Document gitattributes(5)
[gitweb.git] / convert.c
index 898bfe3eb219618e746afbe02eb4a8756eccc6aa..da64253a16e51881decd3e31d5763892f4847e5f 100644 (file)
--- a/convert.c
+++ b/convert.c
@@ -1,4 +1,6 @@
 #include "cache.h"
+#include "attr.h"
+
 /*
  * convert.c - convert a file when checking it out and checking it in.
  *
@@ -8,6 +10,11 @@
  * 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;
@@ -72,18 +79,13 @@ static int is_binary(unsigned long size, struct text_stat *stats)
        return 0;
 }
 
-int convert_to_git(const char *path, char **bufp, unsigned long *sizep)
+static int crlf_to_git(const char *path, char **bufp, unsigned long *sizep, int action)
 {
        char *buffer, *nbuf;
        unsigned long size, nsize;
        struct text_stat stats;
 
-       /*
-        * FIXME! Other pluggable conversions should go here,
-        * based on filename patterns. Right now we just do the
-        * stupid auto-CRLF one.
-        */
-       if (!auto_crlf)
+       if ((action == CRLF_BINARY) || (action == CRLF_GUESS && !auto_crlf))
                return 0;
 
        size = *sizep;
@@ -97,19 +99,21 @@ int convert_to_git(const char *path, char **bufp, unsigned long *sizep)
        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;
+       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(size, &stats))
+                       return 0;
+       }
 
        /*
         * Ok, allocate a new buffer, fill it in, and return true
@@ -119,28 +123,38 @@ int convert_to_git(const char *path, char **bufp, unsigned long *sizep)
        nbuf = xmalloc(nsize);
        *bufp = nbuf;
        *sizep = nsize;
-       do {
-               unsigned char c = *buffer++;
-               if (c != '\r')
-                       *nbuf++ = c;
-       } while (--size);
+
+       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 = *buffer++;
+                       if (c != '\r')
+                               *nbuf++ = c;
+               } while (--size);
+       } else {
+               do {
+                       unsigned char c = *buffer++;
+                       if (! (c == '\r' && (1 < size && *buffer == '\n')))
+                               *nbuf++ = c;
+               } while (--size);
+       }
 
        return 1;
 }
 
-int convert_to_working_tree(const char *path, char **bufp, unsigned long *sizep)
+static int crlf_to_worktree(const char *path, char **bufp, unsigned long *sizep, int action)
 {
        char *buffer, *nbuf;
        unsigned long size, nsize;
        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) ||
+           (action == CRLF_GUESS && auto_crlf <= 0))
                return 0;
 
        size = *sizep;
@@ -158,12 +172,14 @@ int convert_to_working_tree(const char *path, char **bufp, unsigned long *sizep)
        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)
-               return 0;
+       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(size, &stats))
-               return 0;
+               if (is_binary(size, &stats))
+                       return 0;
+       }
 
        /*
         * Ok, allocate a new buffer, fill it in, and return true
@@ -184,3 +200,43 @@ int convert_to_working_tree(const char *path, char **bufp, unsigned long *sizep)
 
        return 1;
 }
+
+static void setup_crlf_check(struct git_attr_check *check)
+{
+       static struct git_attr *attr_crlf;
+
+       if (!attr_crlf)
+               attr_crlf = git_attr("crlf", 4);
+       check->attr = attr_crlf;
+}
+
+static int git_path_check_crlf(const char *path)
+{
+       struct git_attr_check attr_crlf_check;
+
+       setup_crlf_check(&attr_crlf_check);
+
+       if (!git_checkattr(path, 1, &attr_crlf_check)) {
+               const char *value = attr_crlf_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;
+               /* fallthru */
+       }
+       return CRLF_GUESS;
+}
+
+int convert_to_git(const char *path, char **bufp, unsigned long *sizep)
+{
+       return crlf_to_git(path, bufp, sizep, git_path_check_crlf(path));
+}
+
+int convert_to_working_tree(const char *path, char **bufp, unsigned long *sizep)
+{
+       return crlf_to_worktree(path, bufp, sizep, git_path_check_crlf(path));
+}