Annotation of parser3/src/types/pa_vform.C, revision 1.106
1.27 paf 1: /** @file
2: Parser: @b form class.
1.10 paf 3:
1.102 moko 4: Copyright (c) 2001-2012 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:
1.106 ! moko 20: volatile const char * IDENT_PA_VFORM_C="$Id: pa_vform.C,v 1.105 2012/06/15 11:54:18 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;
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;
36: } else if (data[i]!='\r') enter=0;
1.100 misha 37: return 0;
1.3 paf 38: }
39:
1.75 paf 40: static const char* searchAttribute(const char* data,
41: const char* attr, //< expected to be lowercased
1.62 paf 42: size_t len){
1.97 misha 43: size_t i;
44: if (data)
1.62 paf 45: for (i=0;i<len;i++)
1.83 paf 46: if (tolower((unsigned char)data[i])==*attr){
1.62 paf 47: size_t j;
48: for (j=i+1;j<=len;j++)
49: if (!attr[j-i]) return &data[j];
50: else {
51: if (j==len) break;
1.83 paf 52: if (attr[j-i]!=tolower((unsigned char)data[j])) break;
1.62 paf 53: }
54: }
55:
56: return NULL;
1.3 paf 57: }
58:
59: // VForm
60:
1.75 paf 61: extern Methoded* form_base_class;
1.57 paf 62:
1.77 paf 63: VForm::VForm(Request_charsets& acharsets, Request_info& arequest_info): VStateless_class(0, form_base_class),
64: fcharsets(acharsets),
65: frequest_info(arequest_info),
66: filled_source(0),
1.91 misha 67: filled_client(0),
1.96 misha 68: fpost_charset(0)
69: {
70: is_post=(arequest_info.method && StrStartFromNC(arequest_info.method, "post", true));
1.99 misha 71: is_post_charset_detected=false;
1.96 misha 72:
73: post_content_type=UNKNOWN;
74: if(is_post && arequest_info.content_type)
1.99 misha 75: if(StrStartFromNC(arequest_info.content_type, HTTP_CONTENT_TYPE_FORM_URLENCODED))
1.96 misha 76: post_content_type=FORM_URLENCODED;
1.99 misha 77: else if(StrStartFromNC(arequest_info.content_type, HTTP_CONTENT_TYPE_MULTIPART_FORMDATA))
1.96 misha 78: post_content_type=MULTIPART_FORMDATA;
1.3 paf 79: }
80:
1.92 misha 81: char *VForm::strpart(const char* str, size_t len) {
1.96 misha 82: char *result=new(PointerFreeGC) char[len+1];
83: memcpy(result, str, len);
84: result[len]=0;
85: return result;
1.92 misha 86: }
87:
1.106 ! moko 88: char *VForm::getAttributeValue(const char* data, const char *attr, size_t len) {
1.96 misha 89: const char* value=searchAttribute(data, attr, len);
90: if (value){
1.18 paf 91: size_t i;
92: if (!(len-=value-data)) return NULL;
93: if (*value=='"') {
94: for (i=1;i<len;i++) if (value[i]=='"') break;
95: return strpart(&value[1], i-1);
96: } else {
97: for (i=0;i<len;i++) if (strchr(" ;\"\n\r", value[i])) break;
98: return strpart(value, i);
99: }
1.96 misha 100: }
101: return NULL;
1.3 paf 102: }
103:
1.96 misha 104: String::C VForm::transcode(const char* client, size_t client_size, Charset* client_charset) {
1.75 paf 105: return Charset::transcode(
1.96 misha 106: String::C(strdup(client, client_size), client_size),
107: client_charset?*client_charset:fcharsets.client(),
108: fcharsets.source());
1.49 paf 109: }
110:
1.75 paf 111: void VForm::ParseGetFormInput(const char* query_string, size_t length) {
1.42 parser 112: ParseFormInput(query_string, length);
1.3 paf 113: }
114:
1.69 paf 115:
1.75 paf 116: static int atoi(const char* data, size_t alength) {
1.69 paf 117: char buf[MAX_STRING];
1.82 paf 118: size_t length=min(alength, sizeof(buf)-1);
119: memcpy(buf, data, length); buf[length]=0;
1.69 paf 120: return atoi(buf);
121: }
1.96 misha 122:
123: void VForm::ParseFormInput(const char* data, size_t length, Charset* client_charset) {
1.68 paf 124: // cut out ?image_map_tail
125: {
1.67 paf 126: for(size_t pos=0; pos<length; pos++) {
127: if(data[pos]=='?') {
1.69 paf 128: size_t start=pos+1;
129: size_t aftercomma=start;
130: size_t lookingcomma=start;
1.96 misha 131: for(; lookingcomma<length; lookingcomma++)
1.69 paf 132: if(data[lookingcomma]==',') {
133: aftercomma=++lookingcomma;
134: break;
135: }
136:
137: if(aftercomma>start) { // ?x,y
138: int x=atoi(data+start, aftercomma-1-start);
139: int y=atoi(data+aftercomma, length-aftercomma);
1.75 paf 140: imap.put(String("x"), new VInt(x));
141: imap.put(String("y"), new VInt(y));
1.69 paf 142: } else { // ?qtail
1.96 misha 143: AppendFormEntry("qtail", data+start, length-start, client_charset);
1.69 paf 144: }
1.67 paf 145: // cut tail
146: length=pos;
147: break;
148: }
149: }
150: }
1.68 paf 151: // Scan for pairs, unescaping and storing them as they are found
152: for(size_t pos=0; pos<length; ) {
1.16 paf 153: size_t start=pos;
1.68 paf 154: size_t finish=length;
1.96 misha 155: for(; pos<length; pos++)
1.68 paf 156: if(data[pos]=='&') {
157: finish=pos++;
1.3 paf 158: break;
159: }
1.68 paf 160:
161: size_t aftereq=start;
162: size_t lookingeq=start;
1.96 misha 163: for(; lookingeq<finish; lookingeq++)
1.68 paf 164: if(data[lookingeq]=='=') {
165: aftereq=++lookingeq;
1.3 paf 166: break;
167: }
1.45 paf 168:
1.96 misha 169: const char* attr=(aftereq>start)?unescape_chars(data+start, aftereq-1-start, &fcharsets.client()):"nameless";
1.91 misha 170: char *value=unescape_chars(data+aftereq, finish-aftereq, &fcharsets.client());
1.96 misha 171: AppendFormEntry(attr, value, strlen(value), client_charset);
1.3 paf 172: }
173: }
174:
1.92 misha 175: static char* pa_tolower(char *str){
176: if(!str)
177: return 0;
178: for(char *p=str; *p; p++)
179: *p=(char)tolower((unsigned char)*p);
180: return str;
181: }
182:
1.49 paf 183: void VForm::ParseMimeInput(
1.96 misha 184: char *content_type,
185: const char* data, size_t length, Charset* client_charset) {
1.3 paf 186: /* Scan for mime-presented pairs, storing them as they are found. */
1.92 misha 187: const char* boundary=pa_tolower(getAttributeValue(content_type, "boundary=", strlen(content_type)));
188: if(!boundary)
1.54 paf 189: throw Exception(0,
1.18 paf 190: 0,
1.3 paf 191: "VForm::ParseMimeInput no boundary attribute of Content-Type");
192:
1.92 misha 193: const char* lastData=&data[length];
194:
1.3 paf 195: while(true) {
1.66 paf 196: const char
1.18 paf 197: *dataStart=searchAttribute(data, boundary, lastData-data),
198: *dataEnd=searchAttribute(dataStart, boundary, lastData-dataStart);
1.92 misha 199:
1.18 paf 200: size_t headerSize=getHeader(dataStart, lastData-dataStart);
1.3 paf 201:
1.103 moko 202: if(!dataStart || !dataEnd || !headerSize) break;
1.47 paf 203: if(searchAttribute(dataStart, "content-disposition: form-data", headerSize)) {
1.16 paf 204: size_t valueSize=(dataEnd-dataStart)-headerSize-5-strlen(boundary);
1.18 paf 205: char *attr=getAttributeValue(dataStart, " name=", headerSize),
206: *fName=getAttributeValue(dataStart, " filename=", headerSize);
1.3 paf 207:
1.78 paf 208: if(attr) {
1.3 paf 209: /* OK, we have a new pair, add it to the list. */
1.80 paf 210: // fName checks are because MSIE passes unassigned <input type=file> as filename="" and empty body
1.85 misha 211: if( fName && (strlen(fName) || valueSize) ){
1.100 misha 212: AppendFormFileEntry(attr,
213: valueSize? &dataStart[headerSize+1]: "",
214: valueSize,
1.96 misha 215: fName,
1.100 misha 216: client_charset);
1.85 misha 217: } else {
1.100 misha 218: AppendFormEntry(attr,
219: valueSize? &dataStart[headerSize+1]: "",
1.96 misha 220: valueSize,
1.100 misha 221: client_charset);
1.85 misha 222: }
1.3 paf 223: }
224: }
225: data=(dataEnd-strlen(boundary));
226: }
227: }
228:
1.85 misha 229: void VForm::AppendFormFileEntry(const char* cname_cstr,
1.96 misha 230: const char* raw_cvalue_ptr, const size_t raw_cvalue_size,
231: const char* file_name_cstr, Charset* client_charset){
1.85 misha 232:
1.87 misha 233: const char* fname = strdup(file_name_cstr);
1.101 misha 234: const String* sfile_name=new String(transcode(fname, strlen(fname), client_charset));
1.87 misha 235:
1.96 misha 236: const String& sname=*new String(transcode(cname_cstr, strlen(cname_cstr), client_charset));
1.87 misha 237: // maybe transcode text/* files?
1.89 misha 238: // NO!!! some users want to upload file 'as is' or file encoding can be unknown
1.87 misha 239:
1.85 misha 240: VFile* vfile=new VFile;
1.105 moko 241: vfile->set_binary(true/*tainted*/, raw_cvalue_ptr, raw_cvalue_size, sfile_name);
1.85 misha 242:
243: fields.put_dont_replace(sname, vfile);
244:
245: // files
246: Value* vhash=files.get(sname);
247: if(!vhash){
248: // first appearence
249: vhash=new VHash;
250: files.put(sname, vhash);
251: }
252: HashStringValue& hash=*vhash->get_hash();
1.75 paf 253:
1.85 misha 254: hash.put(String::Body::Format(hash.count()), vfile);
255: }
256:
1.96 misha 257: void VForm::AppendFormEntry(const char* cname_cstr, const char* raw_cvalue_ptr, const size_t raw_cvalue_size, Charset* client_charset) {
258: const String& sname=*new String(transcode(cname_cstr, strlen(cname_cstr), client_charset));
1.75 paf 259:
1.85 misha 260: const char* premature_zero_pos=(const char* )memchr(raw_cvalue_ptr, 0, raw_cvalue_size);
261: size_t cvalue_size=premature_zero_pos?premature_zero_pos-(const char* )raw_cvalue_ptr
262: :raw_cvalue_size;
263: char *cvalue_ptr=strdup(raw_cvalue_ptr, cvalue_size);
264: fix_line_breaks(cvalue_ptr, cvalue_size);
1.96 misha 265: String& string=*new String(transcode(cvalue_ptr, cvalue_size, client_charset), String::L_TAINTED);
1.85 misha 266:
267: // tables
268: {
269: Value* vtable=tables.get(sname);
270: if(!vtable) {
271: // first appearence
272: Table::columns_type columns(new ArrayString(1));
273: *columns+=new String("field");
1.75 paf 274:
1.85 misha 275: vtable=new VTable(new Table(columns));
276: tables.put(sname, vtable);
1.36 parser 277: }
1.85 misha 278: Table& table=*vtable->get_table();
279:
280: // this string becomes next row
281: Table::element_type row(new ArrayString(1));
282: *row+=&string;
283: table+=row;
1.3 paf 284: }
1.69 paf 285:
1.85 misha 286: fields.put_dont_replace(sname, new VString(string));
1.69 paf 287: }
288:
1.85 misha 289:
290: void VForm::refill_fields_tables_and_files() {
1.77 paf 291: fields.clear();
292: tables.clear();
1.85 misha 293: files.clear();
1.77 paf 294: imap.clear();
1.75 paf 295:
1.77 paf 296: //frequest_info.query_string="a=123";
1.3 paf 297: // parsing QS [GET and ?name=value from uri rewrite)]
1.77 paf 298: if(frequest_info.query_string) {
299: size_t length=strlen(frequest_info.query_string);
300: char *buf=strdup(frequest_info.query_string, length);
1.42 parser 301: ParseGetFormInput(buf, length);
302: }
1.62 paf 303:
304: #ifdef DEBUG_POST
1.77 paf 305: frequest_info.method="POST";
1.96 misha 306: File_read_result file=file_read(fcharsets, *new String("test.stdin"), true/*as_text*/, 0, true, 0, 0, 0, false/*transcode*/);
307: frequest_info.post_size=file.length;
308: frequest_info.post_data=(char*)file.str;
1.77 paf 309: frequest_info.content_type="multipart/form-data; boundary=----------mcqY2UDNcdEAoN1mLmne2i";
1.62 paf 310:
311: #endif
1.57 paf 312:
313: // parsing POST data
1.96 misha 314: if(is_post && frequest_info.content_type)
315: switch(post_content_type){
316: case FORM_URLENCODED:
317: {
1.99 misha 318: detect_post_charset();
1.96 misha 319: ParseFormInput(frequest_info.post_data, frequest_info.post_size, fpost_charset);
320: break;
321: }
322: case MULTIPART_FORMDATA:
323: {
324: ParseMimeInput(strdup(frequest_info.content_type), frequest_info.post_data, frequest_info.post_size);
325: break;
326: }
327: }
1.56 paf 328:
1.77 paf 329: filled_source=&fcharsets.source();
330: filled_client=&fcharsets.client();
331: }
332:
1.99 misha 333: void VForm::detect_post_charset(){
334: if(is_post && !is_post_charset_detected){
335: fpost_charset=detect_charset(frequest_info.content_type);
336: is_post_charset_detected=true;
337: }
338: }
339:
1.85 misha 340: bool VForm::should_refill_fields_tables_and_files() {
1.96 misha 341: return &fcharsets.source()!=filled_source || &fcharsets.client()!=filled_client;
1.1 paf 342: }
1.33 parser 343:
1.98 misha 344: Value* VForm::get_element(const String& aname) {
1.85 misha 345: if(should_refill_fields_tables_and_files())
346: refill_fields_tables_and_files();
1.56 paf 347:
1.36 parser 348: // $fields
349: if(aname==FORM_FIELDS_ELEMENT_NAME)
1.75 paf 350: return new VHash(fields);
1.36 parser 351:
352: // $tables
353: if(aname==FORM_TABLES_ELEMENT_NAME)
1.75 paf 354: return new VHash(tables);
1.69 paf 355:
1.85 misha 356: // $files
357: if(aname==FORM_FILES_ELEMENT_NAME)
358: return new VHash(files);
359:
1.77 paf 360: // $imap
1.69 paf 361: if(aname==FORM_IMAP_ELEMENT_NAME)
1.75 paf 362: return new VHash(imap);
1.36 parser 363:
1.96 misha 364: // CLASS, CLASS_NAME
1.98 misha 365: if(Value* result=VStateless_class::get_element(aname))
1.33 parser 366: return result;
367:
1.95 misha 368: // $field
1.75 paf 369: return fields.get(aname);
1.72 paf 370: }
1.93 misha 371:
372: Charset* VForm::get_post_charset(){
1.99 misha 373: detect_post_charset();
1.96 misha 374: return fpost_charset;
1.93 misha 375: }
E-mail: