cp_mv.c File Reference

cp and mv actions. More...

#include <fcntl.h>
#include <search.h>
#include "global.h"
#include "cp_mv.h"
#include "status.h"
#include "est_ops.h"
#include "url.h"
#include "checksum.h"
#include "options.h"
#include "props.h"
#include "cache.h"
#include "helper.h"
#include "waa.h"

Include dependency graph for cp_mv.c:

Go to the source code of this file.

Data Structures
struct	cm___candidate_t
	Structure for candidate retrieval. More...
struct	cm___match_t
	-. More...
Defines
#define	MAX_DUPL_ENTRIES (HASH__LIST_MAX -1)
	Maximum number of entries that are stored.
#define	CM___MATCH_NUM (sizeof(cm___match_array)/sizeof(cm___match_array[0]))
Typedefs
typedef int(	cm___register_fn )(struct estat , struct cm___match_t )
	Enters the given entry into the database.
typedef int(	cm___get_list_fn )(struct estat , struct cm___match_t , struct cm___candidate_t *, int count)
	Queries the database for the given entry.
typedef char *(	cm___format_fn )(struct cm___match_t match, struct cm___candidate_t cand)
	Format function for verbose output.
Enumerations
enum	cm___match_e { CM___NAME_F = 0, CM___NAME_D, CM___DIRLIST }
	Enumeration for (some) matching criteria. More...
Functions
datum	cm___md5_datum (const struct estat *sts)
	Gets a datum from a struct estat::md5.
datum	cm___name_datum (const struct estat *sts)
	Gets a datum from the name of an entry; the \0 gets included (for easier dumping).
datum	cm___inode_datum (const struct estat *sts)
	Gets a datum from the filesystem addressing - device and inode.
static int	cm___cand_compare (const void _a, const void _b)
	Compare function for cm___candidate_t.
static int	cm___cand_comp_count (const void _a, const void _b)
	Compare function for cm___candidate_t.
int	cm___hash_register (struct estat sts, struct cm___match_t match)
	-.
int	cm___match_children (struct estat sts, struct cm___match_t match, struct cm___candidate_t *list, int found)
	-.
int	cm___hash_list (struct estat sts, struct cm___match_t match, struct cm___candidate_t *output, int found)
	-.
char *	cm___output_pct (struct cm___match_t match, struct cm___candidate_t cand)
	Puts cm___candidate_t::match_count formatted into buffer.
int	cm___register_entry (struct estat *sts)
	Inserts the given entry in all allowed matching databases.
static int	cm___match (struct estat *entry)
	Shows possible copyfrom sources for the given entry.
int	cm__find_dir_source (struct estat *dir)
int	cm__find_file_source (struct estat *file)
int	cm__find_copied (struct estat *root)
	After loading known entries try to find some match for every new entry.
int	cm__detect (struct estat root, int argc, char argv[])
	-.
int	cm___string_to_rev_path (char string, char out_url, svn_revnum_t orev)
	Converts the (internally stored) string into revision/URL.
int	cm___rev_path_to_string (char url, svn_revnum_t revision, char *string)
	Formats the revision and url for storage in the hash table.
int	cm___absolute_path (char path, char *output)
	Returns the absolute path.
int	cm___not_below_wcpath (char path, char *out)
	Checks whether a path is below `wc_path`, and returns the relative part.
int	cm___dump_list (FILE output, int argc, char normalized[])
	Dump a list of copyfrom relations to the stream.
int	cm___make_copy (struct estat root, char cp_src, svn_revnum_t revision, char *cp_dest, int paths_are_wc_relative)
	Make the copy in the tree started at root.
int	cm___ignore_impl_copied (struct estat *cur)
	Sets all entries that are just implicitly copied to ignored.
int	cm__uncopy (struct estat root, int argc, char argv[])
	-.
int	cm__work (struct estat root, int argc, char argv[])
	-.
int	cm___get_base_source (struct estat sts, char name, char *src_url, svn_revnum_t src_rev, int alloc_extra, int register_for_cleanup)
	Get the source of an entry with `RF_COPY_BASE` set.
int	cm___get_sub_source_rek (struct estat cur, int length_to_add, char dest_buffer, svn_revnum_t src_rev, char **eobuffer)
	Recursively creating the URL.
int	cm___get_sub_source (struct estat sts, char name, char *src_url, svn_revnum_t src_rev)
	Get the source of an entry with `RF_COPY_SUB` set.
int	cm__get_source (struct estat sts, char name, char *src_url, svn_revnum_t src_rev, int register_for_cleanup)
	-.
Variables
int	copydetect_count
	How many entries could be correlated.
cm___register_fn	cm___hash_register
	Inserts into hash tables.
cm___get_list_fn	cm___hash_list
	Queries the database for the given entry.
cm___get_list_fn	cm___match_children
	Match directories by their children.
cm___format_fn	cm___output_pct
	Outputs percent of match.
struct cm___match_t	cm___match_array []
	Array with ways for simple matches.

typedef datum(	cm___to_datum_t )(const struct estat *sts)
	Function and type declarations for entry-to-hash-key conversions.
cm___to_datum_t	cm___md5_datum
cm___to_datum_t	cm___inode_datum
cm___to_datum_t	cm___name_datum

Detailed Description

cp and mv actions.

Various thoughts ...

Can we construct relations between 2 new files? We'd just have to write the MD5 of the new files into the hash, then we'd find the first file on commit of the 2nd file ... and we see that the other one is new, too.
But see creating 2 revisions on commit.

Definition in file cp_mv.c.

Define Documentation

#define CM___MATCH_NUM (sizeof(cm___match_array)/sizeof(cm___match_array[0]))

Definition at line 435 of file cp_mv.c.

Referenced by cm___match(), cm___register_entry(), and cm__detect().

#define MAX_DUPL_ENTRIES (HASH__LIST_MAX -1)

Maximum number of entries that are stored.

The -1 is for overflow detection "...".

Definition at line 304 of file cp_mv.c.

Referenced by cm___match().

Typedef Documentation

typedef char*( cm___format_fn)(struct cm___match_t *match, struct cm___candidate_t *cand)

Format function for verbose output.

Formats the candidate cand in match match into a buffer, and returns this buffer.

Definition at line 353 of file cp_mv.c.

typedef int( cm___get_list_fn)(struct estat *, struct cm___match_t *, struct cm___candidate_t **, int *count)

Queries the database for the given entry.

Output is (the address of) an array of cm___candidates_t, and the number of elements.

Definition at line 348 of file cp_mv.c.

typedef int( cm___register_fn)(struct estat *, struct cm___match_t *)

Enters the given entry into the database.

Definition at line 343 of file cp_mv.c.

typedef datum( cm___to_datum_t)(const struct estat *sts)

Function and type declarations for entry-to-hash-key conversions.

Definition at line 330 of file cp_mv.c.

Enumeration Type Documentation

enum cm___match_e

Enumeration for (some) matching criteria.

Enumerator:

CM___NAME_F
CM___NAME_D
CM___DIRLIST

Definition at line 398 of file cp_mv.c.

Function Documentation

int cm___absolute_path	(	char *	path,
		char **	output
	)

Returns the absolute path.

Definition at line 1056 of file cp_mv.c.

References BUG_ON, cch__add(), DEBUGP, cache_t::entries, hlp__pathcopy(), cache_entry_t::len, cache_t::lru, start_path_len, and STOPIF.

Referenced by cm___make_copy().

Here is the call graph for this function:

Here is the caller graph for this function:

static int cm___cand_comp_count	(	const void *	_a,
		const void *	_b
	)			`[static]`

Compare function for cm___candidate_t.

Definition at line 486 of file cp_mv.c.

References cm___candidate_t::match_count.

Referenced by cm___match_children().

Here is the caller graph for this function:

static int cm___cand_compare	(	const void *	_a,
		const void *	_b
	)			`[static]`

Compare function for cm___candidate_t.

Definition at line 477 of file cp_mv.c.

References cm___candidate_t::sts.

Referenced by cm___match(), and cm___match_children().

Here is the caller graph for this function:

int cm___dump_list	(	FILE *	output,
		int	argc,
		char *	normalized[]
	)

Dump a list of copyfrom relations to the stream.

TODO: filter by wildcards (?)

Definition at line 1102 of file cp_mv.c.

References cm___string_to_rev_path(), hsh__close(), hsh__fetch(), hsh__first(), hsh__new(), hsh__next(), IF_FREE, opt__is_verbose(), STOPIF, STOPIF_CODE_ERR, WAA__COPYFROM_EXT, and wc_path.

Referenced by cm__work().

Here is the call graph for this function:

Here is the caller graph for this function:

int cm___get_base_source	(	struct estat *	sts,
		char *	name,
		char **	src_url,
		svn_revnum_t *	src_rev,
		int	alloc_extra,
		int	register_for_cleanup
	)

Get the source of an entry with RF_COPY_BASE set.

See cm__get_source() for details.

Definition at line 1519 of file cp_mv.c.

References BUG_ON, cm___string_to_rev_path(), DEBUGP, HASH_REMEMBER_FILENAME, hlp__strnalloc(), hsh__close(), hsh__fetch(), hsh__new(), hsh__register_delete(), IF_FREE, ops__build_path(), PATH_SEPARATOR, STOPIF, estat::url, WAA__COPYFROM_EXT, and wc_path.

Referenced by cm___get_sub_source_rek(), and cm__get_source().

Here is the call graph for this function:

Here is the caller graph for this function:

int cm___get_sub_source	(	struct estat *	sts,
		char *	name,
		char **	src_url,
		svn_revnum_t *	src_rev
	)

Get the source of an entry with RF_COPY_SUB set.

See cm__get_source() for details.

This function needs no cleanup.

Definition at line 1668 of file cp_mv.c.

References cm___get_sub_source_rek(), and STOPIF.

Referenced by cm__get_source().

Here is the call graph for this function:

Here is the caller graph for this function:

int cm___get_sub_source_rek	(	struct estat *	cur,
		int	length_to_add,
		char **	dest_buffer,
		svn_revnum_t *	src_rev,
		char **	eobuffer
	)

Recursively creating the URL.

As most of the parameters are constant, we could store them statically ... don't know whether it would make much difference, this function doesn't get called very often. length_to_add is increased while going up the tree; eobuffer gets handed back down.

Definition at line 1612 of file cp_mv.c.

References BUG_ON, cm___get_base_source(), DEBUGP, estat::flags, estat::name, estat::parent, and RF_COPY_BASE.

Referenced by cm___get_sub_source().

Here is the call graph for this function:

Here is the caller graph for this function:

int cm___hash_list	(	struct estat *	sts,
		struct cm___match_t *	match,
		struct cm___candidate_t **	output,
		int *	found
	)

-.

The other members are touched by upper layers, so we have to re-initialize them.

Definition at line 615 of file cp_mv.c.

References cm___match_t::db, hsh__list_get(), cm___match_t::key, cm___candidate_t::sts, and cm___match_t::to_key.

Here is the call graph for this function:

int cm___hash_register	(	struct estat *	sts,
		struct cm___match_t *	match
	)

-.

Definition at line 495 of file cp_mv.c.

References cm___match_t::db, hsh__insert_pointer(), and cm___match_t::to_key.

Here is the call graph for this function:

int cm___ignore_impl_copied ( struct estat * cur )

Sets all entries that are just implicitly copied to ignored.

Explicitly added entries (because of add, or prop-set) are kept.

Returns a 0 or 1, with 1 saying that all entries below are ignored, and so whether cur can (perhaps) be completely ignored, too.

Definition at line 1273 of file cp_mv.c.

References estat::by_inode, DEBUGP, estat::flags, estat::name, ops__has_children(), RF_ADD, RF_CHECK, RF_COPY_SUB, RF_PUSHPROPS, and estat::to_be_ignored.

Referenced by cm__uncopy().

Here is the call graph for this function:

Here is the caller graph for this function:

datum cm___inode_datum ( const struct estat * sts )

Gets a datum from the filesystem addressing - device and inode.

Definition at line 463 of file cp_mv.c.

References sstat_t::dev, sstat_t::ino, and estat::st.

int cm___make_copy	(	struct estat *	root,
		char *	cp_src,
		svn_revnum_t	revision,
		char *	cp_dest,
		int	paths_are_wc_relative
	)

Make the copy in the tree started at root.

The destination must not already exist in the tree; it can exist in the filesystem.

If revision is not 0 (which corresponds to BASE), the correct list of entries must be taken from the corresponding repository.

Uninitialization is done via root==NULL.

If the flag paths_are_wc_relative is set, the paths cp_src and cp_dest are taken as-is. Else they're are converted to wc-relative paths by making them absolute (eventually using start_path as anchor), and cutting the wc-root off.

Definition at line 1178 of file cp_mv.c.

References BUG_ON, cm___absolute_path(), cm___not_below_wcpath(), cm___rev_path_to_string(), estat::flags, hsh__close(), hsh__new(), hsh__store_charp(), OPS__CREATE, ops__traverse(), estat::repos_rev, RF___IS_COPY, RF_ADD, RF_COPY_BASE, RF_COPY_SUB, RF_ISNEW, STOPIF, STOPIF_CODE_ERR, estat::url, url__full_url(), waa__copy_entries(), WAA__COPYFROM_EXT, and wc_path.

Referenced by cm__work().

Here is the call graph for this function:

Here is the caller graph for this function:

static int cm___match ( struct estat * entry ) [static]

Shows possible copyfrom sources for the given entry.

Definition at line 689 of file cp_mv.c.

References BUG_ON, cm___cand_compare(), CM___DIRLIST, CM___MATCH_NUM, copydetect_count, DEBUGP, cm___match_t::entry_type, cm___match_t::format, cm___match_t::get_list, hlp__format_path(), cm___candidate_t::match_count, cm___candidate_t::matches_where, MAX_DUPL_ENTRIES, sstat_t::mode, cm___match_t::name, estat::name, ops__build_path(), opt__is_verbose(), estat::st, STOPIF, STOPIF_CODE_EPIPE, and cm___candidate_t::sts.

Referenced by cm__find_dir_source(), and cm__find_file_source().

Here is the call graph for this function:

Here is the caller graph for this function:

int cm___match_children	(	struct estat *	sts,
		struct cm___match_t *	match,
		struct cm___candidate_t **	list,
		int *	found
	)

-.

The big question is - should this work recursively? Would mean that the topmost directory would be descended, and the results had to be cached.

Definition at line 515 of file cp_mv.c.

References BUG_ON, estat::by_inode, cm___cand_comp_count(), cm___cand_compare(), CM___NAME_D, CM___NAME_F, cm___match_t::db, DEBUGP, estat::entry_count, HASH__LIST_MAX, hsh__list_get(), cm___candidate_t::match_count, sstat_t::mode, estat::name, ops__correlate_dirs(), estat::parent, estat::st, STOPIF, cm___candidate_t::sts, and cm___match_t::to_key.

Here is the call graph for this function:

datum cm___md5_datum ( const struct estat * sts )

Gets a datum from a struct estat::md5.

Definition at line 439 of file cp_mv.c.

References cs__md5tohex_buffered(), and estat::md5.

Here is the call graph for this function:

datum cm___name_datum ( const struct estat * sts )

Gets a datum from the name of an entry; the \0 gets included (for easier dumping).

Definition at line 451 of file cp_mv.c.

References estat::name.

int cm___not_below_wcpath	(	char *	path,
		char **	out
	)			`[inline]`

Checks whether a path is below wc_path, and returns the relative part.

If that isn't possible (because path is not below wc_path), EINVAL is returned. The case path==wc_path is not allowed, either.

Definition at line 1085 of file cp_mv.c.

References PATH_SEPARATOR, wc_path, and wc_path_len.

Referenced by cm___make_copy().

Here is the caller graph for this function:

char* cm___output_pct	(	struct cm___match_t *	match,
		struct cm___candidate_t *	cand
	)

Puts cm___candidate_t::match_count formatted into buffer.

Definition at line 643 of file cp_mv.c.

References BUG_ON, and cm___candidate_t::match_count.

int cm___register_entry ( struct estat * sts )

Inserts the given entry in all allowed matching databases.

Definition at line 658 of file cp_mv.c.

References CM___MATCH_NUM, DEBUGP, estat::entry_status, cm___match_t::entry_type, FS_NEW, cm___match_t::insert, cm___match_t::is_enabled, sstat_t::mode, cm___match_t::name, estat::name, estat::st, and STOPIF.

Referenced by cm__detect().

Here is the caller graph for this function:

int cm___rev_path_to_string	(	char *	url,
		svn_revnum_t	revision,
		char **	string
	)

Formats the revision and url for storage in the hash table.

*string must not be free()ed by the caller.

Definition at line 1029 of file cp_mv.c.

References BUG_ON, cch__entry_set(), and STOPIF.

Referenced by cm___make_copy().

Here is the call graph for this function:

Here is the caller graph for this function:

int cm___string_to_rev_path	(	char *	string,
		char **	out_url,
		svn_revnum_t *	orev
	)

Converts the (internally stored) string into revision/URL.

The string is not freed; that has to be done by the caller.

Definition at line 998 of file cp_mv.c.

References DEBUGP, and hlp__skip_ws().

Referenced by cm___dump_list(), and cm___get_base_source().

Here is the call graph for this function:

Here is the caller graph for this function:

int cm__detect	(	struct estat *	root,
		int	argc,
		char *	argv[]
	)

-.

Todo:

Do we really need to load the URLs here? They're needed for associating the entries - but maybe we should do that two-way:

just read intnum , and store it again
or process to (struct url_t*).

Well, reading the URLs doesn't cost that much ...

Definition at line 918 of file cp_mv.c.

References CHCHECK_NONE, CM___MATCH_NUM, cm___register_entry(), cm__find_copied(), copydetect_count, cm___match_t::db, DEBUGP, cm___match_t::filename, HASH_TEMPORARY, hsh__close(), hsh__new(), cm___match_t::is_enabled, cm___match_t::is_expensive, cm___match_t::name, OPT__CHANGECHECK, OPT__COPYFROM_EXP, opt__is_verbose(), opt__set_int(), opt_recursive, PRIO_MUSTHAVE, STOPIF, STOPIF_CODE_EPIPE, STOPIF_CODE_ERR, url__load_list(), waa__find_common_base(), waa__read_or_build_tree(), and wc_path.

Here is the call graph for this function:

int cm__find_copied ( struct estat * root )

After loading known entries try to find some match for every new entry.

Definition at line 869 of file cp_mv.c.

References estat::by_inode, cm__find_dir_source(), cm__find_file_source(), DEBUGP, estat::entry_status, FS_CHANGED, FS_CHILD_CHANGED, FS_NEW, sstat_t::mode, estat::name, estat::st, and STOPIF.

Referenced by cm__detect().

Here is the call graph for this function:

Here is the caller graph for this function:

int cm__find_dir_source ( struct estat * dir )

Definition at line 834 of file cp_mv.c.

References cm___match(), and STOPIF.

Referenced by cm__find_copied().

Here is the call graph for this function:

Here is the caller graph for this function:

int cm__find_file_source ( struct estat * file )

Definition at line 846 of file cp_mv.c.

References cm___match(), cs__compare_file(), DEBUGP, estat::name, ops__build_path(), and STOPIF.

Referenced by cm__find_copied().

Here is the call graph for this function:

Here is the caller graph for this function:

int cm__get_source	(	struct estat *	sts,
		char *	name,
		char **	src_url,
		svn_revnum_t *	src_rev,
		int	register_for_cleanup
	)

-.

Returns the source of the given entry.

Wrapper around cm___get_base_source() and cm___get_sub_source().

If *src_url is needed, it is allocated and must be free()ed after use.

If name is not given, it has to be calculated.

Both src_name and src_rev are optional. These are always set; if no source is defined, they're set to NULL, NULL and SVN_INVALID_REVNUM.

Uninitializing should be done via calling with sts==NULL; in this case the register_for_cleanup value is used as success flag.

If no source could be found, ENOENT is returned.

Definition at line 1712 of file cp_mv.c.

References cm___get_base_source(), cm___get_sub_source(), DEBUGP, estat::flags, estat::name, RF_COPY_BASE, and RF_COPY_SUB.

Referenced by cm__uncopy(), df__do_diff(), main(), rev__install_file(), st__print_entry_info(), and st__print_status().

Here is the call graph for this function:

Here is the caller graph for this function:

int cm__uncopy	(	struct estat *	root,
		int	argc,
		char *	argv[]
	)

-.

Definition at line 1308 of file cp_mv.c.

References ac__Usage_this, cm___ignore_impl_copied(), cm__get_source(), DEBUGP, estat::flags, OPS__FAIL_NOT_LIST, ops__traverse(), opt_recursive, RF_COPY_BASE, STOPIF, STOPIF_CODE_ERR, url__load_nonempty_list(), waa__find_common_base(), waa__input_tree(), and waa__output_tree().

Here is the call graph for this function:

int cm__work	(	struct estat *	root,
		int	argc,
		char *	argv[]
	)

-.

Definition at line 1368 of file cp_mv.c.

References cm___dump_list(), cm___make_copy(), DEBUGP, hlp__strdup(), hlp__string_from_filep(), opt__is_verbose(), opt_target_revision, opt_target_revisions_given, parm_dump, parm_load, SFF_RESET_LINENUM, SFF_WHITESPACE, STOPIF, STOPIF_CODE_ERR, url__load_nonempty_list(), waa__find_common_base(), waa__input_tree(), and waa__output_tree().

Here is the call graph for this function:

Variable Documentation

cm___get_list_fn cm___hash_list

Queries the database for the given entry.

Definition at line 360 of file cp_mv.c.

cm___register_fn cm___hash_register

Inserts into hash tables.

Definition at line 358 of file cp_mv.c.

cm___to_datum_t cm___inode_datum

Definition at line 332 of file cp_mv.c.

struct cm___match_t cm___match_array[]

Initial value:

{
    [CM___NAME_F] = { .name="name", .to_key=cm___name_datum, 
        .insert=cm___hash_register, .get_list=cm___hash_list,
        .entry_type=S_IFREG,  .filename=WAA__FILE_NAME_EXT},
    [CM___NAME_D] = { .name="name", .to_key=cm___name_datum, 
        .insert=cm___hash_register, .get_list=cm___hash_list,
        .entry_type=S_IFDIR,  .filename=WAA__DIR_NAME_EXT},

    [CM___DIRLIST] = { .name="dirlist", 
        .get_list=cm___match_children, .format=cm___output_pct,
        .entry_type=S_IFDIR, },

    { .name="md5", .to_key=cm___md5_datum, .is_expensive=1,
        .insert=cm___hash_register, .get_list=cm___hash_list,
        .entry_type=S_IFREG, .filename=WAA__FILE_MD5s_EXT},

    { .name="inode", .to_key=cm___inode_datum, 
        .insert=cm___hash_register, .get_list=cm___hash_list,
        .entry_type=S_IFDIR, .filename=WAA__FILE_INODE_EXT},
    { .name="inode", .to_key=cm___inode_datum, 
        .insert=cm___hash_register, .get_list=cm___hash_list,
        .entry_type=S_IFREG, .filename=WAA__DIR_INODE_EXT},
}

Array with ways for simple matches.

We keep file and directory matching separated; a file cannot be the copyfrom source of a directory, and vice-versa.

The important match types are at the start, as they're directly accessed, too.

Definition at line 411 of file cp_mv.c.

cm___get_list_fn cm___match_children

Match directories by their children.

Definition at line 362 of file cp_mv.c.

cm___to_datum_t cm___md5_datum

Definition at line 331 of file cp_mv.c.

cm___to_datum_t cm___name_datum

Definition at line 333 of file cp_mv.c.

cm___format_fn cm___output_pct

Outputs percent of match.

Definition at line 364 of file cp_mv.c.

int copydetect_count

How many entries could be correlated.

Definition at line 314 of file cp_mv.c.

Referenced by cm___match(), and cm__detect().

cp_mv.c File Reference

Data Structures

Defines

Typedefs

Enumerations

Functions

Variables

Detailed Description

Define Documentation

Typedef Documentation

Enumeration Type Documentation

Function Documentation

Variable Documentation