole history you may see different files having the same hash. It does mean the same content, so maybe its OK for your purposes.

Eugene Sajine· Apr 26, 2010, 20:07 UTC · re: Alex Riesen · lore

Re: "git log -- SHA-1" - how to get/simulate this functionality?

On Mon, Apr 26, 2010 at 3:37 PM, Alex Riesen <raa.lkml@gmail.com> wrote:
Show 19 quoted lines
> On Mon, Apr 26, 2010 at 20:57, Alex Riesen <raa.lkml@gmail.com> wrote:
>> On Mon, Apr 26, 2010 at 20:22, Eugene Sajine <euguess@gmail.com> wrote:
>>> Well thank you for that, but that gives me what i already know - the content.
>>> I need to "reverse engineer" the history by having only the content
>>> and its SHA-1 in order to indentify the last commitID and get all
>>> relevant info (author, date etc. )
>>
>> Hmm. In that case, your using of git log look like a good way to get the
>> information. If you wanted an output format which is better suited for
>> automated parsing, you can look at "git rev-list --pretty=format:...",
>> let it output only commit and tree hashes, and grep the output of
>> "git ls-tree" for your hash...
>
> Actually, "git log --no-abbrev --raw --oneline" looks very simple to parse.
> By the way, you may get ambiguous results: any file any time could have
> had the given hash. IOW, by looking at the whole history you may see different
> files having the same hash. It does mean the same content, so maybe its
> OK for your purposes.
>

Yes this looks a bit better, but still not "grep"-able easily as grep SHA-1 cuts off the commit ID. I'm wondering why this "simple" feature which actually solves (almost) the problem of "keyword expansion" substitution for git once and for all is not yet there...:(

While i do realize that i might have ambiguous results and first of all because the same blob can exist in creation commits, modification and move, but mostly people are interested in the last commit of this particular file/blob. So, i would say, it is very rare that at any given point of time there will be two objects with the same hash in repo.

It is a feature request than:)

Please, somebody take a look at the possibility to develop a variant of "git log" or rev-list which would show the commits in which the given blob SHA-1 has participated.

Thanks, Eugene

← back to recent threads