💾 Archived View for gmi.noulin.net › gitRepositories › emoji › file › generateDataForC.c.gmi captured on 2023-01-29 at 13:16:17. Gemini links have been rewritten to link to archived content
-=-=-=-=-=-=-
generateDataForC.c (5005B)
1 #! /usr/bin/env sheepy
2 /* or direct path to sheepy: #! /usr/local/bin/sheepy */
3
4 /* Libsheepy documentation: http://spartatek.se/libsheepy/ */
5 #include "libsheepyObject.h"
6
7 #define lv logVarG
8
9 #define gG(e,i) getG(e, rtChar, i)
10
11 int argc; char **argv;
12
13 /* enable/disable logging */
14 /* #undef pLog */
15 /* #define pLog(...) */
16
17 int main(int ARGC, char** ARGV) {
18
19 argc = ARGC; argv = ARGV;
20
21 initLibsheepy(ARGV[0]);
22 setLogMode(LOG_DATE);
23
24
25 createSmallArray(html);
26 createSmallJson(j);
27 createSmallArray(emojiList);
28 // emojiList elements: strings
29 // [utf8 code point string, c hex string, name lower case, NAME UPPER CASE, bighead, mediumhead]
30 enum {UTF8, CHEX, LOWERC, UPPERC, BIGHEAD, MEDIUMHEAD};
31
32 readFileG(&html, "full-emoji-list-nobin.html");
33
34
35 // Generate emoji list
36
37 char *bighead = NULL;
38 char *mediumhead = NULL;
39 const char *stS[] = {"start", "head", "codes", "name"};
40 typ enum {start, head, codes, name} statust;
41 statust status = start;
42 statust statusCode = start;
43 createSmallArray(elem);
44
45 iter(&html, L) {
46 castS(l,L);
47 /* lv(l); */
48 /* lv(stS[status]); */
49 /* lv(stS[statusCode]); */
50 if (status == codes) {
51 if (statusCode == name and hasG(l, "td class='name'")) {
52 var l_l = extractG(l, ">", "<");
53 var text = getNDupG(l_l, rtSmallStringt, 0);
54 terminateG(l_l);
55 replaceManyG(text, " ", "_",
56 ":", "",
57 "&", "and",
58 ".", "",
59 "-", "",
60 "(", "",
61 ")", "",
62 "é", "e",
63 "ç", "c",
64 "’", "_",
65 "Å", "A",
66 "⊛", "",
67 "“", "",
68 "”", "",
69 "!", "",
70 "*", "x",
71 ",", "",
72 "ô", "o",
73 "ã", "a",
74 "í", "i",
75 "#", "number_sign");
76 lowerG(text);
77 pushNFreeG(&elem, dupG(text));
78 //lv(text);
79 upperG(text);
80 //lv(text);
81 pushNFreeG(&elem, text);
82 pushG(&elem, bighead);
83 pushG(&elem, mediumhead);
84 pushG(&emojiList, &elem);
85 }
86 if (statusCode == codes and hasG(l, "td class='chars'")) {
87 initiateG(&elem);
88 var l_l = extractG(l, ">", "<");
89 //lv(l_l);
90 char *utf8 = getG(l_l, rtChar, 0);
91 char *chex = toHexHeadSepS(utf8, strlen(utf8), "\\x", "");
92 pushG(&elem, utf8);
93 pushNFreeG(&elem, chex);
94 terminateG(l_l);
95 statusCode = name;
96 }
97 if (hasG(l, "td class='rchars'"))
98 statusCode = codes;
99 if (hasG(l, "class='bighead'") or hasG(l, "class='mediumhead'")) {
100 status = head;
101 statusCode = start;
102 }
103 }
104 if (status == head) {
105 if (hasG(l, "class='bighead'")) {
106 free(bighead);
107 var l_l = extractG(l, "name='", "'");
108 bighead = getNDupG(l_l, rtChar, 0);
109 terminateG(l_l);
110 replaceG(&bighead, "&", "and", 1);
111 //lv(bighead);
112 }
113 elif (hasG(l, "class='mediumhead'")) {
114 free(mediumhead);
115 var l_l = extractG(l, "name='", "'");
116 mediumhead= getNDupG(l_l, rtChar, 0);
117 terminateG(l_l);
118 replaceG(&mediumhead, "&", "and", 1);
119 //lv(mediumhead);
120 }
121 elif (hasG(l, "td class='rchars'")) {
122 status = codes;
123 statusCode = codes;
124 }
125 }
126 if (status == start) {
127 if (hasG(l, "<table"))
128 status = head;
129 }
130 }
131
132 freen(bighead);
133 freen(mediumhead);
134 freeG(&html);
135 //lv(&emojiList);
136
137 // generate h header for c
138 createSmallArray(hheader);
139 createSmallArray(lookupList);
140 // lookupList elememts: strings
141 // [name lower case, UTF8]
142
143 iter(&emojiList, E) {
144 cast(smallArrayt*, e, E);
145 initiateG(&elem);
146 pushG(&elem, gG(e, LOWERC));
147 pushG(&elem, gG(e, UTF8));
148 pushG(&lookupList, &elem);
149 if (!eqG(bighead, gG(e, BIGHEAD))) {
150 bighead = gG(e, BIGHEAD);
151 pushNFreeG(&hheader, formatS("\n\n// Category: %s", bighead));
152 }
153 if (!eqG(mediumhead, gG(e, MEDIUMHEAD))) {
154 mediumhead = gG(e, MEDIUMHEAD);
155 pushNFreeG(&hheader, formatS("\n// Subcategory: %s", mediumhead));
156 }
157 pushNFreeG(&hheader, formatS("#define EM_%s \"%s\"", gG(e,UPPERC), gG(e, CHEX)));
158 }
159
160 var s = toStringG(&lookupList);
161 replaceG(&s, "\"", "\\\"", 0);
162
163 pushG(&hheader, "\n\nextern smallJsont *emList;");
164 pushG(&hheader, "bool allocEmList(void);");
165 pushG(&hheader, "void freeEmList(void);");
166 pushNFreeG(&hheader, formatS("\n#define EM_LIST \"%s\"", s));
167 free(s);
168
169
170 writeFileG(&hheader, "emoji.h");
171 freeManyG(&hheader, &lookupList);
172
173 freeG(&emojiList);
174
175 logI("*");
176 }
177 // vim: set expandtab ts=2 sw=2: