Annotation of parser3/src/types/pa_vform.C, revision 1.115
1.27 paf 1: /** @file
2: Parser: @b form class.
1.10 paf 3:
1.112 moko 4: Copyright (c) 2001-2015 Art. Lebedev Studio (http://www.artlebedev.com)
1.52 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.39 parser 6:
1.38 parser 7: based on The CGI_C library, by Thomas Boutell.
1.3 paf 8: */
1.59 paf 9:
1.14 paf 10: #include "pa_sapi.h"
1.1 paf 11: #include "pa_vform.h"
12: #include "pa_vstring.h"
1.3 paf 13: #include "pa_globals.h"
14: #include "pa_request.h"
1.9 paf 15: #include "pa_vfile.h"
1.21 paf 16: #include "pa_common.h"
1.30 parser 17: #include "pa_vtable.h"
1.50 paf 18: #include "pa_charset.h"
1.102 moko 19:
1.115 ! moko 20: volatile const char * IDENT_PA_VFORM_C="$Id: pa_vform.C,v 1.114 2016/04/06 16:08:20 moko Exp $" IDENT_PA_VFORM_H;
1.3 paf 21:
1.62 paf 22: // defines
23:
24: //#define DEBUG_POST
25:
1.3 paf 26: // parse helper funcs
27:
1.75 paf 28: static size_t getHeader(const char* data, size_t len){
1.97 misha 29: size_t i;
30: int enter=-1;
1.111 moko 31: if (data) {
1.62 paf 32: for (i=0;i<len;i++)
33: if (data[i]=='\n'){
34: if (enter>=0) enter++;
35: if (enter>1) return i;
1.111 moko 36: } else {
37: if (data[i]!='\r') enter=0;
38: }
39: }
1.100 misha 40: return 0;
1.3 paf 41: }
42:
1.75 paf 43: static const char* searchAttribute(const char* data,
44: const char* attr, //< expected to be lowercased
1.62 paf 45: size_t len){
1.97 misha 46: size_t i;
47: if (data)
1.62 paf 48: for (i=0;i<len;i++)
1.83 paf 49: if (tolower((unsigned char)data[i])==*attr){
1.62 paf 50: size_t j;
51: for (j=i+1;j<=len;j++)
52: if (!attr[j-i]) return &data[j];
53: else {
54: if (j==len) break;
1.83 paf 55: if (attr[j-i]!=tolower((unsigned char)data[j])) break;
1.62 paf 56: }
57: }
58:
59: return NULL;
1.3 paf 60: }
61:
62: // VForm
63:
1.113 moko 64: extern Methoded* form_class;
1.57 paf 65:
1.114 moko 66: VForm::VForm(Request_charsets& acharsets, Request_info& arequest_info): VStateless_class(form_class),
1.77 paf 67: fcharsets(acharsets),
68: frequest_info(arequest_info),
69: filled_source(0),
1.91 misha 70: filled_client(0),
1.115 ! moko 71: fpost_charset(0),
! 72:
! 73: can_have_body(arequest_info.can_have_body()),
! 74: post_content_type(UNKNOWN),
! 75: charset_detected(false)
1.96 misha 76: {
1.111 moko 77: if(can_have_body && arequest_info.content_type) {
78: if(pa_strncasecmp(arequest_info.content_type, HTTP_CONTENT_TYPE_FORM_URLENCODED)==0) {
1.96 misha 79: post_content_type=FORM_URLENCODED;
1.111 moko 80: } else if(pa_strncasecmp(arequest_info.content_type, HTTP_CONTENT_TYPE_MULTIPART_FORMDATA)==0) {
1.96 misha 81: post_content_type=MULTIPART_FORMDATA;
1.111 moko 82: }
83: }
1.3 paf 84: }
85:
1.92 misha 86: char *VForm::strpart(const char* str, size_t len) {
1.96 misha 87: char *result=new(PointerFreeGC) char[len+1];
88: memcpy(result, str, len);
89: result[len]=0;
90: return result;
1.92 misha 91: }
92:
1.106 moko 93: char *VForm::getAttributeValue(const char* data, const char *attr, size_t len) {
1.96 misha 94: const char* value=searchAttribute(data, attr, len);
95: if (value){
1.18 paf 96: size_t i;
97: if (!(len-=value-data)) return NULL;
98: if (*value=='"') {
99: for (i=1;i<len;i++) if (value[i]=='"') break;
100: return strpart(&value[1], i-1);
101: } else {
102: for (i=0;i<len;i++) if (strchr(" ;\"\n\r", value[i])) break;
103: return strpart(value, i);
104: }
1.96 misha 105: }
106: return NULL;
1.3 paf 107: }
108:
1.96 misha 109: String::C VForm::transcode(const char* client, size_t client_size, Charset* client_charset) {
1.75 paf 110: return Charset::transcode(
1.110 moko 111: String::C(pa_strdup(client, client_size), client_size),
1.96 misha 112: client_charset?*client_charset:fcharsets.client(),
113: fcharsets.source());
1.49 paf 114: }
115:
1.75 paf 116: void VForm::ParseGetFormInput(const char* query_string, size_t length) {
1.42 parser 117: ParseFormInput(query_string, length);
1.3 paf 118: }
119:
1.69 paf 120:
1.75 paf 121: static int atoi(const char* data, size_t alength) {
1.69 paf 122: char buf[MAX_STRING];
1.82 paf 123: size_t length=min(alength, sizeof(buf)-1);
124: memcpy(buf, data, length); buf[length]=0;
1.69 paf 125: return atoi(buf);
126: }
1.96 misha 127:
128: void VForm::ParseFormInput(const char* data, size_t length, Charset* client_charset) {
1.68 paf 129: // cut out ?image_map_tail
130: {
1.67 paf 131: for(size_t pos=0; pos<length; pos++) {
132: if(data[pos]=='?') {
1.69 paf 133: size_t start=pos+1;
134: size_t aftercomma=start;
135: size_t lookingcomma=start;
1.96 misha 136: for(; lookingcomma<length; lookingcomma++)
1.69 paf 137: if(data[lookingcomma]==',') {
138: aftercomma=++lookingcomma;
139: break;
140: }
141:
142: if(aftercomma>start) { // ?x,y
143: int x=atoi(data+start, aftercomma-1-start);
144: int y=atoi(data+aftercomma, length-aftercomma);
1.75 paf 145: imap.put(String("x"), new VInt(x));
146: imap.put(String("y"), new VInt(y));
1.69 paf 147: } else { // ?qtail
1.96 misha 148: AppendFormEntry("qtail", data+start, length-start, client_charset);
1.69 paf 149: }
1.67 paf 150: // cut tail
151: length=pos;
152: break;
153: }
154: }
155: }
1.68 paf 156: // Scan for pairs, unescaping and storing them as they are found
157: for(size_t pos=0; pos<length; ) {
1.16 paf 158: size_t start=pos;
1.68 paf 159: size_t finish=length;
1.96 misha 160: for(; pos<length; pos++)
1.68 paf 161: if(data[pos]=='&') {
162: finish=pos++;
1.3 paf 163: break;
164: }
1.68 paf 165:
166: size_t aftereq=start;
167: size_t lookingeq=start;
1.96 misha 168: for(; lookingeq<finish; lookingeq++)
1.68 paf 169: if(data[lookingeq]=='=') {
170: aftereq=++lookingeq;
1.3 paf 171: break;
172: }
1.45 paf 173:
1.96 misha 174: const char* attr=(aftereq>start)?unescape_chars(data+start, aftereq-1-start, &fcharsets.client()):"nameless";
1.91 misha 175: char *value=unescape_chars(data+aftereq, finish-aftereq, &fcharsets.client());
1.96 misha 176: AppendFormEntry(attr, value, strlen(value), client_charset);
1.3 paf 177: }
178: }
179:
1.92 misha 180: static char* pa_tolower(char *str){
181: if(!str)
182: return 0;
183: for(char *p=str; *p; p++)
184: *p=(char)tolower((unsigned char)*p);
185: return str;
186: }
187:
1.49 paf 188: void VForm::ParseMimeInput(
1.96 misha 189: char *content_type,
190: const char* data, size_t length, Charset* client_charset) {
1.3 paf 191: /* Scan for mime-presented pairs, storing them as they are found. */
1.92 misha 192: const char* boundary=pa_tolower(getAttributeValue(content_type, "boundary=", strlen(content_type)));
193: if(!boundary)
1.54 paf 194: throw Exception(0,
1.18 paf 195: 0,
1.3 paf 196: "VForm::ParseMimeInput no boundary attribute of Content-Type");
197:
1.92 misha 198: const char* lastData=&data[length];
199:
1.3 paf 200: while(true) {
1.66 paf 201: const char
1.18 paf 202: *dataStart=searchAttribute(data, boundary, lastData-data),
203: *dataEnd=searchAttribute(dataStart, boundary, lastData-dataStart);
1.92 misha 204:
1.18 paf 205: size_t headerSize=getHeader(dataStart, lastData-dataStart);
1.3 paf 206:
1.103 moko 207: if(!dataStart || !dataEnd || !headerSize) break;
1.47 paf 208: if(searchAttribute(dataStart, "content-disposition: form-data", headerSize)) {
1.16 paf 209: size_t valueSize=(dataEnd-dataStart)-headerSize-5-strlen(boundary);
1.18 paf 210: char *attr=getAttributeValue(dataStart, " name=", headerSize),
211: *fName=getAttributeValue(dataStart, " filename=", headerSize);
1.3 paf 212:
1.78 paf 213: if(attr) {
1.3 paf 214: /* OK, we have a new pair, add it to the list. */
1.80 paf 215: // fName checks are because MSIE passes unassigned <input type=file> as filename="" and empty body
1.85 misha 216: if( fName && (strlen(fName) || valueSize) ){
1.100 misha 217: AppendFormFileEntry(attr,
218: valueSize? &dataStart[headerSize+1]: "",
219: valueSize,
1.96 misha 220: fName,
1.100 misha 221: client_charset);
1.85 misha 222: } else {
1.100 misha 223: AppendFormEntry(attr,
224: valueSize? &dataStart[headerSize+1]: "",
1.96 misha 225: valueSize,
1.100 misha 226: client_charset);
1.85 misha 227: }
1.3 paf 228: }
229: }
230: data=(dataEnd-strlen(boundary));
231: }
232: }
233:
1.85 misha 234: void VForm::AppendFormFileEntry(const char* cname_cstr,
1.96 misha 235: const char* raw_cvalue_ptr, const size_t raw_cvalue_size,
236: const char* file_name_cstr, Charset* client_charset){
1.85 misha 237:
1.110 moko 238: const char* fname = pa_strdup(file_name_cstr);
1.101 misha 239: const String* sfile_name=new String(transcode(fname, strlen(fname), client_charset));
1.87 misha 240:
1.96 misha 241: const String& sname=*new String(transcode(cname_cstr, strlen(cname_cstr), client_charset));
1.87 misha 242: // maybe transcode text/* files?
1.89 misha 243: // NO!!! some users want to upload file 'as is' or file encoding can be unknown
1.87 misha 244:
1.85 misha 245: VFile* vfile=new VFile;
1.105 moko 246: vfile->set_binary(true/*tainted*/, raw_cvalue_ptr, raw_cvalue_size, sfile_name);
1.85 misha 247:
248: fields.put_dont_replace(sname, vfile);
249:
250: // files
251: Value* vhash=files.get(sname);
252: if(!vhash){
253: // first appearence
254: vhash=new VHash;
255: files.put(sname, vhash);
256: }
257: HashStringValue& hash=*vhash->get_hash();
1.75 paf 258:
1.85 misha 259: hash.put(String::Body::Format(hash.count()), vfile);
260: }
261:
1.96 misha 262: void VForm::AppendFormEntry(const char* cname_cstr, const char* raw_cvalue_ptr, const size_t raw_cvalue_size, Charset* client_charset) {
263: const String& sname=*new String(transcode(cname_cstr, strlen(cname_cstr), client_charset));
1.75 paf 264:
1.85 misha 265: const char* premature_zero_pos=(const char* )memchr(raw_cvalue_ptr, 0, raw_cvalue_size);
266: size_t cvalue_size=premature_zero_pos?premature_zero_pos-(const char* )raw_cvalue_ptr
267: :raw_cvalue_size;
1.110 moko 268: char *cvalue_ptr=pa_strdup(raw_cvalue_ptr, cvalue_size);
1.85 misha 269: fix_line_breaks(cvalue_ptr, cvalue_size);
1.96 misha 270: String& string=*new String(transcode(cvalue_ptr, cvalue_size, client_charset), String::L_TAINTED);
1.85 misha 271:
272: // tables
273: {
274: Value* vtable=tables.get(sname);
275: if(!vtable) {
276: // first appearence
277: Table::columns_type columns(new ArrayString(1));
278: *columns+=new String("field");
1.75 paf 279:
1.85 misha 280: vtable=new VTable(new Table(columns));
281: tables.put(sname, vtable);
1.36 parser 282: }
1.85 misha 283: Table& table=*vtable->get_table();
284:
285: // this string becomes next row
286: Table::element_type row(new ArrayString(1));
287: *row+=&string;
288: table+=row;
1.3 paf 289: }
1.69 paf 290:
1.85 misha 291: fields.put_dont_replace(sname, new VString(string));
1.69 paf 292: }
293:
1.85 misha 294:
295: void VForm::refill_fields_tables_and_files() {
1.77 paf 296: fields.clear();
297: tables.clear();
1.85 misha 298: files.clear();
1.77 paf 299: imap.clear();
1.75 paf 300:
1.77 paf 301: //frequest_info.query_string="a=123";
1.3 paf 302: // parsing QS [GET and ?name=value from uri rewrite)]
1.77 paf 303: if(frequest_info.query_string) {
304: size_t length=strlen(frequest_info.query_string);
1.110 moko 305: char *buf=pa_strdup(frequest_info.query_string, length);
1.42 parser 306: ParseGetFormInput(buf, length);
307: }
1.62 paf 308:
309: #ifdef DEBUG_POST
1.77 paf 310: frequest_info.method="POST";
1.96 misha 311: File_read_result file=file_read(fcharsets, *new String("test.stdin"), true/*as_text*/, 0, true, 0, 0, 0, false/*transcode*/);
312: frequest_info.post_size=file.length;
313: frequest_info.post_data=(char*)file.str;
1.77 paf 314: frequest_info.content_type="multipart/form-data; boundary=----------mcqY2UDNcdEAoN1mLmne2i";
1.62 paf 315:
316: #endif
1.57 paf 317:
318: // parsing POST data
1.115 ! moko 319: switch(post_content_type){
! 320: case FORM_URLENCODED: {
! 321: detect_post_charset();
! 322: ParseFormInput(frequest_info.post_data, frequest_info.post_size, fpost_charset);
! 323: break;
! 324: }
! 325: case MULTIPART_FORMDATA: {
! 326: ParseMimeInput(pa_strdup(frequest_info.content_type), frequest_info.post_data, frequest_info.post_size);
! 327: break;
1.96 misha 328: }
1.115 ! moko 329: case UNKNOWN: break;
! 330: }
1.56 paf 331:
1.77 paf 332: filled_source=&fcharsets.source();
333: filled_client=&fcharsets.client();
334: }
335:
1.99 misha 336: void VForm::detect_post_charset(){
1.107 misha 337: if(can_have_body && !charset_detected){
1.99 misha 338: fpost_charset=detect_charset(frequest_info.content_type);
1.107 misha 339: charset_detected=true;
1.99 misha 340: }
341: }
342:
1.85 misha 343: bool VForm::should_refill_fields_tables_and_files() {
1.96 misha 344: return &fcharsets.source()!=filled_source || &fcharsets.client()!=filled_client;
1.1 paf 345: }
1.33 parser 346:
1.98 misha 347: Value* VForm::get_element(const String& aname) {
1.85 misha 348: if(should_refill_fields_tables_and_files())
349: refill_fields_tables_and_files();
1.56 paf 350:
1.36 parser 351: // $fields
352: if(aname==FORM_FIELDS_ELEMENT_NAME)
1.75 paf 353: return new VHash(fields);
1.36 parser 354:
355: // $tables
356: if(aname==FORM_TABLES_ELEMENT_NAME)
1.75 paf 357: return new VHash(tables);
1.69 paf 358:
1.85 misha 359: // $files
360: if(aname==FORM_FILES_ELEMENT_NAME)
361: return new VHash(files);
362:
1.77 paf 363: // $imap
1.69 paf 364: if(aname==FORM_IMAP_ELEMENT_NAME)
1.75 paf 365: return new VHash(imap);
1.36 parser 366:
1.108 misha 367: // methods (if any)
1.98 misha 368: if(Value* result=VStateless_class::get_element(aname))
1.33 parser 369: return result;
370:
1.95 misha 371: // $field
1.75 paf 372: return fields.get(aname);
1.72 paf 373: }
1.93 misha 374:
1.109 moko 375: Charset* VForm::get_body_charset(){
1.99 misha 376: detect_post_charset();
1.96 misha 377: return fpost_charset;
1.93 misha 378: }
E-mail: