Annotation of libwww/Library/src/HTGuess.c, revision 2.17
2.10 frystyk 1: /* HTGuess.c
2: ** STREAM TO GUESS CONTENT-TYPE
3: **
2.14 frystyk 4: ** (c) COPYRIGHT MIT 1995.
2.10 frystyk 5: ** Please first read the full copyright statement in the file COPYRIGH.
2.1 luotonen 6: **
7: ** This version of the stream object just writes its input
8: ** to its output, but prepends Content-Type: field and an
9: ** empty line after it.
10: **
2.8 duns 11: ** HISTORY:
12: ** 8 Jul 94 FM Insulate free() from _free structure element.
13: **
2.1 luotonen 14: */
15:
2.12 frystyk 16: /* Library include files */
17: #include "tcp.h"
18: #include "HTUtils.h"
19: #include "HTString.h"
2.1 luotonen 20: #include "HTFormat.h"
21: #include "HTAlert.h"
22: #include "HTList.h"
2.13 frystyk 23: #include "HTFWrite.h"
2.12 frystyk 24: #include "HTGuess.h"
25:
26: #define SAMPLE_SIZE 200 /* Number of chars to look at */
2.1 luotonen 27:
28: /* Stream Object
29: ** ------------
30: */
31:
32: struct _HTStream {
33: CONST HTStreamClass * isa;
34:
35: HTRequest * req;
2.13 frystyk 36: HTParentAnchor * anchor;
2.9 frystyk 37: HTFormat output_format;
2.1 luotonen 38: HTStream * output_stream;
2.9 frystyk 39: HTStream * target;
2.1 luotonen 40:
2.13 frystyk 41: BOOL transparent;
2.1 luotonen 42: int cnt;
43: int text_cnt;
44: int lf_cnt;
45: int cr_cnt;
46: int pg_cnt;
47: int ctrl_cnt;
48: int high_cnt;
49: char * write_ptr;
50: char buffer[ SAMPLE_SIZE + 1 ];
51: };
52:
2.13 frystyk 53: #define PUT_CHAR(c) (*me->target->isa->put_character)(me->target,c)
54: #define PUT_STRING(s) (*me->target->isa->put_string)(me->target,s)
55: #define PUT_BLOCK(b,l) (*me->target->isa->put_block)(me->target,b,l)
56:
57: #define CONTENT_TYPE(t) HTAnchor_setFormat(me->anchor, HTAtom_for(t))
58: #define CONTENT_ENCODING(t) HTAnchor_setEncoding(me->anchor, HTAtom_for(t))
59:
60: /* ------------------------------------------------------------------------- */
2.1 luotonen 61:
62: PRIVATE BOOL is_html ARGS1(char *, buf)
63: {
64: char * p = strchr(buf,'<');
65:
66: if (p && (!strncasecomp(p, "<HTML>", 6) ||
67: !strncasecomp(p, "<HEAD", 5) ||
68: !strncasecomp(p, "<TITLE>", 7) ||
69: !strncasecomp(p, "<BODY>", 6) ||
70: !strncasecomp(p, "<PLAINTEXT>", 11) ||
71: (p[0]=='<' && TOUPPER(p[1]) == 'H' && p[3]=='>')))
72: return YES;
73: else
74: return NO;
75: }
76:
2.13 frystyk 77: PRIVATE int HTGuess_flush ARGS1(HTStream *, me)
78: {
79: if (!me->transparent) {
2.17 ! frystyk 80: if (STREAM_TRACE)
2.13 frystyk 81: fprintf(TDEST,"GUESSING.... text=%d newlines=%d ctrl=%d high=%d\n",
82: me->text_cnt, me->lf_cnt, me->ctrl_cnt, me->high_cnt);
83: if (me->cnt) {
2.17 ! frystyk 84: if (STREAM_TRACE) fprintf(TDEST,
2.13 frystyk 85: "Percentages. text=%d%% newlines=%d%% ctrl=%d%% high=%d%%\n",
86: (int)(100*me->text_cnt/me->cnt + 0.5),
87: (int)(100*me->lf_cnt /me->cnt + 0.5),
88: (int)(100*me->ctrl_cnt/me->cnt + 0.5),
89: (int)(100*me->high_cnt/me->cnt + 0.5));
90: }
91:
92: if (!me->ctrl_cnt ||
93: me->text_cnt + me->lf_cnt >= 16 * (me->ctrl_cnt + me->high_cnt)) {
2.15 frystyk 94: char *ptr;
2.13 frystyk 95: /* some kind of text */
96:
97: *me->write_ptr = 0; /* terminate buffer */
98:
99: if (me->high_cnt > 0)
100: CONTENT_ENCODING("8bit");
101: else
102: CONTENT_ENCODING("7bit");
103:
104: if (is_html(me->buffer))
105: CONTENT_TYPE("text/html");
106:
107: else if (!strncmp(me->buffer, "%!", 2))
108: CONTENT_TYPE("application/postscript");
109:
110: else if (strstr(me->buffer, "#define") &&
111: strstr(me->buffer, "_width") &&
112: strstr(me->buffer, "_bits"))
113: CONTENT_TYPE("image/x-xbitmap");
114:
2.15 frystyk 115: else if ((ptr = strstr(me->buffer, "converted with BinHex"))!=NULL)
116: CONTENT_ENCODING("macbinhex");
117:
2.13 frystyk 118: else if (!strncmp(me->buffer, "begin ", 6))
119: CONTENT_ENCODING("base64");
2.15 frystyk 120:
2.13 frystyk 121: else
122: CONTENT_TYPE("text/plain");
123: }
124: else {
125: if (!strncmp(me->buffer, "GIF", 3))
126: CONTENT_TYPE("image/gif");
2.1 luotonen 127:
2.13 frystyk 128: else if (!strncmp(me->buffer, "\377\330\377\340", 4))
129: CONTENT_TYPE("image/jpeg");
2.1 luotonen 130:
2.13 frystyk 131: else if (!strcmp(me->buffer, "MM")) /* MM followed by a zero */
132: CONTENT_TYPE("image/tiff");
2.1 luotonen 133:
2.16 frystyk 134: else if (!strncmp(me->buffer, "\211PNG\r\n\032\n", 8))
135: CONTENT_TYPE("image/x-png");
136:
2.13 frystyk 137: else if (!strncmp(me->buffer, ".snd", 4))
138: CONTENT_TYPE("audio/basic");
2.1 luotonen 139:
2.13 frystyk 140: else if (!strncmp(me->buffer, "\037\235", 2))
141: CONTENT_ENCODING("x-compress");
2.1 luotonen 142:
2.13 frystyk 143: else if (!strncmp(me->buffer, "\037\213", 2))
144: CONTENT_ENCODING("x-gzip");
2.1 luotonen 145:
2.13 frystyk 146: else
147: CONTENT_TYPE("application/octet-stream");
148: }
149:
150: if (!me->anchor->content_type) CONTENT_TYPE("www/unknown");
151: if (!me->anchor->content_encoding) CONTENT_ENCODING("binary");
152:
2.17 ! frystyk 153: if (STREAM_TRACE) fprintf(TDEST,"Guessed..... %s\n",
2.13 frystyk 154: HTAtom_name(me->anchor->content_type));
2.17 ! frystyk 155: if (STREAM_TRACE) fprintf(TDEST,"Encoding.... %s\n",
2.13 frystyk 156: HTAtom_name(me->anchor->content_encoding));
157: if ((me->target = HTStreamStack(me->anchor->content_type,
158: me->output_format, me->output_stream,
159: me->req, NO)) == NULL) {
2.17 ! frystyk 160: if (STREAM_TRACE)
2.13 frystyk 161: fprintf(TDEST, "HTGuess..... Can't convert media type\n");
162: me->target = HTBlackHole();
163: }
2.1 luotonen 164: }
2.13 frystyk 165: return PUT_BLOCK(me->buffer, me->cnt);
2.1 luotonen 166: }
167:
168:
2.13 frystyk 169: PRIVATE int HTGuess_put_block ARGS3(HTStream *, me, CONST char*, b, int, l)
2.1 luotonen 170: {
2.13 frystyk 171: while (!me->transparent && l-- > 0) {
172: int status;
173: if (me->target) {
174: if ((status = HTGuess_flush(me)) != HT_OK)
175: return status;
176: } else {
177: me->cnt++;
178: if (*b == LF)
179: me->lf_cnt++;
180: else if (*b == CR)
181: me->cr_cnt++;
182: else if (*b == 12)
183: me->pg_cnt++;
184: else if (*b =='\t')
185: me->text_cnt++;
186: else if ((unsigned char)*b < 32)
187: me->ctrl_cnt++;
188: else if ((unsigned char)*b < 128)
189: me->text_cnt++;
190: else
191: me->high_cnt++;
192: *me->write_ptr++ = *b++;
193: if (me->cnt >= SAMPLE_SIZE) {
194: if ((status = HTGuess_flush(me)) != HT_OK)
195: return status;
196: else
197: me->transparent = YES;
198: }
199: }
2.1 luotonen 200: }
2.13 frystyk 201: if (l > 0)
202: return PUT_BLOCK(b, l);
203: return HT_OK;
2.1 luotonen 204: }
205:
2.13 frystyk 206: PRIVATE int HTGuess_put_character ARGS2(HTStream *, me, char, c)
2.1 luotonen 207: {
2.13 frystyk 208: return HTGuess_put_block(me, &c, 1);
2.1 luotonen 209: }
210:
2.13 frystyk 211: PRIVATE int HTGuess_put_string ARGS2(HTStream *, me, CONST char*, s)
2.1 luotonen 212: {
2.13 frystyk 213: return HTGuess_put_block(me, s, (int) strlen(s));
2.1 luotonen 214: }
215:
2.9 frystyk 216: PRIVATE int HTGuess_free ARGS1(HTStream *, me)
2.1 luotonen 217: {
2.13 frystyk 218: int status;
219: if (!me->transparent && (status = HTGuess_flush(me)) != HT_OK)
220: return status;
221: else
222: me->transparent = YES;
223: if ((status = (*me->target->isa->_free)(me->target)) != HT_OK)
224: return status;
2.1 luotonen 225: free(me);
2.13 frystyk 226: return HT_OK;
2.1 luotonen 227: }
228:
2.9 frystyk 229: PRIVATE int HTGuess_abort ARGS2(HTStream *, me, HTError, e)
2.1 luotonen 230: {
2.9 frystyk 231: if (me->target)
232: (*me->target->isa->abort)(me,e);
2.1 luotonen 233: free(me);
2.13 frystyk 234: return HT_ERROR;
2.1 luotonen 235: }
236:
237:
238: /* Guessing stream
239: ** ---------------
240: */
241: PRIVATE CONST HTStreamClass HTGuessClass =
242: {
2.13 frystyk 243: "GuessWhat",
244: HTGuess_flush,
2.1 luotonen 245: HTGuess_free,
246: HTGuess_abort,
247: HTGuess_put_character,
248: HTGuess_put_string,
249: HTGuess_put_block
250: };
251:
252:
2.9 frystyk 253: PUBLIC HTStream * HTGuess_new ARGS5(HTRequest *, req,
254: void *, param,
255: HTFormat, input_format,
256: HTFormat, output_format,
257: HTStream *, output_stream)
2.1 luotonen 258: {
2.13 frystyk 259: HTStream * me = (HTStream *) calloc(1,sizeof(HTStream));
2.1 luotonen 260: if (!me) outofmem(__FILE__, "HTGuess_new");
261:
262: me->isa = &HTGuessClass;
2.9 frystyk 263: me->req = req;
2.17 ! frystyk 264: me->anchor = HTRequest_anchor(req);
2.9 frystyk 265: me->output_format = output_format;
266: me->output_stream = output_stream;
2.1 luotonen 267: me->write_ptr = me->buffer;
268: return me;
269: }
Webmaster