pull --rebase: exit early when the working directory is dirty
[gitweb.git] / convert.c
index 20c744aa23652b8b93ea96137a668592c6e28c2d..d8c94cb3edeceff7a79b2f5e6ba7cad250e6b301 100644 (file)
--- a/convert.c
+++ b/convert.c
@@ -1,5 +1,6 @@
 #include "cache.h"
 #include "attr.h"
+#include "run-command.h"
 
 /*
  * convert.c - convert a file when checking it out and checking it in.
  * translation when the "auto_crlf" option is set.
  */
 
+#define CRLF_GUESS     (-1)
+#define CRLF_BINARY    0
+#define CRLF_TEXT      1
+#define CRLF_INPUT     2
+
 struct text_stat {
-       /* CR, LF and CRLF counts */
-       unsigned cr, lf, crlf;
+       /* NUL, CR, LF and CRLF counts */
+       unsigned nul, cr, lf, crlf;
 
        /* These are just approximations! */
        unsigned printable, nonprintable;
@@ -45,6 +51,9 @@ static void gather_stats(const char *buf, unsigned long size, struct text_stat *
                        case '\b': case '\t': case '\033': case '\014':
                                stats->printable++;
                                break;
+                       case 0:
+                               stats->nul++;
+                               /* fall through */
                        default:
                                stats->nonprintable++;
                        }
@@ -60,6 +69,8 @@ static void gather_stats(const char *buf, unsigned long size, struct text_stat *
 static int is_binary(unsigned long size, struct text_stat *stats)
 {
 
+       if (stats->nul)
+               return 1;
        if ((stats->printable >> 7) < stats->nonprintable)
                return 1;
        /*
@@ -74,78 +85,110 @@ static int is_binary(unsigned long size, struct text_stat *stats)
        return 0;
 }
 
-static int autocrlf_to_git(const char *path, char **bufp, unsigned long *sizep)
+static void check_safe_crlf(const char *path, int action,
+                            struct text_stat *stats, enum safe_crlf checksafe)
+{
+       if (!checksafe)
+               return;
+
+       if (action == CRLF_INPUT || auto_crlf <= 0) {
+               /*
+                * 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);
+                       else /* i.e. SAFE_CRLF_FAIL */
+                               die("CRLF would be replaced by LF in %s.", path);
+               }
+       } else if (auto_crlf > 0) {
+               /*
+                * 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);
+                       else /* i.e. SAFE_CRLF_FAIL */
+                               die("LF would be replaced by CRLF in %s", path);
+               }
+       }
+}
+
+static int crlf_to_git(const char *path, const char *src, size_t len,
+                       struct strbuf *buf, int action, enum safe_crlf checksafe)
 {
-       char *buffer, *nbuf;
-       unsigned long size, nsize;
        struct text_stat stats;
+       char *dst;
 
-       if (!auto_crlf)
+       if ((action == CRLF_BINARY) || !auto_crlf || !len)
                return 0;
 
-       size = *sizep;
-       if (!size)
-               return 0;
-       buffer = *bufp;
+       gather_stats(src, len, &stats);
+
+       if (action == CRLF_GUESS) {
+               /*
+                * We're currently not going to even try to convert stuff
+                * that has bare CR characters. Does anybody do that crazy
+                * stuff?
+                */
+               if (stats.cr != stats.crlf)
+                       return 0;
+
+               /*
+                * And add some heuristics for binary vs text, of course...
+                */
+               if (is_binary(len, &stats))
+                       return 0;
+       }
 
-       gather_stats(buffer, size, &stats);
+       check_safe_crlf(path, action, &stats, checksafe);
 
-       /* No CR? Nothing to convert, regardless. */
+       /* Optimization: No CR? Nothing to convert, regardless. */
        if (!stats.cr)
                return 0;
 
-       /*
-        * We're currently not going to even try to convert stuff
-        * that has bare CR characters. Does anybody do that crazy
-        * stuff?
-        */
-       if (stats.cr != stats.crlf)
-               return 0;
-
-       /*
-        * And add some heuristics for binary vs text, of course...
-        */
-       if (is_binary(size, &stats))
-               return 0;
-
-       /*
-        * Ok, allocate a new buffer, fill it in, and return true
-        * to let the caller know that we switched buffers on it.
-        */
-       nsize = size - stats.crlf;
-       nbuf = xmalloc(nsize);
-       *bufp = nbuf;
-       *sizep = nsize;
-       do {
-               unsigned char c = *buffer++;
-               if (c != '\r')
-                       *nbuf++ = c;
-       } while (--size);
-
+       /* only grow if not in place */
+       if (strbuf_avail(buf) + buf->len < len)
+               strbuf_grow(buf, len - buf->len);
+       dst = buf->buf;
+       if (action == CRLF_GUESS) {
+               /*
+                * If we guessed, we already know we rejected a file with
+                * lone CR, and we can strip a CR without looking at what
+                * follow it.
+                */
+               do {
+                       unsigned char c = *src++;
+                       if (c != '\r')
+                               *dst++ = c;
+               } while (--len);
+       } else {
+               do {
+                       unsigned char c = *src++;
+                       if (! (c == '\r' && (1 < len && *src == '\n')))
+                               *dst++ = c;
+               } while (--len);
+       }
+       strbuf_setlen(buf, dst - buf->buf);
        return 1;
 }
 
-static int autocrlf_to_working_tree(const char *path, char **bufp, unsigned long *sizep)
+static int crlf_to_worktree(const char *path, const char *src, size_t len,
+                            struct strbuf *buf, int action)
 {
-       char *buffer, *nbuf;
-       unsigned long size, nsize;
+       char *to_free = NULL;
        struct text_stat stats;
-       unsigned char last;
 
-       /*
-        * FIXME! Other pluggable conversions should go here,
-        * based on filename patterns. Right now we just do the
-        * stupid auto-CRLF one.
-        */
-       if (auto_crlf <= 0)
+       if ((action == CRLF_BINARY) || (action == CRLF_INPUT) ||
+           auto_crlf <= 0)
                return 0;
 
-       size = *sizep;
-       if (!size)
+       if (!len)
                return 0;
-       buffer = *bufp;
 
-       gather_stats(buffer, size, &stats);
+       gather_stats(src, len, &stats);
 
        /* No LF? Nothing to convert, regardless. */
        if (!stats.lf)
@@ -155,67 +198,435 @@ static int autocrlf_to_working_tree(const char *path, char **bufp, unsigned long
        if (stats.lf == stats.crlf)
                return 0;
 
-       /* If we have any bare CR characters, we're not going to touch it */
-       if (stats.cr != stats.crlf)
+       if (action == CRLF_GUESS) {
+               /* If we have any bare CR characters, we're not going to touch it */
+               if (stats.cr != stats.crlf)
+                       return 0;
+
+               if (is_binary(len, &stats))
+                       return 0;
+       }
+
+       /* are we "faking" in place editing ? */
+       if (src == buf->buf)
+               to_free = strbuf_detach(buf, NULL);
+
+       strbuf_grow(buf, len + stats.lf - stats.crlf);
+       for (;;) {
+               const char *nl = memchr(src, '\n', len);
+               if (!nl)
+                       break;
+               if (nl > src && nl[-1] == '\r') {
+                       strbuf_add(buf, src, nl + 1 - src);
+               } else {
+                       strbuf_add(buf, src, nl - src);
+                       strbuf_addstr(buf, "\r\n");
+               }
+               len -= nl + 1 - src;
+               src  = nl + 1;
+       }
+       strbuf_add(buf, src, len);
+
+       free(to_free);
+       return 1;
+}
+
+struct filter_params {
+       const char *src;
+       unsigned long size;
+       const char *cmd;
+};
+
+static int filter_buffer(int fd, void *data)
+{
+       /*
+        * Spawn cmd and feed the buffer contents through its stdin.
+        */
+       struct child_process child_process;
+       struct filter_params *params = (struct filter_params *)data;
+       int write_err, status;
+       const char *argv[] = { "sh", "-c", params->cmd, NULL };
+
+       memset(&child_process, 0, sizeof(child_process));
+       child_process.argv = argv;
+       child_process.in = -1;
+       child_process.out = fd;
+
+       if (start_command(&child_process))
+               return error("cannot fork to run external filter %s", params->cmd);
+
+       write_err = (write_in_full(child_process.in, params->src, params->size) < 0);
+       if (close(child_process.in))
+               write_err = 1;
+       if (write_err)
+               error("cannot feed the input to external filter %s", params->cmd);
+
+       status = finish_command(&child_process);
+       if (status)
+               error("external filter %s failed %d", params->cmd, -status);
+       return (write_err || status);
+}
+
+static int apply_filter(const char *path, const char *src, size_t len,
+                        struct strbuf *dst, const char *cmd)
+{
+       /*
+        * Create a pipeline to have the command filter the buffer's
+        * contents.
+        *
+        * (child --> cmd) --> us
+        */
+       int ret = 1;
+       struct strbuf nbuf;
+       struct async async;
+       struct filter_params params;
+
+       if (!cmd)
                return 0;
 
-       if (is_binary(size, &stats))
+       memset(&async, 0, sizeof(async));
+       async.proc = filter_buffer;
+       async.data = &params;
+       params.src = src;
+       params.size = len;
+       params.cmd = cmd;
+
+       fflush(NULL);
+       if (start_async(&async))
+               return 0;       /* error was already reported */
+
+       strbuf_init(&nbuf, 0);
+       if (strbuf_read(&nbuf, async.out, len) < 0) {
+               error("read from external filter %s failed", cmd);
+               ret = 0;
+       }
+       if (close(async.out)) {
+               error("read from external filter %s failed", cmd);
+               ret = 0;
+       }
+       if (finish_async(&async)) {
+               error("external filter %s failed", cmd);
+               ret = 0;
+       }
+
+       if (ret) {
+               strbuf_swap(dst, &nbuf);
+       }
+       strbuf_release(&nbuf);
+       return ret;
+}
+
+static struct convert_driver {
+       const char *name;
+       struct convert_driver *next;
+       char *smudge;
+       char *clean;
+} *user_convert, **user_convert_tail;
+
+static int read_convert_config(const char *var, const char *value)
+{
+       const char *ep, *name;
+       int namelen;
+       struct convert_driver *drv;
+
+       /*
+        * External conversion drivers are configured using
+        * "filter.<name>.variable".
+        */
+       if (prefixcmp(var, "filter.") || (ep = strrchr(var, '.')) == var + 6)
                return 0;
+       name = var + 7;
+       namelen = ep - name;
+       for (drv = user_convert; drv; drv = drv->next)
+               if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
+                       break;
+       if (!drv) {
+               drv = xcalloc(1, sizeof(struct convert_driver));
+               drv->name = xmemdupz(name, namelen);
+               *user_convert_tail = drv;
+               user_convert_tail = &(drv->next);
+       }
+
+       ep++;
 
        /*
-        * Ok, allocate a new buffer, fill it in, and return true
-        * to let the caller know that we switched buffers on it.
+        * filter.<name>.smudge and filter.<name>.clean specifies
+        * the command line:
+        *
+        *      command-line
+        *
+        * The command-line will not be interpolated in any way.
         */
-       nsize = size + stats.lf - stats.crlf;
-       nbuf = xmalloc(nsize);
-       *bufp = nbuf;
-       *sizep = nsize;
-       last = 0;
-       do {
-               unsigned char c = *buffer++;
-               if (c == '\n' && last != '\r')
-                       *nbuf++ = '\r';
-               *nbuf++ = c;
-               last = c;
-       } while (--size);
 
-       return 1;
+       if (!strcmp("smudge", ep)) {
+               if (!value)
+                       return config_error_nonbool(var);
+               drv->smudge = strdup(value);
+               return 0;
+       }
+
+       if (!strcmp("clean", ep)) {
+               if (!value)
+                       return config_error_nonbool(var);
+               drv->clean = strdup(value);
+               return 0;
+       }
+       return 0;
 }
 
-static void setup_crlf_check(struct git_attr_check *check)
+static void setup_convert_check(struct git_attr_check *check)
 {
        static struct git_attr *attr_crlf;
+       static struct git_attr *attr_ident;
+       static struct git_attr *attr_filter;
 
-       if (!attr_crlf)
+       if (!attr_crlf) {
                attr_crlf = git_attr("crlf", 4);
-       check->attr = attr_crlf;
+               attr_ident = git_attr("ident", 5);
+               attr_filter = git_attr("filter", 6);
+               user_convert_tail = &user_convert;
+               git_config(read_convert_config);
+       }
+       check[0].attr = attr_crlf;
+       check[1].attr = attr_ident;
+       check[2].attr = attr_filter;
 }
 
-static int git_path_is_binary(const char *path)
+static int count_ident(const char *cp, unsigned long size)
 {
-       struct git_attr_check attr_crlf_check;
-
-       setup_crlf_check(&attr_crlf_check);
-
        /*
-        * If crlf is not mentioned, default to autocrlf;
-        * disable autocrlf only when crlf attribute is explicitly
-        * unset.
+        * "$Id: 0000000000000000000000000000000000000000 $" <=> "$Id$"
         */
-       return (!git_checkattr(path, 1, &attr_crlf_check) &&
-               (0 == attr_crlf_check.isset));
+       int cnt = 0;
+       char ch;
+
+       while (size) {
+               ch = *cp++;
+               size--;
+               if (ch != '$')
+                       continue;
+               if (size < 3)
+                       break;
+               if (memcmp("Id", cp, 2))
+                       continue;
+               ch = cp[2];
+               cp += 3;
+               size -= 3;
+               if (ch == '$')
+                       cnt++; /* $Id$ */
+               if (ch != ':')
+                       continue;
+
+               /*
+                * "$Id: ... "; scan up to the closing dollar sign and discard.
+                */
+               while (size) {
+                       ch = *cp++;
+                       size--;
+                       if (ch == '$') {
+                               cnt++;
+                               break;
+                       }
+               }
+       }
+       return cnt;
 }
 
-int convert_to_git(const char *path, char **bufp, unsigned long *sizep)
+static int ident_to_git(const char *path, const char *src, size_t len,
+                        struct strbuf *buf, int ident)
 {
-       if (git_path_is_binary(path))
+       char *dst, *dollar;
+
+       if (!ident || !count_ident(src, len))
                return 0;
-       return autocrlf_to_git(path, bufp, sizep);
+
+       /* only grow if not in place */
+       if (strbuf_avail(buf) + buf->len < len)
+               strbuf_grow(buf, len - buf->len);
+       dst = buf->buf;
+       for (;;) {
+               dollar = memchr(src, '$', len);
+               if (!dollar)
+                       break;
+               memcpy(dst, src, dollar + 1 - src);
+               dst += dollar + 1 - src;
+               len -= dollar + 1 - src;
+               src  = dollar + 1;
+
+               if (len > 3 && !memcmp(src, "Id:", 3)) {
+                       dollar = memchr(src + 3, '$', len - 3);
+                       if (!dollar)
+                               break;
+                       memcpy(dst, "Id$", 3);
+                       dst += 3;
+                       len -= dollar + 1 - src;
+                       src  = dollar + 1;
+               }
+       }
+       memcpy(dst, src, len);
+       strbuf_setlen(buf, dst + len - buf->buf);
+       return 1;
 }
 
-int convert_to_working_tree(const char *path, char **bufp, unsigned long *sizep)
+static int ident_to_worktree(const char *path, const char *src, size_t len,
+                             struct strbuf *buf, int ident)
 {
-       if (git_path_is_binary(path))
+       unsigned char sha1[20];
+       char *to_free = NULL, *dollar;
+       int cnt;
+
+       if (!ident)
                return 0;
-       return autocrlf_to_working_tree(path, bufp, sizep);
+
+       cnt = count_ident(src, len);
+       if (!cnt)
+               return 0;
+
+       /* are we "faking" in place editing ? */
+       if (src == buf->buf)
+               to_free = strbuf_detach(buf, NULL);
+       hash_sha1_file(src, len, "blob", sha1);
+
+       strbuf_grow(buf, len + cnt * 43);
+       for (;;) {
+               /* step 1: run to the next '$' */
+               dollar = memchr(src, '$', len);
+               if (!dollar)
+                       break;
+               strbuf_add(buf, src, dollar + 1 - src);
+               len -= dollar + 1 - src;
+               src  = dollar + 1;
+
+               /* step 2: does it looks like a bit like Id:xxx$ or Id$ ? */
+               if (len < 3 || memcmp("Id", src, 2))
+                       continue;
+
+               /* step 3: skip over Id$ or Id:xxxxx$ */
+               if (src[2] == '$') {
+                       src += 3;
+                       len -= 3;
+               } else if (src[2] == ':') {
+                       /*
+                        * It's possible that an expanded Id has crept its way into the
+                        * repository, we cope with that by stripping the expansion out
+                        */
+                       dollar = memchr(src + 3, '$', len - 3);
+                       if (!dollar) {
+                               /* incomplete keyword, no more '$', so just quit the loop */
+                               break;
+                       }
+
+                       len -= dollar + 1 - src;
+                       src  = dollar + 1;
+               } else {
+                       /* it wasn't a "Id$" or "Id:xxxx$" */
+                       continue;
+               }
+
+               /* step 4: substitute */
+               strbuf_addstr(buf, "Id: ");
+               strbuf_add(buf, sha1_to_hex(sha1), 40);
+               strbuf_addstr(buf, " $");
+       }
+       strbuf_add(buf, src, len);
+
+       free(to_free);
+       return 1;
+}
+
+static int git_path_check_crlf(const char *path, struct git_attr_check *check)
+{
+       const char *value = check->value;
+
+       if (ATTR_TRUE(value))
+               return CRLF_TEXT;
+       else if (ATTR_FALSE(value))
+               return CRLF_BINARY;
+       else if (ATTR_UNSET(value))
+               ;
+       else if (!strcmp(value, "input"))
+               return CRLF_INPUT;
+       return CRLF_GUESS;
+}
+
+static struct convert_driver *git_path_check_convert(const char *path,
+                                            struct git_attr_check *check)
+{
+       const char *value = check->value;
+       struct convert_driver *drv;
+
+       if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
+               return NULL;
+       for (drv = user_convert; drv; drv = drv->next)
+               if (!strcmp(value, drv->name))
+                       return drv;
+       return NULL;
+}
+
+static int git_path_check_ident(const char *path, struct git_attr_check *check)
+{
+       const char *value = check->value;
+
+       return !!ATTR_TRUE(value);
+}
+
+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[3];
+       int crlf = CRLF_GUESS;
+       int ident = 0, ret = 0;
+       char *filter = NULL;
+
+       setup_convert_check(check);
+       if (!git_checkattr(path, ARRAY_SIZE(check), check)) {
+               struct convert_driver *drv;
+               crlf = git_path_check_crlf(path, check + 0);
+               ident = git_path_check_ident(path, check + 1);
+               drv = git_path_check_convert(path, check + 2);
+               if (drv && drv->clean)
+                       filter = drv->clean;
+       }
+
+       ret |= apply_filter(path, src, len, dst, filter);
+       if (ret) {
+               src = dst->buf;
+               len = dst->len;
+       }
+       ret |= crlf_to_git(path, src, len, dst, crlf, checksafe);
+       if (ret) {
+               src = dst->buf;
+               len = dst->len;
+       }
+       return ret | ident_to_git(path, src, len, dst, ident);
+}
+
+int convert_to_working_tree(const char *path, const char *src, size_t len, struct strbuf *dst)
+{
+       struct git_attr_check check[3];
+       int crlf = CRLF_GUESS;
+       int ident = 0, ret = 0;
+       char *filter = NULL;
+
+       setup_convert_check(check);
+       if (!git_checkattr(path, ARRAY_SIZE(check), check)) {
+               struct convert_driver *drv;
+               crlf = git_path_check_crlf(path, check + 0);
+               ident = git_path_check_ident(path, check + 1);
+               drv = git_path_check_convert(path, check + 2);
+               if (drv && drv->smudge)
+                       filter = drv->smudge;
+       }
+
+       ret |= ident_to_worktree(path, src, len, dst, ident);
+       if (ret) {
+               src = dst->buf;
+               len = dst->len;
+       }
+       ret |= crlf_to_worktree(path, src, len, dst, crlf);
+       if (ret) {
+               src = dst->buf;
+               len = dst->len;
+       }
+       return ret | apply_filter(path, src, len, dst, filter);
 }