21 |
|
|
22 |
#include "highlight.h" |
#include "highlight.h" |
23 |
|
|
24 |
|
#include <assert.h> |
25 |
#include <stdlib.h> |
#include <stdlib.h> |
26 |
#include <string.h> |
#include <string.h> |
27 |
|
|
81 |
} |
} |
82 |
} |
} |
83 |
|
|
84 |
static enum Class ** classes = NULL; |
static const char * g_id; |
85 |
static jschar ** lines = NULL; |
static const jschar * g_characters; |
86 |
|
static size_t g_num_characters; |
87 |
static uint16_t num_characters_in_line(jschar * line) { |
static Stream * g_output; |
88 |
uint16_t result = 0; |
static size_t character_offset; |
89 |
while (line[result] != '\0') { |
static uint16_t line_num; |
90 |
result++; |
static uint16_t column_num; |
91 |
|
static enum Class current_class; |
92 |
|
|
93 |
|
static void output_character(jschar c, enum Class class) { |
94 |
|
if (class != current_class) { |
95 |
|
/* output the end tag */ |
96 |
|
if (current_class != CLASS_NONE) { |
97 |
|
Stream_write_string(g_output, "</span>"); |
98 |
|
} |
99 |
|
|
100 |
|
current_class = class; |
101 |
|
|
102 |
|
/* output the start tag */ |
103 |
|
if (current_class != CLASS_NONE) { |
104 |
|
Stream_printf(g_output, "<span class=\"%s\">", get_class_name(class)); |
105 |
|
} |
106 |
} |
} |
|
return result; |
|
|
} |
|
107 |
|
|
108 |
static void mark_token_chars(enum Class class, uint16_t start_line, uint16_t start_column, uint16_t end_line, uint16_t end_column) { |
switch (c) { |
109 |
for (uint16_t i = start_line; i <= end_line; i++) { |
case '&': |
110 |
uint16_t c1 = i == start_line? start_column: 0; |
Stream_write_string(g_output, "&"); |
111 |
uint16_t c2 = i == end_line? end_column: num_characters_in_line(lines[i - 1]); |
break; |
112 |
for (uint16_t j = c1; j < c2; j++) { |
case '<': |
113 |
classes[i - 1][j] = class; |
Stream_write_string(g_output, "<"); |
114 |
|
break; |
115 |
|
case '>': |
116 |
|
Stream_write_string(g_output, ">"); |
117 |
|
break; |
118 |
|
case '\n': |
119 |
|
Stream_write_char(g_output, c); |
120 |
|
break; |
121 |
|
default: |
122 |
|
if (c == '\t' || (32 <= c && c <= 126)) { |
123 |
|
Stream_write_char(g_output, c); |
124 |
|
} |
125 |
|
else { |
126 |
|
Stream_printf(g_output, "&#%d;", c); |
127 |
} |
} |
128 |
|
break; |
129 |
} |
} |
130 |
} |
} |
131 |
|
|
132 |
static void mark_nontoken_chars(uint16_t start_line, uint16_t start_column, uint16_t end_line, uint16_t end_column) { |
static void mark_nontoken_chars(uint16_t end_line, uint16_t end_column) { |
133 |
enum State { |
enum State { |
134 |
STATE_NORMAL, |
STATE_NORMAL, |
135 |
STATE_LINE_COMMENT, |
STATE_LINE_COMMENT, |
137 |
}; |
}; |
138 |
|
|
139 |
enum State state = STATE_NORMAL; |
enum State state = STATE_NORMAL; |
140 |
for (uint16_t i = start_line; i <= end_line; i++) { |
while (character_offset < g_num_characters) { |
141 |
uint16_t c1 = i == start_line? start_column: 0; |
if (end_line != 0 && line_num > end_line) { |
142 |
uint16_t c2 = i == end_line? end_column: num_characters_in_line(lines[i - 1]); |
break; |
|
for (uint16_t j = c1; j < c2; j++) { |
|
|
jschar c = lines[i - 1][j]; |
|
|
switch (state) { |
|
|
case STATE_NORMAL: |
|
|
if (c == '/' && j + 1 < c2 && lines[i - 1][j + 1] == '/') { |
|
|
state = STATE_LINE_COMMENT; |
|
|
classes[i - 1][j] = CLASS_COMMENT; |
|
|
} |
|
|
else if (c == '/' && j + 1 < c2 && lines[i - 1][j + 1] == '*') { |
|
|
state = STATE_MULTILINE_COMMENT; |
|
|
classes[i - 1][j] = CLASS_COMMENT; |
|
|
j++; |
|
|
classes[i - 1][j] = CLASS_COMMENT; |
|
|
} |
|
|
else { |
|
|
classes[i - 1][j] = CLASS_NONE; |
|
|
} |
|
|
break; |
|
|
case STATE_LINE_COMMENT: |
|
|
if (c == '\r' || c == '\n' || c == 0x2028 || c == 0x2029) { |
|
|
state = STATE_NORMAL; |
|
|
classes[i - 1][j] = CLASS_NONE; |
|
|
} |
|
|
else { |
|
|
classes[i - 1][j] = CLASS_COMMENT; |
|
|
} |
|
|
break; |
|
|
case STATE_MULTILINE_COMMENT: |
|
|
classes[i - 1][j] = CLASS_COMMENT; |
|
|
if (c == '*' && j + 1 < c2 && lines[i - 1][j + 1] == '/') { |
|
|
j++; |
|
|
classes[i - 1][j] = CLASS_COMMENT; |
|
|
state = STATE_NORMAL; |
|
|
} |
|
|
break; |
|
|
} |
|
143 |
} |
} |
144 |
/* end of the line */ |
else if (line_num == end_line && column_num >= end_column) { |
145 |
if (state == STATE_LINE_COMMENT) { |
break; |
|
state = STATE_NORMAL; |
|
146 |
} |
} |
|
} |
|
|
} |
|
147 |
|
|
148 |
void jscoverage_highlight_js(JSContext * context, const char * id, const jschar * characters, size_t num_characters, Stream * output) { |
jschar c = g_characters[character_offset]; |
149 |
/* count the lines - see GetChar in jsscan.c */ |
if (c == '\0') { |
150 |
size_t i = 0; |
fatal("%s: script contains NULL character", g_id); |
151 |
uint16_t num_lines = 0; |
} |
152 |
while (i < num_characters) { |
|
153 |
if (num_lines == UINT16_MAX) { |
switch (state) { |
154 |
fatal("%s: script has more than 65535 lines", id); |
case STATE_NORMAL: |
155 |
} |
if (c == '/' && character_offset + 1 < g_num_characters && g_characters[character_offset + 1] == '/') { |
156 |
num_lines++; |
state = STATE_LINE_COMMENT; |
|
jschar c; |
|
|
while (i < num_characters) { |
|
|
c = characters[i]; |
|
|
if (c == '\0') { |
|
|
fatal("%s: script contains NULL character", id); |
|
157 |
} |
} |
158 |
|
else if (c == '/' && character_offset + 1 < g_num_characters && g_characters[character_offset + 1] == '*') { |
159 |
|
state = STATE_MULTILINE_COMMENT; |
160 |
|
output_character('/', CLASS_COMMENT); |
161 |
|
output_character('*', CLASS_COMMENT); |
162 |
|
character_offset += 2; |
163 |
|
if (column_num >= UINT16_MAX - 1) { |
164 |
|
fatal("%s: script contains line with more than 65,535 characters", g_id); |
165 |
|
} |
166 |
|
column_num += 2; |
167 |
|
continue; |
168 |
|
} |
169 |
|
break; |
170 |
|
case STATE_LINE_COMMENT: |
171 |
if (c == '\r' || c == '\n' || c == 0x2028 || c == 0x2029) { |
if (c == '\r' || c == '\n' || c == 0x2028 || c == 0x2029) { |
172 |
break; |
state = STATE_NORMAL; |
173 |
} |
} |
174 |
i++; |
break; |
175 |
} |
case STATE_MULTILINE_COMMENT: |
176 |
if (i < num_characters) { |
if (c == '*' && character_offset + 1 < g_num_characters && g_characters[character_offset + 1] == '/') { |
177 |
i++; |
output_character('*', CLASS_COMMENT); |
178 |
if (c == '\r' && i < num_characters && characters[i] == '\n') { |
output_character('/', CLASS_COMMENT); |
179 |
i++; |
state = STATE_NORMAL; |
180 |
|
character_offset += 2; |
181 |
|
if (column_num >= UINT16_MAX - 1) { |
182 |
|
fatal("%s: script contains line with more than 65,535 characters", g_id); |
183 |
|
} |
184 |
|
column_num += 2; |
185 |
|
continue; |
186 |
} |
} |
187 |
|
break; |
188 |
} |
} |
|
} |
|
|
|
|
|
lines = xnew(jschar *, num_lines); |
|
|
classes = xnew(enum Class *, num_lines); |
|
189 |
|
|
190 |
uint16_t line_num = 0; |
character_offset++; |
191 |
i = 0; |
if (c == '\r' || c == '\n' || c == 0x2028 || c == 0x2029) { |
192 |
while (i < num_characters) { |
if (line_num == UINT16_MAX) { |
193 |
size_t line_start = i; |
fatal("%s: script contains more than 65,535 lines", g_id); |
194 |
jschar c; |
} |
195 |
while (i < num_characters) { |
line_num++; |
196 |
c = characters[i]; |
column_num = 0; |
197 |
if (c == '\r' || c == '\n' || c == 0x2028 || c == 0x2029) { |
if (c == '\r' && character_offset < g_num_characters && g_characters[character_offset] == '\n') { |
198 |
break; |
character_offset++; |
199 |
} |
} |
200 |
i++; |
output_character('\n', CLASS_NONE); |
201 |
} |
} |
202 |
size_t line_end = i; |
else { |
203 |
if (i < num_characters) { |
if (column_num == UINT16_MAX) { |
204 |
i++; |
fatal("%s: script contains line with more than 65,535 characters", g_id); |
205 |
if (c == '\r' && i < num_characters && characters[i] == '\n') { |
} |
206 |
i++; |
column_num++; |
207 |
|
if (state == STATE_NORMAL) { |
208 |
|
output_character(c, CLASS_NONE); |
209 |
|
} |
210 |
|
else { |
211 |
|
output_character(c, CLASS_COMMENT); |
212 |
} |
} |
213 |
} |
} |
|
size_t line_length = line_end - line_start; |
|
|
if (line_length >= UINT16_MAX) { |
|
|
fatal("%s: script has line with 65535 characters or more", id); |
|
|
} |
|
|
jschar * line = xnew(jschar, line_length + 1); |
|
|
memcpy(line, characters + line_start, sizeof(jschar) * line_length); |
|
|
line[line_length] = '\0'; |
|
|
lines[line_num] = line; |
|
|
classes[line_num] = xnew(enum Class, line_length); |
|
|
line_num++; |
|
214 |
} |
} |
215 |
|
} |
216 |
|
|
217 |
|
void jscoverage_highlight_js(JSContext * context, const char * id, const jschar * characters, size_t num_characters, Stream * output) { |
218 |
|
g_id = id; |
219 |
|
g_characters = characters; |
220 |
|
g_num_characters = num_characters; |
221 |
|
g_output = output; |
222 |
|
|
223 |
|
character_offset = 0; |
224 |
|
line_num = 1; |
225 |
|
column_num = 0; |
226 |
|
current_class = CLASS_NONE; |
227 |
|
|
228 |
/* tokenize the JavaScript */ |
/* tokenize the JavaScript */ |
229 |
JSTokenStream * token_stream = js_NewTokenStream(context, characters, num_characters, NULL, 1, NULL); |
JSTokenStream * token_stream = js_NewTokenStream(context, characters, num_characters, NULL, 1, NULL); |
265 |
*/ |
*/ |
266 |
JS_KEEP_ATOMS(cx->runtime); |
JS_KEEP_ATOMS(cx->runtime); |
267 |
|
|
|
line_num = 1; |
|
|
uint16_t column_num = 0; |
|
268 |
for (;;) { |
for (;;) { |
269 |
JSTokenType tt = js_GetToken(context, token_stream); |
JSTokenType tt = js_GetToken(context, token_stream); |
270 |
|
|
273 |
} |
} |
274 |
|
|
275 |
if (tt == TOK_EOF) { |
if (tt == TOK_EOF) { |
276 |
/* it seems t.pos is invalid for TOK_EOF??? */ |
mark_nontoken_chars(0, 0); |
|
/* mark the remaining chars */ |
|
|
if (num_lines == 0) { |
|
|
break; |
|
|
} |
|
|
uint16_t end_line = num_lines; |
|
|
uint16_t end_column = num_characters_in_line(lines[num_lines - 1]); |
|
|
mark_nontoken_chars(line_num, column_num, end_line, end_column); |
|
277 |
break; |
break; |
278 |
} |
} |
279 |
|
|
280 |
/* mark the chars before the token */ |
/* mark the chars before the token */ |
281 |
JSToken t = CURRENT_TOKEN(token_stream); |
JSToken t = CURRENT_TOKEN(token_stream); |
282 |
mark_nontoken_chars(line_num, column_num, t.pos.begin.lineno, t.pos.begin.index); |
mark_nontoken_chars(t.pos.begin.lineno, t.pos.begin.index); |
283 |
|
|
284 |
/* mark the token */ |
/* mark the token */ |
285 |
enum Class class; |
enum Class class; |
458 |
abort(); |
abort(); |
459 |
break; |
break; |
460 |
} |
} |
|
mark_token_chars(class, t.pos.begin.lineno, t.pos.begin.index, t.pos.end.lineno, t.pos.end.index); |
|
|
if (tt == TOK_STRING) { |
|
|
for (uint16_t i = t.pos.begin.index + 1; i < t.pos.end.index - 1; i++) { |
|
|
jschar c = lines[t.pos.begin.lineno - 1][i]; |
|
|
if (c == '\\') { |
|
|
mark_token_chars(CLASS_SPECIALCHAR, t.pos.begin.lineno, i, t.pos.begin.lineno, i + 2); |
|
|
i++; |
|
|
} |
|
|
} |
|
|
} |
|
461 |
|
|
462 |
line_num = t.pos.end.lineno; |
assert(t.pos.begin.lineno == t.pos.end.lineno); |
463 |
column_num = t.pos.end.index; |
if (t.pos.begin.index > t.pos.end.index) { |
464 |
} |
fatal("%s: script contains line with more than 65,535 characters", id); |
465 |
|
} |
466 |
/* output the highlighted code */ |
for (uint16_t i = t.pos.begin.index; i < t.pos.end.index; i++) { |
467 |
enum Class class = CLASS_NONE; |
assert(character_offset < num_characters); |
468 |
for (uint16_t i = 0; i < num_lines; i++) { |
jschar c = characters[character_offset]; |
469 |
uint16_t length = num_characters_in_line(lines[i]); |
if (tt == TOK_STRING && c == '\\') { |
470 |
for (uint16_t j = 0; j < length; j++) { |
output_character(c, CLASS_SPECIALCHAR); |
471 |
jschar c = lines[i][j]; |
character_offset++; |
472 |
if (classes[i][j] != class) { |
i++; |
473 |
if (class != CLASS_NONE) { |
assert(character_offset < num_characters); |
474 |
Stream_write_string(output, "</span>"); |
c = characters[character_offset]; |
475 |
} |
output_character(c, CLASS_SPECIALCHAR); |
476 |
class = classes[i][j]; |
character_offset++; |
|
if (class != CLASS_NONE) { |
|
|
Stream_printf(output, "<span class=\"%s\">", get_class_name(class)); |
|
|
} |
|
|
} |
|
|
if (c == '&') { |
|
|
Stream_write_string(output, "&"); |
|
|
} |
|
|
else if (c == '<') { |
|
|
Stream_write_string(output, "<"); |
|
|
} |
|
|
else if (c == '>') { |
|
|
Stream_write_string(output, ">"); |
|
|
} |
|
|
else if (c == '\t' || (32 <= c && c <= 126)) { |
|
|
Stream_write_char(output, c); |
|
477 |
} |
} |
478 |
else { |
else { |
479 |
Stream_printf(output, "&#%d;", c); |
output_character(c, class); |
480 |
|
character_offset++; |
481 |
} |
} |
482 |
} |
} |
483 |
if (class != CLASS_NONE) { |
|
484 |
Stream_write_string(output, "</span>"); |
line_num = t.pos.end.lineno; |
485 |
class = CLASS_NONE; |
column_num = t.pos.end.index; |
|
} |
|
|
Stream_write_char(output, '\n'); |
|
486 |
} |
} |
487 |
|
|
488 |
for (uint16_t i = 0; i < num_lines; i++) { |
if (current_class != CLASS_NONE) { |
489 |
free(lines[i]); |
output_character('\n', CLASS_NONE); |
|
free(classes[i]); |
|
490 |
} |
} |
|
free(lines); |
|
|
free(classes); |
|
491 |
|
|
492 |
/* cleanup */ |
/* cleanup */ |
493 |
JS_UNKEEP_ATOMS(cx->runtime); |
JS_UNKEEP_ATOMS(cx->runtime); |