X-Git-Url: http://www.privoxy.org/gitweb/?p=privoxy.git;a=blobdiff_plain;f=filters.c;h=292db4bcceb0c546c5aec930198b63321e46bbe8;hp=a4757be3f64cc38d72a1f4231a7afab219608d1c;hb=9775a52f3c99fb826c2dcae1919a37cc53851b3a;hpb=f8d34ce7f912a2e026df77647444ee07f9d061ae diff --git a/filters.c b/filters.c index a4757be3..292db4bc 100644 --- a/filters.c +++ b/filters.c @@ -1,4 +1,4 @@ -const char filters_rcs[] = "$Id: filters.c,v 1.91 2007/09/02 15:31:20 fabiankeil Exp $"; +const char filters_rcs[] = "$Id: filters.c,v 1.96 2007/10/19 16:53:28 fabiankeil Exp $"; /********************************************************************* * * File : $Source: /cvsroot/ijbswa/current/filters.c,v $ @@ -40,6 +40,30 @@ const char filters_rcs[] = "$Id: filters.c,v 1.91 2007/09/02 15:31:20 fabiankeil * * Revisions : * $Log: filters.c,v $ + * Revision 1.96 2007/10/19 16:53:28 fabiankeil + * Add helper function to check if any content filters are enabled. + * + * Revision 1.95 2007/10/17 19:31:20 fabiankeil + * Omitting the zero chunk that ends the chunk transfer encoding seems + * to be the new black. Log the problem and continue filtering anyway. + * + * Revision 1.94 2007/09/29 13:20:20 fabiankeil + * Remove two redundant and one useless log messages. + * + * Revision 1.93 2007/09/29 10:21:16 fabiankeil + * - Move get_filter_function() from jcc.c to filters.c + * so the filter functions can be static. + * - Don't bother filtering body-less responses. + * + * Revision 1.92 2007/09/28 16:38:55 fabiankeil + * - Execute content filters through execute_content_filter(). + * - Add prepare_for_filtering() so filter functions don't have to + * care about de-chunking and decompression. As a side effect this enables + * decompression for gif_deanimate_response() and jpeg_inspect_response(). + * - Change remove_chunked_transfer_coding()'s return type to jb_err. + * Some clowns feel like chunking empty responses in which case + * (size == 0) is valid but previously would be interpreted as error. + * * Revision 1.91 2007/09/02 15:31:20 fabiankeil * Move match_portlist() from filter.c to urlmatch.c. * It's used for url matching, not for filtering. @@ -725,7 +749,7 @@ int acl_addr(const char *aspec, struct access_control_addr *aca) *p++ = '\0'; if (ijb_isdigit(*p) == 0) { - free(acl_spec); + freez(acl_spec); return(-1); } masklength = atoi(p); @@ -733,7 +757,7 @@ int acl_addr(const char *aspec, struct access_control_addr *aca) if ((masklength < 0) || (masklength > 32)) { - free(acl_spec); + freez(acl_spec); return(-1); } @@ -746,7 +770,7 @@ int acl_addr(const char *aspec, struct access_control_addr *aca) if (port <= 0 || port > 65535 || *endptr != '\0') { - free(acl_spec); + freez(acl_spec); return(-1); } } @@ -754,7 +778,7 @@ int acl_addr(const char *aspec, struct access_control_addr *aca) aca->port = (unsigned long)port; aca->addr = ntohl(resolve_hostname_to_ip(acl_spec)); - free(acl_spec); + freez(acl_spec); if (aca->addr == INADDR_NONE) { @@ -1701,7 +1725,7 @@ int is_untrusted_url(const struct client_state *csp) log_error(LOG_LEVEL_ERROR, "Failed to append \'%s\' to trustfile \'%s\': %E", new_entry, csp->config->trustfile); } - free(new_entry); + freez(new_entry); } else { @@ -1740,7 +1764,7 @@ int is_untrusted_url(const struct client_state *csp) * or NULL if there were no hits or something went wrong * *********************************************************************/ -char *pcrs_filter_response(struct client_state *csp) +static char *pcrs_filter_response(struct client_state *csp) { int hits=0; size_t size, prev_size; @@ -1852,7 +1876,7 @@ char *pcrs_filter_response(struct client_state *csp) current_hits += job_hits; if (old != csp->iob->cur) { - free(old); + freez(old); } old = new; } @@ -1898,7 +1922,7 @@ char *pcrs_filter_response(struct client_state *csp) */ if (!hits) { - free(new); + freez(new); return(NULL); } @@ -1926,7 +1950,7 @@ char *pcrs_filter_response(struct client_state *csp) * or NULL in case something went wrong. * *********************************************************************/ -char *gif_deanimate_response(struct client_state *csp) +static char *gif_deanimate_response(struct client_state *csp) { struct binbuffer *in, *out; char *p; @@ -1947,7 +1971,7 @@ char *gif_deanimate_response(struct client_state *csp) if (gif_deanimate(in, out, strncmp("last", csp->action->string[ACTION_STRING_DEANIMATE], 4))) { log_error(LOG_LEVEL_DEANIMATE, "failed! (gif parsing)"); - free(in); + freez(in); buf_free(out); return(NULL); } @@ -1964,8 +1988,8 @@ char *gif_deanimate_response(struct client_state *csp) csp->content_length = out->offset; csp->flags |= CSP_FLAG_MODIFIED; p = out->buffer; - free(in); - free(out); + freez(in); + freez(out); return(p); } @@ -1985,7 +2009,7 @@ char *gif_deanimate_response(struct client_state *csp) * or NULL in case something went wrong. * *********************************************************************/ -char *jpeg_inspect_response(struct client_state *csp) +static char *jpeg_inspect_response(struct client_state *csp) { struct binbuffer *in = NULL; struct binbuffer *out = NULL; @@ -2016,7 +2040,7 @@ char *jpeg_inspect_response(struct client_state *csp) if (jpeg_inspect(in, out)) { log_error(LOG_LEVEL_DEANIMATE, "failed! (jpeg parsing)"); - free(in); + freez(in); buf_free(out); return(NULL); @@ -2026,14 +2050,95 @@ char *jpeg_inspect_response(struct client_state *csp) csp->content_length = out->offset; csp->flags |= CSP_FLAG_MODIFIED; p = out->buffer; - free(in); - free(out); + freez(in); + freez(out); return(p); } } +/********************************************************************* + * + * Function : get_filter_function + * + * Description : Decides which content filter function has + * to be applied (if any). + * + * XXX: Doesn't handle filter_popups() + * because of the different prototype. Probably + * we should ditch filter_popups() anyway, it's + * even less reliable than popup blocking based + * on pcrs filters. + * + * Parameters : + * 1 : csp = Current client state (buffers, headers, etc...) + * + * Returns : The content filter function to run, or + * NULL if no content filter is active + * + *********************************************************************/ +filter_function_ptr get_filter_function(struct client_state *csp) +{ + filter_function_ptr filter_function = NULL; + + /* + * Are we enabling text mode by force? + */ + if (csp->action->flags & ACTION_FORCE_TEXT_MODE) + { + /* + * Do we really have to? + */ + if (csp->content_type & CT_TEXT) + { + log_error(LOG_LEVEL_HEADER, "Text mode is already enabled."); + } + else + { + csp->content_type |= CT_TEXT; + log_error(LOG_LEVEL_HEADER, "Text mode enabled by force. Take cover!"); + } + } + + if (!(csp->content_type & CT_DECLARED)) + { + /* + * The server didn't bother to declare a MIME-Type. + * Assume it's text that can be filtered. + * + * This also regulary happens with 304 responses, + * therefore logging anything here would cause + * too much noise. + */ + csp->content_type |= CT_TEXT; + } + + /* + * Choose the applying filter function based on + * the content type and action settings. + */ + if ((csp->content_type & CT_TEXT) && + (csp->rlist != NULL) && + (!list_is_empty(csp->action->multi[ACTION_MULTI_FILTER]))) + { + filter_function = pcrs_filter_response; + } + else if ((csp->content_type & CT_GIF) && + (csp->action->flags & ACTION_DEANIMATE)) + { + filter_function = gif_deanimate_response; + } + else if ((csp->content_type & CT_JPEG) && + (csp->action->flags & ACTION_JPEG_INSPECT)) + { + filter_function = jpeg_inspect_response; + } + + return filter_function; +} + + /********************************************************************* * * Function : remove_chunked_transfer_coding @@ -2087,16 +2192,11 @@ static jb_err remove_chunked_transfer_coding(char *buffer, size_t *size) if (sscanf(from_p, "%x", &chunksize) != 1) { - log_error(LOG_LEVEL_ERROR, "Parse error while stripping \"chunked\" transfer coding"); - return JB_ERR_PARSE; + log_error(LOG_LEVEL_INFO, "Invalid \"chunked\" transfer encoding detected and ignored."); + break; } } - if (0 == newsize) - { - log_error(LOG_LEVEL_RE_FILTER, "Need to de-chunk first"); - } - /* XXX: Should get its own loglevel. */ log_error(LOG_LEVEL_RE_FILTER, "De-chunking successful. Shrunk from %d to %d", *size, newsize); @@ -2142,7 +2242,6 @@ static jb_err prepare_for_filtering(struct client_state *csp) } else { - log_error(LOG_LEVEL_ERROR, "Failed to de-chunk content."); return JB_ERR_PARSE; } } @@ -2154,6 +2253,12 @@ static jb_err prepare_for_filtering(struct client_state *csp) */ if (csp->content_type & (CT_GZIP|CT_DEFLATE)) { + if (0 == csp->iob->eod - csp->iob->cur) + { + /* Nothing left after de-chunking. */ + return JB_ERR_OK; + } + err = decompress_iob(csp); if (JB_ERR_OK == err) @@ -2169,7 +2274,6 @@ static jb_err prepare_for_filtering(struct client_state *csp) */ csp->content_type &= ~CT_GZIP; csp->content_type &= ~CT_DEFLATE; - log_error(LOG_LEVEL_ERROR, "Failed to decompress content."); } } #endif @@ -2188,24 +2292,34 @@ static jb_err prepare_for_filtering(struct client_state *csp) * 1 : csp = Current client state (buffers, headers, etc...) * 2 : content_filter = The filter function to execute * - * Returns : JB_ERR_OK for success, - * JB_ERR_PARSE otherwise + * Returns : Pointer to the modified buffer, or + * NULL if filtering failed or wasn't necessary. * *********************************************************************/ char *execute_content_filter(struct client_state *csp, filter_function_ptr content_filter) { + if (0 == csp->iob->eod - csp->iob->cur) + { + /* + * No content (probably status code 301, 302 ...), + * no filtering necessary. + */ + return NULL; + } + if (JB_ERR_OK != prepare_for_filtering(csp)) { /* * failed to de-chunk or decompress. - * XXX: if possible, we should continue anyway. */ return NULL; } if (0 == csp->iob->eod - csp->iob->cur) { - /* Empty buffer, nothing to do. */ + /* + * Clown alarm: chunked and/or compressed nothing delivered. + */ return NULL; } @@ -2539,6 +2653,26 @@ struct http_response *direct_response(struct client_state *csp) } +/********************************************************************* + * + * Function : content_filters_enabled + * + * Description : Checks whether there are any content filters + * enabled for the current request. + * + * Parameters : + * 1 : csp = Current client state (buffers, headers, etc...) + * + * Returns : TRUE for yes, FALSE otherwise + * + *********************************************************************/ +inline int content_filters_enabled(const struct client_state *csp) +{ + return (((csp->rlist != NULL) && + (!list_is_empty(csp->action->multi[ACTION_MULTI_FILTER]))) || + (csp->action->flags & (ACTION_DEANIMATE|ACTION_JPEG_INSPECT|ACTION_NO_POPUPS))); +} + /* Local Variables: tab-width: 3