Skip to content

Commit 24e115c

Browse files
CodesAwayCodesAway
committed
Commit before refactor parsing code into separate package
1 parent 7d5bf99 commit 24e115c

File tree

17 files changed

+455
-108
lines changed

17 files changed

+455
-108
lines changed

‎BEXCodeCompare/src/main/java/info/codesaway/bex/BEXPair.java‎

Lines changed: 13 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -3,6 +3,7 @@
33
import static info.codesaway.bex.BEXSide.LEFT;
44
import static info.codesaway.bex.BEXSide.RIGHT;
55

6+
import java.util.Objects;
67
import java.util.function.BiConsumer;
78
import java.util.function.BiFunction;
89
import java.util.function.BiPredicate;
@@ -99,7 +100,7 @@ public default IntBEXPair mapToInt(final ToIntFunction<T> function) {
99100

100101
/**
101102
* Applies the specified BiFunction, passing {@link #getLeft()} and {@link #getRight()} as arguments
102-
103+
103104
* @param function the BiFunction to apply
104105
* @return the result of applying the BiFunction
105106
*/
@@ -109,7 +110,7 @@ public default <R> R apply(final BiFunction<T, T, R> function) {
109110

110111
/**
111112
* Applies the specified ToIntBiFunction, passing {@link #getLeft()} and {@link #getRight()} as arguments
112-
113+
113114
* @param function the ToIntBiFunction to apply
114115
* @return the result of applying the ToIntBiFunction
115116
*/
@@ -119,7 +120,7 @@ public default int applyAsInt(final ToIntBiFunction<T, T> function) {
119120

120121
/**
121122
* Evaluates the specified BiPredicate, passing {@link #getLeft()} and {@link #getRight()} as arguments
122-
123+
123124
* @param predicate the BiPredicate to apply
124125
* @return <code>true</code> if the predicate matches when applying the arugments; otherwise, <code>false</code>
125126
*/
@@ -279,4 +280,13 @@ public default int compareTo(final BEXPair<T> o) {
279280

280281
return right1.compareTo(o.getRight());
281282
}
283+
284+
/**
285+
* Indicates if the left value equals the right value
286+
* @return <code>true</code> if the left value equals the right value
287+
* @since 0.13
288+
*/
289+
public default boolean hasEqualValues() {
290+
return Objects.equals(this.getLeft(), this.getRight());
291+
}
282292
}

‎BEXCodeCompare/src/main/java/info/codesaway/bex/IntBEXRange.java‎

Lines changed: 0 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -126,5 +126,4 @@ public boolean equals(final Object obj) {
126126
IntBEXRange other = (IntBEXRange) obj;
127127
return this.end == other.end && this.hasInclusiveEnd == other.hasInclusiveEnd && this.start == other.start;
128128
}
129-
130129
}

‎BEXCodeCompare/src/main/java/info/codesaway/bex/IntRange.java‎

Lines changed: 39 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -27,6 +27,33 @@ default int getRight() {
2727

2828
public boolean hasInclusiveEnd();
2929

30+
/**
31+
*
32+
* @return
33+
* @since 0.13
34+
*/
35+
public default int getInclusiveStart() {
36+
return this.hasInclusiveStart() ? this.getStart() : this.getStart() + 1;
37+
}
38+
39+
/**
40+
*
41+
* @return
42+
* @since 0.13
43+
*/
44+
public default int getInclusiveEnd() {
45+
return this.hasInclusiveEnd() ? this.getEnd() : this.getEnd() - 1;
46+
}
47+
48+
/**
49+
*
50+
* @return
51+
* @since 0.13
52+
*/
53+
public default int getCanonicalEnd() {
54+
return this.hasInclusiveEnd() ? this.getEnd() + 1 : this.getEnd();
55+
}
56+
3057
public default boolean contains(final int value) {
3158
return value > this.getStart() && value < this.getEnd()
3259
|| this.hasInclusiveStart() && value == this.getStart()
@@ -38,27 +65,31 @@ public default boolean isEmpty() {
3865
&& !(this.hasInclusiveStart() && this.hasInclusiveEnd());
3966
}
4067

68+
/**
69+
*
70+
* @return
71+
* @since 0.13
72+
*/
73+
public default boolean isSingleValue() {
74+
return this.getStart() == this.getEnd()
75+
&& this.hasInclusiveStart() && this.hasInclusiveEnd();
76+
}
77+
4178
/**
4279
*
4380
* @return the length of the range
4481
* @since 0.10
4582
*/
4683
public default int length() {
47-
// Logic from canonical
48-
int start = this.hasInclusiveStart() ? this.getStart() : this.getStart() + 1;
49-
int end = this.hasInclusiveEnd() ? this.getEnd() + 1 : this.getEnd();
50-
51-
return end - start;
84+
return this.getCanonicalEnd() - this.getInclusiveStart();
5285
}
5386

5487
public default IntRange canonical() {
5588
if (this.hasInclusiveStart() && !this.hasInclusiveEnd()) {
5689
return this;
5790
}
5891

59-
int start = this.hasInclusiveStart() ? this.getStart() : this.getStart() + 1;
60-
int end = this.hasInclusiveEnd() ? this.getEnd() + 1 : this.getEnd();
61-
return IntBEXRange.of(start, end);
92+
return IntBEXRange.of(this.getInclusiveStart(), this.getCanonicalEnd());
6293
}
6394

6495
/**

‎BEXCodeCompare/src/main/java/info/codesaway/bex/matching/BEXMatchingStateOption.java‎

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -9,6 +9,8 @@ public enum BEXMatchingStateOption implements MatchingStateOption {
99
IN_LINE_COMMENT, IN_MULTILINE_COMMENT, IN_SECONDARY_MULTILINE_COMMENT,
1010
// String literal
1111
IN_STRING_LITERAL, IN_SECONDARY_STRING_LITERAL,
12+
// Whitespace
13+
WHITESPACE, LINE_TERMINATOR,
1214
// Other states
1315
MISMATCHED_DELIMITERS;
1416

@@ -26,4 +28,9 @@ public boolean isComment() {
2628
public boolean isStringLiteral() {
2729
return in(this, IN_STRING_LITERAL, IN_SECONDARY_STRING_LITERAL);
2830
}
31+
32+
@Override
33+
public boolean isWhitespace() {
34+
return in(this, WHITESPACE, LINE_TERMINATOR);
35+
}
2936
}

‎BEXCodeCompare/src/main/java/info/codesaway/bex/matching/BEXMatchingUtilities.java‎

Lines changed: 75 additions & 32 deletions
Original file line numberDiff line numberDiff line change
@@ -7,6 +7,8 @@
77
import static info.codesaway.bex.matching.BEXMatchingStateOption.IN_SECONDARY_STRING_LITERAL;
88
import static info.codesaway.bex.matching.BEXMatchingStateOption.IN_STRING_LITERAL;
99
import static info.codesaway.bex.matching.BEXMatchingStateOption.IN_TAG;
10+
import static info.codesaway.bex.matching.BEXMatchingStateOption.LINE_TERMINATOR;
11+
import static info.codesaway.bex.matching.BEXMatchingStateOption.WHITESPACE;
1012

1113
import java.util.ArrayDeque;
1214

@@ -224,6 +226,13 @@ public static ImmutableIntRangeMap<MatchingStateOption> parseJavaTextStates(fina
224226
if (c == '\n' || c == '\r') {
225227
int startTextInfo = startTextInfoStack.pop();
226228
builder.put(IntBEXRange.of(startTextInfo, i), stateStack.pop());
229+
230+
if (c == '\r' && nextChar(text, i) == '\n') {
231+
builder.put(IntBEXRange.closed(i, i + 1), LINE_TERMINATOR);
232+
i++;
233+
} else {
234+
builder.put(IntBEXRange.singleton(i), LINE_TERMINATOR);
235+
}
227236
}
228237
// Other characters don't matter?
229238
} else if (stateStack.peek() == IN_MULTILINE_COMMENT) {
@@ -246,6 +255,33 @@ public static ImmutableIntRangeMap<MatchingStateOption> parseJavaTextStates(fina
246255
} else if (c == '\'') {
247256
stateStack.push(IN_SECONDARY_STRING_LITERAL);
248257
startTextInfoStack.push(i);
258+
} else if (c == '\n' || c == '\r') {
259+
if (c == '\r' && nextChar(text, i) == '\n') {
260+
builder.put(IntBEXRange.closed(i, i + 1), LINE_TERMINATOR);
261+
i++;
262+
} else {
263+
builder.put(IntBEXRange.singleton(i), LINE_TERMINATOR);
264+
}
265+
} else if (Character.isWhitespace(c)) {
266+
char nextChar = nextChar(text, i);
267+
if (hasNextChar(text, i) && Character.isWhitespace(nextChar)) {
268+
// Multiple whitespace
269+
int start = i;
270+
271+
do {
272+
if (nextChar == '\n' || nextChar == '\r') {
273+
break;
274+
}
275+
276+
i++;
277+
nextChar = nextChar(text, i);
278+
} while (hasNextChar(text, i) && Character.isWhitespace(nextChar));
279+
280+
builder.put(IntBEXRange.closed(start, i), WHITESPACE);
281+
} else {
282+
// Single whitespace
283+
builder.put(IntBEXRange.singleton(i), WHITESPACE);
284+
}
249285
}
250286
}
251287

@@ -304,7 +340,7 @@ public static ImmutableIntRangeMap<MatchingStateOption> parseJSPTextStates(final
304340
// HTML tag
305341
boolean isTag = false;
306342
// TODO: should I refactor and use this? how would I use it?
307-
String expectedEnd = "";
343+
// String expectedEnd = "";
308344

309345
for (int i = 0; i < text.length(); i++) {
310346
char c = text.charAt(i);
@@ -347,17 +383,28 @@ public static ImmutableIntRangeMap<MatchingStateOption> parseJSPTextStates(final
347383
// TODO: handle unicode and other escaping in String literal
348384

349385
// TODO: Java comments only valid in <% code block %>
386+
} else if (isJava && hasText(text, i, "%>")) {
387+
isJava = false;
388+
389+
if (stateStack.peek() != IN_EXPRESSION_BLOCK) {
390+
// End the current state on the prior character
391+
popMatchingStateOption(i - 1, builder, stateStack, startTextInfoStack);
392+
}
393+
394+
i++;
395+
popMatchingStateOption(i, builder, stateStack, startTextInfoStack);
350396
} else if (isJava && stateStack.peek() == IN_LINE_COMMENT) {
351397
if (c == '\n' || c == '\r') {
352-
int startTextInfo = startTextInfoStack.pop();
353-
builder.put(IntBEXRange.of(startTextInfo, i), stateStack.pop());
398+
popMatchingStateOption(i, builder, stateStack, startTextInfoStack);
399+
// int startTextInfo = startTextInfoStack.pop();
400+
// builder.put(IntBEXRange.of(startTextInfo, i), stateStack.pop());
354401
}
355402
// Other characters don't matter?
356403
} else if (isJava && stateStack.peek() == IN_MULTILINE_COMMENT) {
357404
if (hasText(text, i, "*/")) {
358405
i++;
359-
int startTextInfo = startTextInfoStack.pop();
360-
builder.put(IntBEXRange.closed(startTextInfo, i), stateStack.pop());
406+
popMatchingStateOption(i, builder, stateStack, startTextInfoStack);// int startTextInfo = startTextInfoStack.pop();
407+
// builder.put(IntBEXRange.closed(startTextInfo, i), stateStack.pop());
361408
}
362409
} else if (stateStack.peek() == IN_MULTILINE_COMMENT) {
363410
if (hasText(text, i, "--%>")) {
@@ -369,27 +416,15 @@ public static ImmutableIntRangeMap<MatchingStateOption> parseJSPTextStates(final
369416
i += 2;
370417
popMatchingStateOption(i, builder, stateStack, startTextInfoStack);
371418
}
372-
} else if (stateStack.peek() == IN_EXPRESSION_BLOCK) {
373-
if (hasText(text, i, "%>")) {
374-
isJava = false;
375-
i++;
376-
popMatchingStateOption(i, builder, stateStack, startTextInfoStack);
377-
}
378-
} else if (hasText(text, i, "<%--")) {
379-
stateStack.push(IN_MULTILINE_COMMENT);
380-
startTextInfoStack.push(i);
381-
i += 3;
382-
} else if (hasText(text, i, "<!--")) {
383-
stateStack.push(IN_SECONDARY_MULTILINE_COMMENT);
384-
startTextInfoStack.push(i);
385-
i += 3;
386419
} else if (isJava && c == '/' && nextChar(text, i) == '/') {
387-
stateStack.push(IN_LINE_COMMENT);
388-
startTextInfoStack.push(i);
420+
pushNextLevelMatchingStateOption(IN_LINE_COMMENT, i, builder, stateStack, startTextInfoStack);
421+
// stateStack.push(IN_LINE_COMMENT);
422+
// startTextInfoStack.push(i);
389423
i++;
390424
} else if (isJava && c == '/' && nextChar(text, i) == '*') {
391-
stateStack.push(IN_MULTILINE_COMMENT);
392-
startTextInfoStack.push(i);
425+
pushNextLevelMatchingStateOption(IN_MULTILINE_COMMENT, i, builder, stateStack, startTextInfoStack);
426+
// stateStack.push(IN_MULTILINE_COMMENT);
427+
// startTextInfoStack.push(i);
393428
i++;
394429
} else if (c == '"' && isTag) {
395430
pushNextLevelMatchingStateOption(IN_STRING_LITERAL, i, builder, stateStack, startTextInfoStack);
@@ -402,6 +437,14 @@ public static ImmutableIntRangeMap<MatchingStateOption> parseJSPTextStates(final
402437
} else if (c == '\'' && isJava) {
403438
stateStack.push(IN_SECONDARY_STRING_LITERAL);
404439
startTextInfoStack.push(i);
440+
} else if (hasText(text, i, "<%--")) {
441+
stateStack.push(IN_MULTILINE_COMMENT);
442+
startTextInfoStack.push(i);
443+
i += 3;
444+
} else if (hasText(text, i, "<!--")) {
445+
stateStack.push(IN_SECONDARY_MULTILINE_COMMENT);
446+
startTextInfoStack.push(i);
447+
i += 3;
405448
} else if (hasText(text, i, "<%=")) {
406449
stateStack.push(IN_EXPRESSION_BLOCK);
407450
startTextInfoStack.push(i);
@@ -442,20 +485,20 @@ public static ImmutableIntRangeMap<MatchingStateOption> parseSQLTextStates(final
442485
if (text.length() == 0) {
443486
return ImmutableIntRangeMap.of();
444487
}
445-
488+
446489
// Parse text to get states
447490
// * Block comment
448491
// * Line comment
449492
// * In String literal
450493
// * Other stuff?
451-
494+
452495
ImmutableIntRangeMap.Builder<MatchingStateOption> builder = ImmutableIntRangeMap.builder();
453496
ArrayDeque<MatchingStateOption> stateStack = new ArrayDeque<>();
454497
ArrayDeque<Integer> startTextInfoStack = new ArrayDeque<>();
455-
498+
456499
for (int i = 0; i < text.length(); i++) {
457500
char c = text.charAt(i);
458-
501+
459502
if (stateStack.peek() == IN_STRING_LITERAL) {
460503
// TODO: how to implement escaping, since cannot escape single quote with '\'
461504
if (c == '\'' && nextChar(text, i) == '\'') {
@@ -501,22 +544,22 @@ public static ImmutableIntRangeMap<MatchingStateOption> parseSQLTextStates(final
501544
i++;
502545
int startTextInfo = startTextInfoStack.pop();
503546
builder.put(IntBEXRange.closed(startTextInfo, i), stateStack.pop());
504-
547+
505548
if (!stateStack.isEmpty()) {
506549
// Inside a first level, so add startTextInfo for after expression blocks ends
507550
startTextInfoStack.push(i + 1);
508551
}
509552
} else if (hasText(text, i, "/*")) {
510553
// SQL supports nested block comments
511-
554+
512555
// Going into second level, so end current level
513556
int startTextInfo = startTextInfoStack.pop();
514557
if (startTextInfo != i) {
515558
// Only add if not empty range
516559
// Would be empty for example if ended one expression then immediately started next one
517560
builder.put(IntBEXRange.of(startTextInfo, i), stateStack.peek());
518561
}
519-
562+
520563
stateStack.push(IN_MULTILINE_COMMENT);
521564
startTextInfoStack.push(i);
522565
i++;
@@ -537,14 +580,14 @@ public static ImmutableIntRangeMap<MatchingStateOption> parseSQLTextStates(final
537580
// startTextInfoStack.push(i);
538581
}
539582
}
540-
583+
541584
if (!stateStack.isEmpty()) {
542585
// TODO: what if there are multiple entries?
543586
// (this would suggest improperly formatted code)
544587
int startTextInfo = startTextInfoStack.pop();
545588
builder.put(IntBEXRange.of(startTextInfo, text.length()), stateStack.pop());
546589
}
547-
590+
548591
return builder.build();
549592
}
550593

0 commit comments

Comments
 (0)