convert: unify the "auto" handling of CRLF
[gitweb.git] / convert.c
index e4e2877129f4d02db41ec9d31e47e507e86c2e5b..67d69b5c0e22a74e6dc46764beaff9978a78b85a 100644 (file)
--- a/convert.c
+++ b/convert.c
@@ -31,7 +31,7 @@ enum crlf_action {
 
 struct text_stat {
        /* NUL, CR, LF and CRLF counts */
-       unsigned nul, cr, lf, crlf;
+       unsigned nul, lonecr, lonelf, crlf;
 
        /* These are just approximations! */
        unsigned printable, nonprintable;
@@ -46,13 +46,15 @@ static void gather_stats(const char *buf, unsigned long size, struct text_stat *
        for (i = 0; i < size; i++) {
                unsigned char c = buf[i];
                if (c == '\r') {
-                       stats->cr++;
-                       if (i+1 < size && buf[i+1] == '\n')
+                       if (i+1 < size && buf[i+1] == '\n') {
                                stats->crlf++;
+                               i++;
+                       } else
+                               stats->lonecr++;
                        continue;
                }
                if (c == '\n') {
-                       stats->lf++;
+                       stats->lonelf++;
                        continue;
                }
                if (c == 127)
@@ -86,7 +88,7 @@ static void gather_stats(const char *buf, unsigned long size, struct text_stat *
  */
 static int convert_is_binary(unsigned long size, const struct text_stat *stats)
 {
-       if (stats->cr != stats->crlf)
+       if (stats->lonecr)
                return 1;
        if (stats->nul)
                return 1;
@@ -98,19 +100,18 @@ static int convert_is_binary(unsigned long size, const struct text_stat *stats)
 static unsigned int gather_convert_stats(const char *data, unsigned long size)
 {
        struct text_stat stats;
+       int ret = 0;
        if (!data || !size)
                return 0;
        gather_stats(data, size, &stats);
        if (convert_is_binary(size, &stats))
-               return CONVERT_STAT_BITS_BIN;
-       else if (stats.crlf && stats.crlf == stats.lf)
-               return CONVERT_STAT_BITS_TXT_CRLF;
-       else if (stats.crlf && stats.lf)
-               return CONVERT_STAT_BITS_TXT_CRLF | CONVERT_STAT_BITS_TXT_LF;
-       else if (stats.lf)
-               return CONVERT_STAT_BITS_TXT_LF;
-       else
-               return 0;
+               ret |= CONVERT_STAT_BITS_BIN;
+       if (stats.crlf)
+               ret |= CONVERT_STAT_BITS_TXT_CRLF;
+       if (stats.lonelf)
+               ret |=  CONVERT_STAT_BITS_TXT_LF;
+
+       return ret;
 }
 
 static const char *gather_convert_stats_ascii(const char *data, unsigned long size)
@@ -175,7 +176,9 @@ static enum eol output_eol(enum crlf_action crlf_action)
                return EOL_LF;
        case CRLF_UNDEFINED:
        case CRLF_AUTO_CRLF:
+               return EOL_CRLF;
        case CRLF_AUTO_INPUT:
+               return EOL_LF;
        case CRLF_TEXT:
        case CRLF_AUTO:
                /* fall through */
@@ -207,7 +210,7 @@ static void check_safe_crlf(const char *path, enum crlf_action crlf_action,
                 * CRLFs would be added by checkout:
                 * check if we have "naked" LFs
                 */
-               if (stats->lf != stats->crlf) {
+               if (stats->lonelf) {
                        if (checksafe == SAFE_CRLF_WARN)
                                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 */
@@ -253,21 +256,19 @@ static int crlf_to_git(const char *path, const char *src, size_t len,
        if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) {
                if (convert_is_binary(len, &stats))
                        return 0;
-
-               if (crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) {
-                       /*
-                        * If the file in the index has any CR in it, do not convert.
-                        * This is the new safer autocrlf handling.
-                        */
-                       if (has_cr_in_index(path))
-                               return 0;
-               }
+               /*
+                * If the file in the index has any CR in it, do not convert.
+                * This is the new safer autocrlf handling.
+                */
+               if (checksafe == SAFE_CRLF_RENORMALIZE)
+                       checksafe = SAFE_CRLF_FALSE;
+               else if (has_cr_in_index(path))
+                       return 0;
        }
-
        check_safe_crlf(path, crlf_action, &stats, checksafe);
 
-       /* Optimization: No CR? Nothing to convert, regardless. */
-       if (!stats.cr)
+       /* Optimization: No CRLF? Nothing to convert, regardless. */
+       if (!stats.crlf)
                return 0;
 
        /*
@@ -314,21 +315,15 @@ static int crlf_to_worktree(const char *path, const char *src, size_t len,
 
        gather_stats(src, len, &stats);
 
-       /* No LF? Nothing to convert, regardless. */
-       if (!stats.lf)
-               return 0;
-
-       /* Was it already in CRLF format? */
-       if (stats.lf == stats.crlf)
+       /* No "naked" LF? Nothing to convert, regardless. */
+       if (!stats.lonelf)
                return 0;
 
        if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) {
-               if (crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) {
-                       /* If we have any CR or CRLF line endings, we do not touch it */
-                       /* This is the new safer autocrlf-handling */
-                       if (stats.cr > 0 || stats.crlf > 0)
-                               return 0;
-               }
+               /* If we have any CR or CRLF line endings, we do not touch it */
+               /* This is the new safer autocrlf-handling */
+               if (stats.lonecr || stats.crlf )
+                       return 0;
 
                if (convert_is_binary(len, &stats))
                        return 0;
@@ -338,7 +333,7 @@ static int crlf_to_worktree(const char *path, const char *src, size_t len,
        if (src == buf->buf)
                to_free = strbuf_detach(buf, NULL);
 
-       strbuf_grow(buf, len + stats.lf - stats.crlf);
+       strbuf_grow(buf, len + stats.lonelf);
        for (;;) {
                const char *nl = memchr(src, '\n', len);
                if (!nl)
@@ -444,7 +439,7 @@ static int apply_filter(const char *path, const char *src, size_t len, int fd,
        struct async async;
        struct filter_params params;
 
-       if (!cmd)
+       if (!cmd || !*cmd)
                return 0;
 
        if (!dst)
@@ -711,7 +706,7 @@ static enum crlf_action git_path_check_crlf(struct git_attr_check *check)
        const char *value = check->value;
 
        if (ATTR_TRUE(value))
-               return text_eol_is_crlf() ? CRLF_TEXT_CRLF : CRLF_TEXT_INPUT;
+               return CRLF_TEXT;
        else if (ATTR_FALSE(value))
                return CRLF_BINARY;
        else if (ATTR_UNSET(value))
@@ -781,20 +776,24 @@ static void convert_attrs(struct conv_attrs *ca, const char *path)
        }
 
        if (!git_check_attr(path, NUM_CONV_ATTRS, ccheck)) {
-               enum eol eol_attr;
                ca->crlf_action = git_path_check_crlf(ccheck + 4);
                if (ca->crlf_action == CRLF_UNDEFINED)
                        ca->crlf_action = git_path_check_crlf(ccheck + 0);
                ca->attr_action = ca->crlf_action;
                ca->ident = git_path_check_ident(ccheck + 1);
                ca->drv = git_path_check_convert(ccheck + 2);
-               if (ca->crlf_action == CRLF_BINARY)
-                       return;
-               eol_attr = git_path_check_eol(ccheck + 3);
-               if (eol_attr == EOL_LF)
-                       ca->crlf_action = CRLF_TEXT_INPUT;
-               else if (eol_attr == EOL_CRLF)
-                       ca->crlf_action = CRLF_TEXT_CRLF;
+               if (ca->crlf_action != CRLF_BINARY) {
+                       enum eol eol_attr = git_path_check_eol(ccheck + 3);
+                       if (ca->crlf_action == CRLF_AUTO && eol_attr == EOL_LF)
+                               ca->crlf_action = CRLF_AUTO_INPUT;
+                       else if (ca->crlf_action == CRLF_AUTO && eol_attr == EOL_CRLF)
+                               ca->crlf_action = CRLF_AUTO_CRLF;
+                       else if (eol_attr == EOL_LF)
+                               ca->crlf_action = CRLF_TEXT_INPUT;
+                       else if (eol_attr == EOL_CRLF)
+                               ca->crlf_action = CRLF_TEXT_CRLF;
+               }
+               ca->attr_action = ca->crlf_action;
        } else {
                ca->drv = NULL;
                ca->crlf_action = CRLF_UNDEFINED;
@@ -848,9 +847,9 @@ const char *get_convert_attr_ascii(const char *path)
        case CRLF_AUTO:
                return "text=auto";
        case CRLF_AUTO_CRLF:
-               return "text=auto eol=crlf"; /* This is not supported yet */
+               return "text=auto eol=crlf";
        case CRLF_AUTO_INPUT:
-               return "text=auto eol=lf"; /* This is not supported yet */
+               return "text=auto eol=lf";
        }
        return "";
 }
@@ -952,7 +951,7 @@ int renormalize_buffer(const char *path, const char *src, size_t len, struct str
                src = dst->buf;
                len = dst->len;
        }
-       return ret | convert_to_git(path, src, len, dst, SAFE_CRLF_FALSE);
+       return ret | convert_to_git(path, src, len, dst, SAFE_CRLF_RENORMALIZE);
 }
 
 /*****************************************************************
@@ -1383,27 +1382,22 @@ static struct stream_filter *ident_filter(const unsigned char *sha1)
 struct stream_filter *get_stream_filter(const char *path, const unsigned char *sha1)
 {
        struct conv_attrs ca;
-       enum crlf_action crlf_action;
        struct stream_filter *filter = NULL;
 
        convert_attrs(&ca, path);
-
        if (ca.drv && (ca.drv->smudge || ca.drv->clean))
-               return filter;
+               return NULL;
+
+       if (ca.crlf_action == CRLF_AUTO || ca.crlf_action == CRLF_AUTO_CRLF)
+               return NULL;
 
        if (ca.ident)
                filter = ident_filter(sha1);
 
-       crlf_action = ca.crlf_action;
-
-       if ((crlf_action == CRLF_BINARY) ||
-                       crlf_action == CRLF_AUTO_INPUT ||
-                       (crlf_action == CRLF_TEXT_INPUT))
-               filter = cascade_filter(filter, &null_filter_singleton);
-
-       else if (output_eol(crlf_action) == EOL_CRLF &&
-                !(crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_CRLF))
+       if (output_eol(ca.crlf_action) == EOL_CRLF)
                filter = cascade_filter(filter, lf_to_crlf_filter());
+       else
+               filter = cascade_filter(filter, &null_filter_singleton);
 
        return filter;
 }