[libgit2.git] / src / diff_driver.c

/*
 * 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.
 */

#include "diff_driver.h"

#include "git2/attr.h"

#include "common.h"
#include "diff.h"
#include "strmap.h"
#include "map.h"
#include "buf_text.h"
#include "config.h"
#include "regexp.h"
#include "repository.h"

typedef enum {
	DIFF_DRIVER_AUTO = 0,
	DIFF_DRIVER_BINARY = 1,
	DIFF_DRIVER_TEXT = 2,
	DIFF_DRIVER_PATTERNLIST = 3,
} git_diff_driver_t;

typedef struct {
	git_regexp re;
	int flags;
} git_diff_driver_pattern;

enum {
	REG_NEGATE = (1 << 15) /* get out of the way of existing flags */
};

/* data for finding function context for a given file type */
struct git_diff_driver {
	git_diff_driver_t type;
	uint32_t binary_flags;
	uint32_t other_flags;
	git_array_t(git_diff_driver_pattern) fn_patterns;
	git_regexp  word_pattern;
	char name[GIT_FLEX_ARRAY];
};

#include "userdiff.h"

struct git_diff_driver_registry {
	git_strmap *drivers;
};

#define FORCE_DIFFABLE (GIT_DIFF_FORCE_TEXT | GIT_DIFF_FORCE_BINARY)

static git_diff_driver global_drivers[3] = {
	{ DIFF_DRIVER_AUTO,   0, 0, },
	{ DIFF_DRIVER_BINARY, GIT_DIFF_FORCE_BINARY, 0 },
	{ DIFF_DRIVER_TEXT,   GIT_DIFF_FORCE_TEXT, 0 },
};

git_diff_driver_registry *git_diff_driver_registry_new(void)
{
	git_diff_driver_registry *reg =
		git__calloc(1, sizeof(git_diff_driver_registry));
	if (!reg)
		return NULL;

	if (git_strmap_new(&reg->drivers) < 0) {
		git_diff_driver_registry_free(reg);
		return NULL;
	}

	return reg;
}

void git_diff_driver_registry_free(git_diff_driver_registry *reg)
{
	git_diff_driver *drv;

	if (!reg)
		return;

	git_strmap_foreach_value(reg->drivers, drv, git_diff_driver_free(drv));
	git_strmap_free(reg->drivers);
	git__free(reg);
}

static int diff_driver_add_patterns(
	git_diff_driver *drv, const char *regex_str, int regex_flags)
{
	int error = 0;
	const char *scan, *end;
	git_diff_driver_pattern *pat = NULL;
	git_buf buf = GIT_BUF_INIT;

	for (scan = regex_str; scan; scan = end) {
		/* get pattern to fill in */
		if ((pat = git_array_alloc(drv->fn_patterns)) == NULL) {
			return -1;
		}

		pat->flags = regex_flags;
		if (*scan == '!') {
			pat->flags |= REG_NEGATE;
			++scan;
		}

		if ((end = strchr(scan, '\n')) != NULL) {
			error = git_buf_set(&buf, scan, end - scan);
			end++;
		} else {
			error = git_buf_sets(&buf, scan);
		}
		if (error < 0)
			break;

		if ((error = git_regexp_compile(&pat->re, buf.ptr, regex_flags)) != 0) {
			/*
			 * TODO: issue a warning
			 */
		}
	}

	if (error && pat != NULL)
		(void)git_array_pop(drv->fn_patterns); /* release last item */
	git_buf_dispose(&buf);

	/* We want to ignore bad patterns, so return success regardless */
	return 0;
}

static int diff_driver_xfuncname(const git_config_entry *entry, void *payload)
{
	return diff_driver_add_patterns(payload, entry->value, 0);
}

static int diff_driver_funcname(const git_config_entry *entry, void *payload)
{
	return diff_driver_add_patterns(payload, entry->value, 0);
}

static git_diff_driver_registry *git_repository_driver_registry(
	git_repository *repo)
{
	if (!repo->diff_drivers) {
		git_diff_driver_registry *reg = git_diff_driver_registry_new();
		reg = git__compare_and_swap(&repo->diff_drivers, NULL, reg);

		if (reg != NULL) /* if we race, free losing allocation */
			git_diff_driver_registry_free(reg);
	}

	if (!repo->diff_drivers)
		git_error_set(GIT_ERROR_REPOSITORY, "unable to create diff driver registry");

	return repo->diff_drivers;
}

static int diff_driver_alloc(
	git_diff_driver **out, size_t *namelen_out, const char *name)
{
	git_diff_driver *driver;
	size_t driverlen = sizeof(git_diff_driver),
		namelen = strlen(name),
		alloclen;

	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, driverlen, namelen);
	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, alloclen, 1);

	driver = git__calloc(1, alloclen);
	GIT_ERROR_CHECK_ALLOC(driver);

	memcpy(driver->name, name, namelen);

	*out = driver;

	if (namelen_out)
		*namelen_out = namelen;

	return 0;
}

static int git_diff_driver_builtin(
	git_diff_driver **out,
	git_diff_driver_registry *reg,
	const char *driver_name)
{
	git_diff_driver_definition *ddef = NULL;
	git_diff_driver *drv = NULL;
	int error = 0;
	size_t idx;

	for (idx = 0; idx < ARRAY_SIZE(builtin_defs); ++idx) {
		if (!strcasecmp(driver_name, builtin_defs[idx].name)) {
			ddef = &builtin_defs[idx];
			break;
		}
	}
	if (!ddef)
		goto done;

	if ((error = diff_driver_alloc(&drv, NULL, ddef->name)) < 0)
		goto done;

	drv->type = DIFF_DRIVER_PATTERNLIST;

	if (ddef->fns &&
		(error = diff_driver_add_patterns(
			drv, ddef->fns, ddef->flags)) < 0)
		goto done;

	if (ddef->words &&
	    (error = git_regexp_compile(&drv->word_pattern, ddef->words, ddef->flags)) < 0)
		goto done;

	if ((error = git_strmap_set(reg->drivers, drv->name, drv)) < 0)
		goto done;

done:
	if (error && drv)
		git_diff_driver_free(drv);
	else
		*out = drv;

	return error;
}

static int git_diff_driver_load(
	git_diff_driver **out, git_repository *repo, const char *driver_name)
{
	int error = 0;
	git_diff_driver_registry *reg;
	git_diff_driver *drv;
	size_t namelen;
	git_config *cfg = NULL;
	git_buf name = GIT_BUF_INIT;
	git_config_entry *ce = NULL;
	bool found_driver = false;

	if ((reg = git_repository_driver_registry(repo)) == NULL)
		return -1;

	if ((drv = git_strmap_get(reg->drivers, driver_name)) != NULL) {
		*out = drv;
		return 0;
	}

	if ((error = diff_driver_alloc(&drv, &namelen, driver_name)) < 0)
		goto done;

	drv->type = DIFF_DRIVER_AUTO;

	/* if you can't read config for repo, just use default driver */
	if (git_repository_config_snapshot(&cfg, repo) < 0) {
		git_error_clear();
		goto done;
	}

	if ((error = git_buf_printf(&name, "diff.%s.binary", driver_name)) < 0)
		goto done;

	switch (git_config__get_bool_force(cfg, name.ptr, -1)) {
	case true:
		/* if diff.<driver>.binary is true, just return the binary driver */
		*out = &global_drivers[DIFF_DRIVER_BINARY];
		goto done;
	case false:
		/* if diff.<driver>.binary is false, force binary checks off */
		/* but still may have custom function context patterns, etc. */
		drv->binary_flags = GIT_DIFF_FORCE_TEXT;
		found_driver = true;
		break;
	default:
		/* diff.<driver>.binary unspecified or "auto", so just continue */
		break;
	}

	/* TODO: warn if diff.<name>.command or diff.<name>.textconv are set */

	git_buf_truncate(&name, namelen + strlen("diff.."));
	if ((error = git_buf_PUTS(&name, "xfuncname")) < 0)
		goto done;

	if ((error = git_config_get_multivar_foreach(
			cfg, name.ptr, NULL, diff_driver_xfuncname, drv)) < 0) {
		if (error != GIT_ENOTFOUND)
			goto done;
		git_error_clear(); /* no diff.<driver>.xfuncname, so just continue */
	}

	git_buf_truncate(&name, namelen + strlen("diff.."));
	if ((error = git_buf_PUTS(&name, "funcname")) < 0)
		goto done;

	if ((error = git_config_get_multivar_foreach(
			cfg, name.ptr, NULL, diff_driver_funcname, drv)) < 0) {
		if (error != GIT_ENOTFOUND)
			goto done;
		git_error_clear(); /* no diff.<driver>.funcname, so just continue */
	}

	/* if we found any patterns, set driver type to use correct callback */
	if (git_array_size(drv->fn_patterns) > 0) {
		drv->type = DIFF_DRIVER_PATTERNLIST;
		found_driver = true;
	}

	git_buf_truncate(&name, namelen + strlen("diff.."));
	if ((error = git_buf_PUTS(&name, "wordregex")) < 0)
		goto done;

	if ((error = git_config__lookup_entry(&ce, cfg, name.ptr, false)) < 0)
		goto done;
	if (!ce || !ce->value)
		/* no diff.<driver>.wordregex, so just continue */;
	else if (!(error = git_regexp_compile(&drv->word_pattern, ce->value, 0)))
		found_driver = true;
	else {
		/* TODO: warn about bad regex instead of failure */
		goto done;
	}

	/* TODO: look up diff.<driver>.algorithm to turn on minimal / patience
	 * diff in drv->other_flags
	 */

	/* if no driver config found at all, fall back on AUTO driver */
	if (!found_driver)
		goto done;

	/* store driver in registry */
	if ((error = git_strmap_set(reg->drivers, drv->name, drv)) < 0)
		goto done;

	*out = drv;

done:
	git_config_entry_free(ce);
	git_buf_dispose(&name);
	git_config_free(cfg);

	if (!*out) {
		int error2 = git_diff_driver_builtin(out, reg, driver_name);
		if (!error)
			error = error2;
	}

	if (drv && drv != *out)
		git_diff_driver_free(drv);

	return error;
}

int git_diff_driver_lookup(
	git_diff_driver **out, git_repository *repo,
	git_attr_session *attrsession, const char *path)
{
	int error = 0;
	const char *values[1], *attrs[] = { "diff" };

	assert(out);
	*out = NULL;

	if (!repo || !path || !strlen(path))
		/* just use the auto value */;
	else if ((error = git_attr_get_many_with_session(values, repo,
			attrsession, 0, path, 1, attrs)) < 0)
		/* return error below */;

	else if (GIT_ATTR_IS_UNSPECIFIED(values[0]))
		/* just use the auto value */;
	else if (GIT_ATTR_IS_FALSE(values[0]))
		*out = &global_drivers[DIFF_DRIVER_BINARY];
	else if (GIT_ATTR_IS_TRUE(values[0]))
		*out = &global_drivers[DIFF_DRIVER_TEXT];

	/* otherwise look for driver information in config and build driver */
	else if ((error = git_diff_driver_load(out, repo, values[0])) < 0) {
		if (error == GIT_ENOTFOUND) {
			error = 0;
			git_error_clear();
		}
	}

	if (!*out)
		*out = &global_drivers[DIFF_DRIVER_AUTO];

	return error;
}

void git_diff_driver_free(git_diff_driver *driver)
{
	size_t i;

	if (!driver)
		return;

	for (i = 0; i < git_array_size(driver->fn_patterns); ++i)
		git_regexp_dispose(& git_array_get(driver->fn_patterns, i)->re);
	git_array_clear(driver->fn_patterns);

	git_regexp_dispose(&driver->word_pattern);

	git__free(driver);
}

void git_diff_driver_update_options(
	uint32_t *option_flags, git_diff_driver *driver)
{
	if ((*option_flags & FORCE_DIFFABLE) == 0)
		*option_flags |= driver->binary_flags;

	*option_flags |= driver->other_flags;
}

int git_diff_driver_content_is_binary(
	git_diff_driver *driver, const char *content, size_t content_len)
{
	git_buf search = GIT_BUF_INIT;

	GIT_UNUSED(driver);

	git_buf_attach_notowned(&search, content,
		min(content_len, GIT_FILTER_BYTES_TO_CHECK_NUL));

	/* TODO: provide encoding / binary detection callbacks that can
	 * be UTF-8 aware, etc.  For now, instead of trying to be smart,
	 * let's just use the simple NUL-byte detection that core git uses.
	 */

	/* previously was: if (git_buf_text_is_binary(&search)) */
	if (git_buf_text_contains_nul(&search))
		return 1;

	return 0;
}

static int diff_context_line__simple(
	git_diff_driver *driver, git_buf *line)
{
	char firstch = line->ptr[0];
	GIT_UNUSED(driver);
	return (git__isalpha(firstch) || firstch == '_' || firstch == '$');
}

static int diff_context_line__pattern_match(
	git_diff_driver *driver, git_buf *line)
{
	size_t i, maxi = git_array_size(driver->fn_patterns);
	git_regmatch pmatch[2];

	for (i = 0; i < maxi; ++i) {
		git_diff_driver_pattern *pat = git_array_get(driver->fn_patterns, i);

		if (!git_regexp_search(&pat->re, line->ptr, 2, pmatch)) {
			if (pat->flags & REG_NEGATE)
				return false;

			/* use pmatch data to trim line data */
			i = (pmatch[1].start >= 0) ? 1 : 0;
			git_buf_consume(line, git_buf_cstr(line) + pmatch[i].start);
			git_buf_truncate(line, pmatch[i].end - pmatch[i].start);
			git_buf_rtrim(line);

			return true;
		}
	}

	return false;
}

static long diff_context_find(
	const char *line,
	long line_len,
	char *out,
	long out_size,
	void *payload)
{
	git_diff_find_context_payload *ctxt = payload;

	if (git_buf_set(&ctxt->line, line, (size_t)line_len) < 0)
		return -1;
	git_buf_rtrim(&ctxt->line);

	if (!ctxt->line.size)
		return -1;

	if (!ctxt->match_line || !ctxt->match_line(ctxt->driver, &ctxt->line))
		return -1;

	if (out_size > (long)ctxt->line.size)
		out_size = (long)ctxt->line.size;
	memcpy(out, ctxt->line.ptr, (size_t)out_size);

	return out_size;
}

void git_diff_find_context_init(
	git_diff_find_context_fn *findfn_out,
	git_diff_find_context_payload *payload_out,
	git_diff_driver *driver)
{
	*findfn_out = driver ? diff_context_find : NULL;

	memset(payload_out, 0, sizeof(*payload_out));
	if (driver) {
		payload_out->driver = driver;
		payload_out->match_line = (driver->type == DIFF_DRIVER_PATTERNLIST) ?
			diff_context_line__pattern_match : diff_context_line__simple;
		git_buf_init(&payload_out->line, 0);
	}
}

void git_diff_find_context_clear(git_diff_find_context_payload *payload)
{
	if (payload) {
		git_buf_dispose(&payload->line);
		payload->driver = NULL;
	}
}
Commit	Line	Data
114f5a6c RB	1	/*
	2	* Copyright (C) the libgit2 contributors. All rights reserved.
	3	*
	4	* This file is part of libgit2, distributed under the GNU GPL v2 with
	5	* a Linking Exception. For full terms see the included COPYING file.
	6	*/
eae0bfdc PP	7
eae0bfdc PP	8	#include "diff_driver.h"
114f5a6c RB	9
	10	#include "git2/attr.h"
	11
0c9c969a	12	#include "common.h"
114f5a6c	13	#include "diff.h"
114f5a6c	14	#include "strmap.h"
114f5a6c RB	15	#include "map.h"
114f5a6c RB	16	#include "buf_text.h"
9f77b3f6	17	#include "config.h"
0c9c969a	18	#include "regexp.h"
5dc98298	19	#include "repository.h"
114f5a6c RB	20
	21	typedef enum {
	22	DIFF_DRIVER_AUTO = 0,
5dc98298 RB	23	DIFF_DRIVER_BINARY = 1,
	24	DIFF_DRIVER_TEXT = 2,
	25	DIFF_DRIVER_PATTERNLIST = 3,
114f5a6c RB	26	} git_diff_driver_t;
114f5a6c RB	27
a5a38643	28	typedef struct {
0c9c969a	29	git_regexp re;
a5a38643 RB	30	int flags;
	31	} git_diff_driver_pattern;
	32
114f5a6c	33	enum {
a5a38643	34	REG_NEGATE = (1 << 15) /* get out of the way of existing flags */
114f5a6c RB	35	};
	36
	37	/* data for finding function context for a given file type */
	38	struct git_diff_driver {
	39	git_diff_driver_t type;
5dc98298 RB	40	uint32_t binary_flags;
5dc98298 RB	41	uint32_t other_flags;
a5a38643	42	git_array_t(git_diff_driver_pattern) fn_patterns;
0c9c969a	43	git_regexp word_pattern;
42e6cf78	44	char name[GIT_FLEX_ARRAY];
114f5a6c RB	45	};
114f5a6c RB	46
c7c260a5	47	#include "userdiff.h"
2c65602e	48
114f5a6c RB	49	struct git_diff_driver_registry {
114f5a6c RB	50	git_strmap *drivers;
114f5a6c RB	51	};
114f5a6c RB	52
5dc98298 RB	53	#define FORCE_DIFFABLE (GIT_DIFF_FORCE_TEXT \| GIT_DIFF_FORCE_BINARY)
5dc98298 RB	54
114f5a6c	55	static git_diff_driver global_drivers[3] = {
5dc98298 RB	56	{ DIFF_DRIVER_AUTO, 0, 0, },
	57	{ DIFF_DRIVER_BINARY, GIT_DIFF_FORCE_BINARY, 0 },
	58	{ DIFF_DRIVER_TEXT, GIT_DIFF_FORCE_TEXT, 0 },
114f5a6c RB	59	};
114f5a6c RB	60
8fd74c08	61	git_diff_driver_registry *git_diff_driver_registry_new(void)
114f5a6c	62	{
3eadfecd RB	63	git_diff_driver_registry *reg =
	64	git__calloc(1, sizeof(git_diff_driver_registry));
	65	if (!reg)
	66	return NULL;
	67
0c9c969a	68	if (git_strmap_new(&reg->drivers) < 0) {
3eadfecd RB	69	git_diff_driver_registry_free(reg);
	70	return NULL;
	71	}
	72
	73	return reg;
114f5a6c RB	74	}
	75
	76	void git_diff_driver_registry_free(git_diff_driver_registry *reg)
	77	{
5dc98298 RB	78	git_diff_driver *drv;
5dc98298 RB	79
3eadfecd RB	80	if (!reg)
	81	return;
	82
5dc98298	83	git_strmap_foreach_value(reg->drivers, drv, git_diff_driver_free(drv));
3eadfecd	84	git_strmap_free(reg->drivers);
114f5a6c RB	85	git__free(reg);
	86	}
	87
a5a38643 RB	88	static int diff_driver_add_patterns(
a5a38643 RB	89	git_diff_driver drv, const char regex_str, int regex_flags)
5dc98298	90	{
a5a38643 RB	91	int error = 0;
	92	const char scan, end;
	93	git_diff_driver_pattern *pat = NULL;
	94	git_buf buf = GIT_BUF_INIT;
	95
	96	for (scan = regex_str; scan; scan = end) {
	97	/* get pattern to fill in */
	98	if ((pat = git_array_alloc(drv->fn_patterns)) == NULL) {
e451cd5c	99	return -1;
a5a38643	100	}
5dc98298	101
a5a38643 RB	102	pat->flags = regex_flags;
	103	if (*scan == '!') {
	104	pat->flags \|= REG_NEGATE;
	105	++scan;
	106	}
	107
	108	if ((end = strchr(scan, '\n')) != NULL) {
	109	error = git_buf_set(&buf, scan, end - scan);
	110	end++;
	111	} else {
	112	error = git_buf_sets(&buf, scan);
	113	}
	114	if (error < 0)
	115	break;
	116
0c9c969a	117	if ((error = git_regexp_compile(&pat->re, buf.ptr, regex_flags)) != 0) {
e451cd5c CMN	118	/*
	119	* TODO: issue a warning
	120	*/
a5a38643	121	}
5dc98298 RB	122	}
5dc98298 RB	123
a5a38643 RB	124	if (error && pat != NULL)
a5a38643 RB	125	(void)git_array_pop(drv->fn_patterns); /* release last item */
ac3d33df	126	git_buf_dispose(&buf);
5dc98298	127
e451cd5c CMN	128	/* We want to ignore bad patterns, so return success regardless */
e451cd5c CMN	129	return 0;
5dc98298 RB	130	}
	131
	132	static int diff_driver_xfuncname(const git_config_entry entry, void payload)
	133	{
0c9c969a	134	return diff_driver_add_patterns(payload, entry->value, 0);
5dc98298 RB	135	}
	136
	137	static int diff_driver_funcname(const git_config_entry entry, void payload)
	138	{
a5a38643	139	return diff_driver_add_patterns(payload, entry->value, 0);
5dc98298 RB	140	}
	141
	142	static git_diff_driver_registry *git_repository_driver_registry(
	143	git_repository *repo)
	144	{
	145	if (!repo->diff_drivers) {
	146	git_diff_driver_registry *reg = git_diff_driver_registry_new();
	147	reg = git__compare_and_swap(&repo->diff_drivers, NULL, reg);
	148
	149	if (reg != NULL) /* if we race, free losing allocation */
	150	git_diff_driver_registry_free(reg);
	151	}
	152
	153	if (!repo->diff_drivers)
ac3d33df	154	git_error_set(GIT_ERROR_REPOSITORY, "unable to create diff driver registry");
5dc98298 RB	155
	156	return repo->diff_drivers;
	157	}
	158
392702ee ET	159	static int diff_driver_alloc(
	160	git_diff_driver *out, size_t namelen_out, const char *name)
	161	{
	162	git_diff_driver *driver;
	163	size_t driverlen = sizeof(git_diff_driver),
f1453c59 ET	164	namelen = strlen(name),
f1453c59 ET	165	alloclen;
392702ee	166
ac3d33df JK	167	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, driverlen, namelen);
ac3d33df JK	168	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, alloclen, 1);
392702ee	169
f1453c59	170	driver = git__calloc(1, alloclen);
ac3d33df	171	GIT_ERROR_CHECK_ALLOC(driver);
392702ee ET	172
	173	memcpy(driver->name, name, namelen);
	174
	175	*out = driver;
	176
	177	if (namelen_out)
	178	*namelen_out = namelen;
	179
	180	return 0;
	181	}
	182
a5a38643 RB	183	static int git_diff_driver_builtin(
	184	git_diff_driver **out,
	185	git_diff_driver_registry *reg,
	186	const char *driver_name)
	187	{
a5a38643 RB	188	git_diff_driver_definition *ddef = NULL;
a5a38643 RB	189	git_diff_driver *drv = NULL;
0c9c969a	190	int error = 0;
392702ee	191	size_t idx;
a5a38643 RB	192
	193	for (idx = 0; idx < ARRAY_SIZE(builtin_defs); ++idx) {
	194	if (!strcasecmp(driver_name, builtin_defs[idx].name)) {
	195	ddef = &builtin_defs[idx];
	196	break;
	197	}
	198	}
	199	if (!ddef)
	200	goto done;
	201
392702ee ET	202	if ((error = diff_driver_alloc(&drv, NULL, ddef->name)) < 0)
392702ee ET	203	goto done;
a5a38643 RB	204
a5a38643 RB	205	drv->type = DIFF_DRIVER_PATTERNLIST;
a5a38643 RB	206
	207	if (ddef->fns &&
	208	(error = diff_driver_add_patterns(
0c9c969a	209	drv, ddef->fns, ddef->flags)) < 0)
a5a38643 RB	210	goto done;
	211
	212	if (ddef->words &&
0c9c969a	213	(error = git_regexp_compile(&drv->word_pattern, ddef->words, ddef->flags)) < 0)
a5a38643	214	goto done;
a5a38643	215
0c9c969a UG	216	if ((error = git_strmap_set(reg->drivers, drv->name, drv)) < 0)
0c9c969a UG	217	goto done;
a5a38643 RB	218
a5a38643 RB	219	done:
2c65602e	220	if (error && drv)
a5a38643	221	git_diff_driver_free(drv);
2c65602e	222	else
a5a38643	223	*out = drv;
a5a38643 RB	224
	225	return error;
	226	}
	227
3eadfecd	228	static int git_diff_driver_load(
5dc98298	229	git_diff_driver *out, git_repository repo, const char *driver_name)
3eadfecd	230	{
9f77b3f6	231	int error = 0;
5dc98298	232	git_diff_driver_registry *reg;
0c9c969a UG	233	git_diff_driver *drv;
0c9c969a UG	234	size_t namelen;
392702ee	235	git_config *cfg = NULL;
5dc98298	236	git_buf name = GIT_BUF_INIT;
9a97f49e	237	git_config_entry *ce = NULL;
42e6cf78	238	bool found_driver = false;
5dc98298	239
a5a38643	240	if ((reg = git_repository_driver_registry(repo)) == NULL)
5dc98298	241	return -1;
a5a38643	242
0c9c969a UG	243	if ((drv = git_strmap_get(reg->drivers, driver_name)) != NULL) {
0c9c969a UG	244	*out = drv;
a5a38643	245	return 0;
5dc98298 RB	246	}
5dc98298 RB	247
392702ee ET	248	if ((error = diff_driver_alloc(&drv, &namelen, driver_name)) < 0)
	249	goto done;
	250
a37aa82e	251	drv->type = DIFF_DRIVER_AUTO;
a37aa82e	252
5dc98298	253	/* if you can't read config for repo, just use default driver */
ac99d86b	254	if (git_repository_config_snapshot(&cfg, repo) < 0) {
ac3d33df	255	git_error_clear();
a5a38643	256	goto done;
5dc98298 RB	257	}
5dc98298 RB	258
5dc98298	259	if ((error = git_buf_printf(&name, "diff.%s.binary", driver_name)) < 0)
54faddd2	260	goto done;
9f77b3f6 RB	261
	262	switch (git_config__get_bool_force(cfg, name.ptr, -1)) {
	263	case true:
5dc98298	264	/* if diff.<driver>.binary is true, just return the binary driver */
5dc98298	265	*out = &global_drivers[DIFF_DRIVER_BINARY];
54faddd2	266	goto done;
9f77b3f6	267	case false:
5dc98298 RB	268	/* if diff.<driver>.binary is false, force binary checks off */
	269	/* but still may have custom function context patterns, etc. */
	270	drv->binary_flags = GIT_DIFF_FORCE_TEXT;
42e6cf78	271	found_driver = true;
9f77b3f6 RB	272	break;
9f77b3f6 RB	273	default:
a5a38643	274	/* diff.<driver>.binary unspecified or "auto", so just continue */
9f77b3f6	275	break;
5dc98298 RB	276	}
	277
	278	/* TODO: warn if diff.<name>.command or diff.<name>.textconv are set */
	279
42e6cf78	280	git_buf_truncate(&name, namelen + strlen("diff.."));
0c9c969a UG	281	if ((error = git_buf_PUTS(&name, "xfuncname")) < 0)
	282	goto done;
	283
4efa3290	284	if ((error = git_config_get_multivar_foreach(
5dc98298 RB	285	cfg, name.ptr, NULL, diff_driver_xfuncname, drv)) < 0) {
5dc98298 RB	286	if (error != GIT_ENOTFOUND)
54faddd2	287	goto done;
ac3d33df	288	git_error_clear(); /* no diff.<driver>.xfuncname, so just continue */
5dc98298	289	}
3eadfecd	290
42e6cf78	291	git_buf_truncate(&name, namelen + strlen("diff.."));
0c9c969a UG	292	if ((error = git_buf_PUTS(&name, "funcname")) < 0)
	293	goto done;
	294
4efa3290	295	if ((error = git_config_get_multivar_foreach(
5dc98298 RB	296	cfg, name.ptr, NULL, diff_driver_funcname, drv)) < 0) {
5dc98298 RB	297	if (error != GIT_ENOTFOUND)
54faddd2	298	goto done;
ac3d33df	299	git_error_clear(); /* no diff.<driver>.funcname, so just continue */
5dc98298 RB	300	}
	301
	302	/* if we found any patterns, set driver type to use correct callback */
42e6cf78	303	if (git_array_size(drv->fn_patterns) > 0) {
5dc98298	304	drv->type = DIFF_DRIVER_PATTERNLIST;
42e6cf78 RB	305	found_driver = true;
42e6cf78 RB	306	}
5dc98298	307
42e6cf78	308	git_buf_truncate(&name, namelen + strlen("diff.."));
0c9c969a UG	309	if ((error = git_buf_PUTS(&name, "wordregex")) < 0)
	310	goto done;
	311
9f77b3f6	312	if ((error = git_config__lookup_entry(&ce, cfg, name.ptr, false)) < 0)
54faddd2	313	goto done;
9f77b3f6 RB	314	if (!ce \|\| !ce->value)
9f77b3f6 RB	315	/* no diff.<driver>.wordregex, so just continue */;
0c9c969a	316	else if (!(error = git_regexp_compile(&drv->word_pattern, ce->value, 0)))
42e6cf78	317	found_driver = true;
9f77b3f6 RB	318	else {
9f77b3f6 RB	319	/* TODO: warn about bad regex instead of failure */
9f77b3f6	320	goto done;
5dc98298 RB	321	}
	322
	323	/* TODO: look up diff.<driver>.algorithm to turn on minimal / patience
	324	* diff in drv->other_flags
	325	*/
	326
54faddd2	327	/* if no driver config found at all, fall back on AUTO driver */
42e6cf78	328	if (!found_driver)
54faddd2	329	goto done;
42e6cf78 RB	330
42e6cf78 RB	331	/* store driver in registry */
0c9c969a	332	if ((error = git_strmap_set(reg->drivers, drv->name, drv)) < 0)
54faddd2	333	goto done;
42e6cf78	334
5dc98298	335	*out = drv;
5dc98298	336
54faddd2	337	done:
9a97f49e	338	git_config_entry_free(ce);
ac3d33df	339	git_buf_dispose(&name);
29c4cb09	340	git_config_free(cfg);
54faddd2	341
a5a38643 RB	342	if (!*out) {
	343	int error2 = git_diff_driver_builtin(out, reg, driver_name);
	344	if (!error)
	345	error = error2;
	346	}
54faddd2 RB	347
	348	if (drv && drv != *out)
	349	git_diff_driver_free(drv);
	350
5dc98298	351	return error;
3eadfecd RB	352	}
3eadfecd RB	353
114f5a6c	354	int git_diff_driver_lookup(
eae0bfdc PP	355	git_diff_driver *out, git_repository repo,
eae0bfdc PP	356	git_attr_session attrsession, const char path)
114f5a6c	357	{
3eadfecd	358	int error = 0;
eae0bfdc	359	const char values[1], attrs[] = { "diff" };
114f5a6c RB	360
114f5a6c RB	361	assert(out);
2c65602e	362	*out = NULL;
114f5a6c RB	363
114f5a6c RB	364	if (!repo \|\| !path \|\| !strlen(path))
2c65602e	365	/* just use the auto value */;
eae0bfdc PP	366	else if ((error = git_attr_get_many_with_session(values, repo,
eae0bfdc PP	367	attrsession, 0, path, 1, attrs)) < 0)
2c65602e	368	/* return error below */;
eae0bfdc	369
0c9c969a	370	else if (GIT_ATTR_IS_UNSPECIFIED(values[0]))
42e6cf78	371	/* just use the auto value */;
0c9c969a	372	else if (GIT_ATTR_IS_FALSE(values[0]))
5dc98298	373	*out = &global_drivers[DIFF_DRIVER_BINARY];
0c9c969a	374	else if (GIT_ATTR_IS_TRUE(values[0]))
5dc98298	375	*out = &global_drivers[DIFF_DRIVER_TEXT];
114f5a6c RB	376
114f5a6c RB	377	/* otherwise look for driver information in config and build driver */
eae0bfdc	378	else if ((error = git_diff_driver_load(out, repo, values[0])) < 0) {
2c65602e RB	379	if (error == GIT_ENOTFOUND) {
2c65602e RB	380	error = 0;
ac3d33df	381	git_error_clear();
2c65602e	382	}
3eadfecd	383	}
114f5a6c	384
42e6cf78 RB	385	if (!*out)
	386	*out = &global_drivers[DIFF_DRIVER_AUTO];
	387
2c65602e	388	return error;
114f5a6c RB	389	}
	390
	391	void git_diff_driver_free(git_diff_driver *driver)
	392	{
5dc98298 RB	393	size_t i;
	394
	395	if (!driver)
	396	return;
	397
54faddd2	398	for (i = 0; i < git_array_size(driver->fn_patterns); ++i)
0c9c969a	399	git_regexp_dispose(& git_array_get(driver->fn_patterns, i)->re);
5dc98298 RB	400	git_array_clear(driver->fn_patterns);
5dc98298 RB	401
0c9c969a	402	git_regexp_dispose(&driver->word_pattern);
5dc98298 RB	403
5dc98298 RB	404	git__free(driver);
114f5a6c RB	405	}
114f5a6c RB	406
5dc98298 RB	407	void git_diff_driver_update_options(
5dc98298 RB	408	uint32_t option_flags, git_diff_driver driver)
114f5a6c	409	{
5dc98298 RB	410	if ((*option_flags & FORCE_DIFFABLE) == 0)
	411	*option_flags \|= driver->binary_flags;
	412
	413	*option_flags \|= driver->other_flags;
114f5a6c RB	414	}
	415
	416	int git_diff_driver_content_is_binary(
	417	git_diff_driver driver, const char content, size_t content_len)
	418	{
d4cf1675	419	git_buf search = GIT_BUF_INIT;
114f5a6c RB	420
	421	GIT_UNUSED(driver);
	422
d4cf1675 ET	423	git_buf_attach_notowned(&search, content,
	424	min(content_len, GIT_FILTER_BYTES_TO_CHECK_NUL));
	425
114f5a6c RB	426	/* TODO: provide encoding / binary detection callbacks that can
	427	* be UTF-8 aware, etc. For now, instead of trying to be smart,
	428	* let's just use the simple NUL-byte detection that core git uses.
	429	*/
	430
	431	/* previously was: if (git_buf_text_is_binary(&search)) */
	432	if (git_buf_text_contains_nul(&search))
	433	return 1;
	434
	435	return 0;
	436	}
	437
5dc98298	438	static int diff_context_line__simple(
a5a38643	439	git_diff_driver driver, git_buf line)
5dc98298	440	{
a5a38643	441	char firstch = line->ptr[0];
5dc98298	442	GIT_UNUSED(driver);
a5a38643	443	return (git__isalpha(firstch) \|\| firstch == '_' \|\| firstch == '$');
5dc98298 RB	444	}
	445
	446	static int diff_context_line__pattern_match(
a5a38643	447	git_diff_driver driver, git_buf line)
5dc98298	448	{
b8e86c62	449	size_t i, maxi = git_array_size(driver->fn_patterns);
0c9c969a	450	git_regmatch pmatch[2];
5dc98298	451
b8e86c62	452	for (i = 0; i < maxi; ++i) {
a5a38643 RB	453	git_diff_driver_pattern *pat = git_array_get(driver->fn_patterns, i);
a5a38643 RB	454
0c9c969a	455	if (!git_regexp_search(&pat->re, line->ptr, 2, pmatch)) {
a5a38643 RB	456	if (pat->flags & REG_NEGATE)
a5a38643 RB	457	return false;
b8e86c62 RB	458
b8e86c62 RB	459	/* use pmatch data to trim line data */
0c9c969a UG	460	i = (pmatch[1].start >= 0) ? 1 : 0;
	461	git_buf_consume(line, git_buf_cstr(line) + pmatch[i].start);
	462	git_buf_truncate(line, pmatch[i].end - pmatch[i].start);
082e82db	463	git_buf_rtrim(line);
b8e86c62	464
5dc98298	465	return true;
a5a38643	466	}
5dc98298 RB	467	}
	468
	469	return false;
	470	}
	471
114f5a6c RB	472	static long diff_context_find(
	473	const char *line,
	474	long line_len,
	475	char *out,
	476	long out_size,
	477	void *payload)
	478	{
5dc98298	479	git_diff_find_context_payload *ctxt = payload;
114f5a6c	480
5dc98298	481	if (git_buf_set(&ctxt->line, line, (size_t)line_len) < 0)
114f5a6c	482	return -1;
5dc98298	483	git_buf_rtrim(&ctxt->line);
114f5a6c	484
5dc98298	485	if (!ctxt->line.size)
114f5a6c RB	486	return -1;
114f5a6c RB	487
a5a38643	488	if (!ctxt->match_line \|\| !ctxt->match_line(ctxt->driver, &ctxt->line))
5dc98298	489	return -1;
114f5a6c	490
a5f9b5f8	491	if (out_size > (long)ctxt->line.size)
584f2d30	492	out_size = (long)ctxt->line.size;
a5f9b5f8	493	memcpy(out, ctxt->line.ptr, (size_t)out_size);
114f5a6c	494
a5f9b5f8	495	return out_size;
5dc98298	496	}
114f5a6c	497
5dc98298 RB	498	void git_diff_find_context_init(
	499	git_diff_find_context_fn *findfn_out,
	500	git_diff_find_context_payload *payload_out,
	501	git_diff_driver *driver)
	502	{
	503	*findfn_out = driver ? diff_context_find : NULL;
	504
	505	memset(payload_out, 0, sizeof(*payload_out));
	506	if (driver) {
	507	payload_out->driver = driver;
	508	payload_out->match_line = (driver->type == DIFF_DRIVER_PATTERNLIST) ?
	509	diff_context_line__pattern_match : diff_context_line__simple;
	510	git_buf_init(&payload_out->line, 0);
	511	}
114f5a6c RB	512	}
114f5a6c RB	513
5dc98298	514	void git_diff_find_context_clear(git_diff_find_context_payload *payload)
114f5a6c	515	{
5dc98298	516	if (payload) {
ac3d33df	517	git_buf_dispose(&payload->line);
5dc98298 RB	518	payload->driver = NULL;
5dc98298 RB	519	}
114f5a6c	520	}