log: allow users to follow a file

Teach the "log" UI to behave in the same way as "git log --follow", when
given a suitable instruction by the user.  The default behaviour remains
to show the log without following renames, but the follow behaviour can
be activated by following a link in the page header.

Follow is not the default because outputting merges in follow mode is
tricky ("git log --follow" will not show merges).  We also disable the
graph in follow mode because the commit graph is not simplified so we
end up with frequent gaps in the graph and many lines that do not
connect with any commits we're actually showing.

We also teach the "diff" and "commit" UIs to respect the follow flag on
URLs, causing the single-file version of these UIs to detect renames.
This feature is needed only for commits that rename the path we're
interested in.

For commits before the file has been renamed (i.e. that appear later in
the log list) we change the file path in the links from the log to point
to the old name; this means that links to commits always limit by the
path known to that commit.  If we didn't do this we would need to walk
down the log diff'ing every commit whenever we want to show a commit.
The drawback is that the "Log" link in the top bar of such a page links
to the log limited by the old name, so it will only show pre-rename
commits.  I consider this a reasonable trade-off since the "Back" button
still works and the log matches the path displayed in the top bar.

Since following renames requires running diff on every commit we
consider, I've added a knob to the configuration file to globally
enable/disable this feature.  Note that we may consider a large number
of commits the revision walking machinery no longer performs any path
limitation so we have to examine every commit until we find a page full
of commits that affect the target path or something related to it.

Suggested-by: René Neumann <necoro@necoro.eu>
Signed-off-by: John Keeping <john@keeping.me.uk>

1 files changed, 120 insertions, 11 deletions

@@ -12,7 +12,7 @@
 #include "ui-shared.h"
 #include "argv-array.h"
 
-static int files, add_lines, rem_lines;
+static int files, add_lines, rem_lines, lines_counted;
 
 /*
  * The list of available column colors in the commit graph.
@@ -67,7 +67,7 @@ void show_commit_decorations(struct commit *commit)
                         strncpy(buf, deco->name + 11, sizeof(buf) - 1);
                         cgit_log_link(buf, NULL, "branch-deco", buf, NULL,
                                       ctx.qry.vpath, 0, NULL, NULL,
-                                      ctx.qry.showmsg);
+                                      ctx.qry.showmsg, 0);
                 }
                 else if (starts_with(deco->name, "tag: refs/tags/")) {
                         strncpy(buf, deco->name + 15, sizeof(buf) - 1);
@@ -84,7 +84,7 @@ void show_commit_decorations(struct commit *commit)
                         cgit_log_link(buf, NULL, "remote-deco", NULL,
                                       sha1_to_hex(commit->object.sha1),
                                       ctx.qry.vpath, 0, NULL, NULL,
-                                      ctx.qry.showmsg);
+                                      ctx.qry.showmsg, 0);
                 }
                 else {
                         strncpy(buf, deco->name, sizeof(buf) - 1);
@@ -98,6 +98,74 @@ next:
         html("</span>");
 }
 
+static void handle_rename(struct diff_filepair *pair)
+{
+        /*
+         * After we have seen a rename, we generate links to the previous
+         * name of the file so that commit & diff views get fed the path
+         * that is correct for the commit they are showing, avoiding the
+         * need to walk the entire history leading back to every commit we
+         * show in order detect renames.
+         */
+        if (0 != strcmp(ctx.qry.vpath, pair->two->path)) {
+                free(ctx.qry.vpath);
+                ctx.qry.vpath = xstrdup(pair->two->path);
+        }
+        inspect_files(pair);
+}
+
+static int show_commit(struct commit *commit, struct rev_info *revs)
+{
+        struct commit_list *parents = commit->parents;
+        struct commit *parent;
+        int found = 0, saved_fmt;
+        unsigned saved_flags = revs->diffopt.flags;
+
+
+        /* Always show if we're not in "follow" mode with a single file. */
+        if (!ctx.qry.follow)
+                return 1;
+
+        /*
+         * In "follow" mode, we don't show merges.  This is consistent with
+         * "git log --follow -- <file>".
+         */
+        if (parents && parents->next)
+                return 0;
+
+        /*
+         * If this is the root commit, do what rev_info tells us.
+         */
+        if (!parents)
+                return revs->show_root_diff;
+
+        /* When we get here we have precisely one parent. */
+        parent = parents->item;
+        parse_commit(parent);
+
+        files = 0;
+        add_lines = 0;
+        rem_lines = 0;
+
+        DIFF_OPT_SET(&revs->diffopt, RECURSIVE);
+        diff_tree_sha1(parent->tree->object.sha1,
+                       commit->tree->object.sha1,
+                       "", &revs->diffopt);
+        diffcore_std(&revs->diffopt);
+
+        found = !diff_queue_is_empty();
+        saved_fmt = revs->diffopt.output_format;
+        revs->diffopt.output_format = DIFF_FORMAT_CALLBACK;
+        revs->diffopt.format_callback = cgit_diff_tree_cb;
+        revs->diffopt.format_callback_data = handle_rename;
+        diff_flush(&revs->diffopt);
+        revs->diffopt.output_format = saved_fmt;
+        revs->diffopt.flags = saved_flags;
+
+        lines_counted = 1;
+        return found;
+}
+
 static void print_commit(struct commit *commit, struct rev_info *revs)
 {
         struct commitinfo *info;
@@ -177,7 +245,8 @@ static void print_commit(struct commit *commit, struct rev_info *revs)
                 cgit_print_age(commit->date, TM_WEEK * 2, FMT_SHORTDATE);
         }
 
-        if (ctx.repo->enable_log_filecount || ctx.repo->enable_log_linecount) {
+        if (!lines_counted && (ctx.repo->enable_log_filecount ||
+                               ctx.repo->enable_log_linecount)) {
                 files = 0;
                 add_lines = 0;
                 rem_lines = 0;
@@ -325,7 +394,17 @@ void cgit_print_log(const char *tip, int ofs, int cnt, char *grep, char *pattern
                         }
                 }
         }
-        if (commit_graph) {
+
+        if (!path || !ctx.cfg.enable_follow_links) {
+                /*
+                 * If we don't have a path, "follow" is a no-op so make sure
+                 * the variable is set to false to avoid needing to check
+                 * both this and whether we have a path everywhere.
+                 */
+                ctx.qry.follow = 0;
+        }
+
+        if (commit_graph && !ctx.qry.follow) {
                 argv_array_push(&rev_argv, "--graph");
                 argv_array_push(&rev_argv, "--color");
                 graph_set_column_colors(column_colors_html,
@@ -337,6 +416,8 @@ void cgit_print_log(const char *tip, int ofs, int cnt, char *grep, char *pattern
         else if (commit_sort == 2)
                 argv_array_push(&rev_argv, "--topo-order");
 
+        if (path && ctx.qry.follow)
+                argv_array_push(&rev_argv, "--follow");
         argv_array_push(&rev_argv, "--");
         if (path)
                 argv_array_push(&rev_argv, path);
@@ -347,10 +428,17 @@ void cgit_print_log(const char *tip, int ofs, int cnt, char *grep, char *pattern
         rev.verbose_header = 1;
         rev.show_root_diff = 0;
         rev.ignore_missing = 1;
+        rev.simplify_history = 1;
         setup_revisions(rev_argv.argc, rev_argv.argv, &rev, NULL);
         load_ref_decorations(DECORATE_FULL_REFS);
         rev.show_decorations = 1;
         rev.grep_filter.regflags |= REG_ICASE;
+
+        rev.diffopt.detect_rename = 1;
+        rev.diffopt.rename_limit = ctx.cfg.renamelimit;
+        if (ctx.qry.ignorews)
+                DIFF_XDL_SET(&rev.diffopt, IGNORE_WHITESPACE);
+
         compile_grep_patterns(&rev.grep_filter);
         prepare_revision_walk(&rev);
 
@@ -368,11 +456,12 @@ void cgit_print_log(const char *tip, int ofs, int cnt, char *grep, char *pattern
                 cgit_log_link(ctx.qry.showmsg ? "Collapse" : "Expand", NULL,
                               NULL, ctx.qry.head, ctx.qry.sha1,
                               ctx.qry.vpath, ctx.qry.ofs, ctx.qry.grep,
-                              ctx.qry.search, ctx.qry.showmsg ? 0 : 1);
+                              ctx.qry.search, ctx.qry.showmsg ? 0 : 1,
+                              ctx.qry.follow);
                 html(")");
         }
         html("</th><th class='left'>Author</th>");
-        if (commit_graph)
+        if (rev.graph)
                 html("<th class='left'>Age</th>");
         if (ctx.repo->enable_log_filecount) {
                 html("<th class='left'>Files</th>");
@@ -388,13 +477,30 @@ void cgit_print_log(const char *tip, int ofs, int cnt, char *grep, char *pattern
                 ofs = 0;
 
         for (i = 0; i < ofs && (commit = get_revision(&rev)) != NULL; i++) {
+                if (show_commit(commit, &rev))
+                        i++;
                 free_commit_buffer(commit);
                 free_commit_list(commit->parents);
                 commit->parents = NULL;
         }
 
         for (i = 0; i < cnt && (commit = get_revision(&rev)) != NULL; i++) {
-                print_commit(commit, &rev);
+                /*
+                 * In "follow" mode, we must count the files and lines the
+                 * first time we invoke diff on a given commit, and we need
+                 * to do that to see if the commit touches the path we care
+                 * about, so we do it in show_commit.  Hence we must clear
+                 * lines_counted here.
+                 *
+                 * This has the side effect of avoiding running diff twice
+                 * when we are both following renames and showing file
+                 * and/or line counts.
+                 */
+                lines_counted = 0;
+                if (show_commit(commit, &rev)) {
+                        i++;
+                        print_commit(commit, &rev);
+                }
                 free_commit_buffer(commit);
                 free_commit_list(commit->parents);
                 commit->parents = NULL;
@@ -406,7 +512,8 @@ void cgit_print_log(const char *tip, int ofs, int cnt, char *grep, char *pattern
                         cgit_log_link("[prev]", NULL, NULL, ctx.qry.head,
                                       ctx.qry.sha1, ctx.qry.vpath,
                                       ofs - cnt, ctx.qry.grep,
-                                      ctx.qry.search, ctx.qry.showmsg);
+                                      ctx.qry.search, ctx.qry.showmsg,
+                                      ctx.qry.follow);
                         html("</li>");
                 }
                 if ((commit = get_revision(&rev)) != NULL) {
@@ -414,14 +521,16 @@ void cgit_print_log(const char *tip, int ofs, int cnt, char *grep, char *pattern
                         cgit_log_link("[next]", NULL, NULL, ctx.qry.head,
                                       ctx.qry.sha1, ctx.qry.vpath,
                                       ofs + cnt, ctx.qry.grep,
-                                      ctx.qry.search, ctx.qry.showmsg);
+                                      ctx.qry.search, ctx.qry.showmsg,
+                                      ctx.qry.follow);
                         html("</li>");
                 }
                 html("</ul>");
         } else if ((commit = get_revision(&rev)) != NULL) {
                 htmlf("<tr class='nohover'><td colspan='%d'>", columns);
                 cgit_log_link("[...]", NULL, NULL, ctx.qry.head, NULL,
-                              ctx.qry.vpath, 0, NULL, NULL, ctx.qry.showmsg);
+                              ctx.qry.vpath, 0, NULL, NULL, ctx.qry.showmsg,
+                              ctx.qry.follow);
                 html("</td></tr>\n");
         }