file: restore old behavior for file:////foo/bar URLs

curl 7.57.0 and up interpret this according to Appendix E.3.2 of RFC
8089 but then returns an error saying this is unimplemented. This is
actually a regression in behavior on both Windows and Unix.

Before curl 7.57.0 this URL was treated as a path of "//foo/bar" and
then passed to the relevant OS API. This means that the behavior of this
case is actually OS dependent.

The Unix path resolution rules say that the OS must handle swallowing
the extra "/" and so this path is the same as "/foo/bar"

The Windows path resolution rules say that this is a UNC path and
automatically handles the SMB access for the program. So curl on Windows
was already doing Appendix E.3.2 without any special code in curl.

Regression

Closes #2438

diff --git a/lib/url.c b/lib/url.c
index 945d4e3272e7f924dfb7c0befd33bca95a1dd306..feb1fa70f44b6a7cc645c387326289f17a4e74a7 100644 (file)
--- a/lib/url.c
+++ b/lib/url.c
@@ -2067,15 +2067,6 @@ static CURLcode parseurlandfillconn(struct Curl_easy *data,
       return CURLE_URL_MALFORMAT;
     }
 
-    if(url_has_scheme && path[0] == '/' && path[1] == '/' &&
-       path[2] == '/' && path[3] == '/') {
-      /* This appears to be a UNC string (usually indicating a SMB share).
-       * We don't do SMB in file: URLs. (TODO?)
-       */
-      failf(data, "SMB shares are not supported in file: URLs.");
-      return CURLE_URL_MALFORMAT;
-    }
-
     /* Extra handling URLs with an authority component (i.e. that start with
      * "file://")
      *
@@ -2114,25 +2105,6 @@ static CURLcode parseurlandfillconn(struct Curl_easy *data,
         ptr += 9; /* now points to the slash after the host */
       }
 
-      /*
-       * RFC 8089, Appendix D, Section D.1, says:
-       *
-       * > In a POSIX file system, the root of the file system is represented
-       * > as a directory with a zero-length name, usually written as "/"; the
-       * > presence of this root in a file URI can be taken as given by the
-       * > initial slash in the "path-absolute" rule.
-       *
-       * i.e. the first slash is part of the path.
-       *
-       * However in RFC 1738 the "/" between the host (or port) and the
-       * URL-path was NOT part of the URL-path.  Any agent that followed the
-       * older spec strictly, and wanted to refer to a file with an absolute
-       * path, would have included a second slash.  So if there are two
-       * slashes, swallow one.
-       */
-      if('/' == ptr[1]) /* note: the only way ptr[0]!='/' is if ptr[1]==':' */
-        ptr++;
-
       /* This cannot be done with strcpy, as the memory chunks overlap! */
       memmove(path, ptr, strlen(ptr) + 1);
     }

diff --git a/tests/data/test2072 b/tests/data/test2072
index 2949c250282b96a31db57a404c42bff02a7dd772..0d2489ff15f8e11db8909b78262c786b8f1bc853 100644 (file)
--- a/tests/data/test2072
+++ b/tests/data/test2072
@@ -6,6 +6,13 @@ FILE
 </info>
 
 <reply>
+<data>
+foo
+   bar
+bar
+   foo
+moo
+</data>
 </reply>
 
 # Client-side
@@ -14,12 +21,15 @@ FILE
 file
 </server>
 <name>
-file:// with SMB path
+file:// with unix path resolution behavior for the case of extra slashes
 </name>
 <command>
-file:////bad-host%PWD/log/test1145.txt
+file:////%PWD/log/test2072.txt
 </command>
-<file name="log/test1145.txt">
+<precheck>
+perl -e "print 'Test requires a unix system' if ( $^O eq 'MSWin32' || $^O eq 'cygwin' || $^O eq 'dos');"
+</precheck>
+<file name="log/test2072.txt">
 foo
    bar
 bar
@@ -30,9 +40,5 @@ moo
 
 # Verify data after the test has been "shot"
 <verify>
-# CURLE_URL_MALFORMAT is error code 3
-<errorcode>
-3
-</errorcode>
 </verify>
 </testcase>