diff: support regexp ignoresethomson/diff_regexp_ignore

Emulate git's `-I` (`--ignore-matching-lines`) with a regular
expression.

7 files changed, 133 insertions, 14 deletions

diff --git a/include/git2/diff.h b/include/git2/diff.h
index 9424ffd06..334614680 100644
--- a/include/git2/diff.h
+++ b/include/git2/diff.h
@@ -444,6 +444,13 @@ typedef struct {
 	 * Defaults to "b".
 	 */
 	const char *new_prefix;
+
+	/**
+	 * Ignore lines matching the given regular expression(s); both
+	 * the preimage and postimage lines must match.
+	 */
+	const char **ignore_regexp;
+	size_t ignore_regexp_count;
 } git_diff_options;
 
 /* The current version of the diff options structure */
diff --git a/src/diff_xdiff.c b/src/diff_xdiff.c
index 3f6eccac1..3425f33a3 100644
--- a/src/diff_xdiff.c
+++ b/src/diff_xdiff.c
@@ -230,9 +230,12 @@ static int git_xdiff(git_patch_generated_output *output, git_patch_generated *pa
 	return xo->output.error;
 }
 
-void git_xdiff_init(git_xdiff_output *xo, const git_diff_options *opts)
+int git_xdiff_init(git_xdiff_output *xo, const git_diff_options *opts)
 {
+	git_regexp **regexen;
 	uint32_t flags = opts ? opts->flags : 0;
+	int regexp_flags = GIT_REGEXP_EXTENDED | GIT_REGEXP_NEWLINE;
+	size_t i;
 
 	xo->output.diff_cb = git_xdiff;
 
@@ -256,5 +259,42 @@ void git_xdiff_init(git_xdiff_output *xo, const git_diff_options *opts)
 	if (flags & GIT_DIFF_IGNORE_BLANK_LINES)
 		xo->params.flags |= XDF_IGNORE_BLANK_LINES;
 
+	if (opts && opts->ignore_regexp_count) {
+		regexen = git__calloc(opts->ignore_regexp_count, sizeof(git_regexp *));
+		GIT_ERROR_CHECK_ALLOC(regexen);
+
+		xo->params.ignore_regex = regexen;
+		xo->params.ignore_regex_nr = opts->ignore_regexp_count;
+
+		for (i = 0; i < opts->ignore_regexp_count; i++) {
+			regexen[i] = git__malloc(sizeof(git_regexp));
+			GIT_ERROR_CHECK_ALLOC(regexen[i]);
+
+			if (git_regexp_compile(regexen[i],
+					opts->ignore_regexp[i],
+					regexp_flags) < 0) {
+				git_error_set(GIT_ERROR_INVALID, "could not compile regular expression");
+				git_xdiff_dispose(xo);
+				return -1;
+			}
+		}
+	}
+
 	xo->callback.out_line = git_xdiff_cb;
+
+	return 0;
+}
+
+void git_xdiff_dispose(git_xdiff_output *xo)
+{
+	size_t i;
+
+	for (i = 0; i < xo->params.ignore_regex_nr; i++) {
+		if (xo->params.ignore_regex[i]) {
+			git_regexp_dispose(xo->params.ignore_regex[i]);
+			git__free(xo->params.ignore_regex[i]);
+		}
+	}
+
+	git__free(xo->params.ignore_regex);
 }
diff --git a/src/diff_xdiff.h b/src/diff_xdiff.h
index 9b303e9dc..3e051d1e1 100644
--- a/src/diff_xdiff.h
+++ b/src/diff_xdiff.h
@@ -30,6 +30,7 @@ typedef struct {
 	xdemitcb_t   callback;
 } git_xdiff_output;
 
-void git_xdiff_init(git_xdiff_output *xo, const git_diff_options *opts);
+int git_xdiff_init(git_xdiff_output *xo, const git_diff_options *opts);
+void git_xdiff_dispose(git_xdiff_output *xo);
 
 #endif
diff --git a/src/patch_generate.c b/src/patch_generate.c
index bc598fea8..1d80c036e 100644
--- a/src/patch_generate.c
+++ b/src/patch_generate.c
@@ -540,13 +540,16 @@ static int diff_from_sources(
 	memset(&xo, 0, sizeof(xo));
 	diff_output_init(
 		&xo.output, opts, file_cb, binary_cb, hunk_cb, data_cb, payload);
-	git_xdiff_init(&xo, opts);
+
+	if ((error = git_xdiff_init(&xo, opts)) < 0)
+		return error;
 
 	memset(&pd, 0, sizeof(pd));
 
 	error = patch_generated_from_sources(&pd, &xo, oldsrc, newsrc, opts);
 
 	git_patch_free(&pd.patch.base);
+	git_xdiff_dispose(&xo);
 
 	return error;
 }
@@ -570,13 +573,16 @@ static int patch_from_sources(
 
 	memset(&xo, 0, sizeof(xo));
 	diff_output_to_patch(&xo.output, &pd->patch);
-	git_xdiff_init(&xo, opts);
+
+	if ((error = git_xdiff_init(&xo, opts)) < 0)
+		return error;
 
 	if (!(error = patch_generated_from_sources(pd, &xo, oldsrc, newsrc, opts)))
 		*out = (git_patch *)pd;
 	else
 		git_patch_free((git_patch *)pd);
 
+	git_xdiff_dispose(&xo);
 	return error;
 }
 
@@ -724,7 +730,9 @@ int git_patch_generated_from_diff(
 
 	memset(&xo, 0, sizeof(xo));
 	diff_output_to_patch(&xo.output, patch);
-	git_xdiff_init(&xo, &diff->opts);
+
+	if ((error = git_xdiff_init(&xo, &diff->opts)) < 0)
+		return error;
 
 	error = patch_generated_invoke_file_callback(patch, &xo.output);
 
@@ -741,6 +749,7 @@ int git_patch_generated_from_diff(
 	else
 		*patch_ptr = &patch->base;
 
+	git_xdiff_dispose(&xo);
 	return error;
 }
 
diff --git a/src/regexp.h b/src/regexp.h
index 61824bf5c..f4810c70b 100644
--- a/src/regexp.h
+++ b/src/regexp.h
@@ -62,7 +62,7 @@ int git_regexp_compile(git_regexp *r, const char *pattern, int flags);
 /**
  * Free memory associated with the regular expression
  *
- * @param r The regular expression structure to dispose.
+ e @param r The regular expression structure to dispose.
  */
 void git_regexp_dispose(git_regexp *r);
 
diff --git a/src/xdiff/git-xdiff.h b/src/xdiff/git-xdiff.h
index 0b4f0bc04..cdfb315bd 100644
--- a/src/xdiff/git-xdiff.h
+++ b/src/xdiff/git-xdiff.h
@@ -29,14 +29,9 @@ GIT_INLINE(int) xdl_regexec_buf(
 	const xdl_regex_t *preg, const char *buf, size_t size,
 	size_t nmatch, xdl_regmatch_t pmatch[], int eflags)
 {
-	GIT_UNUSED(preg);
-	GIT_UNUSED(buf);
-	GIT_UNUSED(size);
-	GIT_UNUSED(nmatch);
-	GIT_UNUSED(pmatch);
-	GIT_UNUSED(eflags);
-	GIT_ASSERT("not implemented");
-	return -1;
+	GIT_ASSERT(nmatch > 0 && pmatch && eflags == 0);
+
+	return git_regexp_search_n(preg, buf, size, nmatch, pmatch);
 }
 
 #endif
diff --git a/tests/diff/blob.c b/tests/diff/blob.c
index d2f42207d..88aeeebd3 100644
--- a/tests/diff/blob.c
+++ b/tests/diff/blob.c
@@ -1061,3 +1061,70 @@ void test_diff_blob__can_compare_buffer_to_buffer(void)
 		diff_file_cb, diff_binary_cb, diff_hunk_cb, diff_line_cb, &expected));
 	assert_one_modified(4, 9, 0, 5, 4, &expected);
 }
+
+#define EMPTY_DIFF "diff --git a/file b/file\n" \
+	"index 92dfa21..d5a68b2 100644\n" \
+	"--- a/file\n" \
+	"+++ b/file\n"
+
+#define NOT_EMPTY_DIFF \
+	"diff --git a/file b/file\n" \
+	"index 92dfa21..d5a68b2 100644\n" \
+	"--- a/file\n" \
+	"+++ b/file\n" \
+	"@@ -5 +5 @@ d\n" \
+	"-e\n" \
+	"+E\n"
+
+void test_diff_blob__patch_with_ignore_regexp(void)
+{
+	git_patch *p;
+	git_buf buf = GIT_BUF_INIT;
+	const char *a = "a\nb\nc\nd\ne\nf\ng\nh\ni\nj\n";
+	const char *b = "a\nb\nc\nd\nE\nf\ng\nh\ni\nj\n";
+	const char *ignore_lower = "e";
+	const char *ignore_upper = "E";
+	const char *ignore_both[] = { ignore_lower, ignore_upper };
+	const char *ignore_match = "^(e|E)$";
+
+	opts.interhunk_lines = 0;
+	opts.context_lines = 0;
+	opts.ignore_regexp = ignore_both;
+	opts.ignore_regexp_count = 2;
+
+	/* match both e and E */
+	cl_git_pass(git_patch_from_buffers(&p, a, strlen(a), NULL,
+		b, strlen(b), NULL, &opts));
+
+	cl_git_pass(git_patch_to_buf(&buf, p));
+	cl_assert_equal_s(EMPTY_DIFF, buf.ptr);
+
+	git_patch_free(p);
+	git_buf_dispose(&buf);
+
+	/* match ^(e|E)$ */
+	opts.ignore_regexp = &ignore_match;
+	opts.ignore_regexp_count = 1;
+
+	cl_git_pass(git_patch_from_buffers(&p, a, strlen(a), NULL,
+		b, strlen(b), NULL, &opts));
+
+	cl_git_pass(git_patch_to_buf(&buf, p));
+	cl_assert_equal_s(EMPTY_DIFF, buf.ptr);
+
+	git_patch_free(p);
+	git_buf_dispose(&buf);
+
+	/* matching only E does not match preimage */
+	opts.ignore_regexp = &ignore_upper;
+	opts.ignore_regexp_count = 1;
+
+	cl_git_pass(git_patch_from_buffers(&p, a, strlen(a), NULL,
+		b, strlen(b), NULL, &opts));
+
+	cl_git_pass(git_patch_to_buf(&buf, p));
+	cl_assert_equal_s(NOT_EMPTY_DIFF, buf.ptr);
+
+	git_patch_free(p);
+	git_buf_dispose(&buf);
+}