Code

fast-import: treat filemodify with empty tree as delete
[git.git] / convert.c
index 1144e0b4f5539172eb4043edc8013fb8e69f135d..e41a31e4807e92e210854214d46767b9752c5181 100644 (file)
--- a/convert.c
+++ b/convert.c
@@ -8,7 +8,7 @@
  * This should use the pathname to decide on whether it wants to do some
  * more interesting conversions (automatic gzip/unzip, general format
  * conversions etc etc), but by default it just does automatic CRLF<->LF
- * translation when the "crlf" attribute or "auto_crlf" option is set.
+ * translation when the "text" attribute or "auto_crlf" option is set.
  */
 
 enum action {
@@ -20,12 +20,6 @@ enum action {
        CRLF_AUTO,
 };
 
-enum eol {
-       EOL_UNSET,
-       EOL_LF,
-       EOL_CRLF,
-};
-
 struct text_stat {
        /* NUL, CR, LF and CRLF counts */
        unsigned nul, cr, lf, crlf;
@@ -99,33 +93,55 @@ static int is_binary(unsigned long size, struct text_stat *stats)
        return 0;
 }
 
+static enum eol determine_output_conversion(enum action action) {
+       switch (action) {
+       case CRLF_BINARY:
+               return EOL_UNSET;
+       case CRLF_CRLF:
+               return EOL_CRLF;
+       case CRLF_INPUT:
+               return EOL_LF;
+       case CRLF_GUESS:
+               if (!auto_crlf)
+                       return EOL_UNSET;
+               /* fall through */
+       case CRLF_TEXT:
+       case CRLF_AUTO:
+               if (auto_crlf == AUTO_CRLF_TRUE)
+                       return EOL_CRLF;
+               else if (auto_crlf == AUTO_CRLF_INPUT)
+                       return EOL_LF;
+               else if (eol == EOL_UNSET)
+                       return EOL_NATIVE;
+       }
+       return eol;
+}
+
 static void check_safe_crlf(const char *path, enum action action,
                             struct text_stat *stats, enum safe_crlf checksafe)
 {
        if (!checksafe)
                return;
 
-       if (action == CRLF_INPUT ||
-           (action == CRLF_GUESS && auto_crlf == AUTO_CRLF_INPUT)) {
+       if (determine_output_conversion(action) == EOL_LF) {
                /*
                 * CRLFs would not be restored by checkout:
                 * check if we'd remove CRLFs
                 */
                if (stats->crlf) {
                        if (checksafe == SAFE_CRLF_WARN)
-                               warning("CRLF will be replaced by LF in %s.", path);
+                               warning("CRLF will be replaced by LF in %s.\nThe file will have its original line endings in your working directory.", path);
                        else /* i.e. SAFE_CRLF_FAIL */
                                die("CRLF would be replaced by LF in %s.", path);
                }
-       } else if (action == CRLF_CRLF ||
-                  (action == CRLF_GUESS && auto_crlf == AUTO_CRLF_TRUE)) {
+       } else if (determine_output_conversion(action) == EOL_CRLF) {
                /*
                 * CRLFs would be added by checkout:
                 * check if we have "naked" LFs
                 */
                if (stats->lf != stats->crlf) {
                        if (checksafe == SAFE_CRLF_WARN)
-                               warning("LF will be replaced by CRLF in %s", path);
+                               warning("LF will be replaced by CRLF in %s.\nThe file will have its original line endings in your working directory.", path);
                        else /* i.e. SAFE_CRLF_FAIL */
                                die("LF would be replaced by CRLF in %s", path);
                }
@@ -244,11 +260,7 @@ static int crlf_to_worktree(const char *path, const char *src, size_t len,
        char *to_free = NULL;
        struct text_stat stats;
 
-       if ((action == CRLF_BINARY) || (action == CRLF_INPUT) ||
-           (action != CRLF_CRLF && auto_crlf != AUTO_CRLF_TRUE))
-               return 0;
-
-       if (!len)
+       if (!len || determine_output_conversion(action) != EOL_CRLF)
                return 0;
 
        gather_stats(src, len, &stats);
@@ -307,7 +319,7 @@ struct filter_params {
        const char *cmd;
 };
 
-static int filter_buffer(int fd, void *data)
+static int filter_buffer(int in, int out, void *data)
 {
        /*
         * Spawn cmd and feed the buffer contents through its stdin.
@@ -315,13 +327,15 @@ static int filter_buffer(int fd, void *data)
        struct child_process child_process;
        struct filter_params *params = (struct filter_params *)data;
        int write_err, status;
-       const char *argv[] = { params->cmd, NULL };
+       const char *argv[] = { NULL, NULL };
+
+       argv[0] = params->cmd;
 
        memset(&child_process, 0, sizeof(child_process));
        child_process.argv = argv;
        child_process.use_shell = 1;
        child_process.in = -1;
-       child_process.out = fd;
+       child_process.out = out;
 
        if (start_command(&child_process))
                return error("cannot fork to run external filter %s", params->cmd);
@@ -358,6 +372,7 @@ static int apply_filter(const char *path, const char *src, size_t len,
        memset(&async, 0, sizeof(async));
        async.proc = filter_buffer;
        async.data = &params;
+       async.out = -1;
        params.src = src;
        params.size = len;
        params.cmd = cmd;
@@ -439,12 +454,14 @@ static int read_convert_config(const char *var, const char *value, void *cb)
 
 static void setup_convert_check(struct git_attr_check *check)
 {
+       static struct git_attr *attr_text;
        static struct git_attr *attr_crlf;
        static struct git_attr *attr_eol;
        static struct git_attr *attr_ident;
        static struct git_attr *attr_filter;
 
-       if (!attr_crlf) {
+       if (!attr_text) {
+               attr_text = git_attr("text");
                attr_crlf = git_attr("crlf");
                attr_eol = git_attr("eol");
                attr_ident = git_attr("ident");
@@ -456,6 +473,7 @@ static void setup_convert_check(struct git_attr_check *check)
        check[1].attr = attr_ident;
        check[2].attr = attr_filter;
        check[3].attr = attr_eol;
+       check[4].attr = attr_text;
 }
 
 static int count_ident(const char *cp, unsigned long size)
@@ -493,6 +511,8 @@ static int count_ident(const char *cp, unsigned long size)
                                cnt++;
                                break;
                        }
+                       if (ch == '\n')
+                               break;
                }
        }
        return cnt;
@@ -523,6 +543,11 @@ static int ident_to_git(const char *path, const char *src, size_t len,
                        dollar = memchr(src + 3, '$', len - 3);
                        if (!dollar)
                                break;
+                       if (memchr(src + 3, '\n', dollar - src - 3)) {
+                               /* Line break before the next dollar. */
+                               continue;
+                       }
+
                        memcpy(dst, "Id$", 3);
                        dst += 3;
                        len -= dollar + 1 - src;
@@ -538,7 +563,7 @@ static int ident_to_worktree(const char *path, const char *src, size_t len,
                              struct strbuf *buf, int ident)
 {
        unsigned char sha1[20];
-       char *to_free = NULL, *dollar;
+       char *to_free = NULL, *dollar, *spc;
        int cnt;
 
        if (!ident)
@@ -574,7 +599,10 @@ static int ident_to_worktree(const char *path, const char *src, size_t len,
                } else if (src[2] == ':') {
                        /*
                         * It's possible that an expanded Id has crept its way into the
-                        * repository, we cope with that by stripping the expansion out
+                        * repository, we cope with that by stripping the expansion out.
+                        * This is probably not a good idea, since it will cause changes
+                        * on checkout, which won't go away by stash, but let's keep it
+                        * for git-style ids.
                         */
                        dollar = memchr(src + 3, '$', len - 3);
                        if (!dollar) {
@@ -582,6 +610,20 @@ static int ident_to_worktree(const char *path, const char *src, size_t len,
                                break;
                        }
 
+                       if (memchr(src + 3, '\n', dollar - src - 3)) {
+                               /* Line break before the next dollar. */
+                               continue;
+                       }
+
+                       spc = memchr(src + 4, ' ', dollar - src - 4);
+                       if (spc && spc < dollar-1) {
+                               /* There are spaces in unexpected places.
+                                * This is probably an id from some other
+                                * versioning system. Keep it for now.
+                                */
+                               continue;
+                       }
+
                        len -= dollar + 1 - src;
                        src  = dollar + 1;
                } else {
@@ -651,32 +693,34 @@ static int git_path_check_ident(const char *path, struct git_attr_check *check)
        return !!ATTR_TRUE(value);
 }
 
-enum action determine_action(enum action crlf_attr, enum eol eol_attr) {
-       if (crlf_attr == CRLF_BINARY)
+enum action determine_action(enum action text_attr, enum eol eol_attr) {
+       if (text_attr == CRLF_BINARY)
                return CRLF_BINARY;
        if (eol_attr == EOL_LF)
                return CRLF_INPUT;
        if (eol_attr == EOL_CRLF)
                return CRLF_CRLF;
-       return crlf_attr;
+       return text_attr;
 }
 
 int convert_to_git(const char *path, const char *src, size_t len,
                    struct strbuf *dst, enum safe_crlf checksafe)
 {
-       struct git_attr_check check[4];
+       struct git_attr_check check[5];
        enum action action = CRLF_GUESS;
-       enum eol eol = EOL_UNSET;
+       enum eol eol_attr = EOL_UNSET;
        int ident = 0, ret = 0;
        const char *filter = NULL;
 
        setup_convert_check(check);
        if (!git_checkattr(path, ARRAY_SIZE(check), check)) {
                struct convert_driver *drv;
-               action = git_path_check_crlf(path, check + 0);
+               action = git_path_check_crlf(path, check + 4);
+               if (action == CRLF_GUESS)
+                       action = git_path_check_crlf(path, check + 0);
                ident = git_path_check_ident(path, check + 1);
                drv = git_path_check_convert(path, check + 2);
-               eol = git_path_check_eol(path, check + 3);
+               eol_attr = git_path_check_eol(path, check + 3);
                if (drv && drv->clean)
                        filter = drv->clean;
        }
@@ -686,7 +730,7 @@ int convert_to_git(const char *path, const char *src, size_t len,
                src = dst->buf;
                len = dst->len;
        }
-       action = determine_action(action, eol);
+       action = determine_action(action, eol_attr);
        ret |= crlf_to_git(path, src, len, dst, action, checksafe);
        if (ret) {
                src = dst->buf;
@@ -697,19 +741,21 @@ int convert_to_git(const char *path, const char *src, size_t len,
 
 int convert_to_working_tree(const char *path, const char *src, size_t len, struct strbuf *dst)
 {
-       struct git_attr_check check[4];
+       struct git_attr_check check[5];
        enum action action = CRLF_GUESS;
-       enum eol eol = EOL_UNSET;
+       enum eol eol_attr = EOL_UNSET;
        int ident = 0, ret = 0;
        const char *filter = NULL;
 
        setup_convert_check(check);
        if (!git_checkattr(path, ARRAY_SIZE(check), check)) {
                struct convert_driver *drv;
-               action = git_path_check_crlf(path, check + 0);
+               action = git_path_check_crlf(path, check + 4);
+               if (action == CRLF_GUESS)
+                       action = git_path_check_crlf(path, check + 0);
                ident = git_path_check_ident(path, check + 1);
                drv = git_path_check_convert(path, check + 2);
-               eol = git_path_check_eol(path, check + 3);
+               eol_attr = git_path_check_eol(path, check + 3);
                if (drv && drv->smudge)
                        filter = drv->smudge;
        }
@@ -719,7 +765,7 @@ int convert_to_working_tree(const char *path, const char *src, size_t len, struc
                src = dst->buf;
                len = dst->len;
        }
-       action = determine_action(action, eol);
+       action = determine_action(action, eol_attr);
        ret |= crlf_to_worktree(path, src, len, dst, action);
        if (ret) {
                src = dst->buf;