annotate trunk/src/dil/doc/Doc.d @ 712:f8875ef9a66d

Added function sanitize() to dil.doc.Doc.
author Aziz K?ksal <aziz.koeksal@gmail.com>
date Wed, 30 Jan 2008 18:29:23 +0100
parents 6b3e397229c5
children 140469ecb90e
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
676
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
1 /++
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
2 Author: Aziz Köksal
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
3 License: GPL3
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
4 +/
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
5 module dil.doc.Doc;
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
6
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
7 import dil.ast.Node;
712
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
8 import dil.lexer.Funcs;
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
9 import common;
676
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
10
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
11 bool isDoxygenComment(Token* token)
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
12 { // Doxygen: '/+!' '/*!' '//!'
679
ff6971637f88 Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 676
diff changeset
13 return token.kind == TOK.Comment && token.start[2] == '!';
676
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
14 }
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
15
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
16 bool isDDocComment(Token* token)
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
17 { // DDOC: '/++' '/**' '///'
679
ff6971637f88 Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 676
diff changeset
18 return token.kind == TOK.Comment && token.start[1] == token.start[2];
676
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
19 }
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
20
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
21 /++
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
22 Returns the surrounding documentation comment tokens.
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
23 Note: this function works correctly only if
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
24 the source text is syntactically correct.
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
25 +/
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
26 Token*[] getDocComments(Node node, bool function(Token*) isDocComment = &isDDocComment)
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
27 {
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
28 Token*[] comments;
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
29 // Get preceding comments.
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
30 auto token = node.begin;
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
31 // Scan backwards until we hit another declaration.
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
32 while (1)
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
33 {
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
34 token = token.prev;
679
ff6971637f88 Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 676
diff changeset
35 if (token.kind == TOK.LBrace ||
ff6971637f88 Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 676
diff changeset
36 token.kind == TOK.RBrace ||
ff6971637f88 Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 676
diff changeset
37 token.kind == TOK.Semicolon ||
ff6971637f88 Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 676
diff changeset
38 token.kind == TOK.HEAD ||
680
6b3e397229c5 Renamed Statements, Declarations and EnumMember.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 679
diff changeset
39 (node.kind == NodeKind.EnumMemberDeclaration && token.kind == TOK.Comma))
676
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
40 break;
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
41
679
ff6971637f88 Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 676
diff changeset
42 if (token.kind == TOK.Comment)
676
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
43 {
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
44 // Check that this comment doesn't belong to the previous declaration.
680
6b3e397229c5 Renamed Statements, Declarations and EnumMember.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 679
diff changeset
45 if (node.kind == NodeKind.EnumMemberDeclaration && token.kind == TOK.Comma)
676
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
46 break;
679
ff6971637f88 Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 676
diff changeset
47 switch (token.prev.kind)
676
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
48 {
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
49 case TOK.Semicolon, TOK.RBrace:
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
50 break;
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
51 default:
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
52 if (isDocComment(token))
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
53 comments ~= token;
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
54 }
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
55 }
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
56 }
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
57 // Get single comment to the right.
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
58 token = node.end.next;
679
ff6971637f88 Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 676
diff changeset
59 if (token.kind == TOK.Comment && isDocComment(token))
676
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
60 comments ~= token;
680
6b3e397229c5 Renamed Statements, Declarations and EnumMember.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 679
diff changeset
61 else if (node.kind == NodeKind.EnumMemberDeclaration)
676
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
62 {
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
63 token = node.end.nextNWS;
679
ff6971637f88 Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 676
diff changeset
64 if (token.kind == TOK.Comma)
676
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
65 {
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
66 token = token.next;
679
ff6971637f88 Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 676
diff changeset
67 if (token.kind == TOK.Comment && isDocComment(token))
676
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
68 comments ~= token;
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
69 }
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
70 }
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
71 return comments;
c4e3a34e40f1 Added new module dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
72 }
712
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
73
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
74 bool isspace(char c)
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
75 {
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
76 return c == ' ' || c == '\t' || c == '\v' || c == '\f';
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
77 }
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
78
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
79 /// Sanitizes a DDoc comment string.
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
80 /// The various newline types are converted to '\n'.
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
81 /// Params:
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
82 /// comment = the string to be sanitized.
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
83 /// commentChar = '/', '+', or '*'
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
84 string sanitize(string comment, char commentChar)
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
85 {
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
86 string result = comment.dup ~ '\0';
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
87
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
88 assert(result[$-1] == '\0');
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
89 bool newline = true; // Indicates whether a newline has been encountered.
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
90 uint i, j;
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
91 for (; i < result.length; i++)
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
92 {
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
93 if (newline)
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
94 { // Ignore commentChars at the beginning of each new line.
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
95 newline = false;
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
96 while (isspace(result[i]))
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
97 { i++; }
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
98 while (result[i] == commentChar)
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
99 { i++; }
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
100 }
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
101 // Check for Newline.
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
102 switch (result[i])
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
103 {
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
104 case '\r':
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
105 if (result[i+1] == '\n')
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
106 i++;
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
107 case '\n':
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
108 result[j++] = '\n'; // Copy Newline as '\n'.
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
109 newline = true;
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
110 continue;
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
111 default:
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
112 if (isUnicodeNewline(result.ptr + i))
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
113 {
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
114 i++; i++;
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
115 goto case '\n';
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
116 }
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
117 }
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
118 // Copy character.
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
119 result[j++] = result[i];
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
120 }
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
121 result.length = j; // Adjust length.
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
122 // Lastly, strip trailing commentChars.
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
123 i = result.length - (1 + 1);
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
124 while (i && result[i] == commentChar)
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
125 { i--; }
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
126 return result;
f8875ef9a66d Added function sanitize() to dil.doc.Doc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 680
diff changeset
127 }