-const char parsers_rcs[] = "$Id: parsers.c,v 1.274 2013/01/04 12:20:31 fabiankeil Exp $";
+const char parsers_rcs[] = "$Id: parsers.c,v 1.287 2014/07/25 11:55:11 fabiankeil Exp $";
/*********************************************************************
*
* File : $Source: /cvsroot/ijbswa/current/parsers.c,v $
*
* Purpose : Declares functions to parse/crunch headers and pages.
*
- * Copyright : Written by and Copyright (C) 2001-2012 the
+ * Copyright : Written by and Copyright (C) 2001-2014 the
* Privoxy team. http://www.privoxy.org/
*
* Based on the Internet Junkbuster originally written
static jb_err crunch_client_header (struct client_state *csp, char **header);
static jb_err client_x_filter (struct client_state *csp, char **header);
static jb_err client_range (struct client_state *csp, char **header);
+static jb_err client_expect (struct client_state *csp, char **header);
static jb_err server_set_cookie (struct client_state *csp, char **header);
static jb_err server_connection (struct client_state *csp, char **header);
static jb_err server_content_type (struct client_state *csp, char **header);
#if 0
{ "Transfer-Encoding:", 18, client_transfer_encoding },
#endif
+ { "Expect:", 7, client_expect },
{ "*", 0, crunch_client_header },
{ "*", 0, filter_header },
{ NULL, 0, NULL }
* to a single space in accordance with RFC 2616 2.2.
* This simplifies parsing and filtering later on.
*
- * XXX: Remove log messages before
- * the next stable release?
- *
* Parameters :
* 1 : header = A header with linear white space to reduce.
*
return;
}
+
/*********************************************************************
*
* Function : sed
const add_header_func_ptr *f;
jb_err err = JB_ERR_OK;
+ scan_headers(csp);
+
if (filter_server_headers)
{
v = server_patterns;
f = add_server_headers;
+ check_negative_tag_patterns(csp, PATTERN_SPEC_NO_RESPONSE_TAG_PATTERN);
}
else
{
v = client_patterns;
f = add_client_headers;
+ check_negative_tag_patterns(csp, PATTERN_SPEC_NO_REQUEST_TAG_PATTERN);
}
- scan_headers(csp);
-
while ((err == JB_ERR_OK) && (v->str != NULL))
{
for (p = csp->headers->first; (err == JB_ERR_OK) && (p != NULL); p = p->next)
*********************************************************************/
static jb_err header_tagger(struct client_state *csp, char *header)
{
- int wanted_filter_type;
+ enum filter_type wanted_filter_type;
int multi_action_index;
- int i;
pcrs_job *job;
- struct file_list *fl;
struct re_filterfile_spec *b;
struct list_entry *tag_name;
multi_action_index = ACTION_MULTI_CLIENT_HEADER_TAGGER;
}
- if (filters_available(csp) == FALSE)
+ if (list_is_empty(csp->action->multi[multi_action_index])
+ || filters_available(csp) == FALSE)
{
- log_error(LOG_LEVEL_ERROR, "Inconsistent configuration: "
- "tagging enabled, but no taggers available.");
+ /* Return early if no taggers apply or if none are available. */
return JB_ERR_OK;
}
- for (i = 0; i < MAX_AF_FILES; i++)
+ /* Execute all applying taggers */
+ for (tag_name = csp->action->multi[multi_action_index]->first;
+ NULL != tag_name; tag_name = tag_name->next)
{
- fl = csp->rlist[i];
- if ((NULL == fl) || (NULL == fl->f))
+ char *modified_tag = NULL;
+ char *tag = header;
+ size_t size = header_length;
+ pcrs_job *joblist;
+
+ b = get_filter(csp, tag_name->str, wanted_filter_type);
+ if (b == NULL)
{
- /*
- * Either there are no filter files
- * left, or this filter file just
- * contains no valid filters.
- *
- * Continue to be sure we don't miss
- * valid filter files that are chained
- * after empty or invalid ones.
- */
continue;
}
- /* For all filters, */
- for (b = fl->f; b; b = b->next)
+ joblist = b->joblist;
+
+ if (b->dynamic) joblist = compile_dynamic_pcrs_job_list(csp, b);
+
+ if (NULL == joblist)
{
- if (b->type != wanted_filter_type)
+ log_error(LOG_LEVEL_RE_FILTER,
+ "Tagger %s has empty joblist. Nothing to do.", b->name);
+ continue;
+ }
+
+ /* execute their pcrs_joblist on the header. */
+ for (job = joblist; NULL != job; job = job->next)
+ {
+ const int hits = pcrs_execute(job, tag, size, &modified_tag, &size);
+
+ if (0 < hits)
{
- /* skip the ones we don't care about, */
- continue;
+ /* Success, continue with the modified version. */
+ if (tag != header)
+ {
+ freez(tag);
+ }
+ tag = modified_tag;
}
- /* leaving only taggers that could apply, of which we use the ones, */
- for (tag_name = csp->action->multi[multi_action_index]->first;
- NULL != tag_name; tag_name = tag_name->next)
+ else
{
- /* that do apply, and */
- if (strcmp(b->name, tag_name->str) == 0)
+ /* Tagger doesn't match */
+ if (0 > hits)
{
- char *modified_tag = NULL;
- char *tag = header;
- size_t size = header_length;
- pcrs_job *joblist = b->joblist;
+ /* Regex failure, log it but continue anyway. */
+ assert(NULL != header);
+ log_error(LOG_LEVEL_ERROR,
+ "Problems with tagger \'%s\' and header \'%s\': %s",
+ b->name, *header, pcrs_strerror(hits));
+ }
+ freez(modified_tag);
+ }
+ }
+
+ if (b->dynamic) pcrs_free_joblist(joblist);
- if (b->dynamic) joblist = compile_dynamic_pcrs_job_list(csp, b);
+ /* If this tagger matched */
+ if (tag != header)
+ {
+ if (0 == size)
+ {
+ /*
+ * There is no technical limitation which makes
+ * it impossible to use empty tags, but I assume
+ * no one would do it intentionally.
+ */
+ freez(tag);
+ log_error(LOG_LEVEL_INFO,
+ "Tagger \'%s\' created an empty tag. Ignored.", b->name);
+ continue;
+ }
- if (NULL == joblist)
+ if (!list_contains_item(csp->tags, tag))
+ {
+ if (JB_ERR_OK != enlist(csp->tags, tag))
+ {
+ log_error(LOG_LEVEL_ERROR,
+ "Insufficient memory to add tag \'%s\', "
+ "based on tagger \'%s\' and header \'%s\'",
+ tag, b->name, *header);
+ }
+ else
+ {
+ char *action_message;
+ /*
+ * update the action bits right away, to make
+ * tagging based on tags set by earlier taggers
+ * of the same kind possible.
+ */
+ if (update_action_bits_for_tag(csp, tag))
{
- log_error(LOG_LEVEL_RE_FILTER,
- "Tagger %s has empty joblist. Nothing to do.", b->name);
- continue;
+ action_message = "Action bits updated accordingly.";
}
-
- /* execute their pcrs_joblist on the header. */
- for (job = joblist; NULL != job; job = job->next)
+ else
{
- const int hits = pcrs_execute(job, tag, size, &modified_tag, &size);
-
- if (0 < hits)
- {
- /* Success, continue with the modified version. */
- if (tag != header)
- {
- freez(tag);
- }
- tag = modified_tag;
- }
- else
- {
- /* Tagger doesn't match */
- if (0 > hits)
- {
- /* Regex failure, log it but continue anyway. */
- assert(NULL != header);
- log_error(LOG_LEVEL_ERROR,
- "Problems with tagger \'%s\' and header \'%s\': %s",
- b->name, *header, pcrs_strerror(hits));
- }
- freez(modified_tag);
- }
+ action_message = "No action bits update necessary.";
}
- if (b->dynamic) pcrs_free_joblist(joblist);
-
- /* If this tagger matched */
- if (tag != header)
- {
- if (0 == size)
- {
- /*
- * There is to technical limitation which makes
- * it impossible to use empty tags, but I assume
- * no one would do it intentionally.
- */
- freez(tag);
- log_error(LOG_LEVEL_INFO,
- "Tagger \'%s\' created an empty tag. Ignored.",
- b->name);
- continue;
- }
-
- if (!list_contains_item(csp->tags, tag))
- {
- if (JB_ERR_OK != enlist(csp->tags, tag))
- {
- log_error(LOG_LEVEL_ERROR,
- "Insufficient memory to add tag \'%s\', "
- "based on tagger \'%s\' and header \'%s\'",
- tag, b->name, *header);
- }
- else
- {
- char *action_message;
- /*
- * update the action bits right away, to make
- * tagging based on tags set by earlier taggers
- * of the same kind possible.
- */
- if (update_action_bits_for_tag(csp, tag))
- {
- action_message = "Action bits updated accordingly.";
- }
- else
- {
- action_message = "No action bits update necessary.";
- }
-
- log_error(LOG_LEVEL_HEADER,
- "Tagger \'%s\' added tag \'%s\'. %s",
- b->name, tag, action_message);
- }
- }
- else
- {
- /* XXX: Is this log-worthy? */
- log_error(LOG_LEVEL_HEADER,
- "Tagger \'%s\' didn't add tag \'%s\'. "
- "Tag already present", b->name, tag);
- }
- freez(tag);
- } /* if the tagger matched */
- } /* if the tagger applies */
- } /* for every tagger that could apply */
- } /* for all filters */
- } /* for all filter files */
+ log_error(LOG_LEVEL_HEADER,
+ "Tagger \'%s\' added tag \'%s\'. %s",
+ b->name, tag, action_message);
+ }
+ }
+ else
+ {
+ /* XXX: Is this log-worthy? */
+ log_error(LOG_LEVEL_HEADER,
+ "Tagger \'%s\' didn't add tag \'%s\'. Tag already present",
+ b->name, tag);
+ }
+ freez(tag);
+ }
+ }
return JB_ERR_OK;
}
char *newheader = NULL;
pcrs_job *job;
- struct file_list *fl;
struct re_filterfile_spec *b;
struct list_entry *filtername;
- int i;
- int wanted_filter_type;
+ enum filter_type wanted_filter_type;
int multi_action_index;
if (csp->flags & CSP_FLAG_NO_FILTERING)
multi_action_index = ACTION_MULTI_CLIENT_HEADER_FILTER;
}
- if (filters_available(csp) == FALSE)
+ if (list_is_empty(csp->action->multi[multi_action_index])
+ || filters_available(csp) == FALSE)
{
- log_error(LOG_LEVEL_ERROR, "Inconsistent configuration: "
- "header filtering enabled, but no matching filters available.");
+ /* Return early if no filters apply or if none are available. */
return JB_ERR_OK;
}
- for (i = 0; i < MAX_AF_FILES; i++)
+ /* Execute all applying header filters */
+ for (filtername = csp->action->multi[multi_action_index]->first;
+ filtername != NULL; filtername = filtername->next)
{
- fl = csp->rlist[i];
- if ((NULL == fl) || (NULL == fl->f))
+ int current_hits = 0;
+ pcrs_job *joblist;
+
+ b = get_filter(csp, filtername->str, wanted_filter_type);
+ if (b == NULL)
{
- /*
- * Either there are no filter files
- * left, or this filter file just
- * contains no valid filters.
- *
- * Continue to be sure we don't miss
- * valid filter files that are chained
- * after empty or invalid ones.
- */
continue;
}
- /*
- * For all applying +filter actions, look if a filter by that
- * name exists and if yes, execute its pcrs_joblist on the
- * buffer.
- */
- for (b = fl->f; b; b = b->next)
- {
- if (b->type != wanted_filter_type)
- {
- /* Skip other filter types */
- continue;
- }
-
- for (filtername = csp->action->multi[multi_action_index]->first;
- filtername ; filtername = filtername->next)
- {
- if (strcmp(b->name, filtername->str) == 0)
- {
- int current_hits = 0;
- pcrs_job *joblist = b->joblist;
- if (b->dynamic) joblist = compile_dynamic_pcrs_job_list(csp, b);
+ joblist = b->joblist;
- if (NULL == joblist)
- {
- log_error(LOG_LEVEL_RE_FILTER, "Filter %s has empty joblist. Nothing to do.", b->name);
- continue;
- }
+ if (b->dynamic) joblist = compile_dynamic_pcrs_job_list(csp, b);
- log_error(LOG_LEVEL_RE_FILTER, "filtering \'%s\' (size %d) with \'%s\' ...",
- *header, size, b->name);
-
- /* Apply all jobs from the joblist */
- for (job = joblist; NULL != job; job = job->next)
- {
- matches = pcrs_execute(job, *header, size, &newheader, &size);
- if (0 < matches)
- {
- current_hits += matches;
- log_error(LOG_LEVEL_HEADER, "Transforming \"%s\" to \"%s\"", *header, newheader);
- freez(*header);
- *header = newheader;
- }
- else if (0 == matches)
- {
- /* Filter doesn't change header */
- freez(newheader);
- }
- else
- {
- /* RegEx failure */
- log_error(LOG_LEVEL_ERROR, "Filtering \'%s\' with \'%s\' didn't work out: %s",
- *header, b->name, pcrs_strerror(matches));
- if (newheader != NULL)
- {
- log_error(LOG_LEVEL_ERROR, "Freeing what's left: %s", newheader);
- freez(newheader);
- }
- }
- }
+ if (NULL == joblist)
+ {
+ log_error(LOG_LEVEL_RE_FILTER, "Filter %s has empty joblist. Nothing to do.", b->name);
+ continue;
+ }
- if (b->dynamic) pcrs_free_joblist(joblist);
+ log_error(LOG_LEVEL_RE_FILTER, "filtering \'%s\' (size %d) with \'%s\' ...",
+ *header, size, b->name);
- log_error(LOG_LEVEL_RE_FILTER, "... produced %d hits (new size %d).", current_hits, size);
- hits += current_hits;
+ /* Apply all jobs from the joblist */
+ for (job = joblist; NULL != job; job = job->next)
+ {
+ matches = pcrs_execute(job, *header, size, &newheader, &size);
+ if (0 < matches)
+ {
+ current_hits += matches;
+ log_error(LOG_LEVEL_HEADER, "Transforming \"%s\" to \"%s\"", *header, newheader);
+ freez(*header);
+ *header = newheader;
+ }
+ else if (0 == matches)
+ {
+ /* Filter doesn't change header */
+ freez(newheader);
+ }
+ else
+ {
+ /* RegEx failure */
+ log_error(LOG_LEVEL_ERROR, "Filtering \'%s\' with \'%s\' didn't work out: %s",
+ *header, b->name, pcrs_strerror(matches));
+ if (newheader != NULL)
+ {
+ log_error(LOG_LEVEL_ERROR, "Freeing what's left: %s", newheader);
+ freez(newheader);
}
}
}
+
+ if (b->dynamic) pcrs_free_joblist(joblist);
+
+ log_error(LOG_LEVEL_RE_FILTER, "... produced %d hits (new size %d).", current_hits, size);
+ hits += current_hits;
}
/*
static jb_err client_keep_alive(struct client_state *csp, char **header)
{
unsigned int keep_alive_timeout;
- const char *timeout_position = strstr(*header, ": ");
+ char *timeout_position;
if (!(csp->config->feature_flags & RUNTIME_FEATURE_CONNECTION_KEEP_ALIVE))
{
return JB_ERR_OK;
}
+ /* Check for parameter-less format "Keep-Alive: 100" */
+ timeout_position = strstr(*header, ": ");
if ((NULL == timeout_position)
|| (1 != sscanf(timeout_position, ": %u", &keep_alive_timeout)))
{
- log_error(LOG_LEVEL_ERROR, "Couldn't parse: %s", *header);
- }
- else
- {
- if (keep_alive_timeout < csp->config->keep_alive_timeout)
+ /* Assume parameter format "Keep-Alive: timeout=100" */
+ timeout_position = strstr(*header, "timeout=");
+ if ((NULL == timeout_position)
+ || (1 != sscanf(timeout_position, "timeout=%u", &keep_alive_timeout)))
{
log_error(LOG_LEVEL_HEADER,
- "Reducing keep-alive timeout from %u to %u.",
- csp->config->keep_alive_timeout, keep_alive_timeout);
- csp->server_connection.keep_alive_timeout = keep_alive_timeout;
- }
- else
- {
- /* XXX: Is this log worthy? */
- log_error(LOG_LEVEL_HEADER,
- "Client keep-alive timeout is %u. Sticking with %u.",
- keep_alive_timeout, csp->config->keep_alive_timeout);
+ "Couldn't parse: '%s'. Using default timeout %u",
+ *header, csp->config->keep_alive_timeout);
+ freez(*header);
+
+ return JB_ERR_OK;
}
}
+ if (keep_alive_timeout < csp->config->keep_alive_timeout)
+ {
+ log_error(LOG_LEVEL_HEADER,
+ "Reducing keep-alive timeout from %u to %u.",
+ csp->config->keep_alive_timeout, keep_alive_timeout);
+ csp->server_connection.keep_alive_timeout = keep_alive_timeout;
+ }
+ else
+ {
+ /* XXX: Is this log worthy? */
+ log_error(LOG_LEVEL_HEADER,
+ "Client keep-alive timeout is %u. Sticking with %u.",
+ keep_alive_timeout, csp->config->keep_alive_timeout);
+ freez(*header);
+ }
+
return JB_ERR_OK;
}
}
+/*********************************************************************
+ *
+ * Function : client_expect
+ *
+ * Description : Raise the CSP_FLAG_UNSUPPORTED_CLIENT_EXPECTATION
+ * if the Expect header value is unsupported.
+ *
+ * Rejecting unsupported expectations is a RFC 7231 5.1.1
+ * MAY and a RFC 2616 (obsolete) MUST.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ * 2 : header = On input, pointer to header to modify.
+ * On output, pointer to the modified header, or NULL
+ * to remove the header. This function frees the
+ * original string if necessary.
+ *
+ * Returns : JB_ERR_OK on success, or
+ *
+ *********************************************************************/
+jb_err client_expect(struct client_state *csp, char **header)
+{
+ if (0 != strcmpic(*header, "Expect: 100-continue"))
+ {
+ csp->flags |= CSP_FLAG_UNSUPPORTED_CLIENT_EXPECTATION;
+ log_error(LOG_LEVEL_HEADER,
+ "Unsupported client expectaction: %s", *header);
+ }
+
+ return JB_ERR_OK;
+
+}
+
+
/*********************************************************************
*
* Function : crumble
* port information, parse and evaluate the Host
* header field.
*
- * Also, kill ill-formed HOST: headers as sent by
- * Apple's iTunes software when used with a proxy.
- *
* Parameters :
* 1 : csp = Current client state (buffers, headers, etc...)
* 2 : header = On input, pointer to header to modify.
{
char *p, *q;
- /*
- * If the header field name is all upper-case, chances are that it's
- * an ill-formed one from iTunes. BTW, killing innocent headers here is
- * not a problem -- they are regenerated later.
- */
- if ((*header)[1] == 'O')
- {
- log_error(LOG_LEVEL_HEADER, "Killed all-caps Host header line: %s", *header);
- freez(*header);
- return JB_ERR_OK;
- }
-
if (!csp->http->hostport || (*csp->http->hostport == '*') ||
*csp->http->hostport == ' ' || *csp->http->hostport == '\0')
{
projects / privoxy.git / blobdiff