| blob | f59b2d64167993fccd8ef96c86e395dc35822642 |
1 \r
2 struct FileData\r
3 length dq ?\r
4 cache dd ? ; pointer to FileCache\r
5 ends\r
6 \r
7 struct FileCache\r
8 offset dq ?\r
9 length dd ?\r
10 next dd ? ; pointer to another FileCache\r
11 ends\r
12 \r
13 read_source:\r
14 ; in:\r
15 ; esi - source path\r
16 ; out:\r
17 ; eax - tokenized source, null when file not found\r
18 ; esi - source path in persistent storage\r
19 mov ebx,[file_source_cache]\r
20 xor ecx,ecx\r
21 call get_from_map\r
22 jc read_source_file\r
23 cmp eax,-1\r
24 je get_erroneous_source\r
25 retn\r
26 read_source_file:\r
27 mov edx,esi\r
28 call open\r
29 jc source_file_not_found\r
30 xor eax,eax\r
31 mov edx,eax\r
32 mov cl,2\r
33 call lseek\r
34 jc error_reading_file\r
35 test edx,edx\r
36 jnz out_of_memory\r
37 push eax\r
38 xor eax,eax\r
39 mov edx,eax\r
40 mov cl,al\r
41 call lseek\r
42 jc error_reading_file\r
43 pop ecx\r
44 inc ecx\r
45 mov [source_text_length],ecx\r
46 call malloc\r
47 mov [source_text],eax\r
48 mov edx,eax\r
49 mov ecx,[source_text_length]\r
50 dec ecx\r
51 mov byte [edx+ecx],0\r
52 call read\r
53 jc error_reading_file\r
54 call close\r
55 push esi\r
56 call tokenize_source\r
57 mov eax,[source_text]\r
58 call mfree\r
59 pop esi\r
60 mov eax,[tokenization_buffer]\r
61 xor ecx,ecx\r
62 mov ebx,[file_source_cache]\r
63 call put_into_map\r
64 mov eax,[tokenization_buffer]\r
65 retn\r
66 source_file_not_found:\r
67 xor eax,eax\r
68 xor ecx,ecx\r
69 mov ebx,[file_source_cache]\r
70 call put_into_map\r
71 xor eax,eax\r
72 retn\r
73 error_reading_file:\r
74 or eax,-1\r
75 xor ecx,ecx\r
76 mov ebx,[file_source_cache]\r
77 call put_into_map\r
78 get_erroneous_source:\r
79 mov ebx,esi\r
80 mov edx,_error_reading_file\r
81 call register_error\r
82 mov eax,zero_value\r
83 retn\r
84 \r
85 use_source:\r
86 ; in:\r
87 ; esi - ASCIIZ source string\r
88 ; out:\r
89 ; eax - tokenized source\r
90 ; esi - source text in persistent storage\r
91 mov edi,esi\r
92 xor al,al\r
93 or ecx,-1\r
94 repne scasb\r
95 not ecx\r
96 mov [source_text_length],ecx\r
97 mov ebx,[memory_source_cache]\r
98 xor eax,eax\r
99 call get_from_map\r
100 jc adapt_memory_source\r
101 retn\r
102 adapt_memory_source:\r
103 mov [source_text],esi\r
104 call tokenize_source\r
105 mov eax,[tokenization_buffer]\r
106 mov esi,[source_text]\r
107 mov ecx,[source_text_length]\r
108 mov ebx,[memory_source_cache]\r
109 call put_into_map\r
110 mov eax,[tokenization_buffer]\r
111 retn\r
112 \r
113 tokenize_source:\r
114 ; in:\r
115 ; [source_text] - ASCIIZ text\r
116 ; [source_text_length] = length of text (including terminating character)\r
117 ; out:\r
118 ; [tokenization_buffer] - tokenized source\r
119 ; [tokenization_buffer_length] = length of tokenized source\r
120 mov ecx,[source_text_length]\r
121 shl ecx,1\r
122 add ecx,18\r
123 call malloc_growable\r
124 mov [tokenization_buffer],eax\r
125 mov [tokenization_buffer_length],ecx\r
126 add eax,ecx\r
127 sub eax,[source_text]\r
128 sub eax,[source_text_length]\r
129 mov [buffer_end_offset],eax\r
130 mov esi,[source_text]\r
131 mov edi,[tokenization_buffer]\r
132 mov [last_token],0Ah\r
133 tokenize:\r
134 mov eax,[buffer_end_offset]\r
135 add eax,esi\r
136 sub eax,edi\r
137 cmp eax,18\r
138 jae tokenization_buffer_reserve_ok\r
139 mov ecx,esi\r
140 sub ecx,[source_text]\r
141 mov eax,[source_text_length]\r
142 mul [tokenization_buffer_length]\r
143 div ecx\r
144 mov ecx,eax\r
145 add ecx,18\r
146 mov eax,[tokenization_buffer]\r
147 call realloc\r
148 sub edi,[tokenization_buffer]\r
149 add edi,eax\r
150 mov [tokenization_buffer],eax\r
151 mov [tokenization_buffer_length],ecx\r
152 add eax,ecx\r
153 sub eax,[source_text]\r
154 sub eax,[source_text_length]\r
155 mov [buffer_end_offset],eax\r
156 tokenization_buffer_reserve_ok:\r
157 movzx eax,byte [esi]\r
158 inc esi\r
159 mov ah,[characters+eax]\r
160 cmp ah,20h\r
161 je control_character\r
162 test ah,ah\r
163 jnz make_name_token\r
164 character_token:\r
165 stosb\r
166 mov [last_token],al\r
167 jmp tokenize\r
168 make_string_token:\r
169 mov dl,al\r
170 mov byte [edi],22h\r
171 mov [last_token],22h\r
172 add edi,5\r
173 xor ecx,ecx\r
174 copy_string:\r
175 mov al,[esi]\r
176 cmp al,0Dh\r
177 je broken_string\r
178 cmp al,0Ah\r
179 je broken_string\r
180 cmp al,1Ah\r
181 je broken_string\r
182 test al,al\r
183 jz broken_string\r
184 inc esi\r
185 cmp al,dl\r
186 jne copy_string_character\r
187 cmp byte [esi],al\r
188 jne finish_string_token\r
189 inc esi\r
190 copy_string_character:\r
191 mov [edi+ecx],al\r
192 inc ecx\r
193 jmp copy_string\r
194 broken_string:\r
195 mov byte [edi-5],27h\r
196 finish_string_token:\r
197 mov al,[edi-5]\r
198 mov [edi-4],ecx\r
199 add edi,ecx\r
200 jmp tokenize\r
201 make_name_token:\r
202 cmp al,22h\r
203 je make_string_token\r
204 cmp al,27h\r
205 je make_string_token\r
206 mov byte [edi],1Ah\r
207 mov [last_token],1Ah\r
208 add edi,5\r
209 xor ebx,ebx\r
210 mov ecx,FNV_OFFSET\r
211 mov edx,ecx\r
212 hash_name:\r
213 mov [edi+ebx],al\r
214 inc ebx\r
215 xor cl,al\r
216 xor dl,ah\r
217 imul ecx,FNV_PRIME\r
218 imul edx,FNV_PRIME\r
219 movzx eax,byte [esi]\r
220 inc esi\r
221 mov ah,[characters+eax]\r
222 cmp ah,20h\r
223 je finish_name_token\r
224 test ah,ah\r
225 jnz hash_name\r
226 finish_name_token:\r
227 mov [edi-4],ebx\r
228 add edi,ebx\r
229 mov [edi],ecx\r
230 mov [edi+4],edx\r
231 xor ecx,ecx\r
232 mov [edi+8],ecx\r
233 add edi,12\r
234 cmp ah,20h\r
235 jne character_token\r
236 control_character:\r
237 cmp al,20h\r
238 je whitespace\r
239 cmp al,9\r
240 je whitespace\r
241 cmp [last_token],20h\r
242 je mark_end_of_line\r
243 inc edi\r
244 mark_end_of_line:\r
245 mov byte [edi-1],0Ah\r
246 mov [last_token],0Ah\r
247 cmp al,0Dh\r
248 je cr\r
249 cmp al,0Ah\r
250 je lf\r
251 xor al,al\r
252 stosb\r
253 mov ecx,edi\r
254 mov eax,[tokenization_buffer]\r
255 sub ecx,eax\r
256 call realloc\r
257 mov [tokenization_buffer],eax\r
258 mov [tokenization_buffer_length],ecx\r
259 retn\r
260 cr:\r
261 cmp byte [esi],0Ah\r
262 jne tokenize\r
263 inc esi\r
264 jmp tokenize\r
265 lf:\r
266 cmp byte [esi],0Dh\r
267 jne tokenize\r
268 inc esi\r
269 jmp tokenize\r
270 whitespace:\r
271 cmp [last_token],0Ah\r
272 je tokenize\r
273 cmp [last_token],20h\r
274 je tokenize\r
275 mov al,20h\r
276 stosb\r
277 mov [last_token],al\r
278 jmp tokenize\r
279 \r
280 get_file_data:\r
281 ; in:\r
282 ; esi - file path\r
283 ; out:\r
284 ; ebx - FileData, null when file not found\r
285 ; esi - file path in persistent storage\r
286 ; preserves: edi\r
287 mov ebx,[file_data_cache]\r
288 xor ecx,ecx\r
289 call get_from_map\r
290 jc initialize_file_data\r
291 mov ebx,eax\r
292 retn\r
293 initialize_file_data:\r
294 mov edx,esi\r
295 call open\r
296 jc remember_file_not_found\r
297 push edi\r
298 mov ecx,sizeof.FileData\r
299 call malloc_fixed\r
300 mov edi,eax\r
301 xor eax,eax\r
302 mov edx,eax\r
303 mov cl,2\r
304 call lseek\r
305 jc file_not_seekable\r
306 mov dword [edi+FileData.length],eax\r
307 mov dword [edi+FileData.length+4],edx\r
308 call close\r
309 mov eax,edi\r
310 xor ecx,ecx\r
311 mov [eax+FileData.cache],ecx\r
312 mov ebx,[file_data_cache]\r
313 call put_into_map\r
314 mov ebx,edi\r
315 pop edi\r
316 retn\r
317 file_not_seekable:\r
318 pop edi\r
319 remember_file_not_found:\r
320 xor eax,eax\r
321 mov ecx,eax\r
322 mov ebx,[file_data_cache]\r
323 call put_into_map\r
324 xor ebx,ebx\r
325 retn\r
326 \r
327 read_file_data:\r
328 ; in:\r
329 ; esi - file path\r
330 ; ebx - FileData\r
331 ; edi - buffer for data\r
332 ; [file_offset] = offset of data\r
333 ; [data_length] = length of data\r
334 ; out:\r
335 ; cf set when read failed\r
336 ; preserves: esi\r
337 mov [file_data],ebx\r
338 lea eax,[ebx+FileData.cache]\r
339 mov [file_cache_pointer],eax\r
340 mov ebx,[eax]\r
341 read_from_file_cache:\r
342 mov ecx,[data_length]\r
343 test ecx,ecx\r
344 jz file_data_read\r
345 test ebx,ebx\r
346 jz new_trailing_file_cache_entry\r
347 mov eax,dword [file_offset]\r
348 mov edx,dword [file_offset+4]\r
349 sub eax,dword [ebx+FileCache.offset]\r
350 sbb edx,dword [ebx+FileCache.offset+4]\r
351 jc new_file_cache_entry\r
352 jnz next_entry\r
353 mov edx,[ebx+FileCache.length]\r
354 sub edx,eax\r
355 jbe next_entry\r
356 cmp ecx,edx\r
357 jbe length_to_read_ok\r
358 mov ecx,edx\r
359 length_to_read_ok:\r
360 sub [data_length],ecx\r
361 add dword [file_offset],ecx\r
362 adc dword [file_offset+4],0\r
363 mov edx,esi\r
364 lea esi,[ebx+sizeof.FileCache+eax]\r
365 mov al,cl\r
366 shr ecx,2\r
367 rep movsd\r
368 mov cl,al\r
369 and cl,11b\r
370 rep movsb\r
371 mov esi,edx\r
372 next_entry:\r
373 lea eax,[ebx+FileCache.next]\r
374 mov [file_cache_pointer],eax\r
375 mov ebx,[eax]\r
376 jmp read_from_file_cache\r
377 file_data_read:\r
378 clc\r
379 retn\r
380 new_trailing_file_cache_entry:\r
381 mov ebx,[file_data]\r
382 mov ecx,dword [ebx+FileData.length]\r
383 mov edx,dword [ebx+FileData.length+4]\r
384 cmp ecx,dword [file_offset]\r
385 jne measure_cache_gap\r
386 cmp edx,dword [file_offset+4]\r
387 jne measure_cache_gap\r
388 stc\r
389 retn\r
390 new_file_cache_entry:\r
391 mov ecx,dword [ebx+FileCache.offset]\r
392 mov edx,dword [ebx+FileCache.offset+4]\r
393 measure_cache_gap:\r
394 mov eax,dword [file_offset]\r
395 and eax,not 0FFFh\r
396 sub ecx,eax\r
397 sbb edx,dword [file_offset+4]\r
398 jnz compute_aligned_length\r
399 cmp ecx,[data_length]\r
400 jbe read_into_cache\r
401 compute_aligned_length:\r
402 mov eax,dword [file_offset]\r
403 and eax,0FFFh\r
404 add eax,[data_length]\r
405 dec eax\r
406 shr eax,12\r
407 inc eax\r
408 shl eax,12\r
409 test edx,edx\r
410 jnz use_aligned_length\r
411 cmp eax,ecx\r
412 jae read_into_cache\r
413 use_aligned_length:\r
414 mov ecx,eax\r
415 read_into_cache:\r
416 push ecx\r
417 add ecx,sizeof.FileCache\r
418 call malloc_fixed\r
419 mov ebx,eax\r
420 mov eax,[file_cache_pointer]\r
421 mov edx,ebx\r
422 xchg edx,[eax]\r
423 mov [ebx+FileCache.next],edx\r
424 pop [ebx+FileCache.length]\r
425 mov eax,dword [file_offset]\r
426 and eax,not 0FFFh\r
427 mov edx,dword [file_offset+4]\r
428 mov dword [ebx+FileCache.offset],eax\r
429 mov dword [ebx+FileCache.offset+4],edx\r
430 push ebx edi\r
431 mov edi,ebx\r
432 mov edx,esi\r
433 call open\r
434 jc file_access_error\r
435 mov eax,dword [edi+FileCache.offset]\r
436 mov edx,dword [edi+FileCache.offset+4]\r
437 xor cl,cl\r
438 call lseek\r
439 jc file_access_error\r
440 lea edx,[edi+sizeof.FileCache]\r
441 mov ecx,[edi+FileCache.length]\r
442 call read\r
443 jc file_access_error\r
444 call close\r
445 pop edi ebx\r
446 jmp read_from_file_cache\r
447 file_access_error:\r
448 pop edi ebx\r
449 stc\r
450 retn\r