aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* ui-plain: fix resource leak: free before assigning NULLChristian Hesse2015-10-091-1/+3
| | | | | Coverity-id: 13939 Signed-off-by: Christian Hesse <[email protected]>
* ui-plain: fix resource leak: free before returnChristian Hesse2015-10-091-0/+1
| | | | | Coverity-id: 13940 Signed-off-by: Christian Hesse <[email protected]>
* ui-repolist: fix resource leak: free allocation from cgit_currenturlChristian Hesse2015-10-091-1/+3
| | | | | Coverity-id: 13930 Signed-off-by: Christian Hesse <[email protected]>
* ui-repolist: fix resource leak: free before returnChristian Hesse2015-10-091-1/+3
| | | | | Coverity-id: 13931 Signed-off-by: Christian Hesse <[email protected]>
* filters: Simplify convertersJason A. Donenfeld2015-10-094-1734/+284
| | | | Signed-off-by: Jason A. Donenfeld <[email protected]>
* ui-shared: fix resource leak: free allocation from cgit_hosturlChristian Hesse2015-10-091-2/+3
| | | | Signed-off-by: Christian Hesse <[email protected]>
* ui-shared: return value of cgit_hosturl is not constChristian Hesse2015-10-092-4/+4
| | | | Signed-off-by: Christian Hesse <[email protected]>
* cmd: fix resource leak: free allocation from cgit_currenturl and fmtallocChristian Hesse2015-10-091-3/+7
| | | | Signed-off-by: Christian Hesse <[email protected]>
* ui-shared: fix resource leak: free allocation from cgit_currenturlChristian Hesse2015-10-091-3/+8
| | | | | Coverity-id: 13927 Signed-off-by: Christian Hesse <[email protected]>
* ui-shared: return value of cgit_currenturl is not constChristian Hesse2015-10-092-3/+3
| | | | Signed-off-by: Christian Hesse <[email protected]>
* ui-shared: fix resource leak: free allocation from cgit_fileurlChristian Hesse2015-10-091-5/+11
| | | | | Coverity-id: 13918 Signed-off-by: Christian Hesse <[email protected]>
* ui-ssdiff: fix resource leak: free allocation from cgit_fileurlChristian Hesse2015-10-091-2/+6
| | | | | Coverity-id: 13929 Signed-off-by: Christian Hesse <[email protected]>
* ui-tree: fix resource leak: free before returnChristian Hesse2015-10-091-0/+1
| | | | | Coverity-id: 13938 Signed-off-by: Christian Hesse <[email protected]>
* Avoid use of non-reentrant functionsJason A. Donenfeld2015-10-091-3/+3
| | | | Signed-off-by: Jason A. Donenfeld <[email protected]>
* Makefile: fix MAKEFLAGS tests with multiple flagsJohn Keeping2015-10-091-1/+1
| | | | | | | | findstring is defined as $(findstring FIND,IN) so if multiple flags are set these tests do the wrong thing unless $(MAKEFLAGS) is the second argument. Signed-off-by: John Keeping <[email protected]>
* ui-refs: remove useless null checkJohn Keeping2015-10-091-1/+1
| | | | | | | There is no way that "tag" can be null here. Coverity-id: 13950 Signed-off-by: John Keeping <[email protected]>
* ui-blob: remove useless null checkJohn Keeping2015-10-091-1/+1
| | | | | | | | We have already called strlen() on "path" by the time we get here, so we know it can't be null. Coverity-id: 13954 Signed-off-by: John Keeping <[email protected]>
* scan-tree: remove useless strdup()John Keeping2015-10-091-1/+1
| | | | | | | | | parse_configfile() takes a "const char *" and doesn't hold any references to it after it returns; there is no reason to pass it a duplicate. Coverity-id: 13941 Signed-off-by: John Keeping <[email protected]>
* cgit.c: remove useless null checkJohn Keeping2015-10-091-1/+1
| | | | | | | | Everywhere else in this function we do not check whether the value is null and parse_configfile() never passes a null value to this callback. Coverity-id: 13846 Signed-off-by: John Keeping <[email protected]>
* git: update to v2.6.1Christian Hesse2015-10-062-1/+1
| | | | | | Update to git version v2.6.1, no changes required. Signed-off-by: Christian Hesse <[email protected]>
* mime: rewrite detection functionJason A. Donenfeld2015-08-171-36/+26
| | | | Signed-off-by: Jason A. Donenfeld <[email protected]>
* ui-summary: send images plain for about pageChristian Hesse2015-08-171-2/+13
| | | | | | | | | The about page used to display just fine, but images were broken: The binary image data was embedded in html code. Use cgit_print_plain() to send images in plain mode and make them available on about page. Signed-off-by: Christian Hesse <[email protected]>
* refactor get_mimetype_from_file() to get_mimetype_for_filename()Christian Hesse2015-08-173-51/+44
| | | | | | | * handle mimetype within a single function * return allocated memory on success Signed-off-by: Christian Hesse <[email protected]>
* move get_mimetype_from_file() to sharedChristian Hesse2015-08-173-40/+42
| | | | Signed-off-by: Christian Hesse <[email protected]>
* cmd: fix command definitionJohn Keeping2015-08-141-1/+1
| | | | | | | | The previous commit removed the "pre" field from "struct cgit_cmd" but forgot to update this macro. Signed-off-by: John Keeping <[email protected]> Reviewed-by: Christian Hesse <[email protected]>
* cmd: no need for pre function hook nowJason A. Donenfeld2015-08-143-20/+9
| | | | Signed-off-by: Jason A. Donenfeld <[email protected]>
* ui-shared: cache errors for "dynamic TTL"John Keeping2015-08-141-0/+1
| | | | | | | Most errors we generate are (potentially) transient, such as non-existent object IDs so we don't want them to be cached forever. Signed-off-by: John Keeping <[email protected]>
* cmd: remove "want_layout" fieldJohn Keeping2015-08-143-34/+24
| | | | | | No commands use this any more. Signed-off-by: John Keeping <[email protected]>
* tree: move layout into page functionJohn Keeping2015-08-142-6/+19
| | | | | | | | This also allows us to return proper HTTP error codes when the requested tree is not found and display an error message in one case (invalid path inside valid commit) where we previously just displayed an empty page. Signed-off-by: John Keeping <[email protected]>
* tag: move layout into page functionJohn Keeping2015-08-142-4/+11
| | | | | | | This also allows us to return proper HTTP error codes when something goes wrong. Signed-off-by: John Keeping <[email protected]>
* summary: move layout into page functionJohn Keeping2015-08-142-1/+3
| | | | Signed-off-by: John Keeping <[email protected]>
* stats: move layout into page functionJohn Keeping2015-08-142-3/+7
| | | | | | | This also allows us to return proper HTTP error codes for invalid requests. Signed-off-by: John Keeping <[email protected]>
* refs: move layout to page functionJohn Keeping2015-08-142-2/+3
| | | | Signed-off-by: John Keeping <[email protected]>
* log: move layout into page functionJohn Keeping2015-08-142-2/+5
| | | | Signed-off-by: John Keeping <[email protected]>
* diff: move layout to page functionJohn Keeping2015-08-142-6/+15
| | | | | | | The existing "show_ctrls" flag is used to control whether we are running in an existing page or control the page ourselves. Signed-off-by: John Keeping <[email protected]>
* commit: move layout into page functionJohn Keeping2015-08-142-3/+7
| | | | | | | This allows us to return a proper HTTP status code when an object is not found by switching from cgit_print_error() to cgit_print_error_page(). Signed-off-by: John Keeping <[email protected]>
* about: move layout into page functionsJohn Keeping2015-08-143-4/+11
| | | | Signed-off-by: John Keeping <[email protected]>
* ui-shared: add cgit_print_layout_{start,end}()John Keeping2015-08-142-0/+15
| | | | | | | | These will avoid needing to call three functions to start page layout in subsequent patches when we move the layout setup into each individual page. Signed-off-by: John Keeping <[email protected]>
* html: remove html_status()John Keeping2015-08-142-8/+0
| | | | | | This is now unused. Signed-off-by: John Keeping <[email protected]>
* snapshot: don't reimplement cgit_print_error_page()John Keeping2015-08-141-17/+4
| | | | | | | | | | cgit_print_error_page() has the advantage that it sets a suitable HTTP status code for the response. Note that setting "mimetype" is redundant here since it cannot have changed since being initialized in cgit.c::prepare_context(), so we do not need to worry that cgit_print_error_page() does not set it. Signed-off-by: John Keeping <[email protected]>
* snapshot: use cgit_print_error_page() for HTTP status codesJohn Keeping2015-08-141-2/+4
| | | | | | | | This is a bugfix as well as an improvement to the HTTP status code handling since previously we would not print HTTP headers on any of these code paths. Signed-off-by: John Keeping <[email protected]>
* patch: use cgit_print_error_page() for HTTP status codesJohn Keeping2015-08-141-8/+8
| | | | Signed-off-by: John Keeping <[email protected]>
* blob: use cgit_print_error_page() to add HTTP headersJohn Keeping2015-08-141-4/+8
| | | | | | | | This is a bugfix as well as an improvement to the HTTP status code handling since previously we would not print HTTP headers on any of these code paths. Signed-off-by: John Keeping <[email protected]>
* snapshot: use cgit_print_error_page() instead of html_status()John Keeping2015-08-141-1/+1
| | | | | | | This provides a formatted error response rather than a simple HTTP error. Signed-off-by: John Keeping <[email protected]>
* plain: use cgit_print_error_page() instead of html_status()John Keeping2015-08-141-5/+5
| | | | | | | This provides a formatted error response rather than a simple HTTP error. Signed-off-by: John Keeping <[email protected]>
* clone: use cgit_print_error_page() instead of html_status()John Keeping2015-08-141-5/+5
| | | | | | | This provides a formatted error response rather than a simple HTTP error. Signed-off-by: John Keeping <[email protected]>
* cgit: use cgit_print_error_page() where appropriateJohn Keeping2015-08-141-20/+7
| | | | | | | | These are more-or-less one-to-one translations but in the final hunk we gain an HTTP error code where we used to send "200 OK", which is an improvement. Signed-off-by: John Keeping <[email protected]>
* ui-shared: add cgit_print_error_page() functionJohn Keeping2015-08-142-0/+16
| | | | | | | This will allow us to generate error responses with the correct HTTP response code without needing all of the layout boilerplate. Signed-off-by: John Keeping <[email protected]>
* ui-patch: make sure to send http headersChristian Hesse2015-08-141-0/+4
| | | | | | | | | | Requesting a text/plain patch with bad commit id made cgit send text without proper http headers. This results in "500 Internal Server Error" with "Premature end of script headers" in server logs. So print http headers before error message and return. Signed-off-by: Christian Hesse <[email protected]> Reviewed-by: John Keeping <[email protected]>
* Makefile: make "git/config.mak.uname" inclusion optionalJohn Keeping2015-08-131-1/+1
| | | | | | | If we haven't got a "git" directory, it should still be possible to run "make get-git", so we cannot include this file unconditionally. Signed-off-by: John Keeping <[email protected]>
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-03-11 05:05:23 +0000