Changeset 383 for trunk/ICeCoffEE

Timestamp:

02/11/08 06:35:36 (17 years ago)

Author:

Nicholas Riley

Message:

ICeCoffEEParser.m: Work much better, though still no multiline URLs.

File:

• trunk/ICeCoffEE/ICeCoffEE/ICeCoffEEParser.m (modified) (6 diffs)

trunk/ICeCoffEE/ICeCoffEE/ICeCoffEEParser.m

@@ -23 +23 @@
         [set formUnionWithCharacterSet: [NSCharacterSet punctuationCharacterSet]];
         // XXX obsoleted by RFC 3986 now... use §2.1, 2.2, 2.3
-        [set removeCharactersInString: @";/?:@&=+$,-_.!~*'()%#"]; // RFC 2396 §2.2, 2.3, 2.4, plus % and # from "delims" set
+        [set removeCharactersInString: @";/?:@&=+$,-_.!~*'(){}[]%#"]; // RFC 2396 §2.2, 2.3, 2.4, plus % and # from "delims" set and {}, []
         
         tmpSet = [[set mutableCopy] autorelease];
-        [tmpSet formUnionWithCharacterSet: [NSCharacterSet characterSetWithCharactersInString: @"><("]];
+        [tmpSet formUnionWithCharacterSet: [NSCharacterSet characterSetWithCharactersInString: @"><(["]];
         urlLeftDelimiters = [tmpSet copy]; // make immutable again - for efficiency
         
         tmpSet = [[set mutableCopy] autorelease];
-        [tmpSet formUnionWithCharacterSet: [NSCharacterSet characterSetWithCharactersInString: @"><)"]];
+        [tmpSet formUnionWithCharacterSet: [NSCharacterSet characterSetWithCharactersInString: @"><)]"]];
         urlRightDelimiters = [tmpSet copy]; // make immutable again - for efficiency
     }
@@ -114 +114 @@
 }
 
+static BOOL ICCF_StringIncludesCharacter(NSString *s, unichar character, NSRange range) {
+    return ([s rangeOfCharacterFromSet: [NSCharacterSet characterSetWithCharactersInString:
+                                         [NSString stringWithCharacters: &character length: 1]]
+                               options: NSLiteralSearch range: range].location != NSNotFound);
+}
+
+static BOOL ICCF_IsLikelyURI(NSString *s, NSRange range) {
+    return ([s rangeOfCharacterFromSet: [NSCharacterSet characterSetWithCharactersInString: @":/.@"]
+                               options: NSLiteralSearch range: range].location != NSNotFound);
+}
+
+static BOOL ICCF_IsLikelyIPv6Address(NSString *s, NSRange range) {
+    return ([s rangeOfCharacterFromSet:
+             [[NSCharacterSet characterSetWithCharactersInString: @"ABCDEFabcdef0123456789:"] invertedSet]
+                               options: NSLiteralSearch range: range].location == NSNotFound);
+}
+
 NSRange ICCF_URLEnclosingRange(NSString *s, NSRange range) {
     NSCharacterSet *urlLeftDelimiters = nil, *urlRightDelimiters = nil;
@@ -123 +140 @@
     ICCF_Delimiters(&urlLeftDelimiters, &urlRightDelimiters);
     
+    // right delimiter selected?  Yes, this can break with ...)URL(....  Oh well.
+    if (range.location > 0 && [urlRightDelimiters characterIsMember: [s characterAtIndex: range.location]]) {
+        --range.location;
+        ++range.length;
+        ICLog(@"expanding past initial %c, now |%@|", [s characterAtIndex: range.location + 1],
+              [s substringWithRange: range]);
+    }
+        
 expandFront:
     // XXX instead of 0, make this stop at the max URL length to prevent protracted searches
-    
-    // XXX here's how this is supposed to work:
-    // (http://web.sabi.net/) and <http://web.sabi.net/> should work if they are the entire document, even if clicking at the end/beginning of the document, not barfing with "no URL" (correct, as now) or selecting the final >, or ) (what would happen if we remove this "add 1" accommodation).  But how about "http://web.sabi.net/(foo)"?  That should work too, as long as it's not preceded by a (.
-    // Should probably backport to ICeCoffEETerminal, now I finally understand the method to this madness.
+    // XXX backport to ICeCoffEETerminal
     // add 1 to range to trap delimiters that are on the edge of the selection (i.e., <...)
     delimiterRange = [s rangeOfCharacterFromSet: urlLeftDelimiters
@@ -141 +163 @@
         range.length += range.location - delimiterRange.location - 1;
         range.location = delimiterRange.location + 1;
-        
-        // in url/(parens)stuff, handle clicking inside or after (parens).
-        /*if ([s characterAtIndex: delimiterRange.location] == '(' &&
-            range.location > 2) { // prevent wrapping, ordinarily not necessary
-            if ([s rangeOfString: @")" options: NSLiteralSearch range: range].location != NSNotFound ||
-                [s rangeOfCharacterFromSet: [NSCharacterSet characterSetWithCharactersInString: @"/."]
-                                   options: NSLiteralSearch range: range].location == NSNotFound) {
-                range.location -= 2;
-                range.length += 2;
-                ICLog(@"expanding past (, now |%@|", [s substringWithRange: range]);
-                goto expandFront;
-            }
-        } */       
-    }
-    
-    ICCF_CheckRange(range);
+    }
     
 expandBack:
@@ -169 +176 @@
         // extend to end of string
         range.length += extraLen;
+        extraLen = 0;
     } else {
         NSCAssert(delimiterRange.length == 1, @"Internal error: delimiter matched range is not of length 1");
         range.length += delimiterRange.location - range.location - range.length;
-        
-        // grow URL past closing paren/brace if we've seen an open paren/brace
-        unichar closing = [s characterAtIndex: delimiterRange.location];
-        NSString *opening;
-        if (closing == ')') opening = @"(";
-        else if (closing == '}') opening = @"{";
-        else goto checkRange;
-        if ([s rangeOfString: opening options: NSLiteralSearch range: range].location == NSNotFound)
-            goto checkRange;
-        
-        if (extraLen == 0) {
+        extraLen = [s length] - NSMaxRange(range);
+
+        // grow URL past closing paren/brace/bracket if we've seen an open paren/brace/bracket
+        unichar opening, closing = [s characterAtIndex: delimiterRange.location];
+        if (closing == ')') opening = '(';
+        else if (closing == '}') opening = '{';
+        else if (closing == ']') opening = '[';
+        else goto expandBoth;
+        if (!ICCF_StringIncludesCharacter(s, opening, range))
+            goto expandBoth;
+        
+        if (extraLen == 1) {
             range.length += 1;
+            --extraLen;
             ICLog(@"expanding past %c, now |%@|", closing, [s substringWithRange: range]);
         } else {
@@ -192 +202 @@
     }
     
+expandBoth:
+    if (range.location <= 1)
+        goto checkRange; // nowhere to expand
+    unichar opening = [s characterAtIndex: range.location - 1], closing;
+    if (opening == '(') closing = ')';
+    else if (opening == '{') closing = '}';
+    else if (opening == '[') closing = ']';
+    else goto checkRange;
+
+    ICLog(@"extraLen = %d", extraLen);
+    // check if we're inside a partial delimited URL: not foolproof, but handles (foo), {UUID} and [IPv6]
+    if (delimiterRange.location != NSNotFound && [s characterAtIndex: delimiterRange.location] == closing && 
+        ((opening == '[' && ICCF_IsLikelyIPv6Address(s, range)) || !ICCF_IsLikelyURI(s, range))) {
+        ICLog(@"expanding past %c...%c, was |%@|", opening, closing, [s substringWithRange: range]);
+        range.location -= 2;
+        if (extraLen > 1)
+            range.length += 4;
+        else
+            range.length += 2 + extraLen;
+        ICLog(@"expanding past %c...%c, now |%@|", opening, closing, [s substringWithRange: range]);
+        goto expandFront;
+    }
+
+    if (ICCF_StringIncludesCharacter(s, closing, range) &&
+        ((opening == '[' &&
+          ICCF_IsLikelyIPv6Address(s, NSMakeRange(range.location, 
+                                                  [s rangeOfString: @"]"].location - range.location)))
+         || !ICCF_IsLikelyURI(s, range))) {
+        ICLog(@"expanding past %c, was |%@|", opening, [s substringWithRange: range]);
+        ICLog(@"remaining: |%@|", [s substringWithRange: NSMakeRange(range.location, [s rangeOfString: [NSString stringWithCharacters: &closing length: 1]].location - range.location)]);
+        range.location -= 2;
+        range.length += 2;
+        ICLog(@"expanding past %c, now |%@|", opening, [s substringWithRange: range]);
+        goto expandFront;
+    }
+
 checkRange:
     ICCF_CheckRange(range);