Allow http:// wrapper to automatically apply correct unicode.from.* filter based...

diff --git a/ext/standard/http_fopen_wrapper.c b/ext/standard/http_fopen_wrapper.c
index c8ab32a3fcb71c126e07202c10eb3c69d3e4bcb5..81bc0be112dab88f65931106b9ad5731cd7a875f 100644 (file)
--- a/ext/standard/http_fopen_wrapper.c
+++ b/ext/standard/http_fopen_wrapper.c
@@ -81,6 +81,47 @@
 #define HTTP_HEADER_CONTENT_LENGTH     16
 #define HTTP_HEADER_TYPE                       32
 
+static inline char *php_http_detect_charset(char *http_header_line)
+{
+       char *s;
+
+       /* Note: This is a fairly remedial parser which could be easily confused by invalid data
+          The worst case scenario from such confusion should only result in the unicode filter not
+          being applied.  While unfortunate, it's more an issue of the server sending a bad header */
+       for (s = strchr(http_header_line, ';'); s; s = strchr(s + 1, ';')) {
+               char *p = s;
+
+               while (*(++p) == ' ');
+               if (strncmp(p, "charset", sizeof("charset") - 1) != 0) {
+                       continue;
+               }
+               p += sizeof("charset") - 1;
+
+               while (*p == ' ') p++;
+               if (*p != '=') {
+                       continue;
+               }
+
+               while (*(++p) == ' ');
+               if (*p == '"') {
+                       s = p + 1;
+                       if (!(p = strchr(s, '"'))) {
+                               /* Bad things, unmatched quote */
+                               return NULL;
+                       }
+                       return estrndup(s, p - s);
+                       break;
+               }
+
+               /* Unquoted value */
+               s = p;
+               while (*p && *p != ' ' && *p != ';') p++;
+               return estrndup(s, p - s);
+       }
+
+       return NULL;
+}
+
 php_stream *php_stream_url_wrap_http_ex(php_stream_wrapper *wrapper, char *path, char *mode, int options, char **opened_path, php_stream_context *context, int redirect_max, int header_init STREAMS_DC TSRMLS_DC)
 {
        php_stream *stream = NULL;
@@ -104,6 +145,7 @@ php_stream *php_stream_url_wrap_http_ex(php_stream_wrapper *wrapper, char *path,
        int transport_len, have_header = 0, request_fulluri = 0;
        char *protocol_version = NULL;
        int protocol_version_len = 3; /* Default: "1.0" */
+       char *charset = NULL;
 
        tmp_line[0] = '\0';
 
@@ -543,6 +585,11 @@ php_stream *php_stream_url_wrap_http_ex(php_stream_wrapper *wrapper, char *path,
                        if (!strncasecmp(http_header_line, "Location: ", 10)) {
                                strlcpy(location, http_header_line + 10, sizeof(location));
                        } else if (!strncasecmp(http_header_line, "Content-Type: ", 14)) {
+
+                               if (UG(unicode) && strchr(mode, 't')) {
+                                       charset = php_http_detect_charset(http_header_line + sizeof("Content-type: "));
+                               }
+
                                php_stream_notify_info(context, PHP_STREAM_NOTIFY_MIME_TYPE_IS, http_header_line + 14, 0);
                        } else if (!strncasecmp(http_header_line, "Content-Length: ", 16)) {
                                file_size = atoi(http_header_line + 16);
@@ -572,6 +619,11 @@ php_stream *php_stream_url_wrap_http_ex(php_stream_wrapper *wrapper, char *path,
                php_stream_close(stream);
                stream = NULL;
 
+               if (charset) {
+                       efree(charset);
+                       charset = NULL;
+               }
+
                if (location[0] != '\0')        {
 
                        char new_path[HTTP_HEADER_BLOCK_SIZE];
@@ -684,6 +736,13 @@ out:
 
        }
 
+       if (charset) {
+               if (stream && UG(unicode) && strchr(mode, 't')) {
+                       php_stream_encoding_apply(stream, 0, charset, UG(to_error_mode), NULL);
+               }
+               efree(charset);
+       }
+
        return stream;
 }
 

diff --git a/main/streams/filter.c b/main/streams/filter.c
index 1e74cda667c549f94f5b35decbeff62eb82d3d17..79f3841d6f8c4aaca128f6316aada236fe04c5a2 100644 (file)
--- a/main/streams/filter.c
+++ b/main/streams/filter.c
@@ -550,6 +550,35 @@ PHPAPI int _php_stream_filter_output_prefer_unicode(php_stream_filter *filter TS
        return preferred ^ inverted;
 }
 
+PHPAPI int _php_stream_filter_product(php_stream_filter_chain *chain, int type TSRMLS_DC)
+{
+       php_stream_filter *f;   
+
+       for (f = chain->head; f; f = f->next) {
+               if ((type == IS_STRING && (f->fops->flags & PSFO_FLAG_ACCEPTS_STRING) == 0) ||
+                       (type == IS_UNICODE && (f->fops->flags & PSFO_FLAG_ACCEPTS_UNICODE) == 0)) {
+                       /* At some point, the type produced conflicts with the type accepted */
+                       return 0;
+               }
+
+               if (f->fops->flags & PSFO_FLAG_OUTPUTS_OPPOSITE) {
+                       type = (type == IS_STRING) ? IS_UNICODE : IS_STRING;
+                       continue;
+               }
+               if ((f->fops->flags & PSFO_FLAG_OUTPUTS_SAME) ||
+                       (f->fops->flags & PSFO_FLAG_OUTPUTS_ANY)) {
+                       continue;
+               }
+               if (f->fops->flags & PSFO_FLAG_OUTPUTS_UNICODE) {
+                       type = IS_UNICODE;
+                       continue;
+               }
+               type = IS_STRING;
+       }
+
+       return type;
+}
+
 PHPAPI int _php_stream_filter_flush(php_stream_filter *filter, int finish TSRMLS_DC)
 {
        php_stream_bucket_brigade brig_a = { NULL, NULL }, brig_b = { NULL, NULL }, *inp = &brig_a, *outp = &brig_b, *brig_temp;

diff --git a/main/streams/php_stream_filter_api.h b/main/streams/php_stream_filter_api.h
index 1c131a3eb6cdae0e9b0ab436bb5d41b82c00606c..a4c999a9f9e2de348dbabfb8e94551dbdd0fd640 100644 (file)
--- a/main/streams/php_stream_filter_api.h
+++ b/main/streams/php_stream_filter_api.h
@@ -153,6 +153,7 @@ PHPAPI void _php_stream_filter_prepend(php_stream_filter_chain *chain, php_strea
 PHPAPI void _php_stream_filter_append(php_stream_filter_chain *chain, php_stream_filter *filter TSRMLS_DC);
 PHPAPI int _php_stream_filter_check_chain(php_stream_filter_chain *chain TSRMLS_DC);
 PHPAPI int _php_stream_filter_output_prefer_unicode(php_stream_filter *filter TSRMLS_DC);
+PHPAPI int _php_stream_filter_product(php_stream_filter_chain *chain, int type TSRMLS_DC);
 PHPAPI int _php_stream_filter_flush(php_stream_filter *filter, int finish TSRMLS_DC);
 PHPAPI php_stream_filter *php_stream_filter_remove(php_stream_filter *filter, int call_dtor TSRMLS_DC);
 PHPAPI void php_stream_filter_free(php_stream_filter *filter TSRMLS_DC);
@@ -166,6 +167,7 @@ END_EXTERN_C()
 #define php_stream_filter_flush(filter, finish) _php_stream_filter_flush((filter), (finish) TSRMLS_CC)
 #define php_stream_filter_check_chain(chain) _php_stream_filter_check_chain((chain) TSRMLS_CC)
 #define php_stream_filter_output_prefer_unicode(filter) _php_stream_filter_output_prefer_unicode((filter) TSRMLS_CC)
+#define php_stream_filter_product(chain, type) _php_stream_filter_product((chain), (type) TSRMLS_CC)
 #define php_stream_encoding_apply(stream, writechain, encoding, error_mode, subst) \
                _php_stream_encoding_apply((stream), (writechain), (encoding), (error_mode), (subst) TSRMLS_CC)
 

diff --git a/main/streams/streams.c b/main/streams/streams.c
index 1a6b13d533f35032daeb7567a0974b55aa3af30b..eafecc1a895f67388102b0bee714294c8b4a2e16 100755 (executable)
--- a/main/streams/streams.c
+++ b/main/streams/streams.c
@@ -2305,13 +2305,17 @@ PHPAPI php_stream *_php_stream_open_wrapper_ex(char *path, char *mode, int optio
 
        /* Output encoding on text mode streams defaults to utf8 unless specified in context parameter */
        if (stream && strchr(implicit_mode, 't') && UG(unicode)) {
-               if (strchr(implicit_mode, 'w') || strchr(implicit_mode, 'a') || strchr(implicit_mode, '+')) {
+               /* Only apply implicit unicode.to. filter if the wrapper didn't do it for us */
+               if ((php_stream_filter_product(&stream->writefilters, IS_UNICODE) == IS_STRING) && 
+                       (strchr(implicit_mode, 'w') || strchr(implicit_mode, 'a') || strchr(implicit_mode, '+'))) {
                        char *encoding = (context && context->output_encoding) ? context->output_encoding : "utf8";
 
                        /* UTODO: (Maybe?) Allow overriding the default error handlers on a per-stream basis via context params */
                        php_stream_encoding_apply(stream, 1, encoding, UG(from_error_mode), UG(from_subst_char));
                }
-               if (strchr(implicit_mode, 'r') || strchr(implicit_mode, '+')) {
+
+               /* Only apply implicit unicode.from. filter if the wrapper didn't do it for us */
+               if ((stream->readbuf_type == IS_STRING) && (strchr(implicit_mode, 'r') || strchr(implicit_mode, '+'))) {
                        char *encoding = (context && context->input_encoding) ? context->input_encoding : "utf8";
 
                        /* UTODO: (Maybe?) Allow overriding the default error handlers on a per-stream basis via context params */