Annotation of parser3/src/types/pa_vform.C, revision 1.98
1.27 paf 1: /** @file
2: Parser: @b form class.
1.10 paf 3:
1.95 misha 4: Copyright(c) 2001-2009 ArtLebedev Group (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:
1.98 ! misha 10: static const char * const IDENT_VFORM_C="$Date: 2009-06-24 09:02:22 $";
1.3 paf 11:
1.14 paf 12: #include "pa_sapi.h"
1.1 paf 13: #include "pa_vform.h"
14: #include "pa_vstring.h"
1.3 paf 15: #include "pa_globals.h"
16: #include "pa_request.h"
1.9 paf 17: #include "pa_vfile.h"
1.21 paf 18: #include "pa_common.h"
1.30 parser 19: #include "pa_vtable.h"
1.50 paf 20: #include "pa_charset.h"
1.86 misha 21: //#include "pa_charsets.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;
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;
37: } else if (data[i]!='\r') enter=0;
38: return 0;
1.3 paf 39: }
40:
1.75 paf 41: static const char* searchAttribute(const char* data,
42: const char* attr, //< expected to be lowercased
1.62 paf 43: size_t len){
1.97 misha 44: size_t i;
45: if (data)
1.62 paf 46: for (i=0;i<len;i++)
1.83 paf 47: if (tolower((unsigned char)data[i])==*attr){
1.62 paf 48: size_t j;
49: for (j=i+1;j<=len;j++)
50: if (!attr[j-i]) return &data[j];
51: else {
52: if (j==len) break;
1.83 paf 53: if (attr[j-i]!=tolower((unsigned char)data[j])) break;
1.62 paf 54: }
55: }
56:
57: return NULL;
1.3 paf 58: }
59:
60: // VForm
61:
1.75 paf 62: extern Methoded* form_base_class;
1.57 paf 63:
1.77 paf 64: VForm::VForm(Request_charsets& acharsets, Request_info& arequest_info): VStateless_class(0, form_base_class),
65: fcharsets(acharsets),
66: frequest_info(arequest_info),
67: filled_source(0),
1.91 misha 68: filled_client(0),
1.96 misha 69: fpost_charset(0)
70: {
71: is_post=(arequest_info.method && StrStartFromNC(arequest_info.method, "post", true));
72:
73: post_content_type=UNKNOWN;
74: if(is_post && arequest_info.content_type)
75: if(StrStartFromNC(arequest_info.content_type, HTTP_CONTENT_TYPE_FORM_URLENCODED)){
76: post_content_type=FORM_URLENCODED;
77: fpost_charset=detect_charset(arequest_info.content_type);
78: } else if(StrStartFromNC(arequest_info.content_type, HTTP_CONTENT_TYPE_MULTIPART_FORMDATA)) {
79: post_content_type=MULTIPART_FORMDATA;
80: }
1.3 paf 81: }
82:
1.92 misha 83: char *VForm::strpart(const char* str, size_t len) {
1.96 misha 84: char *result=new(PointerFreeGC) char[len+1];
85: memcpy(result, str, len);
86: result[len]=0;
87: return result;
1.92 misha 88: }
89:
1.75 paf 90: char *VForm::getAttributeValue(const char* data, char *attr, size_t len) {
1.96 misha 91: const char* value=searchAttribute(data, attr, len);
92: if (value){
1.18 paf 93: size_t i;
94: if (!(len-=value-data)) return NULL;
95: if (*value=='"') {
96: for (i=1;i<len;i++) if (value[i]=='"') break;
97: return strpart(&value[1], i-1);
98: } else {
99: for (i=0;i<len;i++) if (strchr(" ;\"\n\r", value[i])) break;
100: return strpart(value, i);
101: }
1.96 misha 102: }
103: return NULL;
1.3 paf 104: }
105:
1.96 misha 106: String::C VForm::transcode(const char* client, size_t client_size, Charset* client_charset) {
1.75 paf 107: return Charset::transcode(
1.96 misha 108: String::C(strdup(client, client_size), client_size),
109: client_charset?*client_charset:fcharsets.client(),
110: fcharsets.source());
1.49 paf 111: }
112:
1.75 paf 113: void VForm::ParseGetFormInput(const char* query_string, size_t length) {
1.42 parser 114: ParseFormInput(query_string, length);
1.3 paf 115: }
116:
1.69 paf 117:
1.75 paf 118: static int atoi(const char* data, size_t alength) {
1.69 paf 119: char buf[MAX_STRING];
1.82 paf 120: size_t length=min(alength, sizeof(buf)-1);
121: memcpy(buf, data, length); buf[length]=0;
1.69 paf 122: return atoi(buf);
123: }
1.96 misha 124:
125: void VForm::ParseFormInput(const char* data, size_t length, Charset* client_charset) {
1.68 paf 126: // cut out ?image_map_tail
127: {
1.67 paf 128: for(size_t pos=0; pos<length; pos++) {
129: if(data[pos]=='?') {
1.69 paf 130: size_t start=pos+1;
131: size_t aftercomma=start;
132: size_t lookingcomma=start;
1.96 misha 133: for(; lookingcomma<length; lookingcomma++)
1.69 paf 134: if(data[lookingcomma]==',') {
135: aftercomma=++lookingcomma;
136: break;
137: }
138:
139: if(aftercomma>start) { // ?x,y
140: int x=atoi(data+start, aftercomma-1-start);
141: int y=atoi(data+aftercomma, length-aftercomma);
1.75 paf 142: imap.put(String("x"), new VInt(x));
143: imap.put(String("y"), new VInt(y));
1.69 paf 144: } else { // ?qtail
1.96 misha 145: AppendFormEntry("qtail", data+start, length-start, client_charset);
1.69 paf 146: }
1.67 paf 147: // cut tail
148: length=pos;
149: break;
150: }
151: }
152: }
1.68 paf 153: // Scan for pairs, unescaping and storing them as they are found
154: for(size_t pos=0; pos<length; ) {
1.16 paf 155: size_t start=pos;
1.68 paf 156: size_t finish=length;
1.96 misha 157: for(; pos<length; pos++)
1.68 paf 158: if(data[pos]=='&') {
159: finish=pos++;
1.3 paf 160: break;
161: }
1.68 paf 162:
163: size_t aftereq=start;
164: size_t lookingeq=start;
1.96 misha 165: for(; lookingeq<finish; lookingeq++)
1.68 paf 166: if(data[lookingeq]=='=') {
167: aftereq=++lookingeq;
1.3 paf 168: break;
169: }
1.45 paf 170:
1.96 misha 171: const char* attr=(aftereq>start)?unescape_chars(data+start, aftereq-1-start, &fcharsets.client()):"nameless";
1.91 misha 172: char *value=unescape_chars(data+aftereq, finish-aftereq, &fcharsets.client());
1.96 misha 173: AppendFormEntry(attr, value, strlen(value), client_charset);
1.3 paf 174: }
175: }
176:
1.92 misha 177: static char* pa_tolower(char *str){
178: if(!str)
179: return 0;
180: for(char *p=str; *p; p++)
181: *p=(char)tolower((unsigned char)*p);
182: return str;
183: }
184:
1.49 paf 185: void VForm::ParseMimeInput(
1.96 misha 186: char *content_type,
187: const char* data, size_t length, Charset* client_charset) {
1.3 paf 188: /* Scan for mime-presented pairs, storing them as they are found. */
1.92 misha 189: const char* boundary=pa_tolower(getAttributeValue(content_type, "boundary=", strlen(content_type)));
190: if(!boundary)
1.54 paf 191: throw Exception(0,
1.18 paf 192: 0,
1.3 paf 193: "VForm::ParseMimeInput no boundary attribute of Content-Type");
194:
1.92 misha 195: const char* lastData=&data[length];
196:
1.3 paf 197: while(true) {
1.66 paf 198: const char
1.18 paf 199: *dataStart=searchAttribute(data, boundary, lastData-data),
200: *dataEnd=searchAttribute(dataStart, boundary, lastData-dataStart);
1.92 misha 201:
1.18 paf 202: size_t headerSize=getHeader(dataStart, lastData-dataStart);
1.3 paf 203:
204: if(!dataStart|!dataEnd|!headerSize) break;
1.47 paf 205: if(searchAttribute(dataStart, "content-disposition: form-data", headerSize)) {
1.16 paf 206: size_t valueSize=(dataEnd-dataStart)-headerSize-5-strlen(boundary);
1.18 paf 207: char *attr=getAttributeValue(dataStart, " name=", headerSize),
208: *fName=getAttributeValue(dataStart, " filename=", headerSize);
1.3 paf 209:
1.78 paf 210: if(attr) {
1.3 paf 211: /* OK, we have a new pair, add it to the list. */
1.80 paf 212: // fName checks are because MSIE passes unassigned <input type=file> as filename="" and empty body
1.85 misha 213: if( fName && (strlen(fName) || valueSize) ){
214: AppendFormFileEntry(attr,
1.80 paf 215: valueSize? &dataStart[headerSize+1]: "",
216: valueSize,
1.96 misha 217: fName,
218: client_charset);
1.85 misha 219: } else {
220: AppendFormEntry(attr,
221: valueSize? &dataStart[headerSize+1]: "",
1.96 misha 222: valueSize,
223: client_charset);
1.85 misha 224: }
1.3 paf 225: }
226: }
227: data=(dataEnd-strlen(boundary));
228: }
229: }
230:
1.85 misha 231: void VForm::AppendFormFileEntry(const char* cname_cstr,
1.96 misha 232: const char* raw_cvalue_ptr, const size_t raw_cvalue_size,
233: const char* file_name_cstr, Charset* client_charset){
1.85 misha 234:
1.87 misha 235: const char* fname = strdup(file_name_cstr);
1.96 misha 236: const String& sfile_name=*new String(transcode(fname, strlen(fname), client_charset));
1.87 misha 237:
1.96 misha 238: const String& sname=*new String(transcode(cname_cstr, strlen(cname_cstr), client_charset));
1.87 misha 239: // maybe transcode text/* files?
1.89 misha 240: // NO!!! some users want to upload file 'as is' or file encoding can be unknown
1.87 misha 241:
1.85 misha 242: VFile* vfile=new VFile;
243: vfile->set(true/*tainted*/, raw_cvalue_ptr, raw_cvalue_size, sfile_name.cstr());
244:
245: fields.put_dont_replace(sname, vfile);
246:
247: // files
248: Value* vhash=files.get(sname);
249: if(!vhash){
250: // first appearence
251: vhash=new VHash;
252: files.put(sname, vhash);
253: }
254: HashStringValue& hash=*vhash->get_hash();
1.75 paf 255:
1.85 misha 256: hash.put(String::Body::Format(hash.count()), vfile);
257: }
258:
1.96 misha 259: void VForm::AppendFormEntry(const char* cname_cstr, const char* raw_cvalue_ptr, const size_t raw_cvalue_size, Charset* client_charset) {
260: const String& sname=*new String(transcode(cname_cstr, strlen(cname_cstr), client_charset));
1.75 paf 261:
1.85 misha 262: const char* premature_zero_pos=(const char* )memchr(raw_cvalue_ptr, 0, raw_cvalue_size);
263: size_t cvalue_size=premature_zero_pos?premature_zero_pos-(const char* )raw_cvalue_ptr
264: :raw_cvalue_size;
265: char *cvalue_ptr=strdup(raw_cvalue_ptr, cvalue_size);
266: fix_line_breaks(cvalue_ptr, cvalue_size);
1.96 misha 267: String& string=*new String(transcode(cvalue_ptr, cvalue_size, client_charset), String::L_TAINTED);
1.85 misha 268:
269: // tables
270: {
271: Value* vtable=tables.get(sname);
272: if(!vtable) {
273: // first appearence
274: Table::columns_type columns(new ArrayString(1));
275: *columns+=new String("field");
1.75 paf 276:
1.85 misha 277: vtable=new VTable(new Table(columns));
278: tables.put(sname, vtable);
1.36 parser 279: }
1.85 misha 280: Table& table=*vtable->get_table();
281:
282: // this string becomes next row
283: Table::element_type row(new ArrayString(1));
284: *row+=&string;
285: table+=row;
1.3 paf 286: }
1.69 paf 287:
1.85 misha 288: fields.put_dont_replace(sname, new VString(string));
1.69 paf 289: }
290:
1.85 misha 291:
292: void VForm::refill_fields_tables_and_files() {
1.77 paf 293: fields.clear();
294: tables.clear();
1.85 misha 295: files.clear();
1.77 paf 296: imap.clear();
1.75 paf 297:
1.77 paf 298: //frequest_info.query_string="a=123";
1.3 paf 299: // parsing QS [GET and ?name=value from uri rewrite)]
1.77 paf 300: if(frequest_info.query_string) {
301: size_t length=strlen(frequest_info.query_string);
302: char *buf=strdup(frequest_info.query_string, length);
1.42 parser 303: ParseGetFormInput(buf, length);
304: }
1.62 paf 305:
306: #ifdef DEBUG_POST
1.77 paf 307: frequest_info.method="POST";
1.96 misha 308: File_read_result file=file_read(fcharsets, *new String("test.stdin"), true/*as_text*/, 0, true, 0, 0, 0, false/*transcode*/);
309: frequest_info.post_size=file.length;
310: frequest_info.post_data=(char*)file.str;
1.77 paf 311: frequest_info.content_type="multipart/form-data; boundary=----------mcqY2UDNcdEAoN1mLmne2i";
1.62 paf 312:
313: #endif
1.57 paf 314:
315: // parsing POST data
1.96 misha 316: if(is_post && frequest_info.content_type)
317: switch(post_content_type){
318: case FORM_URLENCODED:
319: {
320: ParseFormInput(frequest_info.post_data, frequest_info.post_size, fpost_charset);
321: break;
322: }
323: case MULTIPART_FORMDATA:
324: {
325: ParseMimeInput(strdup(frequest_info.content_type), frequest_info.post_data, frequest_info.post_size);
326: break;
327: }
328: }
1.56 paf 329:
1.77 paf 330: filled_source=&fcharsets.source();
331: filled_client=&fcharsets.client();
332: }
333:
1.85 misha 334: bool VForm::should_refill_fields_tables_and_files() {
1.96 misha 335: return &fcharsets.source()!=filled_source || &fcharsets.client()!=filled_client;
1.1 paf 336: }
1.33 parser 337:
1.98 ! misha 338: Value* VForm::get_element(const String& aname) {
1.85 misha 339: if(should_refill_fields_tables_and_files())
340: refill_fields_tables_and_files();
1.56 paf 341:
1.36 parser 342: // $fields
343: if(aname==FORM_FIELDS_ELEMENT_NAME)
1.75 paf 344: return new VHash(fields);
1.36 parser 345:
346: // $tables
347: if(aname==FORM_TABLES_ELEMENT_NAME)
1.75 paf 348: return new VHash(tables);
1.69 paf 349:
1.85 misha 350: // $files
351: if(aname==FORM_FILES_ELEMENT_NAME)
352: return new VHash(files);
353:
1.77 paf 354: // $imap
1.69 paf 355: if(aname==FORM_IMAP_ELEMENT_NAME)
1.75 paf 356: return new VHash(imap);
1.36 parser 357:
1.96 misha 358: // CLASS, CLASS_NAME
1.98 ! misha 359: if(Value* result=VStateless_class::get_element(aname))
1.33 parser 360: return result;
361:
1.95 misha 362: // $field
1.75 paf 363: return fields.get(aname);
1.72 paf 364: }
1.93 misha 365:
366: Charset* VForm::get_post_charset(){
1.96 misha 367: return fpost_charset;
1.93 misha 368: }
E-mail: