1 //===--- CommentBriefParser.cpp - Dumb comment parser ---------------------===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8
9 #include "clang/AST/CommentBriefParser.h"
10 #include "clang/AST/CommentCommandTraits.h"
11 #include "clang/Basic/CharInfo.h"
12
13 namespace clang {
14 namespace comments {
15
16 namespace {
17
18 /// Convert all whitespace into spaces, remove leading and trailing spaces,
19 /// compress multiple spaces into one.
cleanupBrief(std::string & S)20 void cleanupBrief(std::string &S) {
21 bool PrevWasSpace = true;
22 std::string::iterator O = S.begin();
23 for (std::string::iterator I = S.begin(), E = S.end();
24 I != E; ++I) {
25 const char C = *I;
26 if (clang::isWhitespace(C)) {
27 if (!PrevWasSpace) {
28 *O++ = ' ';
29 PrevWasSpace = true;
30 }
31 } else {
32 *O++ = C;
33 PrevWasSpace = false;
34 }
35 }
36 if (O != S.begin() && *(O - 1) == ' ')
37 --O;
38
39 S.resize(O - S.begin());
40 }
41
isWhitespace(StringRef Text)42 bool isWhitespace(StringRef Text) {
43 return llvm::all_of(Text, clang::isWhitespace);
44 }
45 } // unnamed namespace
46
BriefParser(Lexer & L,const CommandTraits & Traits)47 BriefParser::BriefParser(Lexer &L, const CommandTraits &Traits) :
48 L(L), Traits(Traits) {
49 // Get lookahead token.
50 ConsumeToken();
51 }
52
Parse()53 std::string BriefParser::Parse() {
54 std::string FirstParagraphOrBrief;
55 std::string ReturnsParagraph;
56 bool InFirstParagraph = true;
57 bool InBrief = false;
58 bool InReturns = false;
59
60 while (Tok.isNot(tok::eof)) {
61 if (Tok.is(tok::text)) {
62 if (InFirstParagraph || InBrief)
63 FirstParagraphOrBrief += Tok.getText();
64 else if (InReturns)
65 ReturnsParagraph += Tok.getText();
66 ConsumeToken();
67 continue;
68 }
69
70 if (Tok.is(tok::backslash_command) || Tok.is(tok::at_command)) {
71 const CommandInfo *Info = Traits.getCommandInfo(Tok.getCommandID());
72 if (Info->IsBriefCommand) {
73 FirstParagraphOrBrief.clear();
74 InBrief = true;
75 ConsumeToken();
76 continue;
77 }
78 if (Info->IsReturnsCommand) {
79 InReturns = true;
80 InBrief = false;
81 InFirstParagraph = false;
82 ReturnsParagraph += "Returns ";
83 ConsumeToken();
84 continue;
85 }
86 // Block commands implicitly start a new paragraph.
87 if (Info->IsBlockCommand) {
88 // We found an implicit paragraph end.
89 InFirstParagraph = false;
90 if (InBrief)
91 break;
92 }
93 }
94
95 if (Tok.is(tok::newline)) {
96 if (InFirstParagraph || InBrief)
97 FirstParagraphOrBrief += ' ';
98 else if (InReturns)
99 ReturnsParagraph += ' ';
100 ConsumeToken();
101
102 // If the next token is a whitespace only text, ignore it. Thus we allow
103 // two paragraphs to be separated by line that has only whitespace in it.
104 //
105 // We don't need to add a space to the parsed text because we just added
106 // a space for the newline.
107 if (Tok.is(tok::text)) {
108 if (isWhitespace(Tok.getText()))
109 ConsumeToken();
110 }
111
112 if (Tok.is(tok::newline)) {
113 ConsumeToken();
114 // We found a paragraph end. This ends the brief description if
115 // \command or its equivalent was explicitly used.
116 // Stop scanning text because an explicit \paragraph is the
117 // preferred one.
118 if (InBrief)
119 break;
120 // End first paragraph if we found some non-whitespace text.
121 if (InFirstParagraph && !isWhitespace(FirstParagraphOrBrief))
122 InFirstParagraph = false;
123 // End the \\returns paragraph because we found the paragraph end.
124 InReturns = false;
125 }
126 continue;
127 }
128
129 // We didn't handle this token, so just drop it.
130 ConsumeToken();
131 }
132
133 cleanupBrief(FirstParagraphOrBrief);
134 if (!FirstParagraphOrBrief.empty())
135 return FirstParagraphOrBrief;
136
137 cleanupBrief(ReturnsParagraph);
138 return ReturnsParagraph;
139 }
140
141 } // end namespace comments
142 } // end namespace clang
143
144
145