1 // Scintilla source code edit control
2 /** @file LexPO.cxx
3 ** Lexer for GetText Translation (PO) files.
4 **/
5 // Copyright 2012 by Colomban Wendling <ban@herbesfolles.org>
6 // The License.txt file describes the conditions under which this software may be distributed.
7
8 // see https://www.gnu.org/software/gettext/manual/gettext.html#PO-Files for the syntax reference
9 // some details are taken from the GNU msgfmt behavior (like that indent is allows in front of lines)
10
11 // TODO:
12 // * add keywords for flags (fuzzy, c-format, ...)
13 // * highlight formats inside c-format strings (%s, %d, etc.)
14 // * style for previous untranslated string? ("#|" comment)
15
16 #include <stdlib.h>
17 #include <string.h>
18 #include <stdio.h>
19 #include <stdarg.h>
20 #include <assert.h>
21 #include <ctype.h>
22
23 #include "ILexer.h"
24 #include "Scintilla.h"
25 #include "SciLexer.h"
26
27 #include "WordList.h"
28 #include "LexAccessor.h"
29 #include "Accessor.h"
30 #include "StyleContext.h"
31 #include "CharacterSet.h"
32 #include "LexerModule.h"
33
34 using namespace Scintilla;
35
ColourisePODoc(Sci_PositionU startPos,Sci_Position length,int initStyle,WordList * [],Accessor & styler)36 static void ColourisePODoc(Sci_PositionU startPos, Sci_Position length, int initStyle, WordList *[], Accessor &styler) {
37 StyleContext sc(startPos, length, initStyle, styler);
38 bool escaped = false;
39 Sci_Position curLine = styler.GetLine(startPos);
40 // the line state holds the last state on or before the line that isn't the default style
41 int curLineState = curLine > 0 ? styler.GetLineState(curLine - 1) : SCE_PO_DEFAULT;
42
43 for (; sc.More(); sc.Forward()) {
44 // whether we should leave a state
45 switch (sc.state) {
46 case SCE_PO_COMMENT:
47 case SCE_PO_PROGRAMMER_COMMENT:
48 case SCE_PO_REFERENCE:
49 case SCE_PO_FLAGS:
50 case SCE_PO_FUZZY:
51 if (sc.atLineEnd)
52 sc.SetState(SCE_PO_DEFAULT);
53 else if (sc.state == SCE_PO_FLAGS && sc.Match("fuzzy"))
54 // here we behave like the previous parser, but this should probably be highlighted
55 // on its own like a keyword rather than changing the whole flags style
56 sc.ChangeState(SCE_PO_FUZZY);
57 break;
58
59 case SCE_PO_MSGCTXT:
60 case SCE_PO_MSGID:
61 case SCE_PO_MSGSTR:
62 if (isspacechar(sc.ch))
63 sc.SetState(SCE_PO_DEFAULT);
64 break;
65
66 case SCE_PO_ERROR:
67 if (sc.atLineEnd)
68 sc.SetState(SCE_PO_DEFAULT);
69 break;
70
71 case SCE_PO_MSGCTXT_TEXT:
72 case SCE_PO_MSGID_TEXT:
73 case SCE_PO_MSGSTR_TEXT:
74 if (sc.atLineEnd) { // invalid inside a string
75 if (sc.state == SCE_PO_MSGCTXT_TEXT)
76 sc.ChangeState(SCE_PO_MSGCTXT_TEXT_EOL);
77 else if (sc.state == SCE_PO_MSGID_TEXT)
78 sc.ChangeState(SCE_PO_MSGID_TEXT_EOL);
79 else if (sc.state == SCE_PO_MSGSTR_TEXT)
80 sc.ChangeState(SCE_PO_MSGSTR_TEXT_EOL);
81 sc.SetState(SCE_PO_DEFAULT);
82 escaped = false;
83 } else {
84 if (escaped)
85 escaped = false;
86 else if (sc.ch == '\\')
87 escaped = true;
88 else if (sc.ch == '"')
89 sc.ForwardSetState(SCE_PO_DEFAULT);
90 }
91 break;
92 }
93
94 // whether we should enter a new state
95 if (sc.state == SCE_PO_DEFAULT) {
96 // forward to the first non-white character on the line
97 bool atLineStart = sc.atLineStart;
98 if (atLineStart) {
99 // reset line state if it is set to comment state so empty lines don't get
100 // comment line state, and the folding code folds comments separately,
101 // and anyway the styling don't use line state for comments
102 if (curLineState == SCE_PO_COMMENT)
103 curLineState = SCE_PO_DEFAULT;
104
105 while (sc.More() && ! sc.atLineEnd && isspacechar(sc.ch))
106 sc.Forward();
107 }
108
109 if (atLineStart && sc.ch == '#') {
110 if (sc.chNext == '.')
111 sc.SetState(SCE_PO_PROGRAMMER_COMMENT);
112 else if (sc.chNext == ':')
113 sc.SetState(SCE_PO_REFERENCE);
114 else if (sc.chNext == ',')
115 sc.SetState(SCE_PO_FLAGS);
116 else
117 sc.SetState(SCE_PO_COMMENT);
118 } else if (atLineStart && sc.Match("msgid")) { // includes msgid_plural
119 sc.SetState(SCE_PO_MSGID);
120 } else if (atLineStart && sc.Match("msgstr")) { // includes [] suffixes
121 sc.SetState(SCE_PO_MSGSTR);
122 } else if (atLineStart && sc.Match("msgctxt")) {
123 sc.SetState(SCE_PO_MSGCTXT);
124 } else if (sc.ch == '"') {
125 if (curLineState == SCE_PO_MSGCTXT || curLineState == SCE_PO_MSGCTXT_TEXT)
126 sc.SetState(SCE_PO_MSGCTXT_TEXT);
127 else if (curLineState == SCE_PO_MSGID || curLineState == SCE_PO_MSGID_TEXT)
128 sc.SetState(SCE_PO_MSGID_TEXT);
129 else if (curLineState == SCE_PO_MSGSTR || curLineState == SCE_PO_MSGSTR_TEXT)
130 sc.SetState(SCE_PO_MSGSTR_TEXT);
131 else
132 sc.SetState(SCE_PO_ERROR);
133 } else if (! isspacechar(sc.ch))
134 sc.SetState(SCE_PO_ERROR);
135
136 if (sc.state != SCE_PO_DEFAULT)
137 curLineState = sc.state;
138 }
139
140 if (sc.atLineEnd) {
141 // Update the line state, so it can be seen by next line
142 curLine = styler.GetLine(sc.currentPos);
143 styler.SetLineState(curLine, curLineState);
144 }
145 }
146 sc.Complete();
147 }
148
FindNextNonEmptyLineState(Sci_PositionU startPos,Accessor & styler)149 static int FindNextNonEmptyLineState(Sci_PositionU startPos, Accessor &styler) {
150 Sci_PositionU length = styler.Length();
151 for (Sci_PositionU i = startPos; i < length; i++) {
152 if (! isspacechar(styler[i])) {
153 return styler.GetLineState(styler.GetLine(i));
154 }
155 }
156 return 0;
157 }
158
FoldPODoc(Sci_PositionU startPos,Sci_Position length,int,WordList * [],Accessor & styler)159 static void FoldPODoc(Sci_PositionU startPos, Sci_Position length, int, WordList *[], Accessor &styler) {
160 if (! styler.GetPropertyInt("fold"))
161 return;
162 bool foldCompact = styler.GetPropertyInt("fold.compact") != 0;
163 bool foldComment = styler.GetPropertyInt("fold.comment") != 0;
164
165 Sci_PositionU endPos = startPos + length;
166 Sci_Position curLine = styler.GetLine(startPos);
167 int lineState = styler.GetLineState(curLine);
168 int nextLineState;
169 int level = styler.LevelAt(curLine) & SC_FOLDLEVELNUMBERMASK;
170 int nextLevel;
171 int visible = 0;
172 int chNext = styler[startPos];
173
174 for (Sci_PositionU i = startPos; i < endPos; i++) {
175 int ch = chNext;
176 chNext = styler.SafeGetCharAt(i+1);
177
178 if (! isspacechar(ch)) {
179 visible++;
180 } else if ((ch == '\r' && chNext != '\n') || ch == '\n' || i+1 >= endPos) {
181 int lvl = level;
182 Sci_Position nextLine = curLine + 1;
183
184 nextLineState = styler.GetLineState(nextLine);
185 if ((lineState != SCE_PO_COMMENT || foldComment) &&
186 nextLineState == lineState &&
187 FindNextNonEmptyLineState(i, styler) == lineState)
188 nextLevel = SC_FOLDLEVELBASE + 1;
189 else
190 nextLevel = SC_FOLDLEVELBASE;
191
192 if (nextLevel > level)
193 lvl |= SC_FOLDLEVELHEADERFLAG;
194 if (visible == 0 && foldCompact)
195 lvl |= SC_FOLDLEVELWHITEFLAG;
196
197 styler.SetLevel(curLine, lvl);
198
199 lineState = nextLineState;
200 curLine = nextLine;
201 level = nextLevel;
202 visible = 0;
203 }
204 }
205 }
206
207 static const char *const poWordListDesc[] = {
208 0
209 };
210
211 LexerModule lmPO(SCLEX_PO, ColourisePODoc, "po", FoldPODoc, poWordListDesc);
212