2 #include "http_config.h"
4 #include "http_protocol.h"
5 #include "http_request.h"
6 //#include "apr_compat.h"
7 #include "apr_strings.h"
8 #include "apr_reslist.h"
10 #include "util_filter.h"
11 #include "opensrf/utils.h"
13 #include <sys/types.h>
17 #define MODULE_NAME "xmlent_module"
19 /* Define the config defaults here */
20 #define MODXMLENT_CONFIG_STRIP_COMMENTS "XMLEntStripComments"
21 #define MODXMLENT_CONFIG_CONTENT_TYPE "XMLEntContentType"
22 #define MODXMLENT_CONFIG_CONTENT_TYPE_DEFAULT "text/html"
23 #define MODXMLENT_CONFIG_STRIP_PI "XMLEntStripPI"
24 #define MODXMLENT_CONFIG_DOCTYPE "XMLEntDoctype"
25 #define MODXMLENT_CONFIG_STRIP_DOCTYPE "XMLEntStripDoctype"
26 #define MODXMLENT_CONFIG_ESCAPE_SCRIPT "XMLEntEscapeScript"
28 module AP_MODULE_DECLARE_DATA xmlent_module;
30 int xmlEntInScript = 0; /* are we in the middle of a <script> tag */
34 apr_bucket_brigade* brigade; /* the bucket brigade we buffer our data into */
35 XML_Parser parser; /* our XML parser */
40 int stripComments; /* should we strip comments on the way out? */
41 int stripPI; /* should we strip processing instructions on the way out? */
43 int escapeScript; /* if true, we html-escape anything text inside a <scritp> tag */
44 char* contentType; /* the content type used to server pages */
45 char* doctype; /* the doctype header to send before any other data */
49 /* get the content type from the config */
50 static const char* xmlEntSetContentType(cmd_parms *params, void *cfg, const char *arg) {
51 xmlEntConfig* config = (xmlEntConfig*) cfg;
52 config->contentType = (char*) arg;
57 /* get the stip PI flag from the config */
58 static const char* xmlEntSetStripPI(cmd_parms *params, void *cfg, const char *arg) {
59 xmlEntConfig* config = (xmlEntConfig*) cfg;
60 char* a = (char*) arg;
61 config->stripPI = (a && !strcasecmp(a, "yes")) ? 1 : 0;
65 /* Get the strip comments flag from the config */
66 static const char* xmlEntSetStripComments(cmd_parms *params, void *cfg, const char *arg) {
67 xmlEntConfig* config = (xmlEntConfig*) cfg;
68 char* a = (char*) arg;
69 config->stripComments = (a && !strcasecmp(a, "yes")) ? 1 : 0;
73 static const char* xmlEntSetEscapeScript(cmd_parms *params, void *cfg, const char *arg) {
74 xmlEntConfig* config = (xmlEntConfig*) cfg;
75 char* a = (char*) arg;
76 config->escapeScript = (a && !strcasecmp(a, "yes")) ? 1 : 0;
80 static const char* xmlEntSetStripDoctype(cmd_parms *params, void *cfg, const char *arg) {
81 xmlEntConfig* config = (xmlEntConfig*) cfg;
82 char* a = (char*) arg;
83 config->stripDoctype = (a && !strcasecmp(a, "yes")) ? 1 : 0;
88 /* Get the user defined doctype from the config */
89 static const char* xmlEntSetDoctype(cmd_parms *params, void *cfg, const char *arg) {
90 xmlEntConfig* config = (xmlEntConfig*) cfg;
91 config->doctype = (char*) arg;
95 /* Tell apache how to set our config variables */
96 static const command_rec xmlEntCommands[] = {
97 AP_INIT_TAKE1( MODXMLENT_CONFIG_STRIP_COMMENTS,
98 xmlEntSetStripComments, NULL, ACCESS_CONF, "XMLENT Strip Comments"),
99 AP_INIT_TAKE1( MODXMLENT_CONFIG_CONTENT_TYPE,
100 xmlEntSetContentType, NULL, ACCESS_CONF, "XMLENT Content Type"),
101 AP_INIT_TAKE1( MODXMLENT_CONFIG_STRIP_PI,
102 xmlEntSetStripPI, NULL, ACCESS_CONF, "XMLENT Strip XML Processing Instructions"),
103 AP_INIT_TAKE1( MODXMLENT_CONFIG_DOCTYPE,
104 xmlEntSetDoctype, NULL, ACCESS_CONF, "XMLENT Doctype Declaration"),
105 AP_INIT_TAKE1( MODXMLENT_CONFIG_STRIP_DOCTYPE,
106 xmlEntSetStripDoctype, NULL, ACCESS_CONF, "XMLENT Strip Doctype Declaration"),
107 AP_INIT_TAKE1( MODXMLENT_CONFIG_ESCAPE_SCRIPT,
108 xmlEntSetEscapeScript, NULL, ACCESS_CONF, "XMLENT Escape data in script tags"),
112 /* Creates a new config object */
113 static void* xmlEntCreateDirConfig( apr_pool_t* p, char* dir ) {
114 xmlEntConfig* config =
115 (xmlEntConfig*) apr_palloc( p, sizeof(xmlEntConfig) );
117 config->stripComments = 0;
119 config->stripDoctype = 0;
120 config->escapeScript = 1;
121 config->contentType = MODXMLENT_CONFIG_CONTENT_TYPE_DEFAULT;
122 config->doctype = NULL;
124 return (void*) config;
127 /* keep for a while in case we ever need it */
129 #define XMLENT_INHERIT(p, c, f) ((c->f) ? c->f : p->f);
130 static void* xmlEntMergeDirConfig(apr_pool_t *p, void *base, void *overrides) {
131 xmlEntConfig* parent = base;
132 xmlEntConfig* child = overrides;
133 xmlEntConfig* newConf = (xmlEntConfig*) apr_pcalloc(p, sizeof(xmlEntConfig));
134 newConf->contentType = XMLENT_INHERIT(parent, child, contentType);
135 newConf->stripComments = XMLENT_INHERIT(parent, child, stripComments);
141 /* We need a global parser object because sub-requests, with different
142 * filter contexts, are parsing part of the same document.
143 * This means that this filter will only work in forked (non-threaded) environments.
144 * XXX Figure out how to share pointers/data accross filters */
145 XML_Parser parser = NULL;
147 /* utility function which passes data to the next filter */
148 static void _fwrite( ap_filter_t* filter, char* data, ... ) {
149 if(!(filter && data)) return;
150 xmlEntContext* ctx = (xmlEntContext*) filter->ctx;
151 VA_LIST_TO_STRING(data);
152 ap_fwrite( filter->next, ctx->brigade, VA_BUF, strlen(VA_BUF));
156 /** XXX move me to opensrf/utils.h */
157 #define OSRF_UTILS_REPLACE_CHAR(str, o, n)\
160 while(str[i] != '\0') {\
167 /* cycles through the attributes attached to an element */
168 static void printAttr( ap_filter_t* filter, const char** atts ) {
171 for( i = 0; atts[i] && atts[i+1]; i++ ) {
172 const char* name = atts[i];
173 const char* value = atts[i+1];
174 char* escaped = ap_escape_html(filter->r->pool, value);
176 /* we make a big assumption here that if the string contains a ',
177 * then the original attribute was wrapped in "s - so recreate that */
178 if( strchr( escaped, '\'' ) ) {
179 OSRF_UTILS_REPLACE_CHAR(escaped,'"','\'');
180 _fwrite( filter, " %s=\"%s\"", name, escaped );
183 OSRF_UTILS_REPLACE_CHAR(escaped,'\'','"');
184 _fwrite( filter, " %s='%s'", name, escaped );
191 /* Starts an XML element */
192 static void XMLCALL startElement(void *userData, const char *name, const char **atts) {
193 ap_filter_t* filter = (ap_filter_t*) userData;
194 _fwrite(filter, "<%s", name );
195 printAttr( filter, atts );
196 _fwrite(filter, ">", name );
197 if(!strcmp(name, "script"))
201 /* Handles the character data */
202 static void XMLCALL charHandler( void* userData, const XML_Char* s, int len ) {
203 ap_filter_t* filter = (ap_filter_t*) userData;
206 memcpy( data, s, len );
208 xmlEntConfig* config = ap_get_module_config(
209 filter->r->per_dir_config, &xmlent_module );
211 if( xmlEntInScript && ! config->escapeScript ) {
212 _fwrite( filter, "%s", data );
215 char* escaped = ap_escape_html(filter->r->pool, data);
216 _fwrite( filter, "%s", escaped );
220 static void XMLCALL handlePI( void* userData, const XML_Char* target, const XML_Char* data) {
221 ap_filter_t* filter = (ap_filter_t*) userData;
222 _fwrite(filter, "<?%s %s?>", target, data);
225 static void XMLCALL handleComment( void* userData, const XML_Char* comment ) {
226 ap_filter_t* filter = (ap_filter_t*) userData;
227 _fwrite(filter, "<!-- %s -->", comment);
230 /* Ends an XML element */
231 static void XMLCALL endElement(void *userData, const char *name) {
232 ap_filter_t* filter = (ap_filter_t*) userData;
233 _fwrite( filter, "</%s>", name );
234 if(!strcmp(name, "script"))
238 static void XMLCALL doctypeHandler( void* userData,
239 const char* name, const char* sysid, const char* pubid, int hasinternal ) {
241 ap_filter_t* filter = (ap_filter_t*) userData;
242 char* s = (sysid) ? (char*) sysid : "";
243 char* p = (pubid) ? (char*) pubid : "";
244 _fwrite( filter, "<!DOCTYPE %s PUBLIC \"%s\" \"%s\">\n", name, p, s );
248 /* The handler. Create a new parser and/or filter context where appropriate
249 * and parse the chunks of data received from the brigade
251 static int xmlEntHandler( ap_filter_t *f, apr_bucket_brigade *brigade ) {
253 xmlEntContext* ctx = f->ctx;
254 apr_bucket* currentBucket = NULL;
255 apr_pool_t* pool = f->r->pool;
259 /* load the per-dir/location config */
260 xmlEntConfig* config = ap_get_module_config(
261 f->r->per_dir_config, &xmlent_module );
263 ap_log_rerror(APLOG_MARK, APLOG_DEBUG,
264 0, f->r, "XMLENT Config:\nContent Type = %s, "
265 "Strip PI = %s, Strip Comments = %s, Doctype = %s",
267 (config->stripPI) ? "yes" : "no",
268 (config->stripComments) ? "yes" : "no",
271 /* set the content type based on the config */
272 ap_set_content_type(f->r, config->contentType);
275 /* create the XML parser */
277 if( parser == NULL ) {
279 parser = XML_ParserCreate("UTF-8");
280 XML_SetUserData(parser, f);
281 XML_SetElementHandler(parser, startElement, endElement);
282 XML_SetCharacterDataHandler(parser, charHandler);
283 if(!config->stripDoctype)
284 XML_SetStartDoctypeDeclHandler( parser, doctypeHandler );
286 XML_SetProcessingInstructionHandler(parser, handlePI);
287 if(!config->stripComments)
288 XML_SetCommentHandler(parser, handleComment);
291 /* create the filter context */
293 f->ctx = ctx = apr_pcalloc( pool, sizeof(*ctx));
294 ctx->brigade = apr_brigade_create( pool, f->c->bucket_alloc );
295 ctx->parser = parser;
299 if(firstrun) { /* we haven't started writing the data to the stream yet */
301 /* go ahead and write the doctype out if we have one defined */
302 if(config->doctype) {
303 ap_log_rerror( APLOG_MARK, APLOG_DEBUG,
304 0, f->r, "XMLENT DOCTYPE => %s", config->doctype);
305 _fwrite(f, "%s\n", config->doctype);
310 /* cycle through the buckets in the brigade */
311 while (!APR_BRIGADE_EMPTY(brigade)) {
313 /* grab the next bucket */
314 currentBucket = APR_BRIGADE_FIRST(brigade);
316 /* clean up when we're done */
317 if (APR_BUCKET_IS_EOS(currentBucket) || APR_BUCKET_IS_FLUSH(currentBucket)) {
318 APR_BUCKET_REMOVE(currentBucket);
319 APR_BRIGADE_INSERT_TAIL(ctx->brigade, currentBucket);
320 ap_pass_brigade(f->next, ctx->brigade);
321 XML_ParserFree(parser);
326 /* read the incoming data */
327 int s = apr_bucket_read(currentBucket, &data, &len, APR_NONBLOCK_READ);
328 if( s != APR_SUCCESS ) {
329 ap_log_rerror( APLOG_MARK, APLOG_ERR, 0, f->r,
330 "XMLENT error reading data from filter with status %d", s);
336 ap_log_rerror( APLOG_MARK, APLOG_DEBUG,
337 0, f->r, "XMLENT read %d bytes", (int)len);
339 /* push data into the XML push parser */
340 if ( XML_Parse(ctx->parser, data, len, 0) == XML_STATUS_ERROR ) {
342 /* log and die on XML errors */
343 ap_log_rerror( APLOG_MARK, APLOG_ERR, 0, f->r, "XMLENT XML Parse Error: %s at line %d\n",
344 XML_ErrorString(XML_GetErrorCode(ctx->parser)),
345 (int) XML_GetCurrentLineNumber(ctx->parser));
347 XML_ParserFree(parser);
349 return HTTP_INTERNAL_SERVER_ERROR;
353 /* so a subrequest doesn't re-read this bucket */
354 apr_bucket_delete(currentBucket);
357 apr_brigade_destroy(brigade);
362 /* Register the filter function as a filter for modifying the HTTP body (content) */
363 static void xmlEntRegisterHook(apr_pool_t *pool) {
364 ap_register_output_filter("XMLENT", xmlEntHandler, NULL, AP_FTYPE_CONTENT_SET);
367 /* Define the module data */
368 module AP_MODULE_DECLARE_DATA xmlent_module = {
369 STANDARD20_MODULE_STUFF,
370 xmlEntCreateDirConfig, /* dir config creater */
371 NULL, /* dir merger --- default is to override */
372 NULL, /* server config */
373 NULL, /* merge server config */
374 xmlEntCommands, /* command apr_table_t */
375 xmlEntRegisterHook /* register hook */