Annotation of parser3/src/types/pa_vform.C, revision 1.130
1.27 paf 1: /** @file
2: Parser: @b form class.
1.10 paf 3:
1.125 moko 4: Copyright (c) 2001-2024 Art. Lebedev Studio (http://www.artlebedev.com)
1.123 moko 5: Authors: Konstantin Morshnev <moko@design.ru>, Alexandr Petrosian <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"
1.126 moko 16: #include "pa_varray.h"
1.21 paf 17: #include "pa_common.h"
1.30 parser 18: #include "pa_vtable.h"
1.50 paf 19: #include "pa_charset.h"
1.102 moko 20:
1.130 ! moko 21: volatile const char * IDENT_PA_VFORM_C="$Id: pa_vform.C,v 1.129 2024/12/23 18:30:55 moko Exp $" IDENT_PA_VFORM_H;
1.3 paf 22:
1.62 paf 23: // defines
24:
25: //#define DEBUG_POST
26:
1.3 paf 27: // parse helper funcs
28:
1.75 paf 29: static size_t getHeader(const char* data, size_t len){
1.97 misha 30: size_t i;
31: int enter=-1;
1.111 moko 32: if (data) {
1.62 paf 33: for (i=0;i<len;i++)
34: if (data[i]=='\n'){
35: if (enter>=0) enter++;
36: if (enter>1) return i;
1.111 moko 37: } else {
38: if (data[i]!='\r') enter=0;
39: }
40: }
1.100 misha 41: return 0;
1.3 paf 42: }
43:
1.75 paf 44: static const char* searchAttribute(const char* data,
45: const char* attr, //< expected to be lowercased
1.62 paf 46: size_t len){
1.97 misha 47: size_t i;
48: if (data)
1.62 paf 49: for (i=0;i<len;i++)
1.83 paf 50: if (tolower((unsigned char)data[i])==*attr){
1.62 paf 51: size_t j;
52: for (j=i+1;j<=len;j++)
53: if (!attr[j-i]) return &data[j];
54: else {
55: if (j==len) break;
1.83 paf 56: if (attr[j-i]!=tolower((unsigned char)data[j])) break;
1.62 paf 57: }
58: }
59:
60: return NULL;
1.3 paf 61: }
62:
63: // VForm
64:
1.113 moko 65: extern Methoded* form_class;
1.57 paf 66:
1.114 moko 67: VForm::VForm(Request_charsets& acharsets, Request_info& arequest_info): VStateless_class(form_class),
1.77 paf 68: fcharsets(acharsets),
69: frequest_info(arequest_info),
1.115 moko 70:
71: can_have_body(arequest_info.can_have_body()),
1.116 moko 72: charset_detected(false),
1.115 moko 73: post_content_type(UNKNOWN),
1.116 moko 74:
1.129 moko 75: filled_source(0), filled_client(0), fpost_charset(0),
1.126 moko 76:
1.129 moko 77: fields(vfields.hash()), tables(vtables.hash()), files(vfiles.hash()), imap(vimap.hash()), elements(velements.hash())
1.96 misha 78: {
1.111 moko 79: if(can_have_body && arequest_info.content_type) {
80: if(pa_strncasecmp(arequest_info.content_type, HTTP_CONTENT_TYPE_FORM_URLENCODED)==0) {
1.96 misha 81: post_content_type=FORM_URLENCODED;
1.111 moko 82: } else if(pa_strncasecmp(arequest_info.content_type, HTTP_CONTENT_TYPE_MULTIPART_FORMDATA)==0) {
1.96 misha 83: post_content_type=MULTIPART_FORMDATA;
1.111 moko 84: }
85: }
1.3 paf 86: }
87:
1.92 misha 88: char *VForm::strpart(const char* str, size_t len) {
1.96 misha 89: char *result=new(PointerFreeGC) char[len+1];
90: memcpy(result, str, len);
91: result[len]=0;
92: return result;
1.92 misha 93: }
94:
1.106 moko 95: char *VForm::getAttributeValue(const char* data, const char *attr, size_t len) {
1.96 misha 96: const char* value=searchAttribute(data, attr, len);
97: if (value){
1.18 paf 98: size_t i;
99: if (!(len-=value-data)) return NULL;
100: if (*value=='"') {
101: for (i=1;i<len;i++) if (value[i]=='"') break;
102: return strpart(&value[1], i-1);
103: } else {
104: for (i=0;i<len;i++) if (strchr(" ;\"\n\r", value[i])) break;
105: return strpart(value, i);
106: }
1.96 misha 107: }
108: return NULL;
1.3 paf 109: }
110:
1.96 misha 111: String::C VForm::transcode(const char* client, size_t client_size, Charset* client_charset) {
1.75 paf 112: return Charset::transcode(
1.110 moko 113: String::C(pa_strdup(client, client_size), client_size),
1.96 misha 114: client_charset?*client_charset:fcharsets.client(),
115: fcharsets.source());
1.49 paf 116: }
117:
1.75 paf 118: void VForm::ParseGetFormInput(const char* query_string, size_t length) {
1.42 parser 119: ParseFormInput(query_string, length);
1.3 paf 120: }
121:
1.69 paf 122:
1.75 paf 123: static int atoi(const char* data, size_t alength) {
1.69 paf 124: char buf[MAX_STRING];
1.82 paf 125: size_t length=min(alength, sizeof(buf)-1);
126: memcpy(buf, data, length); buf[length]=0;
1.69 paf 127: return atoi(buf);
128: }
1.96 misha 129:
130: void VForm::ParseFormInput(const char* data, size_t length, Charset* client_charset) {
1.68 paf 131: // cut out ?image_map_tail
132: {
1.67 paf 133: for(size_t pos=0; pos<length; pos++) {
134: if(data[pos]=='?') {
1.69 paf 135: size_t start=pos+1;
136: size_t aftercomma=start;
137: size_t lookingcomma=start;
1.96 misha 138: for(; lookingcomma<length; lookingcomma++)
1.69 paf 139: if(data[lookingcomma]==',') {
140: aftercomma=++lookingcomma;
141: break;
142: }
143:
144: if(aftercomma>start) { // ?x,y
145: int x=atoi(data+start, aftercomma-1-start);
146: int y=atoi(data+aftercomma, length-aftercomma);
1.75 paf 147: imap.put(String("x"), new VInt(x));
148: imap.put(String("y"), new VInt(y));
1.69 paf 149: } else { // ?qtail
1.96 misha 150: AppendFormEntry("qtail", data+start, length-start, client_charset);
1.69 paf 151: }
1.67 paf 152: // cut tail
153: length=pos;
154: break;
155: }
156: }
157: }
1.68 paf 158: // Scan for pairs, unescaping and storing them as they are found
159: for(size_t pos=0; pos<length; ) {
1.16 paf 160: size_t start=pos;
1.68 paf 161: size_t finish=length;
1.96 misha 162: for(; pos<length; pos++)
1.68 paf 163: if(data[pos]=='&') {
164: finish=pos++;
1.3 paf 165: break;
166: }
1.68 paf 167:
168: size_t aftereq=start;
169: size_t lookingeq=start;
1.96 misha 170: for(; lookingeq<finish; lookingeq++)
1.68 paf 171: if(data[lookingeq]=='=') {
172: aftereq=++lookingeq;
1.3 paf 173: break;
174: }
1.45 paf 175:
1.96 misha 176: const char* attr=(aftereq>start)?unescape_chars(data+start, aftereq-1-start, &fcharsets.client()):"nameless";
1.91 misha 177: char *value=unescape_chars(data+aftereq, finish-aftereq, &fcharsets.client());
1.96 misha 178: AppendFormEntry(attr, value, strlen(value), client_charset);
1.3 paf 179: }
180: }
181:
1.92 misha 182: static char* pa_tolower(char *str){
183: if(!str)
184: return 0;
185: for(char *p=str; *p; p++)
186: *p=(char)tolower((unsigned char)*p);
187: return str;
188: }
189:
1.49 paf 190: void VForm::ParseMimeInput(
1.96 misha 191: char *content_type,
192: const char* data, size_t length, Charset* client_charset) {
1.3 paf 193: /* Scan for mime-presented pairs, storing them as they are found. */
1.92 misha 194: const char* boundary=pa_tolower(getAttributeValue(content_type, "boundary=", strlen(content_type)));
195: if(!boundary)
1.126 moko 196: throw Exception(0, 0, "VForm::ParseMimeInput no boundary attribute of Content-Type");
1.3 paf 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
1.126 moko 251: {
252: Value* vhash=files.get(sname);
253: if(!vhash)
254: files.put(sname, vhash=new VHash);
255: HashStringValue& hash=*vhash->get_hash();
1.130 ! moko 256: hash.put(String::Body::uitoa(hash.count()), vfile);
1.85 misha 257: }
1.75 paf 258:
1.126 moko 259: // elements
260: {
261: Value* varray=elements.get(sname);
262: if(!varray)
263: elements.put(sname, varray=new VArray);
264: if(VArray* array=dynamic_cast<VArray*>(varray))
265: array->array()+=vfile;
266: }
1.85 misha 267: }
268:
1.96 misha 269: void VForm::AppendFormEntry(const char* cname_cstr, const char* raw_cvalue_ptr, const size_t raw_cvalue_size, Charset* client_charset) {
270: const String& sname=*new String(transcode(cname_cstr, strlen(cname_cstr), client_charset));
1.75 paf 271:
1.85 misha 272: const char* premature_zero_pos=(const char* )memchr(raw_cvalue_ptr, 0, raw_cvalue_size);
1.126 moko 273: size_t cvalue_size=premature_zero_pos ? premature_zero_pos - (const char* )raw_cvalue_ptr : raw_cvalue_size;
274: char *cvalue_ptr=pa_strdup(raw_cvalue_ptr, cvalue_size);
1.85 misha 275: fix_line_breaks(cvalue_ptr, cvalue_size);
1.96 misha 276: String& string=*new String(transcode(cvalue_ptr, cvalue_size, client_charset), String::L_TAINTED);
1.126 moko 277: Value *value=new VString(string);
1.85 misha 278:
279: // tables
280: {
281: Value* vtable=tables.get(sname);
282: if(!vtable) {
283: Table::columns_type columns(new ArrayString(1));
284: *columns+=new String("field");
1.75 paf 285:
1.85 misha 286: vtable=new VTable(new Table(columns));
287: tables.put(sname, vtable);
1.36 parser 288: }
1.85 misha 289: Table& table=*vtable->get_table();
290:
291: // this string becomes next row
292: Table::element_type row(new ArrayString(1));
293: *row+=&string;
294: table+=row;
1.3 paf 295: }
1.69 paf 296:
1.126 moko 297: // elements
298: {
299: Value* varray=elements.get(sname);
300: if(!varray)
301: elements.put(sname, varray=new VArray);
302: if(VArray* array=dynamic_cast<VArray*>(varray))
303: array->array()+=value;
304: }
305:
306: fields.put_dont_replace(sname, value);
1.69 paf 307: }
308:
1.85 misha 309:
310: void VForm::refill_fields_tables_and_files() {
1.77 paf 311: fields.clear();
312: tables.clear();
1.85 misha 313: files.clear();
1.77 paf 314: imap.clear();
1.126 moko 315: elements.clear();
1.75 paf 316:
1.77 paf 317: //frequest_info.query_string="a=123";
1.3 paf 318: // parsing QS [GET and ?name=value from uri rewrite)]
1.77 paf 319: if(frequest_info.query_string) {
320: size_t length=strlen(frequest_info.query_string);
1.110 moko 321: char *buf=pa_strdup(frequest_info.query_string, length);
1.42 parser 322: ParseGetFormInput(buf, length);
323: }
1.62 paf 324:
325: #ifdef DEBUG_POST
1.77 paf 326: frequest_info.method="POST";
1.119 moko 327: File_read_result file=file_read_binary(*new String("test.stdin"), true /*fail on problem*/);
1.96 misha 328: frequest_info.post_size=file.length;
329: frequest_info.post_data=(char*)file.str;
1.77 paf 330: frequest_info.content_type="multipart/form-data; boundary=----------mcqY2UDNcdEAoN1mLmne2i";
1.62 paf 331: #endif
1.57 paf 332:
333: // parsing POST data
1.115 moko 334: switch(post_content_type){
335: case FORM_URLENCODED: {
336: detect_post_charset();
337: ParseFormInput(frequest_info.post_data, frequest_info.post_size, fpost_charset);
338: break;
339: }
340: case MULTIPART_FORMDATA: {
341: ParseMimeInput(pa_strdup(frequest_info.content_type), frequest_info.post_data, frequest_info.post_size);
342: break;
1.96 misha 343: }
1.115 moko 344: case UNKNOWN: break;
345: }
1.56 paf 346:
1.77 paf 347: filled_source=&fcharsets.source();
348: filled_client=&fcharsets.client();
349: }
350:
1.99 misha 351: void VForm::detect_post_charset(){
1.107 misha 352: if(can_have_body && !charset_detected){
1.99 misha 353: fpost_charset=detect_charset(frequest_info.content_type);
1.107 misha 354: charset_detected=true;
1.99 misha 355: }
356: }
357:
1.85 misha 358: bool VForm::should_refill_fields_tables_and_files() {
1.96 misha 359: return &fcharsets.source()!=filled_source || &fcharsets.client()!=filled_client;
1.1 paf 360: }
1.33 parser 361:
1.98 misha 362: Value* VForm::get_element(const String& aname) {
1.85 misha 363: if(should_refill_fields_tables_and_files())
364: refill_fields_tables_and_files();
1.56 paf 365:
1.36 parser 366: // $fields
1.126 moko 367: if(SYMBOLS_EQ(aname, FIELDS_SYMBOL))
1.127 moko 368: return new VHash(vfields);
1.36 parser 369:
370: // $tables
1.126 moko 371: if(SYMBOLS_EQ(aname, TABLES_SYMBOL))
1.128 moko 372: return &vtables;
1.69 paf 373:
1.85 misha 374: // $files
1.126 moko 375: if(SYMBOLS_EQ(aname, FILES_SYMBOL))
1.128 moko 376: return &vfiles;
1.85 misha 377:
1.77 paf 378: // $imap
1.126 moko 379: if(SYMBOLS_EQ(aname, IMAP_SYMBOL))
1.127 moko 380: return new VHash(vimap);
1.126 moko 381:
382: // $elements
383: if(SYMBOLS_EQ(aname, ELEMENTS_SYMBOL))
384: return &velements;
1.36 parser 385:
1.108 misha 386: // methods (if any)
1.98 misha 387: if(Value* result=VStateless_class::get_element(aname))
1.33 parser 388: return result;
389:
1.95 misha 390: // $field
1.75 paf 391: return fields.get(aname);
1.72 paf 392: }
1.93 misha 393:
1.109 moko 394: Charset* VForm::get_body_charset(){
1.99 misha 395: detect_post_charset();
1.96 misha 396: return fpost_charset;
1.93 misha 397: }
1.120 moko 398:
399: const VJunction* VForm::put_element(const String& aname, Value* avalue) {
400: if(should_refill_fields_tables_and_files())
401: refill_fields_tables_and_files();
402:
403: fields.put(aname, avalue);
1.121 moko 404: return 0;
1.120 moko 405: }
E-mail: