Merge pull request libgit2#4586 from emilio/mailmap

Add mailmap support.

Author: ethomson

include/git2.h

@@ -29,6 +29,7 @@
 #include "git2/ignore.h"
 #include "git2/index.h"
 #include "git2/indexer.h"
+#include "git2/mailmap.h"
 #include "git2/merge.h"
 #include "git2/message.h"
 #include "git2/net.h"

include/git2/blame.h

@@ -43,6 +43,10 @@ typedef enum {
 	/** Restrict the search of commits to those reachable following only the
 	 * first parents. */
 	GIT_BLAME_FIRST_PARENT = (1<<4),
+	/** Use mailmap file to map author and committer names and email addresses
+	 * to canonical real names and email addresses. The mailmap will be read
+	 * from the working directory, or HEAD in a bare repository. */
+	GIT_BLAME_USE_MAILMAP = (1<<5),
 } git_blame_flag_t;
 
 /**
@@ -108,6 +112,9 @@ GIT_EXTERN(int) git_blame_init_options(
  *   changed.
  * - `final_start_line_number` is the 1-based line number where this hunk
  *   begins, in the final version of the file
+ * - `final_signature` is the author of `final_commit_id`. If
+ *   `GIT_BLAME_USE_MAILMAP` has been specified, it will contain the canonical
+ *    real name and email address.
  * - `orig_commit_id` is the OID of the commit where this hunk was found.  This
  *   will usually be the same as `final_commit_id`, except when
  *   `GIT_BLAME_TRACK_COPIES_ANY_COMMIT_COPIES` has been specified.
@@ -116,6 +123,9 @@ GIT_EXTERN(int) git_blame_init_options(
  * - `orig_start_line_number` is the 1-based line number where this hunk begins
  *   in the file named by `orig_path` in the commit specified by
  *   `orig_commit_id`.
+ * - `orig_signature` is the author of `orig_commit_id`. If
+ *   `GIT_BLAME_USE_MAILMAP` has been specified, it will contain the canonical
+ *    real name and email address.
  * - `boundary` is 1 iff the hunk has been tracked to a boundary commit (the
  *   root, or the commit specified in git_blame_options.oldest_commit)
  */

include/git2/commit.h

@@ -172,6 +172,34 @@ GIT_EXTERN(const git_signature *) git_commit_committer(const git_commit *commit)
  */
 GIT_EXTERN(const git_signature *) git_commit_author(const git_commit *commit);
 
+/**
+ * Get the committer of a commit, using the mailmap to map names and email
+ * addresses to canonical real names and email addresses.
+ *
+ * Call `git_signature_free` to free the signature.
+ *
+ * @param out a pointer to store the resolved signature.
+ * @param commit a previously loaded commit.
+ * @param mailmap the mailmap to resolve with. (may be NULL)
+ * @return 0 or an error code
+ */
+GIT_EXTERN(int) git_commit_committer_with_mailmap(
+	git_signature **out, const git_commit *commit, const git_mailmap *mailmap);
+
+/**
+ * Get the author of a commit, using the mailmap to map names and email
+ * addresses to canonical real names and email addresses.
+ *
+ * Call `git_signature_free` to free the signature.
+ *
+ * @param out a pointer to store the resolved signature.
+ * @param commit a previously loaded commit.
+ * @param mailmap the mailmap to resolve with. (may be NULL)
+ * @return 0 or an error code
+ */
+GIT_EXTERN(int) git_commit_author_with_mailmap(
+	git_signature **out, const git_commit *commit, const git_mailmap *mailmap);
+
 /**
  * Get the full raw text of the commit header.
  *

include/git2/mailmap.h

@@ -0,0 +1,115 @@
+/*
+ * Copyright (C) the libgit2 contributors. All rights reserved.
+ *
+ * This file is part of libgit2, distributed under the GNU GPL v2 with
+ * a Linking Exception. For full terms see the included COPYING file.
+ */
+#ifndef INCLUDE_git_mailmap_h__
+#define INCLUDE_git_mailmap_h__
+
+#include "common.h"
+#include "types.h"
+#include "buffer.h"
+
+/**
+ * @file git2/mailmap.h
+ * @brief Mailmap parsing routines
+ * @defgroup git_mailmap Git mailmap routines
+ * @ingroup Git
+ * @{
+ */
+GIT_BEGIN_DECL
+
+/**
+ * Allocate a new mailmap object.
+ *
+ * This object is empty, so you'll have to add a mailmap file before you can do
+ * anything with it. The mailmap must be freed with 'git_mailmap_free'.
+ *
+ * @param out pointer to store the new mailmap
+ * @return 0 on success, or an error code
+ */
+GIT_EXTERN(int) git_mailmap_new(git_mailmap **out);
+
+/**
+ * Free the mailmap and its associated memory.
+ *
+ * @param mm the mailmap to free
+ */
+GIT_EXTERN(void) git_mailmap_free(git_mailmap *mm);
+
+/**
+ * Add a single entry to the given mailmap object. If the entry already exists,
+ * it will be replaced with the new entry.
+ *
+ * @param mm mailmap to add the entry to
+ * @param real_name the real name to use, or NULL
+ * @param real_email the real email to use, or NULL
+ * @param replace_name the name to replace, or NULL
+ * @param replace_email the email to replace
+ * @return 0 on success, or an error code
+ */
+GIT_EXTERN(int) git_mailmap_add_entry(
+	git_mailmap *mm, const char *real_name, const char *real_email,
+	const char *replace_name, const char *replace_email);
+
+/**
+ * Create a new mailmap instance containing a single mailmap file
+ *
+ * @param out pointer to store the new mailmap
+ * @param buf buffer to parse the mailmap from
+ * @param len the length of the input buffer
+ * @return 0 on success, or an error code
+ */
+GIT_EXTERN(int) git_mailmap_from_buffer(
+	git_mailmap **out, const char *buf, size_t len);
+
+/**
+ * Create a new mailmap instance from a repository, loading mailmap files based
+ * on the repository's configuration.
+ *
+ * Mailmaps are loaded in the following order:
+ *  1. '.mailmap' in the root of the repository's working directory, if present.
+ *  2. The blob object identified by the 'mailmap.blob' config entry, if set.
+ * 	   [NOTE: 'mailmap.blob' defaults to 'HEAD:.mailmap' in bare repositories]
+ *  3. The path in the 'mailmap.file' config entry, if set.
+ *
+ * @param out pointer to store the new mailmap
+ * @param repo repository to load mailmap information from
+ * @return 0 on success, or an error code
+ */
+GIT_EXTERN(int) git_mailmap_from_repository(
+	git_mailmap **out, git_repository *repo);
+
+/**
+ * Resolve a name and email to the corresponding real name and email.
+ *
+ * The lifetime of the strings are tied to `mm`, `name`, and `email` parameters.
+ *
+ * @param real_name pointer to store the real name
+ * @param real_email pointer to store the real email
+ * @param mm the mailmap to perform a lookup with (may be NULL)
+ * @param name the name to look up
+ * @param email the email to look up
+ * @return 0 on success, or an error code
+ */
+GIT_EXTERN(int) git_mailmap_resolve(
+	const char **real_name, const char **real_email,
+	const git_mailmap *mm, const char *name, const char *email);
+
+/**
+ * Resolve a signature to use real names and emails with a mailmap.
+ *
+ * Call `git_signature_free()` to free the data.
+ *
+ * @param out new signature
+ * @param mm mailmap to resolve with
+ * @param sig signature to resolve
+ * @return 0 or an error code
+ */
+GIT_EXTERN(int) git_mailmap_resolve_signature(
+	git_signature **out, const git_mailmap *mm, const git_signature *sig);
+
+/** @} */
+GIT_END_DECL
+#endif

include/git2/types.h

@@ -434,6 +434,9 @@ struct git_writestream {
 	void (*free)(git_writestream *stream);
 };
 
+/** Representation of .mailmap file state. */
+typedef struct git_mailmap git_mailmap;
+
 /** @} */
 GIT_END_DECL
 

src/blame.c

@@ -14,6 +14,7 @@
 #include "git2/diff.h"
 #include "git2/blob.h"
 #include "git2/signature.h"
+#include "git2/mailmap.h"
 #include "util.h"
 #include "repository.h"
 #include "blame_git.h"
@@ -132,6 +133,9 @@ git_blame* git_blame__alloc(
 		return NULL;
 	}
 
+	if (opts.flags & GIT_BLAME_USE_MAILMAP)
+		git_mailmap_from_repository(&gbr->mailmap, repo);
+
 	return gbr;
 }
 
@@ -150,6 +154,8 @@ void git_blame_free(git_blame *blame)
 
 	git_array_clear(blame->line_index);
 
+	git_mailmap_free(blame->mailmap);
+
 	git__free(blame->path);
 	git_blob_free(blame->final_blob);
 	git__free(blame);
@@ -279,7 +285,7 @@ static int index_blob_lines(git_blame *blame)
     return blame->num_lines;
 }
 
-static git_blame_hunk* hunk_from_entry(git_blame__entry *e)
+static git_blame_hunk* hunk_from_entry(git_blame__entry *e, git_blame *blame)
 {
 	git_blame_hunk *h = new_hunk(
 			e->lno+1, e->num_lines, e->s_lno+1, e->suspect->path);
@@ -289,8 +295,9 @@ static git_blame_hunk* hunk_from_entry(git_blame__entry *e)
 
 	git_oid_cpy(&h->final_commit_id, git_commit_id(e->suspect->commit));
 	git_oid_cpy(&h->orig_commit_id, git_commit_id(e->suspect->commit));
-	git_signature_dup(&h->final_signature, git_commit_author(e->suspect->commit));
-	git_signature_dup(&h->orig_signature, git_commit_author(e->suspect->commit));
+	git_commit_author_with_mailmap(
+		&h->final_signature, e->suspect->commit, blame->mailmap);
+	git_signature_dup(&h->orig_signature, h->final_signature);
 	h->boundary = e->is_boundary ? 1 : 0;
 	return h;
 }
@@ -341,7 +348,7 @@ static int blame_internal(git_blame *blame)
 cleanup:
 	for (ent = blame->ent; ent; ) {
 		git_blame__entry *e = ent->next;
-		git_blame_hunk *h = hunk_from_entry(ent);
+		git_blame_hunk *h = hunk_from_entry(ent, blame);
 
 		git_vector_insert(&blame->hunks, h);
 

src/blame.h

@@ -67,6 +67,7 @@ typedef struct git_blame__entry {
 struct git_blame {
 	char *path;
 	git_repository *repository;
+	git_mailmap *mailmap;
 	git_blame_options options;
 
 	git_vector hunks;

src/commit.c

@@ -11,6 +11,7 @@
 #include "git2/object.h"
 #include "git2/repository.h"
 #include "git2/signature.h"
+#include "git2/mailmap.h"
 #include "git2/sys/commit.h"
 
 #include "odb.h"
@@ -889,3 +890,15 @@ int git_commit_create_with_signature(
 	git_buf_dispose(&commit);
 	return error;
 }
+
+int git_commit_committer_with_mailmap(
+	git_signature **out, const git_commit *commit, const git_mailmap *mailmap)
+{
+	return git_mailmap_resolve_signature(out, mailmap, commit->committer);
+}
+
+int git_commit_author_with_mailmap(
+	git_signature **out, const git_commit *commit, const git_mailmap *mailmap)
+{
+	return git_mailmap_resolve_signature(out, mailmap, commit->author);
+}